Introduction

The conversion of intermittent electrical energy into hydrogen with high-energy density (120 MJ kg–1) by water electrolysis strategy is promising for sustainable energy applications and carbon neutrality1,2. The efficiency of water electrolysis is mainly determined by the oxygen evolution reaction (OER), which limits the energy conversion efficiency of electrocatalytic devices due to its sluggish multi-step proton-coupled electron transfer3,4. Although ruthenium- and iridium-based materials are applied as the benchmark catalysts for the OER, the high cost and scarcity render them unfavorable for large-scale industrial applications5,6. The first-row transition metals possess intrinsic electronic structures with unfavorable interactions with oxygen-containing intermediates. Studies have shown that polymetallic or heteroatom doping can adjust the electronic structure of the first-row transition metal and increase its metal active sites (MASs) exposure to accelerate the OER process7,8. However, the involved multi-component system still faces great challenges in determining the true catalytic active site, and the dopant undergoes dynamic leaching during the electrochemical process, which is not conducive to directional design and study of OER electrocatalysts9.

Coordination complexes (CCs) feature adjustable metal sites, permanent porosity, and well-defined periodic coordinated or supramolecular structures, serving as the model catalysts to elucidate the structure-activity relationship. Besides, by ligands designed to modulate the local electron density, it is available to optimize the reactivity of transition metal-based coordination centers10,11. Despite the merits, many state-of-the-art CCs are inferior in conductivity and feature dynamical microenvironment rebuilding during the OER process, leaving the role of intrinsic microstructure inconclusive12,13,14. Recent advances have shown that the stability and conductivity of CCs can be optimized by applying multigradient-conjugated ligands to construct three-dimensional (3D) coordination polymers15,16. However, these implementations bring challenges for tailoring metal centers, in the premise that open coordination sites (OCSs) are always blocked by the skeleton ligands. This is because OCSs usually exist only during coordination competition or configurational transformation in classical coordination chemistry, which provides high energy barriers of central metals to undergo coordination configuration distortion or ligand removal for association with OER intermediates. It is highly desired to reveal the dynamic OCSs behaviors and unlock the optimal coordination configuration with enhanced kinetics for better OER applications17,18.

Water molecules are pivotal reactants in the water electrolysis process19. Given the involved proton/electron-transfer steps, by coordination with central metal sites (as potential OCSs), the water molecules might be directly activated to the excited states for efficient O–O coupling, while re-coordination can be easily achieved with the water molecules in the pores of CCs. The advances in water environment modulation have been witnessed in recent years. For example, a tailored dinuclear copper complex was engineered to deliver the reaction synergy between the coordinated water molecules and the boric acid buffer for enhanced kinetics20. Besides, the “missing linkers” in typical Co-BDC were found to serve as potential water-coordination sites to optimize OER reactivity21. Despite the discoveries, the key behaviors of coordinated water molecules and the potential complex synergy mechanism remain better understood. Identifying the catalytic models with different coordinated water modes is urgently desired to comprehend the structure-performance relationship and the OER mechanism. It will provide profound implications for the development of better CCs and resolve the coordination competition between the intermediates and ligands toward an optimal catalysis process.

Herein, a tripodal molecule, namely tris((pyridin-4-ylmethyl)benzoimidazol-2-ylmethyl)amine (tPBA), derived from a corrosion inhibitor22, was selected as the organic ligand. The pocket of tPBA exhibits strong chelation with the metal center to protect metal ions from leaching, while the flexible methylpyridine group as a bridging agent extends the structure in the 3D space. In addition, the large steric resistance drives pyridine to be connected by head-to-head and renders water molecules that temporarily occupy OCSs to coordinate in the radial direction. We select three bridging carboxylate anionic ligands to enable the nickel metal centers with different coordination water environments. Single crystal X-ray diffraction (SCXRD) measurements demonstrate that the tailored nickel centers show multiple coordination water molecular environments in the π–π stacking support for 3D supramolecular (Ni-tPBA0~2) and coordination polymer frameworks (Ni-tPBA3). As indicated by in situ analysis and theory simulations, the nickel centers with mono-coordinated water molecules may undergo the intermolecular oxygen coupling, while those with neighboring coordinated water molecules will preferably follow intramolecular oxygen coupling mechanism (IOM) during the OER. Moreover, the involvement of more coordinated water molecules would enable intramolecular electron transfer, thus optimizing the kinetic step. As a proof-of-concept, the design endows the as-crafted Ni-tPBA3 with low overpotentials, Tafel slopes (η10 = 248 mV, 49.7 mV dec–1), and long endurance of 200 h. This study expands the catalyst platform for OER and sheds light on the construction of potential OCSs through molecular engineering.

Results

Material Synthesis and Structural Characterizations

Tripodal benzimidazole tPBA with alkali stabilization characteristics was prepared by assigning additional methylpyridine bridging groups to the organic ligand possess a chelating pocket for corrosion inhibition (Fig. 1a). And tPBA was successfully self-assembled with nickel nitrate hexahydrate into a series of CCs, Ni-tPBA0~3, under simple solvothermal conditions in synergy with the co-coordination of terephthalic acid (H2TPA), isophthalic acid (H2IPA), and 1,3,5-benzenetricarboxylic acid (H3BTC), respectively. The crystallographic structures of Ni-tPBAs were collected by SCXRD and the 3D stacking patterns were analyzed and shown in Fig. 1b. The electronic structure of Ni-tPBA0~3 was further evaluated by calculating the frontier orbital (Fig. 1c). Ni-tPBA3 has higher HOMO energy levels compared to Ni-tPBA0~2, indicative of its stronger electron delocalization characteristics. Moreover, a low HOMO-LUMO gap is delivered for Ni-tPBA3, evidencing its better electronic conduction characteristics.

