Journal of Catalysis 315 (2014) 25–32

Contents lists available at ScienceDirect

Journal of Catalysis

journal homepage: www.elsevier .com/locate / jcat

A fast metal–metal bonded water oxidation catalyst

http://dx.doi.org/10.1016/j.jcat.2014.04.0100021-9517/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding authors. Addresses: Departament de Química Física i Inorgànica,Universitat Rovira i Virgili, Marcel.lí Domingo s/n, E-43007 Tarragona, Spain. Fax:+34 977559563 (J. M. Poblet), Department of Chemistry, Texas A&M University,College Station, TX 77843, USA. Fax: +1 9791234567 (K. R. Dunbar), Institute ofChemical Research of Catalonia (ICIQ), Av. Paisos Catalans 16, E-43007 Tarragona,Spain. Fax: +34 977920224 (J.R. Galán-Mascarós).

E-mail addresses: josepmaria.poblet@urv.cat (J.M. Poblet), dunbar@mail.chem.tamu.edu (K.R. Dunbar), jrgalan@iciq.es (J.R. Galán-Mascarós).

1 These authors contributed equally to this paper.

Sara Goberna-Ferrón a,1, Bruno Peña b,1, Joaquín Soriano-López a, Jorge J. Carbó c, Hanhua Zhao b,Josep M. Poblet c,⇑, Kim R. Dunbar b,⇑, José Ramón Galán-Mascarós a,d,⇑a Institute of Chemical Research of Catalonia (ICIQ), Av. Paisos Catalans 16, E-43007 Tarragona, Spainb Department of Chemistry, Texas A&M University, College Station, TX 77843, USAc Departament de Química Física i Inorgànica, Universitat Rovira i Virgili, Marcel.lí Domingo s/n, E-43007 Tarragona, Spaind Catalan Institution for Research and Advanced Studies (ICREA), Passeig Lluis Companys, 23, E-08010 Barcelona, Spain

a r t i c l e i n f o

Article history:Received 5 February 2014Revised 8 April 2014Accepted 14 April 2014

Keywords:Water oxidationMetal–metal bonded compoundsRu catalystComputational methods

a b s t r a c t

The metal–metal bonded diruthenium (II, II) tetraacetate compound, [Ru2(l-O2CCH3)4], is a water oxida-tion catalyst (WOC) that exhibits oxygen evolution rates as high as 77 s�1 at ambient conditions, exhib-iting comparable catalytic activity over a wide pH range (1 < pH < 10). The turnover conditions of thecatalyst remain active for several hours with oxidative deactivation eventually occurring at a slow rate.DFT calculations support a single-site mechanism via water nucleophilic attack, with the most energydemanding process being the very first one electron oxidation step. This is one of the fastest WOCsreported, and the first WOC found in the field of metal–metal bonded chemistry.

� 2014 Elsevier Inc. All rights reserved.

1. Introduction catalysts for the production of solar fuels is vital if solar technology

Dinuclear metal–metal bonded species constitute a large familyof transition metal complexes whose flexibility in tolerating adiverse array of ligands allows for exquisite control of their redoxand electronic properties [1]. Over the course of many decades,metal–metal bonded molecules and materials with specific struc-tural and physical properties have played prominent roles in diversefields ranging from catalysis [2–4], crystal engineering [5], magne-tism [6–10], spectroscopy [11,12] and anticancer drugs [13–15].

From the perspective of the current work, it is important to notethat stable multielectron chemistry of the late transition metal com-plexes has been exploited for numerous catalytic applicationsincluding C–H activation [16], halogen photoelimination [17–19],organometallic transformations [20], photocatalytic hydrogen pro-duction [21,22] and oxygen reduction [23,24]. The latter two pro-cesses are related to the production of solar fuels which are one ofthe most attractive alternatives to fossil fuels for environmentallyfriendly energy solutions. The development of robust and efficient

is to be widely implemented. The lack of a suitable water oxidationEq. (1) catalyst is one of the main bottlenecks to achieving this goal.

2H2O! O2 þ 4Hþ þ 4e� E ¼ 1:229� 0:059pH vs NHE ð1Þ

Many transition metal complexes have been reported as homo-geneous water oxidation catalysts (WOCs) [25–43], but the state ofthe art is defined by single-ion or dinuclear Ru complexes sup-ported by N-heterocycles with pyridyl or imidazole groups [44–48]. In light of the importance of Ru(II) chemistry in this field,we rationalized that metal–metal bonded diruthenium complexescould offer some advantages over their mononuclear counterpartsif they were found to exhibit activity as WOCs. These species areeasily stabilized by available inexpensive ligands such as carboxyl-ates. In addition, the Ru@Ru double bond in the paddlewheel tetra-carboxylate compounds based on the Ru2

4+ core with the electronicconfiguration r2p4d2d*2p*2 increases in strength upon oxidationsince antibonding orbitals are being depopulated [1]. With theseissues in mind, we set out to explore the catalytic activity of[Ru2(l-O2CCH3)4] (Ru2, Fig. 1). Herein, we report the results ofwater electrolysis experiments that lead to oxygen evolution.

