Open AccessCCS ChemistryRESEARCH ARTICLE11 Mar 2022Trinuclear Nickel Catalyst for Water Oxidation: Intramolecular Proton-Coupled Electron Transfer Triggered Trimetallic Cooperative O–O Bond Formation Qi-Fa Chen, Yao Xiao, Rong-Zhen Liao and Ming-Tian Zhang Qi-Fa Chen Center of Basic Molecular Science (CBMS), Department of Chemistry, Tsinghua University, Beijing 100084 Google Scholar More articles by this author , Yao Xiao Center of Basic Molecular Science (CBMS), Department of Chemistry, Tsinghua University, Beijing 100084 Google Scholar More articles by this author , Rong-Zhen Liao *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] Key Laboratory of Material Chemistry for Energy Conversion and Storage, Ministry of Education, School of Chemistry and Chemical Engineering, Huazhong University of Science and Technology, Wuhan 430074 Google Scholar More articles by this author and Ming-Tian Zhang *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] Center of Basic Molecular Science (CBMS), Department of Chemistry, Tsinghua University, Beijing 100084 Google Scholar More articles by this author https://doi.org/10.31635/ccschem.022.202101668 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesDownload CitationsTrack Citations ShareFacebookTwitterLinked InEmail We report a molecular trinuclear nickel (TNC-Ni) catalyst for water oxidation that exhibited high catalytic performance and stability under neutral conditions (pH 7). Electrochemical studies disclosed that cooperation among the three nickel sites plays a vital role in both charge accumulation and O–O bond formation. This TNC-Ni catalyst could accomplish 4e− oxidation of water by involving all three nickel sites and the O–O bond formation was triggered by a charge distribution process from 5 to 5dp via proton-coupled electron transfer. Download figure Download PowerPoint Introduction Water oxidation (2 H2O → O2 + 4 e− + 4 H+) involving multiple protons and electrons transfer is the bottleneck for solar-powered water splitting.1–3 Molecular water oxidation catalysts (WOCs) based on transition metals have attracted extensive attention since the first “blue dimer” catalyst was reported by Meyer and co-workers in 1982.4 Recently, base-metal complexes for water oxidation have undergone rapid development.5–7 Despite the limitation of nickel-based molecular WOCs, compared with others, including Ru,4,8–14 Ir,15,16 Mn,17–21 Fe,22–26 Co,27–29 and Cu,30–37 nickel has been applied extensively in heterogeneous catalytic water oxidation.38–41 Boettcher et al.42 reported that pure NiOx actually exhibited very low activity in electrocatalytic water oxidation unless tiny amounts of Fe3+ were present, indicating that nickel catalysts have a distinctive behavior. Previously reported mononuclear Ni-based catalysts43–46 also disclosed their unique catalytic behaviors in terms of catalytic performance and stability. These findings stimulated us to explore the mystery of Ni-catalyzed water oxidation. According to the heterogeneous NiOx systems47,48 and Ni–O2 chemistry,49–52 multinuclear Ni-complexes are promising candidates for water oxidation because they could avoid charge accumulation only in a single nickel site and promote the formation of O–O bonds via polymetallic cooperation. Previous studies have shown that the oxygen evolution center (OEC) in photosystem II (PSII) is a CaMn4O5 cluster (Figure 1a) that includes a cubic CaMn3O4 and a hanging Mn ion.53 The charge accumulation in the CaMn4O5 cluster could be dispersed on four Mn sites to achieve a formal high oxidation state54 that facilitates the O–O bond formation.55–57 Biomimetic catalysts based on Mn,19,20,58 Fe,25 Co,59 and Cu37,60 have further confirmed the feasibility of polymetallic catalysis. Herein, we developed a molecular trinuclear nickel (TNC-Ni) complex, referring to the active site structure of nickel oxyhydroxides and layered nickelates (Figure 1b). This Ni3 core (Figure 1c) displayed high activity toward electrocatalytic water oxidation at pH 7, and all three