Fig. 1: Design and synthesis of Ni-tPBA0~3.
figure 1

Schematic illustration of a coordination and π–π stacking modes of the tPBA ligand and b the construction of Ni-tPBA0~3 (The blue polyhedron represents the coordination environments of nickel). c Computational results for the electronic structure of Ni-tPBA0~3 (Isosurfaces in purple and red represents the positive and negative phase, respectively).

Full size image

The high phase purity and crystallinity of Ni-tPBA0~3 were studied by comparing the powder X-ray diffraction (PXRD) and simulated results (Supplementary Fig. 1). The light microscopy photographs in the illustration show that Ni-tPBA0~3 exhibits cubic rod, block, layered, and plate macroscopic structure, respectively. Ni-tPBA2 exhibits a violet color feature different from cyan, possibly due to the dimethylacetamide (DMAc) crystallized in the pore. The free DMAc alters the d-orbital splitting in the nickel center, leading to an energy difference in the change of electrons from the excited state back to the ground state that affects the color expression23.

The single-crystal structural variability of Ni-tPBA0~3 was studied, with the detailed descriptions of the individual crystal structures shown in Supplementary Crystal Structure Description. Without auxiliary bridging by carboxylate anions, Ni-tPBA0 exhibits a zero-dimensional (0D) binuclear structure with two tPBA chelated nickel linking through the pyridine foot (Supplementary Fig. 2). The introduction of TPA on Ni-tPBA0 unit successfully replaced the pyridine linkage, but the single assembly mode of the straight auxiliary ligand H2TPA failed to enable the structural expansion of 0D Ni-tPBA0 (Supplementary Fig. 3). Therefore, the V-type ligand H2IPA with a shorter distance between the coordination groups was introduced as an auxiliary unit to Ni-tPBA2. The asymmetric unit still contains the chelated nickel center by tPBA, and the two additional nickel ions were utilized as metal nodes connecting the pyridine to allow the structure to be extended in the two-dimensional (2D) layer by coordinate covalent bonds (Supplementary Fig. 4). Interestingly Ni-tPBA0~2 can be assembled into 3D supramolecular framework through the conjugation of benzimidazole groups and the N–H···O supramolecular forces between the pyridine and the water molecule. The formation of the 3D supramolecular framework with the chain and layer subunits was demonstrated in Supplementary Figs. 5–10.

Clearly, although Ni-tPBA0~2 exhibits unique 3D structures, the weak supramolecular interaction may be fragile during the violent OER process24. Therefore, the three-groups H3BTC was used as an auxiliary ligand to further assemble the high-dimensional framework. As expected, the H3BTC was bridge coordinated in μ2 mode similar to H2IPA. The additional carboxyl group acts as a lattice-filling unit rather than a linking group to provide site resistance, thus leading to a notable 3D crystalline coordination polymer framework for Ni-tPBA3 (Fig. 2).

Fig. 2: Crystal structure of Ni-tPBA3.
figure 2

a The crystallography asymmetric unit of Ni-tPBA3, with five kinds of nickel center and the corresponding coordination environment. b Staggered coordination 2D layer structure constructed by tPBA in Ni-tPBA3. c 1D chain constructed under BTC-assisted extension in Ni-tPBA3. The illustration are Cartesian coordinates, with the a, b, and c axes marked in red, green, and blue. d The sra topology (zeolite ABW) with 4-connect net of Ni-tPBA3. e Mortise and tenon structure in Ni-tPBA3 (The pink triangle represents the supramolecular stacking mode).

Full size image

The asymmetric unit of Ni-tPBA3 consists of five crystallographically independent nickel atoms. Similar to Ni-tPBA0~2, Ni1 and Ni2 in Ni-tPBA3 are chelated at the center of tPBA to enhance the overall structural framework stability. Two parallel pyridines of tPBA are sufficient to anchor the Ni3 and Ni4 centers in a head-to-head manner, forming a stable θ-ring tetranuclear cluster with deprotonated BTC anchoring in the ring center (Fig. 2a). Ni1 and Ni2 on adjacent ring clusters is connected by two BTC bridges with Ni5 coordination. Moreover, pyridine on tPBA, which is not involved in ring formation, fixed Ni5 vertically at the crystallographic interface in the head-to-head connection (Fig. 2b, c). Therefore, Ni5 acts as an important connection center, which enables the tetranuclear θ-ring to assemble into a stable 3D framework with sra topology structure (zeolite ABW, Fig. 2d)25,26. Also as designed, the presence of π–π stacking between each benzimidazole in the inter-θ-rings allows the tPBA molecules placed in an anti-helical symmetry mode on both sides of the ring, further reinforced as stable triangles in a mortise-and-tenon like structure (Fig. 2e and Supplementary Fig. 11). This makes it possible for Ni-tPBA3 to constitute a large conjugated system that facilitates electron transfer. The π–π co-supported coordination polymer structure may have superior structural stability compared to supramolecular structures.

In summary, the chelate pockets of Ni-tPBA0~3 are all filled with a nickel ion to limit the bulk rigidity of the framework, while the introduction of the pyridine foot enriches the structural diversity. Due to the multidentate coordination unit of tPBA, Ni-tPBA0~3 contains nickel metal centers with a variety environments. The nickel metal centers in Ni-tPBA0~3 exhibit typical six-coordinated octahedral configuration. The composition of the coordination environments originating from the κ4 chelate pocket of twisted T-shape tPBA, κ1 or κ2 carboxylate oxygens of bridging carboxylate anions, and coordinated water or OH– ions as potential OCSs, respectively. The bond lengths of Ni–OH2 in Ni-tPBAs are all around 2.05 Å, which suggests that the nickel centers bind these water molecules in a similar coordination force. The relevant crystallographic data as well as bond lengths and bond angles are placed in Supplementary Tables 1–2.

