Introduction

Water electrolysis, capable of producing environmentally friendly hydrogen with a high energy density of 120 MJ kg−1, has emerged as a promising technology to address the pressing issues of fossil fuel shortage and support the goal of carbon neutralization by the 2050s1,2. According to the International Energy Agency, it is intended to capture a significant 22% market share in worldwide hydrogen generation3. However, the large-scale commercialization of this technology is hinged upon the development of highly active and stable electrocatalysts4,5,6. In particular, high-speed ion and electron transfers in electrocatalysis are, in general, achieved in harsh environments, such as strong acidity/alkalinity solutions and high oxidation/reduction potentials7,8. Such extreme conditions lead to severe stability issues for catalysts, including both noble and non-noble metal-based nanostructures9,10, which are the primary bottleneck problems restraining the development of high-performance catalysts for large-scale commercial applications11.

The catalyst stability in water splitting is influenced by the structural and compositional variations upon electrochemical operation in thermodynamics12,13,14. These variations during catalysis are well-accepted and often defined as “reconstruction” or “activation of precatalysts”15. As oxidation potentials more readily induce surface structural and compositional changes of nanostructured precatalysts, extensive research focusing on reconstruction is primarily reported for oxygen evolution reactions (OER) and ignores the hydrogen evolution reaction (HER) catalysts16,17,18,19,20,21,22,23,24,25,26,27,28,29. This situation prompts researchers to shift their focus to the latter field. For example, Peng et al. reported that the reconstructed Co(OH)2@MXene catalyst, derived from a CoC2O4@MXene precursor upon HER operation, can greatly boost catalytic activity for alkaline hydrogen production. This improvement was attributed to the optimization of water dissociation and hydrogen adsorption/desorption energy barriers on Co(OH)2@MXene30. However, the reconstruction mechanism from CoC2O4 to Co(OH)2 was not elucidated, and reconstructions under different conditions were not investigated, posing challenges for guiding the further design of efficient catalysts through reconstruction. Recent research also indicated that soaking SrCoO3-δ in KOH solution induced its surface reconstruction to CoO2-δ, revealing that the reconstruction can also proceed via chemical reactions31. These reactions can be further accelerated or decelerated by tuning the redox potential of precatalysts and altering the applied potentials32. Multiple factors, including the material's redox properties, pH of electrolytes, applied potentials, and reaction temperature, influence the rate and extent of surface transformation on precatalysts, complicating the identification and utilization of the reconstruction mechanisms33,34,35,36,37. To date, a comprehensive methodology integrating these factors has not been implemented to understand the reconstruction mechanisms.

Reconstruction can either enhance or impair the catalytic activity, thus it is crucial to develop strategies that can promote the beneficial reconstruction while inhibiting the detrimental case. As an electrochemically stable structure, the positively reconstructed catalysts ensure high catalytic activity and stability, thereby supporting the development of efficient catalysts. Under working conditions, the variations in the surface structures of precatalysts affect the composition of electrolytes, and the presence of trace anions/cations dissolved from precatalysts into electrolytes can have a significant impact on catalytic performance38,39,40,41. Previous studies have revealed that the introduction of anion additives (e.g. SeO32−) to 1 M KOH solution greatly enhanced the HER activity of the MoSe2 catalyst by facilitating the activation and dissociation of water molecules42. However, most of the current research merely focused on the structure-performance relationship of these restructured catalysts. This might be incomplete and potentially misleading to understand the performance enhancement mechanisms, as it overlooks the role of components dissolved in the electrolyte during the restructuring process. The elucidation of the species etched from precatalysts into the electrolyte and their role in effecting HER activity still poses unsolved questions. Therefore, a comprehensive consideration of the variations in both precatalyst and electrolyte is urgently required for the study of the underlying mechanism as well as the further design of highly active and stable catalysts for electrochemical hydrogen production.

In this study, we have successfully synthesized high-purity hexagonal Co2Mo3O8 nanoparticles exposed with the cobalt-terminated (001) facet and utilized them as a model precatalyst for water splitting, considering their features of metastable phase, high conductivity, and good HER activity. The potential-dependent surface structure evolution mechanism of Co2Mo3O8 has been investigated, and a correlation between the catalyst’s structure, performance, and mechanism with a combined computational and experimental methodology has been established. It was found that the surface structure transformation of Co2Mo3O8 in 1 M KOH at room temperature can be controlled by applied potentials, and was accompanied by the etching of Mo from the precatalyst into electrolytes in the form of MoO42−, finally resulting in the formation of an electrochemically stable Co(OH)2@Co2Mo3O8 heterostructure. Mechanistic study revealed a potential-dependent interaction of MoO42−/Mo2O72− on the reconstructed Co(OH)2@Co2Mo3O8, which led to notably enhanced catalytic activity and stability. Based on these findings, a bias-manipulated method was employed to reconstruct the high amount of Co(OH)2 on Co2Mo3O8. In 1 M KOH with 50 mM MoO42−, the as-obtained MoO42−/Mo2O72−–Co(OH)2@Co2Mo3O8 catalyst demonstrated notable activity and stability for alkaline HER. These findings provide a basis for the understanding of reconstruction behaviors of oxides-based precatalysts and for designing high-performance transition metal catalysts, particularly those incorporating VI B group elements.

Results

Synthesis and characterization of Co2Mo3O8 precatalyst

Pure Co2Mo3O8 nanostructures were synthesized using a hydrothermal-calcination-etching method, as illustrated in Fig. 1a. The synthesis and characterization details of various precursors can be found in the “Methods” section and Supplementary Figs. 1–5. By subjecting the Co/CoO@Co2Mo3O8 intermediate to an acid-etching process, the surface metallic Co and CoO components were eliminated, leaving only the inner Co2Mo3O8 phase. This was confirmed by the Raman spectra (Supplementary Fig. 6) and XRD pattern (Fig. 1b)43,44. The as-obtained hexagonal Co2Mo3O8 exhibited a platelet morphology with an average size of 65.1 ± 18.2 nm and a thickness of ca. 13.4 nm (Fig. 1c, d, and Supplementary Fig. 7). Close observation by transmission electron microscopy (TEM) identified that the Co2Mo3O8 platelets were oriented along the [210] axis with the (001) facet as the exposed surface. The well-defined inter-planar spacing of 5.0 Å corresponds to the (002) lattice plane of Co2Mo3O8 (Fig. 1e, f). The distribution of Co, Mo, and O elements throughout the Co2Mo3O8 matrix was uniform, indicating the good solubility of Co and Mo in this bimetal oxide (Fig. 1g).