nickel sites were involved in the charge accumulation and O–O bond formation. The catalytic performance of this Ni3 complex was 27 times faster than the corresponding binuclear Ni2 complex. Figure 1 | (a) The structure of CaMn4O5 cluster. (b) The fragment of a general nickelate structure: red, oxygen; purple, nickel. (c) The structure of the Ni3O4 investigated in this work: blue, nitrogen. (d) Crystal structure of the TNC-Ni catalyst. Selected bond distances (Å) and angles (deg): Ni1–O1 1.990(7), Ni1–O2 2.074(8), Ni1–O3 2.129(9), Ni1–N1 2.096(6), Ni1–N3 2.142(1), Ni2–O3 1.855(1), Ni2–O4 1.858(8), Ni2–N4 1.869(1), Ni2–N5 1.896(1), Ni3–O1 2.032(7), Ni3–O2 2.046(8), Ni3–O4 2.100(9), Ni3–N6 2.10(1), Ni3–N8 2.118(6). Ni1–O1–Ni3 97.0(3), Ni1–O2–Ni3 93.9(3), Ni1–O3–Ni2 128.2(5), Ni2–O4–Ni3 129.6(4), O3–Ni2–O4 95.1(4), O1–Ni1–O2 84.4(3), O1–Ni3–O2 84.1(3). Download figure Download PowerPoint Experimental Methods Instrumentation Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker AVANCE III HD 400 spectrometer (Shanghai, China) at 1H = 400 MHz and 13C = 100 MHz. Electrospray ionization high-resolution mass spectrometry (ESI-HRMS) spectra were recorded on a liquid chromatography ion trap/time-of-flight mass spectrometry (LCMS-IT/TOF, Shimadzu, Japan) and Thermo scientific ultimate 3000+ system (Shanghai, China). Field emission scanning electron microscopy (FESEM) images were collected using the Hitachi SU-8010 instrument (Hitachi Ltd., Tokyo, Japan). Dynamic light scattering (DLS) was measured using the SZ-100-Z Nanoparticle Analysis System (HORIBA Scientific, Beijing, China). The X-ray photoelectron spectroscopy (XPS) data were collected on ESCALAB 250Xi (Thermo Fisher, Shanghai, China) photoelectron spectrometer. UV–vis spectra were recorded in phosphate-buffered saline (PBS; 0.1 M, pH 7) using Agilent Cary 8454 or Agilent Cary 60 UV–vis spectrometer (Agilent Technologies Co. Ltd., Beijing, China). Electrochemical experiments All the electrochemical experiments were investigated using CHI-630 electrochemical workstation. Boron-doped diamond (BDD; 0.07 cm2) was used as a working electrode (WE). Pt wire and saturated calomel electrode (SCE; Hg/HgCl2, saturated KCl) were used as a counter electrode (CE) and reference electrode (RE), respectively. The polished BDD electrode was cleaned by multiple cycles of the cyclic voltammetry (CV; 0–2 V vs SCE) in blank PBS (0.1 M, pH 7). Tetrabutylammonium hexafluorophosphate (TBA·PF6) was used as a supporting electrolyte for nonaqueous electrochemical measurements. Oxygen evolution The evolved oxygen was detected by a calibrated Ocean Optics FOXY probe during controlled potential electrolysis (CPE). Fluorine-doped tin oxide (FTO; 1 cm2) electrode was used as WE for bulky electrolysis. After the CPE process, the used FTO electrode was kept for further investigation. Synthesis and structure characterization of catalyst Chan’s trinuclear ligand (H2ahp) was used in this work.61,62 Modified synthetic procedures of the ligand and preparation of [(ahp2−)NiII3(μ-OH)2](CF3SO3)2 are summarized in the Supporting Information Schemes S1–S2 and Figures S1–S8. The reaction of 2 equiv Ni(CF3SO3)2 with H2ahp, in the presence of 4 equiv NaOH as a base in CH3OH/H2O (1:1) solution under ambient temperature, induced an instantaneous color change from yellow to pink. This generated tri-nickel complex was recrystallized from CH3OH-diethyl ether. The X-ray crystallography structure of the [(ahp2−)NiII3(μ-OH)2]2+ (Figure 1d) showed that Ni(2) ion had a distorted square-planar N2O2 coordination geometry, Ni(1) and Ni(3) assumed a distorted octahedral N3O3 coordination geometry. This complex was further characterized by XPS and ESI-HRMS, respectively. The ESI-HRMS peak ( Supporting Information Figure S5) at m/z = 267.0544 was assigned to [(ahp2−)NiII3(μ-OH)]3+, [z = 3, m/z = 267.0544 (cal.)], consistent with a complex consisting of three nickel ions. In addition, the reaction between 3 equiv Ni(CF3SO3)2 and 1 equiv H2ahp in the presence of 4 equiv NaOH in CH3OH/H2O (1:1) solution could form a Ni3 complex with μ-Cl bridge [(ahp2−)NiII3(μ-OH)(μ-Cl)](CF3SO3)2], Supporting Information Figure S9), if recrystallizing from acetone-diethyl ether in the presence of 1 equiv KCl. There were no apparent UV–vis absorption differences between [ahp2−]NiII3(μ-OH)(μ-Cl)](CF3SO3)2 and [ahp2−]NiII3(μ-OH)2](CF3SO3)2 in PBS, indicating that [ahp2−]NiII3(μ-OH)(μ-Cl)](CF3SO3)2 could quickly convert to [ahp2−]NiII3(μ-OH)2](CF3SO3)2 in aqueous solution ( Supporting Information Figures S11a and S11b). Results and Discussion Cyclic voltammetry experiments First, the electrochemical behavior of this TNC-Ni catalyst ([(ahp2−)NiII3(μ-OH)2]2+) was studied in anhydrous and Ar-saturated acetonitrile (MeCN). Figure 2a (black line) shows that the CV of TNC-Ni using BDD as the WE had two oxidation waves at 0.97 and 1.35 V (vs Fc/Fc+), respectively. Differential pulse voltammetry (DPV) displayed three oxidation peaks at 0.67, 0.96, and 1.32 V (vs Fc/Fc+) (Figure 2a, inset). These waves were attributed to the continuous oxidation of all three nickel ions. Upon adding 10% H2O into the MeCN solution, the current (blue line in Figure 2a) in the MeCN/H2O system was significantly enhanced after the first oxidation peak. This current enhancement is consistent with an electrocatalytic water oxidation process. Figure 2 | (a) CVs of TNC-Ni ([(ahp2−)NiII3(μ-OH)2]2+, 1 mM) in anhydrous MeCN with tetrabutylammonium hexafluorophosphate (TBA·PF6; (0.1 M) as the supporting electrolyte. The scan rate was 100 mV s−1. WE: BDD electrode, CE: Pt electrode, RE: 0.1 M AgNO3/Ag. (Inset: DPV of TNC-Ni (1 mM) in anhydrous MeCN with TBA·PF6 (0.1 M) as the supporting electrolyte.) (b) CVs in PBS (0.1 M, pH 7) without (black line) and with (blue line) TNC-Ni (1 mM). The scan rate was 100 mV s−1. WE: BDD electrode, CE: Pt electrode, RE: Saturated calomel electrode (SCE). (Inset: DPV of TNC-Ni (1 mM) in 0.1 M PBS, pH 7.) Download figure Download PowerPoint The typical CV graph of TNC-Ni in PBS (0.1 M, pH 7) shown in Figure 2b displays two oxidation waves at Ea1 = 1.16 V and Ea2 = 1.40 V versus the normal hydrogen electrode (NHE), respectively. The onset overpotential of TNC-Ni (1.24 V) is ∼340 mV was lower than that with Ni(CF3SO3)2 salt (1.58 V, Supporting Information Figure S14b). The catalytic performance in this tri-nickel system caused by nickel ion was also ruled out, based on the comparison of electrochemical behavior between the TNC-Ni complex and the simple nickel salt. The typical electrochemical behavior of 3 mM Ni(CF3SO3)2 in PBS (0.1 M, pH 7) had a cross-current feature in CV scan and a prominent reduction peak at 1.19 V, corresponding to the reduction of NiOx ( Supporting Information Figure S14b). The normalized current (i/ν1/2) of the second wave increased while the scan rate decreased ( Supporting Information Figure S11c), indicating that the first wave was a diffusion-controlled redox process and the second wave was a catalytic process. The catalytic process could be identified as catalytic water oxidation according to the following facts: First, bubbles were generated on the BDD electrode surface when the CV scanning potential was above 1.40 V versus NHE. Second, oxygen evolution was further confirmed by the CPE experiment (Figure 3a). O2 bubbles appeared on the FTO electrode surface ( Supporting Information Figure S11d), and the evolved O2 was detected with a calibrated Ocean Optics FOXY probe. The generated O2 was negligible in a blank solution without TNC-Ni over time (black line in Figure 3b). In contrast, the dissolved O2 of the solution containing 0.5 mM TNC-Ni increased from 47 to 214 μM during electrolysis (blue line in Figure 3b). Meanwhile, the electrolysis current was maintained at 0.6 mA/cm2 during long-term electrolysis at a TON value of 13, with a Faradaic efficiency of 93 ± 2%, indicating that TNC-Ni sustained acceptable stability during the catalytic process. Figure 3 | (a) CPEs at 1.54 V in PBS (0.1 M, pH 7) without (green line) and with (blue line) TNC-Ni ([(ahp2−)NiII3(μ-OH)2]2+), 0.5 mM, 3 mL) using FTO as working electrode (WE). The black line represents the CPE curve of the used FTO after 1 h CPE in TNC-Ni solution, rinsed with deionized water, and then moved to the PBS without TNC-Ni. WE: 1 cm2 FTO, CE: Pt electrode, RE: SCE. (b) The O2 evolution during the CPE without (blank line) and with (blue line) TNC-Ni (0.5 mM, 10 mL). Download figure Download PowerPoint The confirmation of homogeneous catalysis It was challenging to identify homogeneous or heterogeneous catalysis on a molecular complex for water oxidation. NiOx is easily generated via molecular precatalysts decomposition; thus, responsible for the catalytic performance.63–65 The CPE current shown in Figure 3a increased from 0.1 to 0.6 mA/cm2 for the first 20 min, which indicated that new catalytic-active species (Ni oxide or molecular active intermediate) were formed.43,44,66–73 We carefully examined the stability of our TNC-Ni complex in PBS (0.1 M, pH 7.0) by a series of controlled experiments ( Supporting Information Figures S12–S20). The multiple CVs with FTO electrode ( Supporting Information Figure S12a) and BDD electrode ( Supporting Information Figure S12b) showed that the catalytic currents decreased over the increased cycles, and no significant change of the CV curve was noted. This agreed with a homogeneous electrochemical behavior and different from the typical electrochemical behavior of NiOx64,65 formation by molecular catalyst decomposition. The FTO and BDD electrodes that have been used for multiple CVs experiment with TNC-Ni (1 mM) did not show catalytic performance in blank solution without catalyst (the blue line in Supporting Information Figure S12), which indicated that there was no active colloidal species formation on the electrode surface in the catalytic process. This was further confirmed by electrode surface analysis. The used FTO electrode was analyzed by UV–vis spectrum after 1-h CPE at 1.54 V with 0.5 mM catalyst in PBS (0.1 M, pH 7). There was no additional absorption observed compared with the fresh electrode, and the absorption spectrum of the solution when the CPE was kept constant ( Supporting Information Figure S13), indicating that the solution was stable and no colloidal species were absorbed on the electrode surface during the electrolysis. The FTO electrode after electrolysis with TNC-Ni showed negligible current in blank solution without catalyst during CPE (Figure 3a) and CV tests ( Supporting Information Figure S14a), which further supported the absence of colloidal species formation and TNC-Ni as a homogeneous catalyst. Additionally, SEM, XPS, and DLS analyses on the electrode after electrolysis further indicate that no nickel oxide formation occurred during long-term electrolysis ( Supporting Information Figures S15–S17). The catalytic performance of nickel oxide attracted our attention, so CPE with Ni(CF3SO3)2 as the precursor was applied as a comparison. After electrolysis, the FTO electrode with Ni(CF3SO3)2 exhibited a high current (∼0.9 mA/cm2) in blank solution, which implied that some active nickel oxide species were generated on the FTO surface ( Supporting Information Figure S18). However, the CPE current with Ni(CF3SO3)2 showed an increased period, similar to TNC-Ni. To further check whether nickel oxide would generate from TNC-Ni on FTO surface, multiple cycles of CPE between TNC-Ni and Ni(CF3SO3)2 were conducted: The FTO electrode was taken out from the solution during each CPE (600 s) at 1.54 V, then rinsed with deionized water and returned to the previous solution for next CPE. For the first CPE with Ni(CF3SO3)2 as the catalyst, the current increased with time, while the currents of the second and third cycles showed a large current at the beginning and kept constant for some time, which differed from the behavior of the first cycle ( Supporting Information Figure S19). These results indicated that the active nickel oxide species could catalyze water oxidation and hardly remove the FTO. However, the current with