Surface-determined OER process prefers a large pore volume and highly accessible surface area for the full exposure of active sites. It is also evidenced that the free and coordinated solvents play a key role in stabilizing the 3D structure17. These renders it of great significance to decouple cavities/channels and filling solvent of materials to understand the intrinsic reactivity. Here, the pore structure of the framework was first simulated by Mercury and CrystalExplorer after solvent removal27,28. The results demonstrate that the solvent-accessible volumes in Ni-tPBA0~3 account for 30.9%, 23.4%, 39.4% and 30.5% of the pore-filled 3D framework, with nickel active sites uniformly exposed in the pores. The simulated pore structures and parameters of Ni-tPBA0~3 were shown in Supplementary Figs. 12–16 and Supplementary Table 3. Subsequently, the adsorption property of fresh Ni-tPBA3 was evaluated using CO2 adsorption with a smaller kinetic radius. The low adsorption amount demonstrates the compact filling of solvent water molecules in the framework structure (Supplementary Fig. 17). The distribution of water molecules in the pore space of the planar portion of Ni-tPBA3(100) obtained by single-crystal resolution is shown in the illustration. These results finely confirm that Ni-tPBA3 can accommodate sufficient water molecules, which is capable of facilitating the rapid alternate release of intermediates during the electrochemical process.

Notably, Ni-tPBA0 was slight soluble in organic solvents such as N,N-dimethylformamide (DMF) and ethanol (EtOH), while the introduction of the auxiliary ligand optimizes the solution stability of Ni-tPBAs, i.e., rendering Ni-tPBA1~3 insoluble in DMF and EtOH. Thermogravimetric analysis (TGA) of Ni-tPBA0~3 demonstrates that the removal of free and coordinated solvents occurs in the interval from 60 to 200 °C (Supplementary Fig. 18). The solvent mass ratio close to 10% in Ni-tPBA0~3 proves that the 3D structure contains a rich cavity unit, conducive to the sufficient exposure of the active sites. It demonstrates that the construction of the π–π co-supported 3D framework improves the thermodynamic stability (up to 342 °C) of the Ni-tPBA3. In addition, Ni-tPBA3 remains its primary crystallinity and electronic structure after one day of immersion in 1 M KOH (Supplementary Fig. 1, Supplementary Figs. 19–22 and Supplementary Note 1), responsible for the sustainable and efficient electrocatalysis process.

Study on the electrocatalytic performance

The initial measurement indicates the intrinsic electrochemical activity of Ni-tPBA3 comparable to ruthenium oxide (Supplementary Figs. 23–27, Supplementary Note 2). To shed more light on the advantages of the materials29, carbon-coupled Ni-tPBA0~3 (Ni-tPBA0~3/CP) were prepared (Supplementary Figs. 28–34, Supplementary Note 3). Their performances were further evaluated in a H-type three-electrode cell with 1 M KOH as the electrolyte (Supplementary Fig. 35). As expected, the optimal η10 overpotential of 248 mV was exhibited in Ni-tPBA3/CP (Fig. 3a). Ni-tPBA0~3/CP showing low Tafel slope of 58.3, 61.0, 53.3 and 49.7 mV dec–1, respectively, indicative of their fast reaction kinetics, especially for Ni-tPBA3/CP (Fig. 3b). This is further evidenced by the lowest overpotential of Ni-tPBA3/CP even at a large current density of 100 mA cm–2 (Supplementary Fig. 37a). The double layer capacitance (Cdl) estimated by CV curves in a non-faradaic region indicates a large electrochemical active surface area (ECSA) of Ni-tPBA0~3/CP (Fig. 3c).

Fig. 3: Electrocatalytic OER performance.
figure 3

a LSV polarization curves of Ni-tPBA0~3/CP at 5 mV/s with iR-correction (the series resistance is 2.4, 2.3, 2.4 and 2.4 Ω for Ni-tPBA0~3/CP, respectively). The inset is the illustration of the electron transfer and catalysis process. b Tafel plots, (c) Cdl values, and d electrochemical impedance spectra at an overpotential of 300 mV. e Stability test for Ni-tPBA3/CP in H-type cell with anion exchange membrane at the constant current density from 10~400 mA cm–2 (without iR-correction).

Full size image

The turnover frequency (TOF) values of the materials were evaluated by calculating the mass percentage of nickel with effective potential OCSs in the structure units. Ni-tPBA3 exhibited a highest turnover rate (0.0351 s–1) at the overpotential of 300 mV, demonstrated the rapid water oxidation efficiency and superior electrocatalytic OER activity (Supplementary Fig. 37b). The intrinsic catalytic activity of Ni-tPBA3 is further demonstrated by the ECSA normalized LSV curve (Supplementary Fig. 39a). The electrochemical impedance spectroscopy (EIS) further shows the low charge-transfer resistance of Ni-tPBA3/CP (1.91 Ω cm2) than that of Ni-tPBA0~2/CP (3.84–4.58 Ω cm2) at the overpotential of 300 mV (Fig. 3d and Supplementary Table 4). It may be attributed to that the π–π stacking driven by the conjugated ligand facilitates the electronic transfer. The tailored Ni-tPBA3 as OER electrocatalyst is superior to many non-complex, modified complex and pure complex-based OER materials in recent years (Supplementary Table 5). More importantly, Ni-tPBA3 has relatively high electrochemical activity and stability in comparison to other MOF-materials that contain labile solvent coordination (Supplementary Table 6). The electrochemical stability was demonstrated by comparing the LSV curves before and after 1000 CV cycles (Supplementary Fig. 40, Supplementary Note 4). In addition, Ni-tPBA3/CP can maintain a stable low overpotential for 200 h under a step current density 10–400 mA cm–2 (Fig. 3e). Besides, it shows a desirable four-electron pathway for oxygen evolution and a high selectivity for the OER process (Supplementary Figs. 41–44 and Supplementary Note 5). These results substantiate that Ni-tPBA3 serves as an efficient OER catalyst with improves the reactivity and electrochemical stability.

