Switching the O-O Bond Formation Pathways of Ru-pda Water Oxidation Catalyst by Third Coordination Sphere Engineering

Water oxidation is a vital anodic reaction for renewable fuel generation via electrochemical- and photoelectrochemical-driven water splitting or CO2 reduction. Ruthenium complexes, such as Ru-bda family, have been shown as highly efficient water-oxidation catalysts (WOCs), particularly when they undergo a bimolecular O-O bond formation pathway. In this study, a novel Ru(pda)-type (pda2– =1,10-phenanthroline-2,9-dicarboxylate) molecular WOC with 4-vinylpyridine axial ligands was immobilized on the glassy carbon electrode surface by electrochemical polymerization. Electrochemical kinetic studies revealed that this homocoupling polymer catalyzes water oxidation through a bimolecular radical coupling pathway, where interaction between two Ru(pda)–oxyl moieties (I2M) forms the O-O bond. The calculated barrier of the I2M pathway by density-functional theory (DFT) is significantly lower than the barrier of a water nucleophilic attack (WNA) pathway. By using this polymerization strategy, the Ru centers are brought closer in the distance, and the O-O bond formation pathway by the Ru (pda) catalyst is switched from WNA in a homogeneous molecular catalytic system to I2M in the polymerized film, providing some deep insights into the importance of third coordination sphere engineering of the water oxidation catalyst.


Introduction
In a typical solar fuel generation device of either electrochemical or photoelectrochemical driven, it often consists of a fuelforming cathodic half-reaction, such as hydrogen evolution reaction (HER), carbon dioxide reduction reaction (CO 2 -RR), or nitrogen reduction reaction (NRR), and acquires protons and electrons. Water oxidation (2H 2 O → O 2 + 4H + + 4e − ), on the other hand, plays the unique role of an anodic half-reaction that provides protons and electrons in the over-all reaction of fuel generation [1,2]. This oxidation process is a four-proton four-electron process, which is not only thermodynamically demanding but also kinetically sluggish. Therefore, developing highly efficient water oxidation catalysts to promote this process is imperative for any practical application of HER, CO 2 -RR, and NRR [3,4]. To meet this demand, a comprehensive understanding of the catalytic mechanism of the water oxidation reaction is required [5,6]. Studying a multielectron and multiproton transfer process is a difficult task where molecular catalysts with tailor-designed organic ligands show their advantages of precise structures and relatively easy identification of intermediates involved in the catalytic processes.
Molecular water oxidation catalysts (MWOCs) containing various transition metal cores have therefore drawn significant attention in recent years, because their structures can be readily tailored for systematic structure-reactivity investigation, and their homogeneous catalytic property allows illustration of reaction mechanisms using classic spectroscopic techniques [7][8][9]. According to the mechanistic study upon numerous MWOCs, the O-O bond formation is often believed to be the rate-determining step (RDS) with the highest energy barrier through the catalytic water oxidation cycle [8,10]. Two major approaches have been proposed and widely accepted for the essential O-O bond formation step (Scheme 1): the water nucleophilic attack mechanism (WNA) and the oxygen radical coupling mechanism (I2M). Briefly, proton-coupled redox reaction upon the metal center (M n ) of a MWOC yields a high valence metal-oxo (M n+2 =O) species, which can either be attacked by a water molecule in a nucleophilic manner, forming the O-O bond (WNA), or transform to a corresponding metal-oxyl radical (M n+1 -O • ) intermediate, immediately affording a peroxo-bond with another adjacent metal-oxyl radical (I2M) [11].
Among Ru-based MWOCs, the family of mononuclear Ru(bda)L 2 (bda 2-= 2,2'-bipyridine-6,6'-dicarboxylic acid, L = N-cyclic aromatic ligands, e.g., pyridine) complexes have shown outstanding performances in terms of catalytic turnover frequency (TOF) and overpotential [12][13][14]. It has been demonstrated that water oxidation by these complexes proceeds through a dinuclear I2M pathway and the O-O bond forms via intermolecular coupling of two formal Ru V =O moieties with significant Ru IV -oxyl character. The hydrophobic nature of the Ru V =O species was believed to be one major force to drive the bimolecular radical coupling of Ru (bda) under aqueous catalytic conditions [15,16]. The series of Ru(pda)L 2 (pda 2-=1,10-phenanthroline-2,9-dicarboxylate) complexes have remarkably similar coordination geometry with the Ru(bda)L 2 catalysts. However, small modification of the backbone organic ligands from bda to a relatively rigid pda dramatically altered the catalytic pathway of the O-O bond formation. Unlike Ru(bda)L 2 , Ru(pda)L 2 prefers a WNA water oxidation pathway when using ceric ammonium nitrate as a sacrificial oxidant [17].
Density-functional theory (DFT) study displayed that, compared with the more flexible bda backbone ligand, the pda ligand increased the steric repulsion among catalysts and thus hindered two formal (pda)Ru V =O species from approaching each other [17]. Meanwhile, the (pda)Ru V =O species possesses less Ru IV -oxyl radical character than the (bda)Ru V =O species, and less likely to proceed through the I2M pathway [18]. For the case study of Ru(bda)L 2 as well as its analogs illustrate that the catalytic behavior of Ru MWOCs can be affected by even a small change of the primary coordination environment [17]. While the orientation of hydrophobic/hydrophilic substituent groups of MWOCs is a key factor contributing to the different reaction mechanisms [18], modifying the third coordination sphere of the catalyst is another approach to adjust their local environment and thus facilitate the I2M O-O bond formation pathway. These findings encouraged us to manipulate the catalytic water oxidation pathway of Ru(pda) complexes using combined strategies of direct ligand modification and local catalytic environment design.
Herein, two Ru(pda) MWOCs were synthesized with 4-picoline (1) and 4-vinylpyridyl (2) as axial ligands, respectively (Scheme 2). Kinetic studies revealed that electrochemical-driven water oxidation by 1 went through a WNA O-O bond formation pathway, similarly to what was observed in the previous Ce IV -driven water oxidation by 1. Nevertheless, complex 2 after immobilized on the glassy carbon (GC) electrode surface by electropolymerization (this homocoupling polymer is denoted as poly-2) in triggers intermolecular O-O radical coupling between metal-oxyl radicals during the electrocatalytic water oxidation process. Our experimental results clearly revealed that the WNA O-O bond formation pathway of single-site MWOCs can switch to the I2M mechanism by changing the local environment, such as shortening the intermolecular distance of catalyst active center, via a homopolymerization method in this case.