TNC-Ni increased from 0.1 mA/cm2 in each cycle ( Supporting Information Figure S20), implying that no nickel oxide species were deposited on the electrode and that the active species remained the molecular catalyst. Therefore, this was a homogeneous catalytic process, and the TNC-Ni might be converted to a new homogeneous active species rather than NiOx under the catalytic conditions. E-pH diagram and kinetic analysis The electrochemical behavior determined by DPV (Figure 2b, inset) displayed three evident oxidation waves at 1.05, 1.35, and 1.84 V, indicating the successive oxidation of three NiII ions, respectively. Figure 4a shows the E-pH relationship in detail. The first oxidation wave at ∼1.16 V was pH-dependent with a slope of ∼−50 mV/pH, corresponding to a proton-coupled electron transfer (PCET) process, NiII3(μ-OH)(μ-OH2) → NiII2NiIII(μ-OH)2 + e− + H+. The second oxidation wave at ∼1.40 V was pH-independent, which was a single electron transfer process, NiII2NiIII(μ-OH)2 → NiIINiIII2(μ-OH)2 + e−. The third oxidation wave at ∼1.84 V was pH-dependent with a slope of −31 mV/pH, indicating a PCET process involving 1H+/2e− transfer process, NiIINiIII2(μ-OH)2 → NiIII2NiIV(μ-OH)(μ-O) + 2e− + H+. The NiIII2NiIV(μ-OH)(μ-O) should be the in situ active species toward water oxidation, according to the oxygen evolution experiments discussed above. Figure 4 | (a) E-pH diagram of TNC-Ni (1 mM) in PBS (0.1 M, pH 6.57–9.39) (the potentials were determined by differential pulse voltammetry (DPV) method). (b) CVs of TNC-Ni complex in PBS (0.1 M, pH 7) with different catalyst concentrations. The scan rate was 100 mV s−1. WE: BDD electrode, CE: Pt electrode, RE: SCE. Inset: Plot of the diffusion currents (id) at ∼1.16 V, the catalytic currents (icat) at ∼1.40 V, and the catalytic currents (icat) at ∼1.84 V vs catalyst concentration ([TNC-Ni]), respectively. Download figure Download PowerPoint To elucidate the catalytic kinetics for this TNC-Ni catalyst, we further explored the relationship between the catalytic current and the different catalyst concentrations (Figure 4b). The catalytic currents in CV varied linearly with catalyst concentrations (Figure 4b, inset), which indicated that the catalytic process only involved a single TNC-Ni molecule rather than a dimer species. Accordingly, the relationship between the catalytic current icat and catalyst concentrations should obey the eq. 1, in which ncat (=4) is the number of the transferred electrons during the catalytic cycle, F (=96485.3 C mol−1) is the Faraday constant, A is the surface area of the WE (0.07 cm2 for BDD electrode used in this work), [Cat.] is the concentration of the TNC-Ni catalyst, Dcat is the diffusion coefficient of the catalyst, and kcat is the catalytic rate constant of the catalyst. i cat = n cat F A [ Cat . ] D cat 1 2 k cat 1 2 (1) The normalized CVs (i/ν1/2) versus scan rates ν showed that the first irreversible oxidation wave at ∼1.16 V was a diffusion-controlled wave ( Supporting Information Figure S11c), which is in accordance with the Randles–Sevcik eq. 2,74 where α (=0.5) is the transfer coefficient of the catalyst, nd (=1) is the number of transferred electrons in this diffusion process, R is the universal gas constant, and T is the absolute temperature. i d = 0.496 α 1 2 n d F A [ Cat . ] ( F v n d D cat / RT ) 1 2 (2) An equation containing icat and id is obtained by dividing eq. 1 by eq. 2. The kcat for catalytic water oxidation could be calculated using the following equation: i cat i d = 0.359 n cat n d 3 / 2 k cat / α ν (3) The normalized CVs (i/ν1/2) show that the normalized currents increased with the decreasing scan rates, which indicates a catalytic process. By plotting the icat/id versus 1/ν1/2, kcat was calculated as 0.54 s−1 by eq. 3 (Figure 5a). The involvement of three nickel sites in the water oxidation process was validated by comparing the performance between Ni3 complex and Ni2 complex. For this purpose, we used the same ligand to assemble the Ni2 complex characterized by HRMS and