Theoretical and in situ Insights into the OER Mechanism

In situ Raman characterization was carried out for Ni-tPBA3 to gain more insights into the structure evolution during the OER process. We first collected the Raman curves at different potentials with a scan rate of 1 mV s–1. Besides, after polarization, the Raman signal was restored for 20 min of relaxation without voltage applied. For each case, the Raman signal was attenuated at high potentials without any shift in the peak position (Fig. 4a, b), for example, the C–H out of plane bending benzimidazole in tPBA (769 cm–1)30, and asymmetric stretches of the carboxylate anion (COO–) in BTC (1591 cm–1)31. The result indicates that Ni-tPBA3 differs from the traditional CCs undergoing drastic ligand dissolution/decomposition and reconstruction, attributed to the robust binding of the nickel center to the ligand molecules with high stability32. The reason for the transient peaks decay may be the disturbance of the feedback due to the rapid release of oxygen at high potentials. Another detail is that the peak of nitrate (totally symmetric N–O stretch 1021 cm–1)33 disappears gradually due to ion exchange during the OER of Ni-tPBA3. Besides, there are no new peak signals observed in the long-wave region (460~600 cm–1)34,35, indicating that no detectable nickel oxide, hydroxy nickel oxide and nickel hydroxide species were generated during the OER process (Supplementary Fig. 45). The repeated in situ measurement under polarization and rest-state also demonstrates well-remained and visible Raman signals of Ni-tPBA3 without shift or emergence of new peaks (Supplementary Fig. 46), further substantiating that the structure stability of Ni-tPBA3 during the OER process.

Fig. 4: Mechanistic insights into the OER process.
figure 4

a In situ Raman contour maps for Ni-tPBA3/CP accompanied by first LSV processes with applied potentials range from 1.1 to 1.8 V vs. RHE and b the corresponding in situ Raman spectra of Ni-tPBA3/CP at different representative voltages and open circuit potential (OCP). c Comparison of the first and 100th CV curves for Ni-tPBA0~3/CP without iR-correction. Insets show the proportion of nickel centers with different potential OCSs coordination modes in Ni-tPBA0~3. d The proposed OER reaction process for nickel centers with different potential OCSs.

Full size image

The catalysts after the reaction were analyzed and characterized in detail (Supplementary Figs. 47–57, Supplementary Note 6). The results show that Ni-tPBA3 retains the crystallinity and electronic structure to a great degree after OER. Moreover, X-ray absorption spectroscopy (XAS) characterization was conducted to gain more insights into the stability (Supplementary Fig. 57). The absorption edge shows a slightly shift toward high energies after OER (Supplementary Fig. 57a), indicative of a little increment in valence state36. As depicted by the extended X-ray absorption fine structure (EXAFS) spectra, Ni-tPBA3 pre-/post-OER showed a similar peak of Ni–N/O in the first shell region (Supplementary Fig. 57b). With the CIF of Ni-tPBA3 single crystal model, the best-fitting analysis showed that the distances from the coordinated solvent and organic ligand (Ni–N/O) remain less changed within 0.2 Å after OER, well consistent with bond lengths derived from SCXRD (Supplementary Fig. 57c–f and Supplementary Tables 8–9). A visual illustration was provided by the wavelet transform (WT) EXAFS analysis (Supplementary Fig. 57g), which further demonstrates a similar structural configuration of Ni centers with negligible change in Ni–N/O contribution.

To delve into the real OER process, the 1st and 100th cycle curves of CV activation were collected, and the number of nickel centers with different types of potential OCSs as a proportion of the total number of metal centers was analyzed (Fig. 4c). Considering the potential OCSs and the coordination environment of Ni-tPBA0~3, it is believed that the nickel centers may undergo five different reaction processes during the OER with the mechanism proposed as follows (Fig. 4d)37,38:

  1. (1)

    For the active nickel center with single potential OCS, a slow process of intermolecular oxygen coupling happens similar to adsorbate evolution mechanism after the generation of –Ni–OH reaction precursors (Path 1).

  2. (2)

    With additional potential OCSs at the neighbor site, the oxidation process occurs under the coupling of the two oxygen occupying the OCSs, which can quickly and directly form O–O bonds to accelerate the reaction process, i.e., the IOM (Path 2)39.

  3. (3)

    Both Path 1 & 2 require re-coordination to generate Intermediate 1 (–Ni–(OH)x, x = 1 for Path 1 and x = 2 for Path 2) for initiating the second catalytic cycle after deoxygenation. By increasing potential OCSs, additional coordinated water at the mirror position can directly participate in electron transfer with the –OO via the nickel active center (Intermediate 4 Cyc1 to 5Cyc1, Paths 3~5), accompanied by deprotonation to produce active –OH as an active intermediate for the next catalytic process. Such an intramolecular electron-transfer pathway facilitates the generation of –Ni–(OH)x intermediate without awaiting deoxygenation before re-coordination. This further breaks the limitations to enable the re-coordination of water molecules happening simultaneously with the deprotonation of active –OH in Intermediate 5, namely the fast kinetics from Intermediate 5Cyc1 to 2Cyc2 directly. It is believed that the synergistic IOM optimizes the kinetics of the OER process to a great extent. (Path 3 to 5).