Fig. 1: Synthesis and characterization of the Co2Mo3O8 precatalyst.
Fig. 1: Synthesis and characterization of the Co2Mo3O8 precatalyst.
Full size image

a Schematic illustration of the preparation route for developing pristine Co2Mo3O8, where CoMo-P, CoMo@ZIF-67-P, and Co/CoO@Co2Mo3O8 represent various precursors or intermediates for synthesizing Co2Mo3O8, details are available in the “Methods” section. b X-ray diffraction (XRD) pattern, c scanning electron microscopy (SEM) image, d atomic force microscopy (AFM) image, e transmission electron microscope (TEM) image, f high-resolution TEM (HRTEM) image (insets are the corresponding structural illustration and the fast Fourier transform (FFT) pattern, Z.A. represents zone axis), and g high-angle annular dark-field scanning TEM (HAADF-STEM) image and energy-dispersive X-ray spectroscopy (EDX) elemental mapping images of Co2Mo3O8. h Density of states (DOS) curves, i ultraviolet photoelectron spectroscopy (UPS) spectrum, j calculated work function (WF), and k charge distribution showing the electron density surrounding Co, Mo, and O atoms of the Co-terminated (001) surface. Evac and Ef means vacuum and Fermi levels, respectively.

Density of theory (DOS) calculations were performed to provide the structure information of Co2Mo3O8 (see “Methods” subsection “Computational details”). The Co, Mo, O projected DOS curves, as well as total DOS, exhibited continuity at the Fermi level, implying the metallic nature of Co2Mo3O8 (Fig. 1h and Supplementary Fig. 8)45. The electronic properties of Co2Mo3O8 were further investigated through work function (Wf) measurements using ultraviolet photoelectron spectroscopy (UPS). The experimental Wf of Co2Mo3O8 was determined to be 3.58 eV, which aligns well with the calculated Wf value for a Co2Mo3O8 (001) surface with Co termination (3.61 eV), rather than with Mo termination (5.21 eV) or O terminations (6.94 and 6.11 eV). This result is consistent with the lowest surface energy associated with Co-termination among all surfaces, confirming the presence of a Co-terminated Co2Mo3O8 (001) surface (Fig. 1i, j, Supplementary Fig. 9 and Supplementary Table 1), which is also supported by TEM observations. The electronic structures of Co, Mo, and O sites on the Co-terminated (001) surface were further studied by analyzing the DOS and charge distribution. It was evident that on Co-terminated (001) surface, the Co projected DOS dominated the majority of the total DOS, indicating a predominant contribution of Co atoms to the electronic density of (001) surface (Supplementary Fig. 10). Furthermore, electrons were primarily localized around the metallic Co site, which contributed fewer electrons to the nearby O atoms compared to the metallic Mo site (Fig. 1k, Supplementary Fig. 11, and the Bader charge analysis results in Supplementary Table. 2). These Co sites, which were rich in electrons, provided ample opportunities for interaction with oxygen-containing species under reaction conditions. Therefore, the hexagonal Co2Mo3O8 nanostructure with a Co-terminated (001) surface is highly suitable as a model precatalyst for investigating the dynamic structure evolution mechanism during water electrocatalysis.

Manipulating the surface states of electrode and composition of electrolyte

In an alkaline medium (1 M KOH), the Pourbaix diagram indicated that Co2Mo3O8 would undergo a nucleophilic attack by electron-rich hydroxide ions at the Mo sites, leading to the generation of soluble [CoOH2]− and MoO42− species in the electrolyte46. Consequently, the Raman intensity associated with the Co2Mo3O8 lattice was notably diminished, with CoOx species predominating on the surface of the Co2Mo3O8 precatalyst (Supplementary Fig. 12). This result indicated that the reconstruction on Co2Mo3O8 could also be realized through a chemical effect. At pH = 14, the Gibbs free energy difference (ΔGpbx) was closely related to the potential applied to the precatalysts, and the stability regions encompassed all potential combinations of solid and aqueous species (Supplementary Fig. 13)47,48,49. A cyclic voltammetry curve of the Co2Mo3O8 precatalyst was performed in 1 M KOH with a scan rate of 5 mV s−1. The HER was observed to commence at potentials below 0 V vs. RHE, while the oxygen evolution reaction (OER) initiated at 1.51 V, respectively. A series of surface reactions occurring between 0 and 1.51 V were also indicated in Supplementary Fig. 14, aligning well with the Pourbaix results. Based on these findings, three protocols (Supplementary Fig. 15a–d) with varying potential settings were established and implemented to control the dynamic structural transformations of the Co2Mo3O8 precatalyst. As confirmed in Supplementary Fig. 15e, the formation of Co(OH)2 nanosheets on Co2Mo3O8 was consistent across the three protocols. However, the Co(OH)2@Co2Mo3O8 catalyst formed by Protocol 3 demonstrated the highest catalytic activity for hydrogen production (Supplementary Fig. 16).

Comprehensive investigations were conducted to elucidate the dynamic structural transformation and activity origin of the developed Co(OH)2@Co2Mo3O8 catalyst by Protocol 3. The methodology of Protocol 3 was displayed in Fig. 2a. The notably enhanced current density at Stage II indicated the structure transformations of Co2Mo3O8 precatalyst at reductive potentials, suggesting that the reconstructed structure was more conducive to HER. In-situ Raman spectroscopy was used to visualize the dynamic structural evolution process. At Stage I (1.51 V for 10 min), the nucleophilic attack by hydroxide ions was more intensive (Fig. 2b)50. Mo species in Co2Mo3O8 were rapidly etched and dissolved into the electrolyte, leading to the weakened or even disappeared Raman translational, asymmetric/symmetric bending vibrations of the [MoxOy]m− group (ranges between 150 and 430 cm−1)51. Mo-etching facilitated the exposure of CoOx on the surface of Co2Mo3O8, as evidenced by its typical and distinct Raman vibrations of A1g (689 cm−1), F2g (521 cm−1), and Eg (484 cm−1)52. As the operation time extended, the vibrations of CoOx disappeared, the A1g (436 cm−1) and A2u (504 cm−1) Raman peaks of Co(OH)2 emerged, indicating a phase change from CoOx to Co(OH)2 at 1.51 V53. A spontaneous decomposition from Co(OH)2 to CoOx was realized, as confirmed by the ex-situ Raman spectrum and TEM results after Stage I (Supplementary Figs. 17, 18). For Stage II, the Raman vibrations (A1g, A2u, and Eg) of Co(OH)2 were detected during the 30-min catalysis (Fig. 2c), suggesting that the formed Co(OH)2 was a stable phase under the electrochemical conditions (applied potential: −0.39 V; electrolyte: 1 M KOH). Ex-situ XRD patterns and SEM images of Co2Mo3O8 precatalyst after the Protocol 3 treatment further demonstrated that a high amount of Co(OH)2 nanosheets grew on the Co2Mo3O8 surface (Fig. 2d, e). The Co(OH)2@Co2Mo3O8 composite featured a well-defined Co2Mo3O8 (002) and Co(OH)2 (002) interface, where a Co2Mo3O8 core and Co(OH)2 shell structure was formed (Fig. 2f, g and Supplementary Figs. 19, 20).