elemental analysis (the synthetic details and characterization are listed in the Supporting Information Figure S10). The electrochemical properties of the Ni2 complex were carried out under the same conditions as the Ni3 complex (Figure 5b). The Ni2 complex showed much lower current intensity, and the normalized CV current (i/ν1/2) was weakly dependent on the scan rate, indicating that the Ni2 complex was not active toward water oxidation (kcat = 0.02 s−1 shown in Supporting Information Figures S21–S23). This finding disclosed a key role played by the Ni3 core in the TNC-Ni triad in the O–O bond formation. Figure 5 | (a) Plot of the catalytic currents (at 1.84 V) icat/id vs ν−1/2 for TNC-Ni complex. (b) The CV graphs of Ni3 (1 mM, TNC-Ni) and Ni2 (1 mM) complex in PBS (0.1 M, pH 7), respectively. Download figure Download PowerPoint Density function theory calculations Density functional theory calculations at the B3LYP-D3 level were performed to elucidate the reaction mechanism of water oxidation catalyzed by this TNC-Ni complex (The computational details were listed in Supporting Information Figures S25–S33 and Tables S1–S2). The calculations showed that a water molecule could coordinate to both Ni1 and Ni3 of this Ni3 complex to form 1 (total charge of +3, Figure 6), and this water-binding process was exergonic by 6.8 kcal/mol. The pKa of 1 was calculated to be 13.0, suggesting that 1 is the most stable form at pH 7. The spin densities on Ni1 and Ni3 in 1 were calculated to be 1.69, while the spin density on Ni2 was 0.01 ( Supporting Information Figure S25), suggesting the presence of two octahedral-coordinated high spin NiII ions and a square-planar low spin NiII ion. In an aqueous solution, 1 underwent a PCET event to form 2 (Ni3II,II,III, total charge of +3, Supporting Information Figure S26), possessing two bridging hydroxide ligands. The spin density on Ni3 decreased from 1.69 in 1 to 0.89 in 2 ( Supporting Information Table S2), indicating that the electron released from Ni3 and a proton released from the binded water molecule. Meanwhile, the symmetric feature of the Ni3O4 core was broken. The bond lengths of Ni3–O1 and Ni3–O2 decreased from 1.996 to 1.888 Å and from 2.322 to 1.879 Å, respectively. These values were shorter than those of Ni1–O1 (2.062 Å) and Ni1–O2 (2.136 Å) in 2, consistent with the formation of NiIII at the Ni3 site ( Supporting Information Figure S26). Subsequently, one-electron oxidation of 2 proceeded to afford 3 (Ni3II,III,III, total charge of +4, Supporting Information Figure S27). The spin density on Ni2 increased from 0.01 to 0.75 ( Supporting Information Table S2), which indicated Ni2 was the oxidized site for this process. Next, 4 (Ni3II,III,III-O•, total charge of +4) was generated from 3 through a PCET process, where the spin density on O2 was 0.96 ( Supporting Information Figure S28). Therefore, this process could be classified as the oxidation of the bridging hydroxide to generate a bridging oxyl radical. Notably, during the one-electron oxidation of 4 to produce 5 (Ni3III,III,IV, the total charge of +5, Supporting Information Figure S29), a remarkable NiIV ion with little spin density was formed by an intramolecular electron transfer from Ni1 to the bridging oxyl radical and an additional electron release from Ni1 to the solution. Meanwhile, the axial Ni1–O3 and Ni1–N bond lengths both decreased significantly from 2.289 to 1.963 Å and from 2.067 to 1.905 Å (Figure 7 and Supporting Information Figure S29), respectively. The potentials for these four oxidation steps at pH 7 were calculated to be 1.32 V (PCET), 1.36 V (electron transfer, ET), 1.82 V (PCET), and 1.65 V (ET, Figure 6), respectively. The nature of the first two oxidation processes agreed with the experimental results (Figure 4a), where the slopes for the first two E-pH curves were −50 and 0 mV/pH, respectively. However, a single peak was observed for the conversion of 3 to form 5 in the CV experiment, with an oxidation process involving two electrons and one proton. The