Driving electrocatalytic water splitting with renewably generated electricity provides a sustainable means of producing hydrogen. Its large-scale deployment, however, is limited by the oxygen evolution reaction (OER), whose four-electron transfer character makes the kinetics sluggish and the energy barriers intrinsically high^[1,2]. Although noble-metal catalysts, particularly Ir- and Ru-based materials, exhibit excellent OER activity, their scarcity and cost have hindered industrial application^[3]. As a result, research has increasingly focused on developing low-cost, earth-abundant transition-metal catalysts based on Mn, Fe, Co, and Ni^[4].

Although numerous non-noble-metal OER catalysts, including metal-organic frameworks (MOFs), sulfides, and borides, have been widely reported recently, many lack stability under strongly oxidative conditions^[5,6]. Rather than remaining structurally invariant during catalysis, these materials often undergo electrochemically induced surface reconstruction^[7], involving ligand or anion leaching and metal oxidation to form high-valence (oxy)hydroxides that constitute the real active species^[5,8]. Despite the ubiquity of dynamic transformation, research largely centers on the static properties of pristine catalysts, with limited insights into how OER-induced reconstruction governs activity. In particular, the mechanisms by which atomic doping directs structural evolution toward optimized active phases are rarely studied, limiting the rational design of high-performance catalysts.

Against this backdrop, the structurally well-defined and compositionally tunable nature of MOFs makes them valuable model precursors for sysstematically investigating how multimetallic interactions influence electrochemical reconstruction^[9,10]. Among them, zeolitic imidazolate framework-67 (ZIF-67), a cobalt-based zeolitic imidazolate framework, has been widely explored for OER because of its ordered Co-N coordination environment, tunable structure, and compositional flexibility. Previous studies have improved the OER performance of ZIF-67-based catalysts through facet engineering, low-temperature deligandation to form quasi-ZIF-67, and construction of ZIF-67-derived Co-based compounds or carbon composites^[11-13]. Meanwhile, in situ spectroelectrochemical studies have shown that ZIF-67 can reconstruct into Co(OH)₂/CoOOH-like species under anodic OER conditions^[14], indicating that the reconstructed cobalt oxyhydroxides rather than the original Co nodes act as the dominant active sites. However, how trace heterometal incorporation regulates the reconstruction pathway of intact ZIF-67 precatalysts remains poorly understood.

Here, we report a Mn-doped ZIF-67 (Mn-ZIF-67) obtained through a facile ambient-temperature self-assembly route and systematically reveal the impact of trace Mn incorporation on the pristine catalyst and its evolution under OER conditions. Unlike previous ZIF-67-based OER studies focusing on precursor optimization or derivative design, this work targets the reconstruction process itself, using trace Mn to regulate the evolution of ZIF-67 into a Co-Mn (oxy)hydroxide active layer and establish a direct link between precursor structure, reconstruction pathway, and catalytic performance. Benefiting from earth-abundant Mn/Co elements and mild solution-phase synthesis, this system offers potential advantages in cost and scalability compared with Ir/Ru-based benchmark catalysts. Mn doping increases electrochemically accessible sites and accelerates charge-transfer kinetics, while steering in situ reconstruction toward a highly active Co-Mn (oxy)hydroxide phase. The resulting catalyst exhibits durable performance in alkaline seawater splitting and is compatible with photovoltaic-powered operation. Overall, this work highlights transition-metal doping as an effective strategy for modulating MOF-based catalysts and their OER-induced reconstruction behavior, providing a rational basis for the design of efficient and robust OER electrocatalysts.

