A Dinuclear Ruthenium-Based Water Oxidation Catalyst: Use of Non-Innocent Ligand Frameworks for Promoting Multi-Electron Reactions

Insight into how H2O is oxidized to O2 is envisioned to facilitate the rational design of artificial water oxidation catalysts, which is a vital component in solar-to-fuel conversion schemes. Herein, we report on the mechanistic features associated with a dinuclear Ru-based water oxidation catalyst. The catalytic action of the designed Ru complex was studied by the combined use of high-resolution mass spectrometry, electrochemistry, and quantum chemical calculations. Based on the obtained results, it is suggested that the designed ligand scaffold in Ru complex 1 has a non-innocent behavior, in which metal–ligand cooperation is an important part during the four-electron oxidation of H2O. This feature is vital for the observed catalytic efficiency and highlights that the preparation of catalysts housing non-innocent molecular frameworks could be a general strategy for accessing efficient catalysts for activation of H2O.


Introduction
The production of solar fuels, such as H 2 ,b yt he splitting of H 2 Oc onstitutes an appealing strategy for attaining sustainable systemsf or energy conversion.I ns uch systems, the catalytic conversion of H 2 Ot oO 2 [Eq. (1)],w hich will generate the necessary electrons and protons, is considered to be the essential step. [1][2][3][4] Due to the thermodynamic and catalytic complexity connected with the four-electron oxidation of H 2 O, the development of artificial water oxidation catalysts (WOCs) has become an important research field. [5] 2H 2 O ! O 2 þ4e À þ4H þ E ¼ 1:23Àð0:059 pHÞ Vv ersus NHE ð1Þ Although there has been considerable progress during the last couple of years with the construction of molecular catalysts based on Ru, [6,7] Ir, [8] Mn, [9] Fe, [10] Co, [11] and Cu, [12] the mechanistic description associated with such WOCs is usually not well defined. It is therefore important to understand how H 2 Oi so xidized by artificial WOCs. This is an intricate scientific goal, which,i fa ccomplished, is expected to facilitate the design of more-efficient catalysts. Through an umber of experimental [13] and theoretical [14] studies, impressive progress has been made in unravelling the different molecular mechanisms by which H 2 Oi sa ctivated by artificial WOCs. It has thus been established that H 2 Oo xidation and OÀOb ond formation mainly take place through two different pathways:1 )Water nucleophilic attack on ah igh-valent metal-oxo unit (WNA mechanism;F igure 1, left), and 2) the interaction of two MÀO units (I2M mechanism;F igure 1, right), which can occur either intermolecularly between two catalyst entities or intramolecularly for dinuclear catalysts. [15] At hird pathway has recently been suggested for aM n-based WOC, in which OÀOb ond formation occurs by coupling aM n IV -bound terminal oxyl radical and ad i-Mn bridging oxo group within at etranuclear Mn species. [16] It is essential that the catalysts are ableo fr eaching high redox states within ar elativelyn arrow potentialw indow.F or this to occur efficiently,i ti st hus important that the designed catalysts undergo concerted removal of protons and electrons to avoid Coulombic chargeb uild-up. The coordinated movement of protons and electrons through proton-coupled electron transfer (PCET) ensures pathways with low-energy profiles, which is particularly beneficial for am ulti-electron reaction such as H 2 Oo xidation. [17,18] We have recently developed the dinuclear Ru 2 II,III complex 1 (Figure 2), which efficiently catalyzes the photochemical conversion of H 2 Oi nto O 2 . [19] The designed electron-rich anionic pyrazole-based ligand scaffold (H 5 L) was found to afford asuitable environment for the Ru centers by placing them in close proximity (4.93 , see the Supporting Information, FigureS6), to give ac omplex with favorable low redox potentials, thus allowing access to multiple redox states. Based on these attractive catalytic features it was decided to study this catalytic system in more detail by combining experimental measurements andq uantum chemical calculations.
In this work, we furthere xplore the mechanistic detailsa ssociated with the dinuclear Ru complex 1 by using acombination of experimental measurements and quantum chemical calculations. In addition to stabilizing the metal centers in high-valent redox states, it was revealed that the constructed ligand motif facilitates protonm ovement and actively participates during the four-electron oxidation of H 2 O. These features collectively account for the observed catalytic efficiency of Ru complex 1 and highlight the concept of using non-innocent ligand scaffolds for carrying out multi-electronr eactions, such as oxidation of H 2 O.