2. Experimental

2.1. Synthesis

All reagents were commercially available (>99.9%, Sigma–Aldrich) and used without further purification. [Ru2(l-O2CCH3)4]

Fig. 1. Schematic representation of the molecular structure of [Ru2(l-O2CCH3)4].The acetate ligands are bound to the Ru2

4+ core in the equatorial positions; L denotesopen axial positions for solvent or anion binding.

26 S. Goberna-Ferrón et al. / Journal of Catalysis 315 (2014) 25–32

(Ru2) and Ru2(l-O2CCH3)4Cl were prepared according to literaturemethods [1,2].

2.2. Electrochemistry

All experiments were performed with a Biologic SP-150 poten-tiostat. Cyclic voltammetry experiments were performed with a0.07-cm2 glassy carbon button working electrode, a Pt wirecounter electrode and a Ag/AgCl (NaCl 3 M) reference electrodein 1 M NaNO3 milli-Q water, with 50 mM sodium phosphate(NaPi) buffer. The working electrodes were polished with 3 lmand 1 lm diamond paste and rinsed with distilled water priorto use. Cyclic voltammograms were collected in the 0–2.0 Vrange (vs normal hydrogen electrode, NHE) at different scanrates between 1 and 100 mV/s and different Ru2 concentrationsbetween 0.1 and 2 mM.

Bulk water electrolyses were carried out with stirring in a two-chamber cell, with a porous frit connecting both chambers. A Ptmesh counter electrode was placed in one chamber. The otherchamber contained a FTO-coated glass working electrode (Pilking-ton NSG TEC 15A 2.3 mm slides with 12–14 X/sq. surface resistiv-ity) and a BASi Ag/AgCl (NaCl 3 M) reference electrode. FTO slideswere cleaned prior use by sonication for 10 min in alkaline soapysolution (LABWASH), deionized water and isopropanol, followedby annealing at 400 �C for 30 min. Typical electrolysis experimentswere carried out in a 50 mM NaPi buffer (pH 7) with 1 M NaNO3.Compensation for ohmic drop was applied by using the positivefeedback compensation implemented in the instrument. Oxygenevolution was determined with an Ocean Optics NeoFOX oxygensensing system equipped with a FOXY probe inserted into thehead-space of the anodic compartment.

2.3. Electrochemical quartz crystal microbalance

Simultaneous bulk water electrolysis and piezoelectric gravim-etry measurements were collected with an SRS QCM-200 microbal-ance and a Biologic SP-150 potentiostat. Both instruments werecomputer-controlled with the BioLogic EC-Lab software. A conven-tional three-electrode cell configuration was used, with a platinumwire as counter electrode, and a BASi Ag/AgCl (NaCl 3 M) as the ref-erence. The working electrode was a 25.4 mm diameter quartzcrystal (5 MHz, AT-cut, plano-plano, provided by MicroVacuumLtd.) coated with a ITO conducting surface (approximate electroac-tive surface area: 1.4 cm2). The quartz crystal disk was mounted inthe SRS QCM-200 quartz crystal holder. The frequency response ofthe quartz crystal was monitored before and during experiments.The frequency response was calibrated by electrodeposition ofmetallic Cu from a 1 mM CuCl2 water solution to calculate theequivalence factor in order to transform frequency variations todeposited mass.

2.4. Methods

Scanning electron microscopy (SEM) and electron dispersive X-ray spectroscopy (EDX) analysis were performed with a JEOL-JSM6400 microscope equipped with an Oxford EDX analyzer(Oxford Instruments). Mass spectrometry measurements weremade using an Omnistar TM GSD 301 C (Pfeiffer) quadrupole massspectrometer apparatus. All solutions were degassed prior to use.The composition of the gas phase (O2 m/z = 32; N2 m/z = 28 andCO2 m/z = 44) was measured during controlled potential electroly-sis at 1.5 V (vs NHE) using a FTO-coated glass working electrode. Atwo compartment H-cell separated by a frit was used for theseexperiments. The working compartment contained the FTO-coatedglass working electrode, the Ag/AgCl (NaCl 3 M) reference elec-trode and the MS-spectrometer cannula and was filled with20 mL of a Ru2 (0.5 or 1 mM) solution in 50 mM NaPi buffer with1 M NaNO3 as the electrolyte. The second compartment of the elec-trochemical cell contained the platinum mesh auxiliary electrodeand was filled with 20 mL of buffer. The solutions were cappedwith a septum and stirred vigorously.