Results and Discussion
Electrochemical properties of complex 1 in solution and complex 2 that is immobilized on glass carbon (GC) electrode via electropolymerization were investigated by cyclic voltammograms (CVs) and differential pulse voltammograms  (DPVs) (as shown in Figure 1).  [17]. Considering the backbone ligand of poly-2 tend to draw Ru(pda) units closer in a face-to-face manner of the active sides than the catalyst 1 in a dilute solution, the O-O bond formation via I2M is more likely to occur in the former situation than the latter one, leading to different catalytic water oxidation pathways by 1/GC and poly-2@GC.
To gain some more deep insights into the different O-O bond formation mechanisms between 1/GC and poly-2@GC, three key kinetic indicators were compared between catalytic water oxidation reactions by these two Ru(pda) catalytic systems [19]: (1) reaction orders of the catalyst, (2) reaction orders of the phosphate proton acceptor, and (3) deuterium kinetic isotope effects (KIEs H/D ). In a WNA scenario as shown in Scheme 1, the catalytic reaction rate would show a first-order relationship with respect to the catalyst concentration, and increasing concentration of an effective proton acceptor, which promotes the formation of metalhydroperoxide species ( Figure 1: Cyclic voltammograms (CVs) of 1/GC (a) and poly-2@GC (b) measured in a pH = 7:0 sodium phosphate buffer (100 mM) at a scan rate of 100 mV s −1 . Insets show the corresponding differential pulse voltammograms (DPVs). the O-H bond cleavage is involved in the rate-determining step (RDS). In an I2M scenario, by contrast, the catalytic reaction rate would exhibit a second-order relationship depending on the catalyst concentration if the binuclear coupling is the RDS. At the same time, the overall rate of water oxidation should be insensitive to the proton acceptor concentration and deuterium substitution because proton transfer is not involved in the RDS.
Kinetic studies for water oxidation by Ru(pda) were firstly conducted by electrochemical methods in the homogeneous system (1/GC). The scan rate-dependent CVs of complex 1 ranging from 50 to 125 mV/s in aqueous sodium phosphate buffer solution (ionic strength = 1 M, Na 2 SO 4 ) are shown in Figure S2a. The position of Ru III/II redox peak at E 1/2 = 0:55 V is independent of the scan rates; however, its peak current (j p ) varies linearly to υ 1/2 ( Figure S2b). Meanwhile, as shown in Figure S2c and S2d, the peak current shows a linear relationship with the concentration of complex 1. This linear relationship is consistent with the Randles-Sevcik relation (Equation S1), indicating a diffusioncontrolled redox process [20,21].
The concentration-dependent catalytic currents were obtained from LSV curves of the 1/GC at various catalyst concentrations (as shown in Figure 2(a)). The apparent electrode reaction order of the catalyst (ρ j cat ½cat ) based on the catalytic current can be calculated according to Equation S2, and all slopes approximately to 1 at different applied potentials indicate a single-site electrocatalytic pathway for the 1/GC system [22,23]. The reaction rate constant of catalyst (k cat ) can be obtained according to Equation S3.
The influence (ρ k cat ½cat ) of the concentration of complex 1 on its k cat for the water oxidation reaction can be evaluated (Equation S4). Because the reaction orders of the catalytic current (ρ j cat ½cat ) and peak current (ρ j p ½cat ) on the concentration of complex 1 are both 1; the calculated ρ k cat ½cat is zero, suggesting the independence of the rate constant (k cat ) on the concentration of the catalyst in a dilute solution. This is in agreement with the single-site WNA O-O bond formation mechanism [17]. In the case of a heterogeneous system catalyzed by the polymerized Ru(pda) (poly-2@GC) as described above, the high local catalyst concentration in the polymer (poly-2) backbone may shorten the intermolecular distance of Ru units and afford a Ru face-to-face environment, which is beneficial to the dinuclear I2M pathway. The Faraday efficiency of the complex 2 modified electrode (poly-2@GC) was measured (Equation S5) before the kinetic study [19]. As shown in Figure S3, the high Faraday efficiency of poly-2@GC indicated that the cumulative charges passing through the electrochemical system were almost quantitatively depleted in the water oxidation reaction, and therefore, the corresponding catalytic currents of the electrodes were directly applied for the kinetics analysis. Meanwhile, because this electrode reaction for the poly-2@GC system is not a catalyst diffusion-controlled process, the catalytic current density of poly-2@GC can represent the rate constant as well as the turnover frequencies