Uniform rhombic dodecahedral ZIF-67 and Mn-ZIF-67 were successfully prepared via a room-temperature self-assembly method [Figure 1A-C]. Mn²⁺ incorporation modulates nucleation and suppresses crystal growth, reducing the average particle size from 0.94 ± 0.25 μm for ZIF-67 [Supplementary Figure 2] to 0.42 ± 0.09 μm for Mn-ZIF-67 [Supplementary Figure 3], which may improve electrolyte accessibility by increasing the external surface-to-volume ratio and shortening ion-diffusion pathways. Consistently, nitrogen adsorption-desorption measurements show that, although Mn-ZIF-67 has a slightly lower Brunauer-Emmett-Teller (BET) surface area than ZIF-67 (1,238.33 vs. 1,264.18 m² g^-1), its t-plot external surface area increases from 23.50 to 40.97 m² g^-1, accompanied by increased Barrett-Joyner-Halenda (BJH) adsorption cumulative pore area and volume from 15.49 m² g^-1 and 0.03025 cm³ g^-1 to 30.96 m² g^-1 and 0.04332 cm³ g^-1, respectively [Supplementary Figures 4 and 5 and Supplementary Table 2]. These results suggest enhanced external surface accessibility and mesoporous/interparticle pore contribution in Mn-ZIF-67, which are favorable for electrolyte penetration, ion transport, and anodic reconstruction. Energy-dispersive X-ray spectroscopy (EDS) mapping in Figure 1D confirms the homogeneous spatial distribution of constituent elements. Transmission electron microscopy (TEM) images retain the characteristic polyhedral outline, and the diffraction pattern recorded from an individual particle is consistent with a highly crystalline single-domain specimen (Figure 1E and inset). High-resolution TEM (HRTEM) exhibits ~ 1.21 nm lattice fringes corresponding to the ZIF-67 (011) plane [Figure 1F]. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) mapping likewise shows Mn and Co distributed across the same particle [Figure 1G]. Collectively, the characterization results demonstrate the incorporation of Mn²⁺ into the metal sites of the MOF without detectable formation of phase-separated Mn oxide particles.

X-ray diffraction (XRD) patterns of both ZIF-67 and Mn-ZIF-67 match the simulated patterns [Figure 2A], confirming high phase purity and preservation of the intrinsic crystal structure^[24]. The absence of secondary diffraction features such as Mn(NO₃)₂·4H₂O and Mn₃O₄ further indicates the successful integration of Mn into the MOF framework without the formation of segregated impurity phases^[25]. X-ray photoelectron spectroscopy (XPS) confirms the expected elemental composition of Mn-ZIF-67, including C, N, O, Co and Mn, as presented in Figure 2B^[26,27]. The high-resolution N 1s spectrum in Figure 2C can be deconvoluted into metal-coordinated pyridinic nitrogen attributed to M-N bonding, where M represents Co or Mn, at approximately 398.8 eV and pyrrolic nitrogen at 400.2 eV^[26]. These features confirm the coordination of 2-methylimidazole ligands and the formation of a stable ZIF-type framework^[27]. To enable a more direct comparison of surface chemical states, the O 1s spectra of pristine ZIF-67 and Mn-ZIF-67 are plotted together in Figure 2D. Two main components are observed in both samples, corresponding to metal hydroxyl or adsorbed water species at 531.9 eV and to C-O species. The C-O component is located at 533.8 eV for pristine ZIF-67 and at 534.2 eV for Mn-ZIF-67, corresponding to a positive shift of 0.4 eV upon Mn incorporation. In parallel, the ratio of metal hydroxyl or adsorbed water species to C-O decreases from 4.59 to 3.28, indicating reduced surface hydroxylation and a more ligand-dominated surface environment. This result suggests that Mn incorporation modifies the local coordination environment of the pristine ZIF-67 framework. The Mn 2p spectrum in Figure 2E exhibits the characteristic Mn 2p_3/2 (~ 641.2 eV) and Mn 2p_1/2 (~ 652.8 eV) spin-orbit components, together with an additional satellite-related feature. It should be noted that Mn 2p spectra generally exhibit complex multiplet and final-state structures; therefore, the resolved satellite contribution is distinguished here from the Mn 2p_3/2 /Mn 2p_1/2 spin-orbit doublet^[28]. Based on the overall spectral envelope and comparison with previously reported Mn²⁺-containing systems, the Mn species in the as-prepared Mn-ZIF-67 are predominantly present in a Mn²⁺-like chemical environment^[29,30]. Nevertheless, considering the complex Mn 2p spectral structure and the surface-sensitive nature of XPS, minor contributions from other Mn chemical states cannot be completely excluded. The directly compared Co 2p spectra of pristine ZIF-67 and Mn-ZIF-67 in Figure 2F further reveal that Mn incorporation significantly modulates the Co electronic structure. In the direct Co 2p comparison, the principal 2p_3/2 feature moves by approximately 0.7 eV, from 781.8 eV for ZIF-67 to 781.1 eV after Mn introduction. This negative shift indicates increased electron density at the Co centers via the ligand-bridged framework, analogous to previously reported electron donation from boron to transition metals^[31,32]. This result provides evidence of Mn-induced electronic modulation within the pristine ZIF framework, which may influence the subsequent electrochemical evolution of Mn-ZIF-67 under OER conditions.