Fig. 2: Manipulating the surface states of the Co2Mo3O8 electrode.
Fig. 2: Manipulating the surface states of the Co2Mo3O8 electrode.
Full size image

a Current density–time (j–t) curves of the pristine Co2Mo3O8 precatalyst under Protocol 3 treatment. Operating conditions: 1.51 V vs. RHE for 10 min, followed by −0.39 V for 15 h (A: pristine Co2Mo3O8, B: operation at 1.51 V for 10 min, C: operation at 1.51 V for 10 min and −0.39 V for 1 h, D: operation at 1.51 V for 10 min and −0.39 V for 15 h). Time-dependent in-situ Raman spectra of the Co2Mo3O8 precatalyst with continuously applied potentials of b 1.51 V and then c −0.39 V. d Ex-situ XRD patterns, and e SEM images of the same piece of Co2Mo3O8 precatalyst under different treatments in Protocol 3. f, g HRTEM image and EDX line-scan profiles of Co, Mo, and O elements of the Co(OH)2@Co2Mo3O8 catalyst developed through Protocol 3. (Inset shows the scanned region marked with an arrow.) h Side views of the Co2Mo3O8 (001) slab showing various adsorbates, such as H-, O- or OH-containing species, at different atomic sites on the surface, e.g. OH–Co represents the coverage of OH species on Co sites. i Two-dimensional (2D) surface Pourbaix diagrams for the Co2Mo3O8 (001) surface at 25 °C, illustrating the dependence of free energies on the potential and pH value of electrolytes, where ML represents monolayer. j The programmed protocols to develop various Co(OH)2@Co2Mo3O8 structures and to elucidate the potential-dependent structure evolution mechanism of the Co2Mo3O8 precatalyst. Details can be seen in Supplementary Fig. 15.

To better understand the dynamic surface structure variation on Co2Mo3O8 under realistic working conditions, a surface Pourbaix diagram based on the Co2Mo3O8 (001) surface was constructed to investigate the surface states as a function of bias voltage and pH of the electrolyte54. All possible cases involving the coverage of *OH, *O, oxygen vacancy (Ov), and *H on the Co2Mo3O8 (001) facet were taken into consideration (Fig. 2h, and Supplementary Fig. 21). Based on the calculated surface Pourbaix diagram results (Fig. 2i and Supplementary Fig. 22), the Co2Mo3O8 (001) surface was predicted to be covered by *O ranging from 1/4 to 6/4 monolayer (ML) when the potential for hydrogen evolution (USHE) exceeded 0.3 V vs. SHE (corresponding to 1.11 V vs. RHE in 1 M KOH). This facilitated the formation of CoOx with the accumulation of operation time. These disordered CoOx would be easily converted into Co(OH)2 at reductive potentials (−0.39 V vs. RHE, corresponding to −1.2 V vs. SHE in 1 M KOH). At a directly reductive potential (e.g. ≤−1 V vs. SHE), the Co2Mo3O8 (001) surface was clearly covered by 5/4 to 6/4 ML *H on O sites. The abundant *H prevented electron-rich OH− ions from approaching the inner Mo sites, effectively shielding the structure from severe etching-induced variations. This theoretic prediction aligns well with the time-dependent in-situ Raman spectra of Co2Mo3O8 at −0.39 V (Supplementary Fig. 23). These simulation results provided a clear explanation for the formation of Co(OH)2@Co2Mo3O8 by these three Protocols (Fig. 2j). The amount of Co(OH)2 on the Co2Mo3O8 surface developed by these protocols was in the order of Protocol 3 > Protocol 1 > Protocol 2 (Fig. 2d, e and Supplementary Fig. 15). The highest amount of Co(OH)2/Co2Mo3O8 interface from Protocol 3 provided the most abundant active sites for HER, thus exhibiting the highest catalytic activity (Supplementary Fig. 16).

According to the aqueous Pourbaix diagram, Mo in Co2Mo3O8 precatalyst was susceptible to nucleophilic attack by hydroxide ions and existed as MoO42− in the electrolyte (1 M KOH)46. In the Raman spectrum of the Co(OH)2@Co2Mo3O8 developed by Protocol 3 (Supplementary Fig. 24), a broad peak centered at 300 cm−1 was detected, which was attributed to the Eg mode of MoO42− adsorbed on the reconstructed catalyst (unless specified otherwise, all the subsequent mentions of Co(OH)2@Co2Mo3O8 refer to the one developed by Protocol 3)55,56. The X-ray photoelectron spectroscopy (XPS) spectra of the Co2Mo3O8 precatalyst before and after the Protocol 3 treatment were displayed in Fig. 3a–c. Following Protocol 3, the characteristic binding energy splitting of ~16 eV between Co2+ 2p3/2 (781.1 eV) and Co2+ 2p1/2, and the greatly increased OH− content in O 1s spectrum confirmed the existence of Co(OH)2 phase in the reconstructed Co(OH)2@Co2Mo3O857. Compared with the pristine Co2Mo3O8, only Mo6+ species at the binding energy positions of approximately 532.5 and 535.2 eV were detected in Co(OH)2@Co2Mo3O8, suggesting the dissolved Mo species re-adsorbed on the surface of the catalyst in the form of MoO42−58. To support this, the concentrations of etched Mo into the electrolyte under various treatment conditions in Protocol 3 were measured by an inductively coupled plasma optical emission spectrometer (ICP-OES). It was observed that dissolved Mo at Stage I reached 27.1 mg L−1 and gradually decreased to 17.2 mg L−1 at point D of Stage II (Fig. 3d). Even when adding MoO42− with concentrations ranging from 10 to 100 mM into 1 M KOH electrolyte, the noticeably reduced Mo concentration in the electrolytes post Protocol 3 convincingly indicated the capture of Mo species by the catalysts (Fig. 3e). The stronger recapturing ability of MoO42− by the reconstructed Co(OH)2@Co2Mo3O8 was further supported theoretically by its more negative adsorption free energy of −7.7 eV (Fig. 3f). MoO42− additives had negligible influence on the concentration of dissolved Co2+ ions from the precatalyst in 1 M KOH (Supplementary Fig. 25), but a higher concentration of MoO42− in electrolytes lowered the content of formed crystalline Co(OH)2 nanosheets on Co2Mo3O8 post-Protocol 3 and reduced their size (Supplementary Figs. 26, 27). The reconstructed Co(OH)2@Co2Mo3O8 in 1 M KOH with a higher concentration of MoO42− consistently exhibited a larger current density (Supplementary Fig. 28), demonstrating that MoO42− played a more critical role in accelerating hydrogen production than the content of surface Co(OH)2.