of the catalyst on the surface. By adding the polarization times of electrochemical polymerization, poly-2@GC with various catalyst coverages (Γ) were obtained ( Figure S4 and Equation S6). As shown in Figure 2(b), the catalytic current density (j cat ) raised with an increasing Γ at the certain applied potential. According to Equation S2, S7, and S8, the reaction orders on the catalyst concentration (ρ j cat ½cat ) were calculated to be 1.8~1.91 regarding applied potentials. Furthermore, the ρ k cat ½cat is equal to the ρ j cat ½cat as discussed above. The approximate value 2 of ρ k cat ½cat indicates that the water oxidation reaction catalyzed by poly-2@GC is a second-order reaction on Ru(pda) unit. This observation is consistent with the I2M O-O bond formation pathway. Contrary to the homogeneous catalytic conditions, the k cat of Ru(pda) on the electrode surface shows a strong correlation with the catalyst coverage. The difference of reaction kinetics between the homogenous and heterogeneous conditions confirms the promotion of I2M pathway by reducing intermolecular distance of Ru(pda) moieties on the electrode surface.
As described previously, atom-proton transfer (APT) can dramatically promote the rate of a WNA process [24]. As shown in Figure 2(c), the catalytic currents of 1/GC increase significantly with the increment of phosphate concentration; correspondingly, the electrode reaction order on phosphate (ρ j cat ½pi ) can be defined and calculated to be 0.3 (Equation S9). The influence of phosphate concentration on the catalytic reaction rate constant ρ k cat ½pi was evaluated in a similar manner (Equation S10). As shown in Figure S5a    on the electrode surface makes the approaching of two formal (pda)Ru V =O species overcome the steric hindrance of the pda ligand and benefits the I2M mechanism reaction rather than the WNA pathway of pristine Ru(pda)-type MWOC. DFT calculations and molecular dynamics (MD) simulations were performed to further illustrate the reaction pathway of electrochemical driven water oxidation by Ru(pda) in poly-2@GC and properties of the catalyst. Our previous study has shown that the face-to-back configuration is also more favored than the face-to-face configuration of two (pda)Ru V =O, due to the hydrophobic pda backbone and the hydrophobic oxo, which results that Ru(pda) complex disfavored the binuclear I2M pathway    for homogeneous water oxidation in aqueous solution [18]. The kinetic study on poly-2@GC showed a binuclear I2M reaction pathway. Hence, a supramolecular dimer with face-to-face configuration linked by two propyl groups was revisited as a reasonable initial structural model for the calculations. The calculated redox potentials of ½Ru II − OH 2 /½Ru III − OH 2 + , ½Ru III − OH 2 /½Ru IV − OH + , ½Ru IV − OH/½Ru V = O + are 0.63, 0.73, and 1.22 V at pH 7.0, respectively, which are close to the experimental values ( Figure S7). For the key intermediate, [Ru V =O] + , MD simulations were performed to understand solvation properties of the oxo in the supramolecular dimer, by using the previously parameterized Ru(pda) model with new partial charge parameters [18]. The H-bond analysis and the radial distribution function (RDF) analysis ( Figure S8) of oxygen atoms with water molecules showed that the oxo of the supramolecular dimer is also hydrophobic, the same as that in the molecular Ru(pda) catalysts. The high spin density of the oxo (0.69) from DFT calculations is similar to that of the oxo in molecular Ru(pda) (0.71). With the O-O bond formation step of two (pda)Ru V =O species, as shown in the relative free energy profile (Figure 4), the reaction is exergonic with a free energy of -21.16 kcal mol -1 from the prereactive to the product, accompanied by an activation free energy of 0.53 kcal mol -1 from the prereactive to the transition state. The low activation free energy of the (pda)Ru V =O coupling process, comparing to the activation free energy of 20.6 kcal mol -1 by WNA pathway [18], supports the experimental results that the Ru(pda) in poly-2@GC catalyzed water oxidation through the I2M mechanism. Once the catalysts overcoming the arrangement step and yielding the prereactive dimer state, the O-O bond is favored to form via coupling of two linked Ru(pda) V =O species. The theoretical evidence implied the possibility of switching the O-O bond formation pathway of the singlesite Ru(pda) catalyst from the WNA mechanism to the I2M by manipulating the third coordination sphere, that is the hydrophobic/hydrophilic directionality of the catalysts. For instance, forcing the Ru(pda) catalyst to be a prereactive face-to-face dimer configuration via structural modification is a good strategy for I2M O-O bond formation pathway.