Electrochemical OER activity was assessed in 1.0 M KOH^[33]. LSV curves shown in Figure 3A reveal that both MOF-based electrodes deliver substantially higher anodic currents than the bare NF substrate, with Mn-ZIF-67 consistently outperforming ZIF-67^[34]. A pre-OER anodic response appears in the 1.3-1.4 V versus RHE region. This feature is consistent with oxidation of Co species and the onset of oxyhydroxide-like reconstruction before substantial oxygen evolution^[35,36]. This redox feature is more pronounced for Mn-ZIF-67, suggesting that Mn incorporation enhances the redox accessibility of Co sites and facilitates the generation of electrochemically active high-valence Co species under alkaline conditions^[37]. Figure 3B displays the potentials needed to sustain the selected geometric current benchmarks. Mn-ZIF-67 requires only 235 mV to reach 10 mA cm^-2, 41 mV lower than ZIF-67, and maintains superior activity at 100 mA cm^-2 with an overpotential of 310 mV compared with 353 mV for ZIF-67. An LSV backscan was further performed to exclude possible interference from Ni foam oxidation. As shown in Supplementary Figure 6, the backscan curve shows no obvious Ni oxidation peak and gives a similar overpotential at 10 mA cm^-2, confirming the reliable OER activity evaluation of Mn-ZIF-67. When both electrodes are compared at 1.53 V versus RHE (η = 300 mV), Mn-ZIF-67 delivers 69.8 mA cm^-2, versus 18.5 mA cm^-2 for ZIF-67, corresponding to a 3.8-fold increase in geometric current response. Reaction-kinetic trends were next compared using the Tafel analysis in Figure 3C. The fitted slopes are 36.49, 50.12, and 132.56 mV dec^-1 for Mn-ZIF-67, ZIF-67, and bare NF, respectively. Combined with the lowered overpotentials, this accelerated electron transfer confirms Mn-ZIF-67 as a highly efficient alkaline OER electrocatalyst. Operational stability is a critical requirement for practical application. During the 12 h chronoamperometric test [Figure 3D], the final current remains at ~ 93% of the initial value, indicating limited activity loss over the measurement period. To further examine the durability under industrially relevant conditions, chronopotentiometry was performed at a high current density of 1 A cm^-2 in the same three-electrode configuration. The Mn-ZIF-67 electrode sustains this current density for 24 h with the potential remaining stable at approximately 1.65 V vs. RHE [Supplementary Figure 7], indicating that the reconstructed active layer withstands the intense gas evolution and mechanical stress associated with high-current operation.

The origin of the distinct catalytic activity of ZIF-67 and Mn-ZIF-67 was further unraveled through combined electrochemical analysis and post-reaction characterization. Under alkaline OER conditions, the measured catalytic activity can reflect several coupled contributions, notably interfacial electron transfer, electrochemical accessibility, transport, and reconstruction of the precursor surface^[38-40]. Recent studies on water/seawater-splitting electrodes have also emphasized that reliable physicochemical characterization, including structural analysis, surface chemical-state evolution, electrochemical impedance, accessible surface-area evaluation, normalized activity comparison, and post-reaction examination, is essential for establishing meaningful structure-activity relationships^[41]. Interfacial kinetic differences were first probed by EIS. As revealed by the Nyquist plots and corresponding fitting curves in Figure 4A, Mn-ZIF-67 exhibits a significantly smaller semicircle diameter than both ZIF-67 and bare NF, indicating a lower interfacial resistance under OER-relevant conditions. Quantitative fitting using an equivalent circuit model, with details provided in Supplementary Figure 8 and Supplementary Table 3, reveals that Mn-ZIF-67 possesses a low charge-transfer resistance (R_ct) of 2.17 Ω. This value is less than half that of ZIF-67 (4.60 Ω) and far lower than that of NF (35.36 Ω). The pronounced decrease in resistance demonstrates that Mn incorporation effectively facilitates interfacial electron transfer and reduces kinetic polarization losses during the OER process^[42]. The density of electrochemically accessible active sites is another critical factor in determining catalytic activity^[38]. Electrochemical accessibility was evaluated next from C_dl values fitted to the non-Faradaic CV data in Figure 4B and C and Supplementary Figures 9 and 10. Based on ECSA = C_dl/C_s, with C_s taken as 40 μF cm^-2, the C_dl values correspond to estimated ECSA values of 120.50, 63.25, and 72.50 cm² for Mn-ZIF-67, ZIF-67, and NF, respectively. These C_dl-derived ECSA values should be regarded as apparent electrochemically accessible areas rather than absolute physical surface areas, and are used here mainly for relative comparison under identical testing conditions. The larger apparent ECSA of Mn-ZIF-67 indicates more electrochemically accessible sites, which contributes to the enhanced OER performance^[43]. To further evaluate the intrinsic activity of individual active sites, the LSV currents were normalized to the respective ECSA values. As displayed in Figure 4D, the ECSA-normalized polarization curves show that Mn-ZIF-67 still outperforms ZIF-67, suggesting that the improved activity is not solely caused by the increased electrochemically accessible area^[44,45]. A comparison with representative ZIF-67/MOF-based and transition-metal OER electrocatalysts is summarized in Supplementary Table 4. Notably, Mn-ZIF-67 delivers a lower overpotential and Tafel slope than pristine ZIF-67 and many previously reported catalysts, highlighting its superior catalytic kinetics. The enhanced activity can be attributed to the synergistic contributions of increased electrochemically accessible surface area, accelerated interfacial charge transfer, and the formation of a Mn-containing Co oxyhydroxide-like active layer with enhanced intrinsic activity during reconstruction.