In order to verify the effect of carboxylate anions on the nickel centers during OER, LSV curves of Ni-tPBA0&3 were further collected at different temperatures (T) (Fig. 5a, b). The activation energy (W) of Ni-tPBA0&3 was calculated by fitting the Arrhenius curves of the natural logarithm of current density (ln j) versus 1/T. It is demonstrated that Ni-tPBA3 exhibits a lower W compared to Ni-tPBA0 and the W difference (ΔW, 8.3 to 1.6 kcal mol–1) is linearly related to the change in overpotential (Fig. 5c), in good accordance with the comparison results of TOF value. As the activation energy determines the intrinsic property of the materials, these difference prove that Ni-tPBA0&3 possesses different intermediates and potential determining step (PDS)37. That is to say, the introduction of carboxylate anions modulates the intrinsic adsorption capacity of the nickel center toward the intermediates40,41.

Fig. 5: Experimental and theoretical investigations into the reaction free energy.
figure 5

LSV curves without iR-correction measured at different temperatures for a Ni-tPBA0 and b Ni-tPBA3. The insets show the fitted Arrhenius curves at different potentials. c Fitted curves of activation energies obtained at different potentials. d, e Free energy diagram and optimized intermediate structures for Ni-tPBA3 via Paths 1 and 2. f Free energy diagram and the optimized intermediate structures for Ni-tPBA3 via Path 4.

Full size image

To understand the role of potential OCSs in nickel coordination environments, the density functional theory (DFT) was employed to understand the plausibility of the hypothesized pathways. The Ni3 atom with 3-coordinated water in Ni-tPBA3 was applied as the catalytic model, and the free energy differences under Paths 1 and 2 as well as the typical adsorbate evolution mechanism (Path AEM) were calculated (Fig. 5d, e and Supplementary Fig. 58), respectively. The PDS was found to be the formation of the Intermediate 3 (–OOH). Compared to Path AEM (2.16 eV) and Path 1 involving a single water molecule (2.07 eV), the IOM between neighboring coordinated water produces –OOHintra with a smaller free energy of 1.65 eV (Path 2). It is therefore inferred that in the presence of neighboring coordinated water molecules, the nickel preferentially undergoes the intramolecular oxygen coupling. Notably, it is distinguished from the lattice oxygen mechanism, because the reaction of coordinated water is not constrained by the loss of lattice oxygen, and thus the materials may be endowed with enhanced electrochemical stability and kinetics.

Further, the free energy change for the catalysis process on Ni3 atoms was calculated, which contains three water molecules as potential OCSs (Fig. 5f). In the first catalytic cycle, the catalytic reaction follows the free energy change of Path 2. It is noted that with additional intramolecular water participating in the reaction, OCS alternation happens after electron transfer and deprotonation (Intermediate 4 Cyc1 to 5Cyc1, Path 4). Under the asymmetric catalytic environment with the carboxylate ligands (Supplementary Fig. 59), the occurrence of OCS alternation contributes to optimizing the free energy in the second catalytic cycle, especially for Intermediate 2Cyc2 to 3Cyc242,43. Therefore, the reaction pathway in the environment of multiple water molecule coordination may tend to a dynamic co-coupling process of multiple potential OCSs (Paths 3~5), contributing to the optimal reaction energy barrier and kinetic steps, and thereby the enhanced catalysis performances of Ni-tPBA3.

Discussion

In summary, we have tailored a series of well-defined and stable CC electrocatalysts Ni-tPBA0~3 under the coordination of bridge neutral nitrogen heterocyclic ligands and different aromatic carboxylic acids. SCXRD results demonstrate that Ni-tPBA0~2 feature the 3D coordination supramolecular structures, while Ni-tPBA3 show a strong 3D coordination polymerization framework with sra topology. The engineered CCs contain a variety of nickel centers with different coordination environments, which possess one to four potential OCSs through the steric effect. Thanks to the potential synergy among the multi-coordinated water molecules, Ni-tPBA3 coupled with carbon delivers a high TOF value (0.0351 s–1) at the overpotential of 300 mV and maintains stable operation for 200 h. This study provides design concept of metal-organic structures and highlights the potential of coordination regulation for efficient electrocatalysis.

Looking forward, many opportunities remain in designing potential OCSs of coordination complexes before widespread applications. On the one hand, the chelating characteristics of a ligand contribute to robust structure stability but consume some available active sites, which deserve more attempts and subtle regulation. The interaction between the central metal and the water-ligand also needs modulation to ensure optimal coordination competition and easy detachment of products. Moreover, it is pursued to achieve precise localization of the multiple metals to advance the OER properties of the water-coordinated complexes. Different small molecules can be applied to replace the water molecules as potential OCSs for modulating the structure assembly and the interaction between metal centers and intermediates.