Results and Discussion
High-resolution mass spectrometry analysisofintermediates involved in H 2 Oo xidation High-resolution mass spectrometry (HRMS)h as previously been employed as at ool for studying and detecting intermediates during H 2 Oo xidation catalysis. [20,21] HRMS was therefore initially assessed as at echnique for investigating the different intermediates involved in the catalytic systemf or Ru complex 1.A na queous solution containing the dinuclearR uc omplex 1 was analyzed in positive mode. This resulted in the observation of ap eak at m/z 1146.2155, which was assigned to the startingc omplex 1,{ [(H 2 L)Ru 2 II,III (pic) 6 ] 2 + ÀH + } + . [19] The one-electron oxidized specieso fc omplex 1,t he corresponding Ru 2 III,III complex, was also possible to detect at m/z 1145.2093 (Figure 3a nd [Eq. (2)]). In accordance with the Pourbaix diagram (see below) the reaction is proton-coupled at pH > 3a nd the product from this oxidation (at pH 7.2) thus results in [(HL)Ru 2 III,III (pic) 6 ] 2 + .
Access to the corresponding Ru-aquac omplexes is essential in H 2 Oo xidation catalysis since it permits PCET redox events to occur,a llowing high-valent redox states to be generated. Allowing the Ru 2 III,III oxidizeds pecies to stand for as hort period of time in aqueous solutionsr esulted in the formation of the formal Ru 2 III,III aqua species[ (HL)Ru 2 III,III (pic) 6 (OH 2 )] 2 + [Eq.
(3)] shown in Figure 4. This is in accordance with quantum chemical calculations (see below,s ee also Figure 8), which show that the oxidized Ru complex 1 and the aqua species have comparable stabilities, Figure S47 (the Supporting Information).
The [(HL)Ru 2 III,III (pic) 6 (OH 2 )] 2 + and [(L)Ru 2 III,IV (pic) 6 (OH)] + speciesw ere initially believed to contain as even-coordinated Ru center. However,q uantum chemical calculations suggest intermediates in which one of the coordinating pyrazolen itrogen atoms has been detached from the Ru center (see below). Thus, the dissociation of one of the Ru centers from the bridging pyrazole ligand scaffold creates an open coordination site for the incoming aqua ligand, thereby bypassing the formation of the unfavorable seven-coordinated intermediate. These results suggest that the designed pyrazole ligand motif actively promotes coordination of H 2 Od uring catalysis and that this metal-ligand cooperation contributes to the catalytic activity of Ru complex 1.
Ap eak at m/z 976.0901 matching the Ru 2 III,IV intermediate [(L)Ru 2 III,IV (pic) 4 (OH)] + ( Figure 6) was also observed and is assumed to arise from the conversion of [(L)Ru 2 III,IV (pic) 6  (pic) 4 (OH)] + species thus containsafree coordination site, which should enable the coordination of as econd aqua ligand.F acilitated by the electronrich ligand scaffold, such ad iaqua Ru species should readily be able to access even higherr edox states,w hich ultimately trig-gers OÀOb ond formation.T he fundamentalc atalytic features, which are thus provided by HRMS,a re important for understanding the properties of the dinuclearR uc omplex 1 as illustrated by the intricate Pourbaixdiagram, see Figure 11.