As-prepared Mn-ZIF-67 undergoes pronounced electrochemical surface reconstruction under OER conditions^[46,47]. To directly monitor this structural evolution, in situ electrochemical Raman spectroscopy was performed under OER-relevant potentials. As shown in Figure 5A, the Raman spectra collected from open-circuit potential (OCP) to 1.80 V show clear potential-dependent changes. At lower potentials, the spectrum is dominated by Co-based oxide/oxide-like species. When the applied potential increases into the OER region, a Raman band around 503 cm^-1 becomes more evident, which is assigned to the Co-O vibration of CoOOH. Meanwhile, the CoO_x-related band around 686 cm^-1 evolves under anodic polarization. The potential-dependent Raman evolution therefore directly supports the formation of a CoOOH-like surface species during anodic polarization. The in situ Raman spectra mainly confirm the potential-induced formation of CoOOH-like species; Mn-related Raman signals are relatively weak and may overlap with Co-based vibrational features. Post-OER morphological analysis in Figure 5B shows that the initial polyhedral crystals evolve into an interconnected nanosheet architecture. Local lattice ordering in the post-OER material is visible in the HRTEM image of Figure 5C. Measured spacings of 0.193 and 0.235 nm are consistent with the reported CoOOH (104) and (101) reflections, respectively^[48]. Two additional spacings, 0.262 and 0.214 nm, are consistent with values reported for MnOOH-related (-102) and (-121) planes^[49]. Together with the post-OER Mn 2p evolution and inductively coupled plasma (ICP) retention data, these lattice features support the presence of Mn-containing oxyhydroxide-related regions in the reconstructed catalyst; they do not, by themselves, resolve the exact atomic arrangement of the working Mn sites^[50]. XPS further clarifies the evolution of surface chemical states. The post-OER Co 2p spectra in Figure 5D show the main Co 2p_3/2 peak shifts from 781.1 eV, characteristic of Co²⁺ with intense satellite features, to 779.9 eV with markedly suppressed satellites. This change indicates oxidation to a Co³⁺-dominated surface, consistent with CoOOH formation^[51,52]. Similarly, the post-OER Mn 2p spectra in Figure 5E show that the Mn 2p_3/2 peak, initially centered at 641.2 eV and characteristic of a Mn²⁺-like environment, shifts positively to 642.0 eV after the OER, accompanied by a marked change in line shape. This 0.8 eV positive shift and the altered line shape indicate oxidation of Mn toward a Mn³⁺-dominated environment and support the participation of Mn in the reconstructed oxyhydroxide layer^[53]. The ex situ XRD patterns before and after OER further show weakened diffraction features after electrolysis, indicating electrochemical reconstruction and partial loss of long-range crystallinity of the pristine ZIF framework during OER [Supplementary Figure 11]. For comparison, the post-OER surface states of pristine ZIF-67 were also examined [Supplementary Figures 12 and 13]. The post-OER Co 2p spectrum of ZIF-67 exhibits Co 2p_3/2 and Co 2p_1/2 peaks at 779.9 and 795.6 eV assignable to Co³⁺-containing oxyhydroxide-like species, while the pronounced satellite features indicate that considerable Co²⁺ character remains. Meanwhile, the post-OER HRTEM image of ZIF-67 reveals lattice fringes with a spacing of 0.45 nm, corresponding to the CoOOH (003) plane, confirming that undoped ZIF-67 likewise undergoes electrochemical reconstruction into CoOOH-like species during the OER. Compared with post-OER Mn-ZIF-67, whose Co 2p_3/2 peak shifts to 779.9 eV with markedly suppressed satellites, the surface oxidation of ZIF-67 is less complete, further indicating that Mn incorporation facilitates the oxidative reconstruction of Co sites toward the active oxyhydroxide phase. To determine the actual Mn doping level and its retention after electrochemical operation, ICP analysis was performed on the catalyst-loaded electrodes before and after long-term OER operation. As summarized in Supplementary Table 5, the as-prepared Mn-ZIF-67 has an Mn/Co atomic ratio of 0.1005 ± 0.0024, close to the nominal feeding ratio of 0.111 and corresponding to an Mn incorporation efficiency of about 91%. After long-term OER operation, the ratio decreases to 0.0901 ± 0.0024, corresponding to a normalized Mn/Co atomic ratio of 89.7% of its initial value. Most Mn species are therefore retained in the reconstructed catalyst layer, consistent with the HRTEM and XPS evidence for the formation of Mn-containing oxyhydroxide species.