Fig. 3: Revealing the interaction of MoO42− with catalyst.
Fig. 3: Revealing the interaction of MoO42− with catalyst.
Full size image

High-resolution X-ray photoelectron spectroscopy (HRXPS) of a Co 2p, b O 1s, c Mo 3d in pristine Co2Mo3O8 and the developed Co(OH)2@Co2Mo3O8 by Protocol 3. d The concentration of Mo in electrolytes after treating Co2Mo3O8 precatalyst under different conditions in Protocol 3, A: fresh electrolyte; B: 1.51 V for 10 min; C: 1.51 V for 10 min and −0.39 V for 1 h; D: 1.51 V for 10 min and −0.39 V for 15 h. e The Mo content in electrolytes with various concentrations of MoO42− additive before and after Protocol 3. f Calculated adsorption energies of MoO42− on Co2Mo3O8 and Co(OH)2@Co2Mo3O8. g Time-dependent current density curves of Co2Mo3O8 and Co(OH)2@Co2Mo3O8 catalysts with the injection of various concentrations of MoO42− in 1 M KOH, and the inset shows the current density changes after altering the electrolyte composition.

To further explore the impact of MoO42− concentration on the HER performance of the pristine Co2Mo3O8 and Co(OH)2@Co2Mo3O8 catalysts, time-dependent current density measurements at −0.39 V (without iR-compensation) were conducted following the injection of MoO42−. The pH values of electrolytes with various concentrations of MoO42− maintained consistently, ruling out the possibility of pH calibration errors (Supplementary Fig. 29). Under all tested conditions, the Co(OH)2@Co2Mo3O8 catalyst consistently delivered notably higher current densities compared to both the pristine Co2Mo3O8 and Co(OH)2 catalysts (Fig. 3g, Supplementary Figs. 30, 31)51. Moreover, there was a substantial increase in current density with the introduction of MoO42− and the current density showed a smaller decrease when returning to 1 M KOH, consistent with the behavior observed for the Co(OH)2@Co2MO3O8 developed by Protocol 2 (Supplementary Fig. 32). These findings further suggested that both the heterointerface within Co(OH)2@Co2Mo3O8 and the interaction with MoO42− notably enhanced HER performance, with the latter one being the predominant factor.

Mechanistic insights

To understand the notably enhanced HER activity of the reconstructed Co(OH)2@Co2Mo3O8 catalyst with the addition of MoO42− in the electrolyte, it was crucial to analyze the synergistic effects between the reconstructed catalyst and the electrolyte’s composition. Characterization results confirmed the structural transformation of the catalyst from Co2Mo3O8 to Co(OH)2@Co2Mo3O8 during the electrochemical operation, as well as the interaction between MoO42− and the Co(OH)2@Co2Mo3O8 catalyst. Raman spectra of Na2MoO4·2H2O powder and MoO42− ions in 1 M KOH identified three distinct vibration modes at approximately 318, 834, and 897 cm−1, which corresponded to the symmetric bending mode (Eg), asymmetric stretching mode (F2g), and symmetric stretching mode (A1g) of MoO42−, respectively (Supplementary Figs. 33, 34). Potential-dependent in-situ Raman spectra of the Co(OH)2@Co2Mo3O8 catalyst were subsequently performed in 1 M KOH with 50 mM MoO42− (Fig. 4a and Supplementary Fig. 35). At open-circuit potential (OCP), the vibration modes at 318 and 897 cm−1 were attributed to Eg and A1g of MoO42−, respectively, with additional peaks arising from Co(OH)2 within the reconstructed catalyst. When the applied potential was more positive than −0.05 V vs. RHE, the Raman vibrations remained unchanged from those observed at OCP, indicating that MoO42− was stably associated with the catalyst, facilitating hydrogen production. However, when the applied potential decreased below −0.05 V, a broad peak at ~267 cm−1 emerged while other vibrations remained consistent. This new peak corresponded to the Mo–O–Mo motif in Mo2O72−59. Thus, a potential-dependent action mechanism for the Co(OH)2@Co2Mo3O8 catalyst in 1 M KOH with MoO42− additive was revealed. Specifically, the cooperative interaction between MoO42− and Co(OH)2@Co2Mo3O8 was responsible for the comparable HER performance at potentials more positive than −0.05 V. Conversely, the interaction of Mo2O72− with Co(OH)2@Co2Mo3O8 accounted for the notable HER activity at potentials more negative than −0.05 V.

Fig. 4: Theoretical calculation.
Fig. 4: Theoretical calculation.
Full size image

a Potential-dependent in-situ Raman spectra of the Co(OH)2@Co2Mo3O8 catalyst in 1 M KOH with 50 mM MoO42−, with a potential amplitude of −50 mV. b Structural models of MoO42−–Co2Mo3O8, MoO42−–Co(OH)2, MoO42−–Co(OH)2@Co2Mo3O8 and Mo2O72−–Co(OH)2@Co2Mo3O8. c Calculated Co 3d projected DOS and corresponding d-band center values. d Charge distributions of MoO42−/Mo2O72−–Co(OH)2@Co2Mo3O8 models in top and side views (Isosurface units: 0.0023 and 0.0018 e Å−3 for Co(OH)2@Co2Mo3O8 with the interaction of MoO42− and Mo2O72−, respectively, color legend for isosurface: cyan indicates charge depletion, yellow indicates charge accumulation). e Calculated kinetic barrier of water dissociation (IS initial state, TS transition state, FS final state). f Free energy diagrams of *H intermediates adsorbed on various nanostructures. The inset shows the structural model of MoO42−/Mo2O72−–Co(OH)2@Co2Mo3O8 with *H. g Proposed mechanism for the enhanced HER performance over the MoO42−/Mo2O72−–Co(OH)2@Co2Mo3O8 catalyst.