Quantum chemical description of the key intermediates
Densityf unctional theory (DFT) calculations [22] have proven to be essential for understanding the processes leadingt oH 2 O oxidation and to study the influence of the ligand on its environment in artificially developed WOCs at different redox states. Aimingt ou nderstand the different redox processes that are involved for Ru complex 1,t he redox potentials for the three complexes ([(H 2 L)Ru 2 (pic) 6 ], [(H 2 L)Ru 2 (pic) 6 (OH 2 )],a nd [(H 2 L)Ru 2 (pic) 4 (OH 2 ) 2 ]) were calculated (Table1). By calculating al arge number of relevant structures housing different numbers of picoline and aqua/hydroxo/oxo ligands in different redox states,different protonation states, isomers, and possible spin states, detailedi nsightw as gained. All the relevant structures are shown in the Supporting Information (Figures S1-S44). The calculated structures of the Ru 2 II,III speciesa re depicted in Figure 7a nd the Supporting Information (Figures S6-S12). In acidic aqueous solutions, in the Ru 2 II,II and Ru 2 II,III redox states, the dominant form of the Ru complex should be [(H 2 L)Ru 2 (pic) 6 ] 2 + (see Table 1), which is also supported experimentally (see Figure 11).
Since the Ru 2 III,III state can easily be accessed under neutral conditions, at which catalysis occurs (see Ta ble 1a nd Figures S13-S25), the possible ligand exchange processes were studied at different redox states.T he variouss tructures are shown in the Supporting Information (FiguresS13-S38). Calculations show that the Ru 2 III,III complex prefers to coordinate six picoline ligands( [(L)Ru 2 III,III (pic) 6 ]), and therefore the energy of this complex was set to be zero. At pH 7.2, this complex has at otal charge of + 1a nd the insertion of an aqua ligand could lead to as even-coordinated Ru center, which has previously been reported for Ru-based WOCs.
[6e] However,t he calculations suggest that this does not occur for Ru complex 1,i nstead one Ru center appears to dissociate from the bridgingp yrazole ligand, thus resultingi na no pen coordination site for an aqua ligand to occupy ( Figure 8).   An additional feature of the developed ligand scaffold, which was seen from calculations, is that several different isomeric structures can co-exist. Transfer of ap roton from the Rubound aqua ligand to the imidazole nitrogen to yield structur-eA (see Figure 8a nd the Supporting Information, Figure S15) was found to result in the most stable isomer.S tructure Bw as found to be slightly higher in energy and have the protona t the pyrazolem oiety.T he overall energetics of this aqua addition process was thus revealed to be endergonic by 2.5 kcal mol À1 (see the Supporting Information, Figure S47). The active participation of the engineered ligand framework in ligand exchange in Ru complex 1 is an intriguing feature that might facilitate coordination processes during the catalytic cyclea nd enhancep roton-coupled events. From the generated formal [(L)Ru 2 III,III (pic) 6 (OH 2 )] + intermediate, one picoline ligand can dissociate from the Ru center housing the "hydroxide" ligand (structure AinF igure 8), whichisconcomitant with recoordina-tion of the pyrazolen itrogen to the Ru center.T his step was calculated to be close to isogonic (see the Supporting Information, Figure S47). The picoline-aqua ligand exchange thus takes place through an associativetype of mechanism,i nw hich the bridging pyrazole ligand plays ak ey role in providing access for coordination of the aqua ligand.
As econd aqual igand can then be coordinated to the other Ru centeri nasimilar fashion. Here it was also shown that the bridging pyrazoleh as an oninnocent role in affording an open coordination site by dissociating from the metal center. The resulting complex (the Supporting Information, Figure S21) thus has five picoline ligands and two aqua ligands ([(L)Ru 2 III,III (pic) 5 (OH 2 ) 2 ] + )a nd is the least favorable species in the ligand exchange Scheme as its energy is + 5.4 kcal mol À1 relative to the startingc omplex [(L)Ru 2 III,III (pic) 6 ] + (see the Supporting Information, Figure S47). Finally,p icoline dissociation from [(L)Ru 2 III,III (pic) 5 (OH 2 ) 2 ] + leads to the expected [(L)Ru 2 III,III (pic) 4 (OH 2 ) 2 ] + complex that may initiate H 2 Oo xidation. As imilarl igand exchange analysis was performed for the developed Ru complex in its Ru 2 III,IV state. Here it was found that the oxidation of the [(L)Ru 2 III,III (pic) 6 ] + complex to its Ru 2 III,IV state was associated with ar elatively high redox potential of 1.07 Vv ersus NHE at pH 7.2 (see Ta ble  (pic) 6 (OH 2 )} 2 + ÀH + ] + ), which furthers upports the non-innocence of the developed  (pic) 4 (OH) 2 ]c omplex is endergonic by 3.2 kcal mol À1 (the Supporting Information, Figure S48). However,c onsidering that the Ru complex concentration (mm) is much lower than the H 2 Oc oncentration (55.5 m), [(L)Ru 2 III,IV (pic) 4 (OH) 2 ]s hould also be an existing speciesa tt he Ru 2 III,IV state. From Ta ble 1i ti so bvious that for the two highest redox processes fort he [(L)Ru 2 (pic) 4 (OH 2 ) 2 ]s pecies, namely the Ru 2 IV,IV /Ru 2 III,IV ,and Ru 2 IV,V /Ru 2 IV,IV redox events, the agreement between the calculated and experimentally obtained potentials is excellent and highlights that access to the higher redox states requires the generation of the diaqua complex [(L)Ru 2 (pic) 4 (OH 2 ) 2 ].
The calculations furtherr evealed that the formal [(L)Ru 2 III,IV (pic) 4 (OH) 2 ]c omplex is better described as [(L)Ru 2 III,IV (pic) 4 (O)(OH 2 )],i nw hicht he oxo moiety has significant oxyl character (Ru 2 III,III ÀOC), with as pin density of 0.70 (see the Supporting Information, Figure S35 A). The subsequent redox process, which has ar edox potentialo f0 .85 V, was also found to be proton-coupled and results in the formation of [(L)Ru 2 IV,IV (pic) 4 (O)(OH)].F inally,t he highest redox process under consideration was ao ne-electron oxidation that generates af ormal Ru 2 IV,V complex at am odest potentialo f1 .14 Vt hanks to the crafted ligand framework in complex 1.H owever,h ere the electronic structure is better described as am ixture of Ru 2 III,V (OC)(O) and Ru III,IV (OC) 2 ,i nw hich the spin densities are % 1.0 on each of the Ru III center and the connecting oxygen, and % 0.5 on the Ru IV centera nd the connected oxygen (structure Ai nF igure 10). From this high-valent species, OÀOb ond formation should readily occur,t hereby affording ap athway for activation of H 2 Oi nw hich the cooperative involvement of the ligand motif has av ital role during catalysis.