To obtain atomic-level insight into Mn-induced activity enhancement, DFT calculations were performed. ZIF-67 and Mn-ZIF-67 precursor models were constructed to evaluate Mn-induced charge redistribution and electronic modulation in the pristine framework [Figure 6A and B]. In addition, considering the post-OER HRTEM and XPS evidence for electrochemical reconstruction into Co/Mn oxyhydroxide species, CoOOH and Mn-doped CoOOH surface models were constructed to evaluate the OER reaction energetics on the reconstructed active surfaces [Figure 6C]. The results indicate that Mn adopts a charge state distinct from Co and induces a redistribution of electron density within the ligand-bridged ZIF framework. Although Mn and Co centers are not directly bonded in the pristine structure, the charge redistribution mediated by imidazolate linkers suggests that Mn incorporation modifies the electronic environment of the framework. This result is consistent with the Co 2p binding-energy shift observed in XPS, indicating Mn-induced electronic perturbation in the pristine Mn-ZIF-67. To further elucidate how this redistribution affects the adsorption of OER intermediates, the d-band center (ε_d) of the active metal sites was evaluated as an electronic descriptor, as shown in Figure 6B. Mn incorporation shifts ε_d upward from -2.00 eV in ZIF-67 to -1.70 eV in Mn-ZIF-67. Within the d-band framework, such an upshift of ε_d indicates stronger metal-oxygen interaction due to modified antibonding state occupancy. This effect optimizes the adsorption energetics and activation of oxygenated intermediates such as *OH, *O, and *OOH, thereby promoting OER kinetics. This trend is consistent with the experimentally observed enhancement in interfacial charge transfer and catalytic activity upon Mn incorporation.

To model the electrochemically reconstructed active surfaces suggested by post-OER characterizations, CoOOH and Mn-modified CoOOH (Mn/CoOOH) were constructed to represent the post-OER surfaces of ZIF-67 and Mn-ZIF-67, respectively^[35,54,55]. The Gibbs free energy (ΔG) profiles for the four-electron OER pathway in Figure 6C show that the potenial-determining step (PDS) on CoOOH is the *O to *OOH transformation, with a ΔG of 2.22 eV. In contrast, Mn incorporation optimizes intermediate adsorption. Although the PDS remains the same, the ΔG of Mn/CoOOH decreases to 1.71 eV, significantly lowering the energetic penalty for OER turnover^[56,57]. These results indicate that Mn acts as an effective electronic promoter by tuning the Co electronic structure and optimizing intermediate binding on the reconstructed oxyhydroxide surface. The free-energy profile further supports that the Mn-containing reconstructed Co oxyhydroxide surface serves as the active phase responsible for the improved OER kinetics.