Discussion
Recently, several MWOCs based on first-row transitionmetal such as Mn, Fe, Co, Ni, and Cu have been developed [25][26][27][28][29][30][31][32][33][34][35]. Although the catalytic mechanism and structureactivity correlation of these WOCs are still not known as much as that of Ru-based WOCs, they are very attractive for opening up the way to develop WOCs based on earthabundant elements. The O-O bond formation is often the RDS for most WOCs; meanwhile, for most single-site molecular WOCs, the O 2 is produced through WNA. How to improve the activity of single-site WNA WOCs attracts attention [11]. Among these, designing binuclear complexes from single-site catalysts has attracted much interest, owning to the potential possibility to obtain complexes that undergo the fast I2M pathway [7]. However, there is no clear experimental evidence that the reported binuclear complexes react via I2M mechanistic pathway. So far, most structure-activity studies of MWOCs have focused on the first and second coordination sphere surrounding the metal center of the catalysts (the chemical structure of the catalyst and proton transfer relay) [36][37][38]. A significant influence of intermolecular environments (the third coordination sphere) over the O-O bond formation pathway has rarely been reported. This is the first successful study on the switching of a WNA O-O bond formation pathway of a catalyst to a binuclear I2M pathway via manipulating the third coordination sphere of a catalyst. Our approaches might inspire related research on nonnoble metal WOCs-based water splitting devices in the future.
In summary, we studied the O-O bond formation mechanisms of electrochemical water oxidation by Ru(pda) type catalysts. Ru(pda) bearing 4-picoline axial ligands (1/GC) electrochemically catalyzed water oxidation through the water nucleophilic attack (WNA) pathway in a homogeneous system. When the intermolecular distance of the Ru(pda) type catalyst was forced to diminish and configurated in a face-to-face manner, by immobilizing complex 2, a Ru(pda) with 4-vinylpyridyl axial ligands, on glassy carbon electrode surface via electropolymerization (poly-2@GC), the O-O bond formation mechanism switched from WNA to intermolecular radical coupling (I2M). Kinetic studies, including the phosphate concentration and deuterium kinetic isotope effects upon reaction orders, revealed that Ru(pda) in the homocoupling polymer (poly-2) triggers two metal-oxyl radicals coupling interaction during water-oxidation catalysis. This I2M reaction pathway was also supported by DFT calculations. Our results suggest that the WNA O-O bond formation pathway of single-site MWOCs can switch to a binuclear I2M mechanism by manipulating the third coordination spheres of catalysts through structural modification. Our results provide a novel perspective for the design of more advanced water-oxidation catalysts by regulating the catalyst environment without changing the structures of catalysts.