Methods

Chemicals and materials

Nickel(II) nitrate hexahydrate (Ni(NO3)2·6H2O, >98%) was purchased from Xilong Scientific Co., Ltd. o-Phenylenediamine (98%), 4-chloromethyl pyridine hydrochloride (97%) and tetrabutylammonium bromide (TBAB, 99%) were purchased from Macklin. Nitrilotriacetic acid (≥98.5%), terephthalic acid (p-C6H4(COOH)2, TPA, ≥99%), isophthalic acid (m-C6H4(COOH)2, IPA, ≥99%) and trimesic Acid (C6H3(COOH)3, BTC, ≥98%) were purchased from Aladin. Acetone (≥99.5%), 1,2-ethandiol (EG, ≥99.5%), methanol (MeOH, ≥98.5%), ethanol (EtOH, ≥99.5%), N,N-dimethylformamide (DMF, ≥99.5%), N,N-dimethylacetamide (DMAc, 99.0%), dimethyl sulfoxide (DMSO, 97%), toluene (≥99.5%), potassium hydroxide (KOH, ≥85.0%) and sodium hydroxide (NaOH, ≥96.0%) were purchased from Sinopharm. RuO2 (99.95%) was purchased from Adamas. Nafion (5%) was purchased from Sigma-Aldrich. All the reagents were used as received without further purification. Carbon fiber paper (CP) was purchased from Toray Industries, Inc. The CP was cut into 0.5 cm × 1 cm and 2 cm × 3 cm for further use. Anion exchange membrane (A20) was purchased from Versogen. Pt/C/CP (1 mg cm–2) was purchased from the Sinero fuel cell store.

Synthesis of tri(2-benzimidazolylmethyl) amine (tBIA)

The synthesis of tBIA was conducted via a modified procedure according to the previous reported22. o-Phenylenediamine (32.4 g, 0.3 mol), nitrilotriacetic acid (19.1 g, 0.1 mol) and 1,2-ethandiol were mixed and refluxed at 160 °C for 24 h. The violet solution was poured into ice water, filtered, and washed several times with deionized water to obtain the pink powder tPBA without further purification. Yield: 95%. 1H NMR (DMSO–d6, 400 MHz): δ (ppm) = 12.46 (3H, s, –NH–), 7.57 (6H, t, Ph–H), 7.18 (6H, dt, Ph–H), 4.49 (6H, s, –CH2–).

Synthesis of tris((pyridine-4-ylmethyl)benzoimidazol-2-ylmethyl)amine (tPBA)

The synthesis of tPBA was prepared by an optimized method reported before44. tBIA (5.1 g, 12.5 mmol), tetrabutylammonium bromide (50 mg), toluene (20 ml), acetone (130 ml) and 50% NaOH aqueous (5 g NaOH in 5 ml H2O) were mixed and heated for about 30 min until the tBIA was dissolved. Then 4-chloromethyl pyridine hydrochloride (6.35 g, 38.7 mmol) was added and the resulting solution was refluxed for 5 h. The dark yellow solution was poured into ice water, filtered, and washed several times with deionized water to obtain the yellow powder tPBA, which did not require further purification. Yield: 92%. 1H NMR (DMSO–d6, 400 MHz): δ (ppm) = 8.21 (6H, m, Ph–H), 7.52 (3H, dt, Ph–H), 7.37 (3H, m, Ph–H), 7.22 (6H, m, Ph–H), 6.52 (6H, m, Ph–H), 5.31 (6H, s, –CH2–), 4.13 (6H, s, –CH2–).

Synthesis of Ni-tPBA0 crystals

Ni(NO3)2·6H2O (29.08 mg, 0.1 mmol), tPBA(68.08 mg, 0.1 mmol) and mixed solutions (DMF: H2O, 1: 4 mL) were encapsulated in a 10 mL vial. The sample was heated at 90 °C for 72 hours, and then cooled to room temperature. After several washes with DMF and EtOH, the Ni-tPBA0~3 crystals were obtained. {[Ni2(tNBP)2(NO3)4(H2O)3]·Solvent} (C84H78N20Ni2O3, without free solvent), cyan cubic rod crystals; yield: 43.8 mg, 45%, based on Ni.

Synthesis of Ni-tPBA1 crystals

Ni(NO3)2·6H2O (29.08 mg, 0.1 mmol), tPBA (34.04 mg, 0.05 mmol), TPA (16.61 mg, 0.1 mmol) and mixed solutions (DMF: H2O, 0.5: 5 mL) were encapsulated in a 10 mL vial. The sample was heated at 90 °C for 72 hours, and then cooled to room temperature. After several washes with DMF and EtOH, the Ni-tPBA0~3 crystals were obtained. {[Ni2(tNBP)2(TPA)(NO3)2(H2O)2]·Solvent} (C92H80N20Ni2O6, without free solvent), cyan bulk crystals; yield: 51.6 mg, 51%. based on Ni.

Synthesis of Ni-tPBA2 crystals

Ni(NO3)2·6H2O (29.08 mg, 0.1 mmol), tPBA(34.04 mg, 0.05 mmol), IPA(16.61 mg, 0.1 mmol) and mixed solutions (DMAc:H2O, 0.5: 3.5 mL) were encapsulated in a 10 mL vial. The sample was heated at 90 °C for 72 hours, and then cooled to room temperature. After several washes with DMAc and EtOH, the Ni-tPBA0~3 crystals were obtained. {[Ni4(tNBP)2(IPA)2(OH)4(H2O)2]·Solvent}n (C50H44N10Ni2O7, without free solvent), purple flaky crystals; yield: 26.3 mg, 50%. based on Ni.

Synthesis of Ni-tPBA3 crystals

Ni(NO3)2·6H2O (29.08 mg, 0.1 mmol), tPBA(34.04 mg, 0.05 mmol), BTC(14.71 mg, 0.07 mmol) and mixed solutions (DMF: H2O, 2: 3 mL) were encapsulated in a 10 mL vial. The sample was heated at 90 °C for 72 hours, and then cooled to room temperature. After several washes with DMF and EtOH, the Ni-tPBA0~3 crystals were obtained. {[Ni5(tNBP)2(BTA)3(OH)(H2O)7]·Solvent}n (C111H96N20Ni5O26, without free solvent), cyanic tabular Crystals; yield: 36.4 mg, 72%. based on Ni.