To evaluate practical feasibility, a two-electrode alkaline seawater electrolyzer was assembled using a Mn-ZIF-67/Ni foam anode, a graphite rod cathode, and alkaline natural seawater collected from Shenzhen Bay^[58]. When the same electrolyzer was connected to a commercial silicon solar panel, continuous gas evolution was observed at both electrodes under natural sunlight [Figure 7A], providing qualitative proof of concept for compatibility with renewable power input^[59]. The quantitative data in Figure 7B and C were acquired separately under workstation control. Overall cell performance was evaluated by LSV, as shown in Figure 7B. The Mn-ZIF-67 electrolyzer significantly outperforms the bare NF control. It delivers current densities of 10 and 50 mA cm^-2 at 1.84 and 2.06 V, respectively, whereas the NF control requires 1.97 and 2.28 V to reach the same current densities, corresponding to reductions of 130 and 220 mV in cell voltage and indicating reduced overall-cell polarization. Chronopotentiometry measurements shown in Figure 7C further confirm excellent operational durability under harsh anodic conditions. At a constant current density of 50 mA cm^-2, the cell voltage remains highly stable at 2.06 V over 100 h of continuous operation without observable decay, demonstrating strong practical viability. During OER, the Mn-ZIF-67 undergoes in situ reconstruction into a nanosheet-like bimetallic (oxy)hydroxide layer composed of CoOOH and MnOOH nanocrystals [Figure 5C]^[60]. The stable operation over 100 h demonstrates the good durability of the Mn-ZIF-67-derived anode in alkaline seawater. To further examine the structural stability after seawater electrolysis, post-reaction XRD was performed. As shown in Supplementary Figure 14, the characteristic ZIF-related diffraction peaks largely disappear after electrolysis, indicating electrochemical reconstruction of the MOF precursor under anodic conditions. No crystalline chloride-containing corrosion products were detected by ex situ XRD after the test^[61]. The surface chemical states after seawater electrolysis were further examined by XPS [Supplementary Figures 15 and 16]. The Co 2p spectrum of Mn-ZIF-67 after alkaline seawater electrolysis exhibits Co 2p_3/2 and Co 2p_1/2 peaks at 780.1 and 795.2 eV with weak satellite features, indicating that the Co³⁺-dominated oxyhydroxide-like surface is well preserved. The Mn 2p spectrum retains well-defined Mn 2p_3/2 and Mn 2p_1/2 peaks at 642.3 and 654.0 eV, consistent with Mn³⁺ oxyhydroxide-like species, indicating that Mn is retained in the reconstructed layer. These chemical states closely resemble those observed after the OER in 1.0 M KOH [Figure 5D and E], further suggesting that the reconstructed Co-Mn (oxy)hydroxide layer maintains its chemical integrity against chloride attack during long-term alkaline seawater electrolysis. Collectively, these findings identify Mn-ZIF-67 as a promising anode material for solar-driven alkaline seawater electrolysis. To benchmark the overall seawater splitting performance, Mn-ZIF-67 was compared with representative ZIF/MOF-derived and Mn-containing electrocatalytic systems [Table 1]. The Mn-ZIF-67/NF || graphite rod cell requires 1.84 V at 10 mA cm^-2 and 2.06 V at 50 mA cm^-2 in alkaline natural seawater, showing competitive performance relative to several related systems. Together with its stable operation for 100 h at 50 mA cm^-2 and photovoltaic-driven demonstration, these results highlight the practical potential of Mn-ZIF-67 for alkaline seawater electrolysis.

This work demonstrates that trace Mn doping can direct the electrochemical reconstruction pathway of ZIF-67 and thereby determine the nature of the true active phase, the OER kinetics, and the catalytic durability. Rather than acting as a static active phase, Mn-ZIF-67 behaves as a structurally dynamic catalyst under OER conditions. During anodic operation, the pristine Mn-ZIF-67 surface reconstructs into a robust bimetallic Co-Mn oxyhydroxide layer, which serves as the catalytically active phase. Density functional theory calculations confirm that Mn-induced electronic redistribution upshifts the Co d-band center, thereby lowering the Gibbs free energys of intermediates, while electrochemical measurements show that Mn incorporation simultaneously increases the density of electrochemically accessible sites and enhances interfacial charge transfer. The practical viability of this dynamically reconstructed interface is demonstrated in a two-electrode alkaline seawater electrolyzer, which sustains a current density of 50 mA cm^-2 at approximately 2.06 V for 100 h, and which can also be operated directly using a commercial silicon solar panel. This work establishes a general mechanistic framework in which trace transition metal doping directs the in situ structural evolution of metal-organic frameworks, offering an effective route toward the rational design of next-generation efficient electrocatalysts.