Materials and Instrumentation.
All commercial chemical reagents were used as received, and the water applied in this work was deionized using the Milli-Q technique. cis-Ru(DM-SO) 4 Cl 2 was prepared based on published works [39]. NMR spectra were measured by Bruker Advance 500 spectrometer. Mass spectra were collected on a Finnigan LCQ Advantage MAX mass spectrometer.

Electrochemistry.
Electrochemical measurements were carried out with a CH Instruments 660E electrochemical workstation at room temperature. A three-electrode configuration, where a glassy carbon was used as the working electrodes, a Ag/AgCl (3 M KCl) was used as the reference electrode, and a platinum wire was used as the counterelectrode, in a single compartment cell. All the measured potentials were converted to NHE according to a previous report [41]. All GC electrodes were polished with 1 μm alumina powder before all experiments. 4.5. Computational Details. All DFT calculations for the estimation of Gibbs free energies were carried out with the Jaguar 8.3 program package by Schrödinger LLC [42]. Molecular geometries were optimized using Becke's threeparameter hybrid functional and the LYP correlation functional (B3LYP) with D3 correction of Grimme et al. with the LACVP * * basis set [43][44][45][46]. To identify the transition states for O-O bond formation, we searched the potential energy surface by scanning the terminal O-O bond distance [Ru V =O•••O=Ru V ] of the antiferromagnetic open-shell singlet. The thermochemical corrections for estimations of the Gibbs free energy barrier from the prereactive dimers were calculated at the B3LYP-D3/LACVP * * level for both the prereactive dimers and transition state structures. Single-point energy corrections were performed with the B3LYP-D3 functional using the LACV3P * * ++ basis set augmented with two f -functions on the metal. Based on the gasphase optimized geometries, the implicit solvation energies were estimated by single-point calculations using the Poisson-Boltzmann reactive field implemented in Jaguar (PBF) in water. The Gibbs free energy was defined by the following equation G = E ðB3LYP − D3/LACV3P * * ++2f on RuÞ + G solv + ZPE + H 298 − TS 298 + 1:9 kcal mol −1 (the value 1.9 kcal mol -1 is a concentration correction to the free energy of solvation, which by default is calculated at 1 M (g) to 1 M (aq) in Jaguar).
MD simulations were performed with the Gromacs 5.1.4 MD software package [47]. A 20 ns MD run was performed in a 44 × 41 × 38 Å3 periodic box filled with TIP3P water molecules and two chloride ions to neutralize the charge [48]. In MD simulations, the resulting systems were subject to 100 000 steps of steepest descent minimization. The periodic boundary condition was applied in the simulation. The cutoff radius for the Lennard-Jones and electrostatic interactions were set to be 10 Å. For an accurate evaluation of the longrange Coulomb interactions, Particle Mesh Ewald (PME) summation method is used for electrostatic interactions beyond the cutoff [49]. The system was heated to 300 K in 100 ps by using a v-rescale thermostat for the canonical ensemble (NVT) simulations [50]. During this process, the Linear Constraint Solver (LINCS) algorithm was used to constrain all the bond lengths [51]. The isothermal isobaric ensemble (NPT) was used in the subsequent simulations, with the pressure set to 1 bar in 100 ps, controlled using a v-rescale thermostat and Parrinello-Rahman barostat [52]. Thereafter, the systems were simulated for 20 ns. Three repeated MD simulations with different initial velocities were also performed. A time step of 2.0 fs was used throughout the simulations.

Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this article.

Authors' Contributions
F. Li and L. Sun conceived the idea, designed the experiments, and guided the whole project as the corresponding author. Y. Li and S. Zhan contributed equally to this work. Y. Li conceived the idea, conducted the experiments, analyzed the data, and wrote the draft. S. Zhan conducted the DFT calculations and MD simulations, analyzed the data, drew the figures, and composed the manuscript. All authors discussed the results and revised the manuscript.

Supplementary Materials
Derivation of equations S1-S15. Figure S1: H-NMR spectrum of complex 2. Figure S2: scan rate-dependent CV measurements of complex 1. Figure S3: Ferriday efficiency of poly-2@GC. Figure S4: the voltammetric waves of poly-2@GC with various catalyst coverage. Figure S5: the concentration of phosphate-dependent CV measurements of 1/GC. Figure  S6: diffusion constant measurements of complex 1 in D 2 O and the corresponding KIEs H/D . Figure S7: DFT optimized geometries of the intermediates. Figure S8: H-bond analysis, Figure S9: the RDF analysis. Table S1: the summary of the electrochemical data of several single-site Ru molecular catalysts. Data files S1-S7, the calculated geometries of intermediates and product and partial charge parameters. (Supplementary Materials)