The mechanism underlying the enhancement of HER performance through the synergistic effects of the purposefully developed Co(OH)2@Co2Mo3O8 heterostructure and the interaction of MoO42−/Mo2O72− was theoretically elucidated using DFT calculations. The calculation models for pristine Co2Mo3O8, Co(OH)2, and Co(OH)2@Co2Mo3O8 were based on our experimental observations. As illustrated in Fig. 4b, MoO42− and Mo2O72− were grafted onto the surfaces of various nanostructures through favorable energy interactions. The d-band theory, widely used to evaluate the adsorption ability of intermediates on catalyst surfaces, was applied in this study. The calculated d-band centers (εd) for the MoO42−–Co(OH)2@Co2Mo3O8 and Mo2O72−–Co(OH)2@Co2Mo3O8 were −0.73 and −1.39 eV, which was closer to the Fermi level than those of MoO42−–Co2Mo3O8 (εd = −1.46 eV), MoO42−–Co(OH)2 (εd = −1.91 eV), and Co(OH)2@Co2Mo3O8 (εd = −1.48 eV). This suggested stronger adsorption of reactants (H2O molecules) on MoO42−/Mo2O72−–Co(OH)2@Co2Mo3O8 (Fig. 4c and Supplementary Fig. 36)60. Charge analysis revealed that the interaction of MoO42−/Mo2O72− with the formed Co(OH)2@Co2Mo3O8 induced charge redistribution on the catalyst surface, where Co atoms near MoO42−/Mo2O72− in Co(OH)2 presented electron-deficient states, and O atoms in MoO42−/Mo2O72− displayed electron-rich states (Fig. 4d and Supplementary Figs. 37, 38). Such a surface configuration altered the behaviors of H2O dissociation and hydrogen adsorption (Supplementary Figs. 39, 40). Figure 4e depicted the free energy diagrams of water dissociation (Volmer step) on Co active sites of Co2Mo3O8, Co(OH)2@Co2Mo3O8, MoO42−–Co(OH)2@Co2Mo3O8, and Mo2O72−–Co(OH)2@Co2Mo3O8. The energy barrier for the Volmer reaction was much higher for Co2Mo3O8 (0.67 eV) compared to the reconstructed catalysts with and without MoO42−/Mo2O72− interaction (0.13 eV for Co(OH)2@Co2Mo3O8, 0.2 eV for MoO42−–Co(OH)2@Co2Mo3O8, and 0.17 eV for Mo2O72−–Co(OH)2@Co2Mo3O8)61. The reduced energy barrier of the Volmer reaction on the Co(OH)2@Co2Mo3O8 was in accordance with the smaller Tafel slope (Supplementary Fig. 41f). At the same applied potential, the Co(OH)2@Co2Mo3O8 catalyst exhibited both lower frequencies and phase angles compared to Co2Mo3O8, suggesting a dominating Volmer process for the former catalyst, while the Heyrovsky step was the primary reaction for the latter during HER (Supplementary Fig. 42)62. The binding energy of hydrogen (∆G*H) was evaluated as an activity descriptor for assessing proton adsorption (*H) and H2 desorption behaviors (Fig. 4f). An ideal HER catalyst should have a ∆G*H value close to zero to balance these processes. Among the considered models, MoO42−–Co(OH)2@Co2Mo3O8 and Mo2O72−–Co(OH)2@Co2Mo3O8 showed very close ∆G*H of 0.16 and 0.06 eV to zero, while Co(OH)2@Co2Mo3O8 displayed a ∆G*H value of −0.81 eV, indicating strong hydrogen adsorption which inhibited *H coupling and the subsequent release of H263,64. Interestingly, the interaction of MoO42−/Mo2O72− with the reconfigured Co(OH)2@Co2Mo3O8 notably optimized the hydrogen behaviors, resulting in a substantial improvement in H2 yield rate compared with the Co2Mo3O8 catalyst, as evidenced by an ~21.2-fold increase in the hydrogen yield rate at −0.4 V (Fig. 5c). Overall, the Co(OH)2@Co2Mo3O8 heterostructure facilitated the water dissociation (Volmer reaction), while the interaction of MoO42−/Mo2O72− with the reconstructed catalyst optimized the hydrogen behaviors (Fig. 4g). This positive synergism between the heterostructure and its interaction with MoO42−/Mo2O72− anions led to a dramatically enhanced activity and stability of MoO42−/Mo2O72−–Co(OH)2@Co2Mo3O8 catalyst towards alkaline HER.

Fig. 5: HER performance.
Fig. 5: HER performance.
Full size image

a Linear sweep voltammetry (LSV) curves of the Co2Mo3O8, Co(OH)2@Co2Mo3O8, and commercial 20 wt% Pt/C catalysts with iR-compensation in 1 M KOH with 50 mM MoO42−. b Comparison of overpotentials (η10 and η50) at current densities of 10 and 50 mA cm−2, mass activities, turnover frequency (TOF) values, charge transfer resistance (Rct) at η = 200 mV, and the double layer capacitance (Cdl) for the Co2Mo3O8 and Co(OH)2@Co2Mo3O8 catalysts. The error bars of overpotentials were derived from three individual measurements. Mass activities, TOF, Cdl, Rct, and Tafel slopes were performed once. c Hydrogen yield rates and Faradaic efficiencies (FEs) at different potentials (V vs. RHE, without iR-compensation) for the Co2Mo3O8 and Co(OH)2@Co2Mo3O8 catalysts. d Chronoamperometry curves of the Co(OH)2@Co2Mo3O8 and commercial 20 wt% Pt/C catalysts at a fixed potential of −0.39 V vs. RHE, without iR-compensation. Insets are the enlarged views at different time scales.