Fabrication of Ni-tPBA0~3/CP with Nafion, RuO2/CP and Ni-tPBA3/GC electrodes

The synthesis of Ni-tPBA0~3/CP with Nafion electrodes was conducted via a modified procedure according to the previous reported45,46. To prepare the catalyst ink, 5 mg of completely ground Ni-tPBA0~3 was dispersed into a mixed solution of water, EtOH, and 0.5% Nafion solution (0.25: 0.7: 0.05 mL), followed by ultrasonication for 1 h. The mixed catalyst ink is evenly dripped onto 1 × 0.5 cm2 carbon paper and dried at room temperature and with a load of about 0.2 mg cm–2. The RuO2/CP electrode was fabricated by replacing only the catalyst with 5 mg RuO2, and the Ni-tPBA3/GC was fabricated by replacing only the CP with a glassy carbon electrode (Φ = 5 mm).

Synthesis of Ni-tPBA0~3/CP electrodes

Adding Ni(NO3)2·6H2O and CP (2 × 3 cm2) to H2O, tPBA and aromatic carboxylic acids to organic solvent gives liquids a and b. After heating liquids a and b to 90 °C until dissolution, liquid b was added into liquid a to initiate further reaction at 90 °C for a certain time, and then cool down to room temperature naturally. The yielding Ni-tPBA0 was cleaned with ultra-pure water, while Ni-tPBA1~3 were cleaned with DMF and ethanol to obtain Ni-tPBA0~3/CP. The active mass loading is about 5 mg cm–2. The specific reaction conditions are as follows:

Ni-tPBA0/CP: Ni(NO3)2·6H2O (116.31 mg, 0.4 mmol), tPBA(272.33 mg, 0.4 mmol), DMF:H2O (0.5: 19 mL), and the reaction time was 20 h.

Ni-tPBA1/CP: Ni(NO3)2·6H2O (116.31 mg, 0.4 mmol), tPBA (136.16 mg, 0.2 mmol), TPA (66.45 mg, 0.4 mmol), DMF:H2O (1.5: 18 mL), and the reaction time was 15 h.

Ni-tPBA2/CP: Ni(NO3)2·6H2O (116.31 mg, 0.4 mmol), tPBA(136.16 mg, 0.2 mmol), IPA(66.45 mg, 0.4 mmol), DMAc:H2O (2: 17 mL), and the reaction time was 17 h.

Ni-tPBA3/CP: Ni(NO3)2·6H2O (116.31 mg, 0.4 mmol), tPBA(136.16 mg, 0.2 mmol), BTC(63.04 mg, 0.3 mmol), DMF: H2O (6: 12 mL), and the reaction time was 12 h.

Characterization

The as-made materials were characterized by field emission SEM (FEI Verios G4), PXRD (Bruker D8 Advance, Cu–Kα, λ = 0.15406 nm), XPS (Kratos AXIS Ultra DLD), FT–IR (Bruker Tensor 27, KBr), diffuse reflectance infrared Fourier transform spectroscopy (iS20 FTIR), TGA (Mettler Toledo TGA 1), HRTEM (FEI Talos F200X, equipped with EDX spectroscopy). In situ Raman measurements of Ni-tPBA3/CP were performed from 1.105–1.805 V vs. RHE at a rate of 1 mV s–1 with 532 nm excitation from an argon ion laser. X-ray absorption spectroscopy (XAS) data was collected by the RapidXAFS 2 M (Anhui Absorption Spectroscopy Analysis Instrument Co., Ltd.) with transmission (or fluorescence) mode at 10 kV and 20 mA, while the Si (440) spherically bent crystal analyzer (radius of curvature, 500 mm) was used for Ni.

Single-crystal X-ray crystallography

A suitable single crystal was mounted on a glass fiber, and the intensity data was collected on a Bruker Apex II area detector with graphite-monochromatized MoKa radiation (λ = 0.71073 Å) at T = 193 K. Data reduction and cell refinement were performed using the programs Smart and Saint47,48. The absorption corrections were carried out by an empirical method49. The crystal structure was analyzed by Direct Methods and refined by full-matrix least-squares against F2 using the program Olex250,51. Hydrogen atoms were located geometrically and treated with the riding model.

Electrochemical evaluations

Electrochemical measurements were conducted on a CHI 760E electrochemical workstation using a three-electrode system. Ni-tPBA0~3/CP and Ni-tPBA0~3/CP with Nafion were served as the working electrodes. Before the test, Ni-tPBA0~3/CP were cut by the surgical blade. The test area is 1 × 0.5 cm2. The carbon rod electrode and Hg/HgO were used as the counter and reference electrodes, respectively. The OER test was carried out in a H-type cell. A fresh 1 M KOH solution with pH = 13.70 ± 0.04 was prepared using a 500 mL volumetric bottle for electrochemical testing. The electrode potentials were calibrated to a reversible hydrogen electrode (RHE) via the equation: ERHE = EHg/HgO + 0.098 V + 0.059 pH. The catalysts were activated by 100 cyclic voltammetry (CV) cycles in 1.16 to 1.66 V vs. RHE at a scan rate of 50 mV s–1 before the OER test.