Mechanistic analysis from electrochemical measurements
To obtain further insight into the catalytic properties of the dinuclear Ru complex 1,e lectrochemical measurements were conducted to construct the Pourbaix diagram. Pourbaixd iagrams are important tools for highlighting possible stable redox states of the studied catalytic system.H ere, the Nernst Equation ([Eq. (5)]: in which m is the number of protons transferreda nd n is the number of electrons transferred) describes the potential dependencies for reactions involvingp roton transfer.T hus the potentialo faone-electron-one-proton redoxe vent will decrease by % 59 mV per pH unit, whereas at wo-electron-oneproton process will decrease by % 29 mV per pH unit. These features make it possible to distinguish between the different processes and to gain insighti nto the catalytic properties of the examined system. Ta ble 1s ummarizes the electrochemical properties for Ru complex 1 at different pH. None of the oxidations appear as reversible peaks in the cyclic voltammogram at pH 7.2. However,r easonably resolved peaks were obtained by differential pulse voltammetry (DPV). The Pourbaixd iagram of the dinuclear Ru complex 1 was subsequently constructed by measuring the redox potentials by DPV at different pH, 1.5 < pH < 8.5, and is depicted in Figure 11.T his diagram is quite complicated, however,b yt he use of quantum chemical calculations and MS it was possible to assign the different redox steps. It was found that Ru complex 1 exhibited severalr edox eventsi n-volvingP CET,a se vident by the decrease of the redox potentials with increasing pH.
On dissolution of Ru complex 1 in aqueous solutions, HRMS suggest that [(H 2 L)Ru 2 II,III (pic) 6 ] 2 + (1)i st he initial species at the Ru 2 II,III state, [19] which is further supported by the quantum chemicalc alculations. The Ru 2 II,III /Ru 2 II,II and Ru 2 III,III /Ru 2 II,III redox couples should therefore correspond to the oxidation of this picoline species. In Table 1, the calculations show that this is true for Ru 2 II,III /Ru 2 II,II at pH < 3, in which the experimentally measured and calculated potentials, 0.24 and 0.19 V, respectively,a gree.H owever,t he calculations suggestt hat the redox process should be pH-independenta tp H>3, which does not agree with what is observed experimentally.I nstead, the redox process becomes proton-coupled at higher pH, with as lope corresponding to ao ne-electron-one-proton process. Although it was only possible to observe [(H 2 L)Ru 2 II,III (pic) 6 ] 2 + (1)a tt he Ru 2 II,III state by HRMS, both the [(H 2 L)Ru 2 II,III (pic) 6 (OH 2 )] 2 + and [(H 2 L)Ru 2 II,III (pic) 4 (OH 2 ) 2 ] 2 + speciescan form relatively readily,see Figure S46 (the SupportingI nformation). In its reduced form (Ru 2 II,II state), the latter species should undergo proton-coupled electron transfer (Table 1), which might explain the experimental results. Alternatively,[ (H 2 L)Ru 2 II,II (pic) 6 (OH 2 )] + could be protonated to give [(H 3 L)Ru 2 II,III (pic) 6 (OH 2 )] 2 + ,which would be capable of undergoing PCET,a ss uggested in the Pourbaix diagram (Figure 11).
The two subsequentr edox processes, occurring at 0.86, and 1.12 Va tn eutral conditions, are also pH-dependent with as lope of À59 mV per pH unit over the whole studied pH range and correspond to the formal oxidations [(L)Ru 2 III,IV (pic) 4 (Table 1). This suggests that it is necessary to access the high-valent Ru 2 IV,V state before OÀOb ond formation can be triggered.
As triking feature that is apparent from Table 1, is that when comparing the calculated and experimental Ru 2 III,III /Ru 2 II,III redox potentials at pH % 5, the Ru complex 1 seems to exist in two forms, namely as [(H 2 L)Ru 2 (pic) 6 ] 2 + and [(H 2 L)Ru 2 (pic) 6 (OH 2 )] 2 + . This phenomenon also seems to prevaila tp H7,i nw hich the complex appearst oc o-exist as [(H 2 L)Ru 2 (pic) 6 ] 2 + and [(H 2 L)Ru 2 (pic) 6 (OH 2 )] 2 + (see Ta ble 1). Collectively,t his shows that aquation of Ru complex 1 occurs more easily at neutral pH, thus causing the complex to exist as different speciesc ontaining varying amountso fp icoline ligands, which explains that more redox peaks are observed in the Pourbaix diagram than are intuitively expected. During the redoxp rocesses the ligand scaffold actively participates, whichf acilitates access to the higherr edox states and ultimately triggersO À Ob ond formation (see Figure 12).