HER performance

To further advance the positively cooperative effect of the reconstructed catalyst with the interaction of anion, the electrochemical HER performance of the pristine Co2Mo3O8 and the reconstructed Co(OH)2@Co2Mo3O8 catalysts was evaluated in 1 M KOH with 50 mM MoO42−. Compared with pristine Co2Mo3O8, the Co(OH)2@Co2Mo3O8 catalyst with the interaction of MoO42−/Mo2O72− exhibited notably improved HER activity. It achieved current densities of 10 and 50 mA cm−2 (j10 and j50) at overpotentials (η) of only 85 and 161 mV, respectively (Fig. 5a, b, and Supplementary Figs. 41, 43). These values were substantially less than those observed for the pristine Co2Mo3O8 catalyst, which required overpotentials of 359 and 479 mV to reach j10 and j50, respectively. Additionally, the Co(OH)2@Co2Mo3O8 catalyst exhibited a much higher mass activity (197.8 A g−1), turnover frequency (TOF) value (0.244 s−1), double layer capacitance (Cdl, 7.9 mF cm−2), and lower charge transfer resistance (Rct, 3.3 Ω) at η = 200 mV, further evidencing its increased active sites and accelerated reaction kinetics in hydrogen production (Fig. 5b and Supplementary Fig. 41). Hydrogen production was quantitatively analyzed using a gas chromatography column, revealing that the Co(OH)2@Co2Mo3O8 catalyst achieved a hydrogen yield rate more than ten times greater than that of pristine Co2Mo3O8 across all tested potentials. Particularly at −0.4 V, the catalyst reached a comparable Faradaic efficiency (FE) of ~99.9% and a hydrogen yield rate of 1.85 mol h−1. The yield rate is ~21.2 times higher than that of the Co2Mo3O8 catalyst, underscoring its notable HER activity (Fig. 5c). Impressively, the Co(OH)2@Co2Mo3O8 catalyst with the interaction of MoO42−/Mo2O72− also demonstrated prolonged stability, maintaining operation at −0.39 V (without iR-compensation, current density of ~100 mA cm−2) for more than one month without any detectable phase transition (Fig. 5d and Supplementary Fig. 44). During this period, the current density showed a modest increase of about 8 mA cm−2 compared to the value at 15 h. In sharp contrast, the current density of commercial 20 wt% Pt/C catalyst drastically declined from ~230 to 62 mA cm−2 within a single day. This catalyst showcased notably better HER stability compared to other reported catalysts of the same kind for alkaline water splitting (Supplementary Table 3). These results highlight the promising potential of the MoO42−/Mo2O72−-Co(OH)2@Co2Mo3O8 catalyst for industrial electrochemical hydrogen production applications, given its comparable activity and reliable stability.

Discussion

In summary, we have successfully synthesized pure hexagonal Co2Mo3O8 with predominantly exposed (001) facets and employed them as a model catalyst to study the dynamic surface structure evolution of precatalysts in water splitting. It was found that the surface structure transformation of Co2Mo3O8 precatalyst is potential-dependent, of which the etching of Mo by nucleophilic attack of hydroxide ions resulted in the exposure of CoOx on precatalyst surface and the existence of MoO42− in electrolytes. A further reductive potential led to the formation of an electrochemically stable Co(OH)2@Co2Mo3O8. Mechanistic investigations demonstrated that the Co(OH)2@Co2Mo3O8 heterostructure accelerated water dissociation, while the potential-dependent interaction of MoO42−/Mo2O72− with Co(OH)2@Co2Mo3O8 activated the energy barrier of H adsorption, H-coupling, and H2 desorption, leading to greatly enhanced HER activity. Inspired by this understanding, a bias-manipulated method (Protocol 3) was employed to construct a heterostructure containing a high amount of Co(OH)2 on Co2Mo3O8 (Co(OH)2@Co2Mo3O8). In 1 M KOH with 50 mM MoO42−, it achieved a notable hydrogen yield rate of 1.85 mol h−1 with an FE of 99.9% at −0.4 V vs. RHE, and stable operation for over one month at a current density of ~100 mA cm−2, showcasing notable HER activity and stability for alkaline water splitting. This work underscores the importance of considering variations in both catalyst and electrolyte in the rational design of high-performance nanostructures for industrial water electrocatalysis.

Methods

Chemicals and materials

Cobalt (II) chloride hexahydrate (CoCl2·6H2O, ≥99.99%), ammonium molybdate (H24Mo7N6O24·4H2O, ≥99.99%), sodium molybdate dehydrate (Na2MoO4·2H2O, ≥99%), Urea (CH4N2O, ≥99.5%), 2-methylimidazole (C4H6N2, ≥98%), dimethyl sulfoxide (C2H6SO, ≥99.9%), isopropanol (C3H8O, ≥99.9%), potassium hydroxide (KOH, ≥99.99%) were purchased from Aladdin Scientific Corp. All chemicals were used as received without any further purification if not specifically indicated. Hydrophilic carbon paper (TGP-H-060) was purchased from Toray Group. 20 wt% Pt/C was obtained from Sigma Aldrich, and 5 wt% Nafion dispersion was purchased from DuPont de Nemours, Inc.

Synthesis of CoMoO4 precursor

In a typical procedure, a solution was prepared by dissolving CoCl2·6H2O (0.119 g), H24Mo7N6O24·4H2O (0.088 g), and CH4N2O (0.3 g) into 36 mL of deionized water. After electromagnetically stirring for 2 h, the mixture was transferred to a 50 mL Teflon-lined stainless-steel autoclave, sealed, and then kept at 120 °C for 12 h. Following the completion of the reaction, the autoclave was cooled down to room temperature naturally, and the precipitates were collected and subjected to a centrifugation process (12,000 rpm, 5 min) for five times using deionized water. The obtained precipitates were then dried in a vacuum oven at 60 °C for 8 h to obtain the CoMoO4 micro-flowers (denoted as CoMo-P).

Synthesis of CoMoO4@ZIF-67 precursor

CoMoO4 micro-flowers (0.2 g) and 2-methylimidazole (2-MIM, 0.5 g) were poured into a beaker containing 10 mL of dimethyl sulfoxide (DMSO) and stirred for 6 h. Then, CoCl2·6H2O (0.119 g) was dissolved into 5 mL of DMSO and added to the aforementioned mixture. The combined solution was further stirred for 12 h at room temperature. The resulting product was centrifugated at 12,000 rpm for 5 min and then washed with ethanol and deionized water, followed by drying at 60 °C for 8 h in a vacuum oven. This resulting product was denoted as CoMo@ZIF-67-P.