Linear sweep voltammetry (LSV) was recorded at a scan rate of 5 mV s–1. iR drop (95%) was corrected using the uncompensated resistance measured at the open circuit potential. The long-term stability tests were measured by chronopotentiometry at the step current density of 10–400 mA cm–2 with anion exchange membrane (1.5 × 1.5 cm2) and Pt/C/CP counter electrode (0.5 × 1 cm2). Before the test, the membrane was soaked in 1 M KOH for 1 h. The double layer capacitance (Cdl) values of the materials were measured by cyclic voltammetry in the non-faraday region at 0.15–0.25 V vs. Hg/HgO at a scan rate of 20, 40, 60, 80, and 100 mV s–1, respectively. The Cdl can be estimated by fitting the linear relationship between half of the current density difference at the intermediate voltage (0.5ΔJ = (Joxidation – Jreduction)) and the scan rate. The roughness factor or electrochemical active surface area (ECSA) of the catalysts can be calculated according to52:

$${{\mbox{ECSA}}}=\frac{C{{\mbox{dl}}}}{C{{\mbox{s}}}}$$
(1)

Here, Cs is the specific capacitance of the CP electrode with a value of 0.04 mF cm–2 in an alkaline system53. Therefore, the current density normalized by ECSA can be calculated by the following formula:

$$J _{{\mbox{ECSA}}}=\frac{J}{{\mbox{ECSA}}}=\frac{I}{A _{{\mbox{geom}}} \times {\mbox{ECSA}}}$$
(2)

Electrochemical impedance spectroscopy (EIS) was measured at a potential of 1.55 V vs. RHE with frequency from 0.1 to 105 Hz.

Turnover frequency (TOF) is calculated via the following equation:

$${{\mbox{TOF}}}=\frac{J\times A}{4\times n\times F}$$
(3)

Here, J is the current density at an overpotential of 300 mV during the LSV measurement in 1 M KOH solution, A stands for the area of the electrode (0.5 cm2), F is the Faradic constant (96485 C mol–1), n represents the quantity of active sites that can be obtained by single crystal structure analysis.

$$n=\frac{ {x}_{{{\mbox{active}}} \, {{\mbox{site}}}} \times m_{{\mbox{catalyst}}}} {M_{{\mbox{catalyst}}}}$$
(4)

Here, xactive site is the number of effective active sites in Ni-tPBA0~3, mcatalyst is the mass of the catalyst loaded on the electrode, and Mcatalyst is the analytical molar mass of Ni-tPBA0~3. In this paper, the number of active sites of Ni-tPBA0~3 is 2, 2, 2 and 4, respectively.

The activation energy is calculated by measuring LSV curves at different temperatures and fitting Arrhenius curves.

$${E}_{{{{\rm{a}}}}}=-{{{\rm{R}}}}\frac{{{{\rm{d}}}}\,({{{\rm{I}}}}{{{\rm{n}}}}\,J)}{{{{\rm{d}}}}\,({T}^{-1})}$$
(5)

Here, E is the calculated activation energy; J is the current density (mA cm–2) at different temperature T (K).

Rotating ring disk electrode test

Typical RRDE were used to probe the electrochemical behavior of Ni-tPBA3 (Pine Research Instrumentation), which is comprised of glassy carbon electrode (0.2475 cm2) and platinum ring electrode (0.1866 cm2). The catalyst ink was prepared by diffusing 5 mg of Ni-tPBA3 in 1 mL of a mixed solution of isopropanol (0.7 mL), water (0.25 mL), and Nafion (5 wt%, 0.05 mL). After that, 10 μL of ink was evenly applied to the glassy carbon electrode to dry naturally. The test was performed in 0.1 M KOH with a rotational speed set from 900 to 1600 rpm and a scan rate of 10 mV s–1. The ring electrode potential was set to 1.3 V vs. RHE to detect the hydrogen peroxide produced (proof that the OER product was free of hydrogen peroxide). The ring electrode was set at 0.4 V vs. RHE to detect the Faraday efficiency of O2, with the disk electrode was set at 1.6 V to oxidize water. The electron transport number (N) was calculated from the response current54:

$$N=4\times \frac{I {{\mbox{d}}}}{I {{\mbox{d}}}+I {{\mbox{r}}} /C}$$
(6)

The Faradaic efficiency (FE) was calculated by the following11:

$${{\mbox{FE}}}=\frac{I {\mbox{r}}}{I {{\mbox{d}}} \times C}$$
(7)

Here, Id is disk current, Ir is ring current, and C is the current collection efficiency of RRDE, which was determined to be 0.37.

Computational methods

We have employed the first-principles55,56 to perform density functional theory (DFT) calculations within the generalized gradient approximation (GGA) using the Perdew-Burke-Ernzerhof (PBE)57 formulation. We have chosen the projected augmented wave (PAW) potentials58,59 to describe the ionic cores and take valence electrons into account using a plane wave basis set with a kinetic energy cutoff of 520 eV. The GGA + U method was adopted in our calculations. The value of the effective Hubbard U was set as 4.669 eV for Ni. Partial occupancies of the Kohn−Sham orbitals were allowed using the Gaussian smearing method with a width of 0.1 eV. The electronic energy was considered self-consistent when the energy change was smaller than 10−5 eV. A geometry optimization was considered convergent when the energy change was smaller than 0.05 eV Å−1. The Brillouin zone integration is performed using 1×1×1 Monkhorst-Pack k-point sampling for a structure. Finally, the adsorption energies (Eads) were calculated as Eads = Ead/sub – Ead – Esub, where Ead/sub, Ead, and Esub are the total energies of the optimized adsorbate/substrate system, the adsorbate in the structure, and the clean substrate, respectively. The free energy was calculated using the equation:

$$G={E}_{{{{\rm{ads}}}}}+{{{\rm{ZPE}}}}-{TS}$$
(8)

Here, G, Eads, ZPE and TS are the free energy, total energy from DFT calculations, zero-point energy and entropic contributions, respectively, where T is set to 300 K.