Conclusion
Introduction of the designed electron-rich pentacoordinating ligand to generate Ru complex 1 significantly lowers the redox potentials for the complex and allows access to aw ide variety of redox states. This work uncovers the fundamentalmechanistic detailsb yw hich the dinuclearR uc omplex 1 mediates H 2 O oxidation, which is essential for the development of more efficient WOCs. The combination of experimental analysisa nd quantum chemical calculations resulted in the discoveryo f vital features associated with Ru complex 1.O ne vital feature of the ligand motif in Ru complex 1 is the non-innocent behavior,s ince it participates in both protont ransfer and accepting/ donating electrons. This metal-ligandc ooperation plays ak ey role during the multi-electron oxidation of H 2 O. These results highlight that the engineering of catalysts for the activation of H 2 Ou sing non-innocent molecular scaffolds may be of value for the future construction of efficient WOCs.

Electrochemistry
Electrochemical measurements were carried out with an Autolab potentiostat with aG PES electrochemical interface (Eco Chemie), using ag lassy carbon disk (diameter 3mm) as the working electrode, and platinum as the counter-electrode. The reference electrode was as aturated calomel electrode (SCE). All potentials reported herein are converted to normal hydrogen electrode (NHE)

Computational details
The geometry optimizations in the present study were performed by using the Gaussian 09 [24] package and the B3LYP [25] functional. The 6-31G(d,p) basis set was applied for the C, N, O, He lements and the SDD [26] pseudopotential for Ru. Frequencies were calculated analytically at the same level of theory as the geometry optimization to obtain the Gibbs free energy corrections and to confirm the nature of various stationary points. Solvation effects from the H 2 Os olvent were calculated by using the SMD [27] continuum solvation model with the larger basis set in which all elements, except Ru, were described by 6-311 + G(2df,2p) at the B3LYP* (15 %e xact exchange) level. [28] It has been shown that B3LYP* gives ab etter description of relative energies in transition metal complexes. [28] For H 2 O, the experimental solvation free energy (À6.3 kcal mol À1 ) was used. [29] The concentration correction of 1.9 kcal mol À1 at room temperature (derived from the free-energy change of 1mol of an ideal gas from 1atm (24.5 Lmol À1 ,2 98.15 K) to 1 m was added for all species except H 2 O, for which the corresponding value is 4.3 kcal mol À1 as the standard state of H 2 Oi s5 5.6 m.U nless otherwise specified, the B3LYP*-D2 energies are reported, including Gibbs free energy corrections from B3LYP and dispersion corrections proposed by Grimme. [30] To calculate the redox potential, the absolute redox potential of the standard hydrogen electrode (4.281 V) was used as the reference, [31] which corresponds to 127.8 kcal mol À1 for ao ne-electron oxidation and 407.9 kcal mol À1 for ap roton-coupled one-electron oxidation at pH 7.2. For the latter case, the gas phase Gibbs free energy of ap roton is À6.3 kcal mol À1 and the experimental solvation free energy of the proton (À264.0 kcal mol À1 )w as used. [29] These values were also used for calculating the absolute pK a .

Acknowledgements
Financials upport from the Knut and Alice Wallenberg Foundation, the Swedish Research Council,t he Carl Trygger Foundation and the Swedish EnergyA gency is gratefully acknowledged.