Synthesis of pristine Co2Mo3O8

The CoMo@ZIF-67-P hybrid (0.3 g) was heated in an argon atmosphere from room temperature to 800 °C with a heating rate of 5 °C min−1 and then maintained for 2 h. After naturally cooling down to room temperature, the product was centrifugated, washed with deionized water, and dried at 60 °C for 8 h in a vacuum oven to obtain a Co/CoO@Co2Mo3O8 intermediate. To develop the pristine Co2Mo3O8 nanostructure, the Co/CoO@@Co2Mo3O8 hybrid was subjected to acid-etching in 0.5 M HNO3 solution for 10 h at room temperature. Then, the acid-etched material underwent the same washing and drying process as described above.

Electrode preparation

Prior to being used as the current collector, carbon paper (CP) was cut into small pieces with a surface area of 1.2 × 1 cm2. These CP species were then sonicated and washed with ethanol, aqua regia, and deionized water in sequential order to remove surface impurities and enhance hydrophilicity. After drying in a vacuum oven, these CP pieces were saved for further use as current collectors. Meanwhile, 10 mg catalysts (including 20 wt% Pt/C) and 30 μL 5 wt% Nafion solution were added into 470 μL of isopropanol. The mixture was then sonicated for 1 h to get a uniformly dispersed ink. Using a pipette, the catalyst ink was transferred and coated onto the surface of the small CP pieces (coating area: 1 × 1 cm2). After coating, the working electrodes were dried at 60 °C for 1 h in a vacuum oven. The mass loading of the catalyst on the electrode was measured and fixed at approximately 1.5 mg cm−2.

The construction of Co(OH)2@Co2Mo3O8 heterostructures involved three different protocols: Protocol 1, Protocol 2, and Protocol 3. Each protocol had specific steps and conditions. Details can be seen in Supplementary Fig. 15.

Materials characterization

The morphologies, structures, and composition of the samples were characterized using a field emission scanning electron microscope (Thermal Fisher Quattro S Environmental SEM) operating at an acceleration voltage of 15 kV and a transmission electron microscopy (TEM, JEM-2100F JEOL Ltd., Japan) at 200 kV. X-ray powder diffraction (XRD) patterns were obtained using a Rigaku D/max 2550 diffractometer with Cu Kα radiation (λ = 0.15405 nm). Raman spectral analysis was conducted using a Reishaw 2000 Raman Spectroscopy system equipped with a laser source of 532 nm. In-situ Raman spectroscopy was performed in a flow cell setup, where an Ag/AgCl filled with 3.5 M KCl solution and a Pt wire served as reference and counter electrode, respectively. The Raman signals from the working electrodes were continuously monitored in 1 M KOH with or without 50 mM MoO42− electrolyte, during a step-wise potential increase of −50 mV to observe microstructural changes. Fourier transform infrared (FTIR) spectroscopy was carried out using a Nicolet 6700 infrared spectroscope (Thermo Nicolet Corporation) spanning a range of 400–3800 cm−1, with KBr pellets used for reference. Surface chemical analyses were performed through X-ray photoelectron spectroscopy (XPS, ESCALAB Xi+, Thermo Fisher Scientific) with binding energies calibrated using the C 1s peak at 284.8 eV as a reference. The concentrations of cobalt and molybdenum in fresh Co2Mo3O8, the reconstructed Co(OH)2@Co2Mo3O8 catalyst, and in the electrolytes upon reconstruction were quantified using an inductively coupled plasma optical emission spectrometer (ICP-OES, Optima 8000).

Electrochemical measurements

Electrochemical experiments were performed at ambient temperature using a standard three-electrode system, controlled by a Chenhua CHI 660E electrochemical workstation. A carbon rod acted as the counter electrode, and an Hg/HgO electrode, filled with 1 M KOH, served as the reference electrode. The working electrode comprised a catalyst-coated carbon paper. The electrolyte solution, containing 1 M KOH and variable concentrations of molybdate (MoO42−, ranging from 0 to 100 mM), was prepared in a 250 mL volumetric flask. Herein, the required quantities of KOH and Na2MoO4·2H2O were weighed, transferred into the flask, and diluted to the mark with deionized water. This solution was used immediately after mixing and was purged with high-purity argon to remove impurities prior to the experiment. Cyclic voltammetry (CV) tests were conducted at various scan rates (50, 100, 200, 300 mV s−1), within a range devoid of faradaic processes, to determine the electrochemical double-layer capacitance (Cdl). The Cdl was calculated by plotting half the difference in current density (Δ j/2) against the scan rates. Linear sweep voltammetry (LSV) curves were acquired at a scan rate of 5 mV s−1. Unless stated otherwise, the reported potentials do not include iR compensation. Electrochemical impedance spectroscopy (EIS) was carried out over a frequency range from 100 kHz to 0.01 Hz at various potentials, with an AC voltage amplitude of 5 mV. Potentials were referenced to the reversible hydrogen electrode (RHE) using the Nernst equation as shown65:

$${E}_{{\rm{RHE}}}={E}_{{\rm{Hg}}/{\rm{HgO}}}+{E}_{{\rm{Hg}}/{\rm{HgO}}}^{0}+0.059\times {\rm{pH}}$$
(1)

where \({E}_{{\rm{Hg}}/{\rm{HgO}}}\) is the measured potential, \({E}_{{\rm{Hg}}/{\rm{HgO}}}^{0}\) is the standard electrode potential of the Hg/HgO electrode (0.098 V at room temperature), and the pH of the electrolyte is ~14.

Mass activity (A g−1) of the catalyst was quantified using the catalyst loading (m) and the observed current density (j, mA cm−2), according to Eq. (2):

$${{\rm {Mass}}\; {\rm {activity}}}=j/m$$
(2)

The turnover frequency (TOF) was determined under the assumption that every cobalt (Co) atom acts as an active site participating in the catalytic process, as illustrated by Eq. (3)66:

$${{\rm {TOF}}}=(\;j\times S)/(z\times F\times n)$$
(3)

where j represents measured the current density (mA cm−2); S is the geometrical surface area of the working electrode (1 cm2); z is the number of electron transfers per molecule of the H2 produced for the HER (z = 2); F is Faraday’s constant (96,485.3 C mol−1), and n is the mole (mol) of Co atoms within the catalyst loading.