calculated potential from this process was 1.74 V at pH 7, and the slope for the E-pH curve was −29.6 mV/pH, which corresponded with the experimental value of −31 mV/pH. Taking together, the calculated potentials of 1.32, 1.36, and 1.74 V were in reasonable agreement with the experimental values of 1.16, 1.40, and 1.84 V derived from the CV experiment. Figure 6 | Gibbs energy diagram (kcal/mol) for water oxidation catalyzed by TNC-Ni complex at the B3LYP-D3 level. Download figure Download PowerPoint Interestingly, further deprotonation of 5 to 5dp (total charge of +4, Figure 7 and Supporting Information Figure S29) with two bridging oxygen groups required minimal energy cost. During this process, an intramolecular electron transfer occurred from one of the bridging oxos to Ni2 to produce a bridging oxyl radical, evidenced by the change of spin density on Ni2 from 0.95 in 5 to 0.01 in 5dp( Supporting Information Table S2). Meanwhile, the bond lengths of Ni3–O1 and Ni1–O1 decreased from 1.975 to 1.854 Å and from 1.879 to 1.835 Å (Figure 7 and Supporting Information Figure S29), respectively. Although the two equatorial Ni–O bond distances in 5dp were very close for Ni1 and Ni3, the axial Ni–O and Ni–N bonds showed a significantly larger difference, resulting in a more compact structure for Ni1. The spin densities on Ni1, Ni3, and the oxyl radical in 5dp are 0, 0.97, and 0.88 ( Supporting Information Table S2), respectively. These results implied that Ni1 in 5dp could be considered a NiIV ion (3d6) and Ni3 as a NiIII ion (3d7). Figure 7 | Structures of key intermediates and transition state for the O–O bond formation, including 5, 5dp, TS1, and Int1. Distances are given in Ångstroms, while spin densities on key atoms are indicated in red italics. The calculation details are included in the Supporting Information. Download figure Download PowerPoint The above calculations showed that the most favorable O–O bond formation was triggered by 5dp, which preferred to be a triplet. The broken-symmetry open-shell singlet and quintet were +1.5 and +5.9 kcal/mol higher in energy, respectively. The electronic structure of 5dp (Figure 7) could be described as featuring a low spin NiIII (S = 1/2, Ni3) ferromagnetically interacting with the bridging oxyl radical (S = 1/2), a low spin NiII (S = 0, Ni2), and a low spin NiIV (S = 0, Ni1). O–O bond was formed by the coupling of the bridging oxyl radical and the bridging oxo ( TS1, Figure 7). The barrier for TS1 was calculated to be 21.1 kcal/mol relative to 5dp in the triplet state ( Supporting Information Figure S30), while the broken-symmetry open-shell singlet gave a lower barrier of 15.3 kcal/mol (Figure 6). TS1 had only one imaginary frequency of 318.2i cm−1, which corresponded to the O–O bond formation, where the O–O distance was 1.92 Å ( Supporting Information Figure S30) and the spin densities on Ni1 and Ni3 were 0.65 and −0.66 (Figure 7). This O–O bond formation pathway was very similar to the O–O bond cleavage mediated by diiron oxidase and multicopper oxidases (MCOs).75–77 A bridging superoxide intermediate Int1 (Ni3II,II,III-O2•−, Figure 7 and Supporting Information Figure S31) was −12.6 kcal/mol relative to 5dp. Int1 proceeded via one-electron oxidation (potential of 1.68 V) with electron release in Ni2 to generate Int1′. Finally, the liberation of a dioxygen molecule, binding of two water molecules, and the proton release could regenerate catalyst 2 for the next water oxidation cycle. CV experiments were also conducted in D2O phosphate buffer ( Supporting Information Figure S24a). A solvent kinetic isotope effect (KIE = kH2O/kD2O) of 1.0 indicated that there was no O–H bond cleavage involved in the O–O bond formation. The electrochemical behaviors of TNC-Ni with different buffer concentrations (the ionic strength of I = 0.462 M) were carried out ( Supporting Information Figure S24b), and there no noticeable buffer effect was observed. These results further supported the O–O bond formation via the direct coupling of two bridging o