Hydrogen production measurement

The electrochemical hydrogen production tests were conducted in an H-type cell with a Chenhua CHI 660E electrochemical workstation. A carbon rod was used as the counter electrode, and a Hg/HgO electrode filled with 1 M KOH served as the reference electrode. The cathodic electrolyte was 1 M KOH with 50 mM MoO42−, while the anodic electrolyte was 1 M KOH. Before the measurements, the electrolyte was bubbled with high-purity argon for 30 min to remove dissolved O2 and H2. The volume of electrolyte in each chamber was maintained at 30 mL, separated by a Nafion 117 membrane. During the constant potential test (without iR compensation), high-purity argon with a flow rate of 50 sccm was continuously supplied to the cathode chamber. The gas outlet was connected to a gas chromatography (GC, Shimadzu GC-2010 plus) for online analysis of the gas products. The produced hydrogen was detected using a thermal conductivity detector (TCD). The Faradaic efficiencies of H2 were calculated using the following equation:

$${\rm{FE}} \%=\frac{2F{P}_{0}}{{RT}}\times \frac{1}{\alpha }\times {peak\; area}\times ({flow\; rate})$$
(4)

where F represents the Faraday constant (96,485 C mol−1), and R means the gas constant (8.314 J mol−1 k−1). P0 is the ambient pressure (101,325 Pa), T is the room temperature (298.15 K), and α is the conversion factor for H2 on the basis of the calibration of the GC with a standard sample.

Computational details

Spin-polarized density functional calculations were conducted via Vienna ab initio simulation package (VASP)67, in which the projector augmented wave (PAW) method was employed68. Kohn–Sham wave functions are expanded in a plane-wave basis set to evaluate the valence electrons with a kinetic energy cutoff fixed at 520 eV69,70. General gradient approximation (GGA) method71 parametrized by the revised Perdew–Burke–Ernzerhof (RPBE) functional72 was employed to compute the electron exchange and correlation interactions. All structures were relaxed until the forces on each atom were less than 0.05 eV Å−1. The Brillouin zone was sampled by a (3 × 3 × 1) k-point mesh. A vacuum slab of 15 Å was introduced along the z-direction to separate two periodic surfaces. The atomic simulation environment (ASE) Libraries were used for crystal structure manipulation, input generation, and surface stability comparison73. Dispersive interactions were described by Grimme’s DFT-D3 method74. Climbing image nudged elastic band (CI-NEB) method was introduced to evaluate the transition state of water dissociation75. The change in Gibbs free energy for each step was computed by taking advantage of the computational hydrogen electrode (CHE) model developed by Nørskov et al.76. The zero-point energy (ZPE) and entropy differences (∆S) between the gas phase and the adsorbed state were calculated from the vibrational frequencies of the object (Supplementary Table 6). An empirical solvation correction value of approximately 0.15 eV, referred to in previous study77, was employed for the H adsorption on the oxygen site. Surface Pourbaix diagrams were calculated based on the method proposed by Hansen et al.78. The equilibrium of water dissociation during electrocatalysis can be represented by

$${{{H}}}_{x}{{{O}}}_{y}^{*}+(2y-x) * ({{{H}}}^{+}+{{{e}}}^{-})\rightleftarrows \, * +{{y{{H}}}}_{2}{{O}}$$
(5)

where y and x stand for the number of O and H adsorbed on the primitive surface, respectively. The energy of different surface states shown in the surface Pourbaix diagram (\({G}_{{\rm {P}}}\)) was computed using the following equation:

$${G}_{{{P}}}={G}_{{{{pristine}}}}+y{G}_{{{{H}}}_{2}{{O}}}-{G}_{{{{pri}}}{{\_}}{cov}}-(2y-x)\left(\frac{1}{2}{G}_{{{{H}}}_{2}}-{U}_{{{{SHE}}}}-2.303{k}_{{{B}}}T * {{{pH}}}\right)$$
(6)

where \({G}_{{{\rm {pristine}}}}\), \({G}_{{{\rm {H}}}_{2}{\rm {O}}}\), \({G}_{{{\rm {pri}}\_}\mathrm{cov}}\), and \({G}_{{{\rm {H}}}_{2}}\) represent the total energies of the pristine surface, a water molecule, the surface with adsorbates, and a hydrogen molecule, respectively. The bias referred to a standard hydrogen electrode (SHE) is denoted by \({U}_{{{\rm {SHE}}}}\), \({k}_{\rm {{B}}}\), and \(T\) are Boltzmann constant and temperature (298.15 K), respectively.

Surface energy determination

To identify the most energetically favorable stoichiometric surface among different terminated configurations, the surface energies (\({E}_{\rm {{S}}}\)) were calculated utilizing the following equation:

$${E}_{{{S}}}=\frac{{E}_{{{{Slab}}}}-{{d}}{E}_{{{{Bulk}}}}}{2S}$$
(7)

where \({E}_{{\rm {{Slab}}}}\) and \({E}_{{\rm {{Bulk}}}}\) are the total energy of the slab cleaved from the bulk structure and the bulk models, respectively. d stands for the atomic ratio between the slab and bulk structures, and S is the area of the slab model.

Ab initio molecular dynamics (AIMD) simulations and evaluation of the solvation effect

To accurately model the metal water interface and examine the solvent effect, AIMD simulations were conducted using the VASP code with the RPBE exchange-correlation functional and D3 dispersion correction79,80. The simulations utilized an energy cutoff of 400 eV within a plane wave basis and employed a Gaussian smearing set at 0.1 eV. We performed these simulations at the Γ-points with 1 fs time step, maintaining the system temperature at 298 K using a Nosé thermostat across 10,000 simulation steps, as shown in Supplementary Data 1. To account for diverse structural complexities, we constructed explicit solvation models for the Co2Mo3O8 (001) surface, incorporating up to 45 water molecules. This model serves as a representative for other configurations, based on the premise that various structures involved in this research share similar solvation corrections. From the AIMD results, 11 structures were selected at regular intervals to assess the solvation effects on hydrogen adsorption energy (Supplementary Fig. 46) by comparing them against values obtained in a vacuum (without a water layer). An average solvation correction value of approximately −0.17 eV was obtained which is close to the empirical value as introduced. Additionally, to further verify the solvation correction value, we developed varying models of MoO42−–Co(OH)2@Co2Mo3O8 and Mo2O72−–Co(OH)2@Co2Mo3O8, incrementally adding water molecules (ranging from 1 to n) to interact with the adsorbed hydrogen (Supplementary Fig. 47). This iterative process continued until the solvation correction values for *H adsorption stabilized, ensuring consistent and reliable data for further analysis.