2.5. Computational methods

Calculations were performed with the Gaussian09 programpackage [49]. All geometries were fully optimized in the gas phaseat the M06-L level [50–52] of density functional theory. The Stutt-gart ECP28MWB contracted pseudopotential basis set [53] wasused on Ru and the 6-31G(d) basis set [54–57] on all other atoms.The nature of all stationary points was verified by analytical com-putation of vibrational frequencies, which were also used for calcu-lation of free energy contributions. The determination of energeticsrequired single-point calculations in the solvent phase employingthe SMD continuum solvation model [58], in order to account theeffects associated with water.

3. Results and discussion

3.1. Electrocatalytic water oxidation

The electrochemical properties of diruthenium tetracarboxyl-ate-bridged species are highly solvent dependent [59]. The pres-ence of coordinating anions or neutral ligands in the axialpositions also exerts a strong influence on the redox potentials.In organic solvents, Ru2 typically exhibits a single Ru2

+5/+4 revers-ible couple in the 0.2–0.5 V range vs NHE [60–62] (all potentialsin this work are referenced to NHE). A perusal of the literaturerevealed that the aqueous electrochemistry of Ru2 has not beenwell investigated, although in one report it was described to beanalogous to that observed in organic solvents [63]. We foundquite different behavior (Fig. 2, left, a). At pH 1 (dilute sulfuric acidsolution), no distinct reversible redox processes were evident inthe cyclic voltammograms of Ru2 (0.5 mM) but a feature indicativeof a catalytic water oxidation wave above 1.40 V was present.When the sweep was extended to cathodic potentials, a broad irre-versible peak was observed below – 0.3 V (Fig. 2, left) which corre-sponds to the irreversible O2/O2

� reduction [64]. The featureappears only after successive scans, in support of effective catalyticO2 evolution.

When the pH was slowly increased by addition of NaOH, a quasi-irreversible redox couple appeared above pH 2 with peak potentialsat Eox = 1.24 V and Ered = 1.05 V (Eox–Ered = 200 mV) which isassigned to the Ru2

+5/+4 couple. The large Eox–Ered difference in watersuggests that electron transfer could be associated with protontransfer and the change in going from H2O to OH� in the axial posi-tion stabilizes the Ru2

+5 redox state. The current associated with the

Fig. 2. (left) Cyclic voltammetry scans (100 mV/s) of Ru2 (0.5 mM) as a function of different pH values, in dilute sulfuric acid solutions adding NaOH to adjust the pH. (right)Cyclic voltammetry scans (100 mV/s) of Ru2 (0.5 mM) as a function of different pH values in a 50 mM NaPi buffer (pH 7) with 1 M NaNO3 and adding H3PO4/NaOH to adjustthe pH.

S. Goberna-Ferrón et al. / Journal of Catalysis 315 (2014) 25–32 27

catalytic water oxidation wave also increased, exhibiting a smallshoulder at � 1.6 V. As the pH was increased to pH 9, no significantchanges were observed. Identical results were also obtained byvarying the pH from basic to acid media or vice versa, supportingthe conclusion that the catalyst is unaffected by pH variations.

The aforementioned experiments were repeated in a 50 mMNaPi buffer (pH 7) with 1 M NaNO3 as a function of pH values(Fig. 2, right). Below pH 5, the cyclic voltammograms are analogousto those observed for the buffer-free experiments. Above this pH,however, the Ru2

+5/+4 couple becomes reversible with iox/ired � 1and 75 mV peak separation values that are independent of scanrate (Figs. 2 and S1, Supplementary information). In addition, theshoulder on the water oxidation wave disappears. The reversibilityof the Ru2

+5/+4 couple at neutral pH values in the presence of NaPi

also supports the participation of proton transfer associated withelectron transfer, given that phosphate axial binding would leadto deprotonation of the phosphate group instead of the water mol-ecule. No significant changes were observed for the voltammo-grams recorded at pH > 6.

It is important to note that the cyclic voltammograms revert tothe original irreversible form when the pH is lowered to less than 5by addition of phosphoric acid. The electrochemical data acquiredin the presence of excess phosphoric or sulfuric acid support thecontention that phosphate or sulfate groups do not replace acetateligands in the equatorial positions since identical CVs are obtained.Acetate substitution by highly charged bridging ligands in waterrequires extreme conditions and, most importantly, the electro-chemical properties of sulfate and phosphate bridged dirutheniumcomplexes do not match the present data [65,66].

3.2. Ru2 as a Homogeneous catalyst for bulk water electrolysis

Bulk water electrolysis experiments were performed with aFTO-coated glass working electrode by applying a potential of1.5 V to a well-stirred neutral 50 mM NaPi buffer solution contain-ing 1 mM Ru2 (Fig. 3) in a two-chamber cell to separate the reduc-tion and oxidation electrodes. Oxygen evolution was monitored inthe oxidation compartment with the use of a fluorescence probe.

The oxygen levels match the expected yield calculated from Fara-day’s law, an indication of quantitative efficiency for the catalyticprocess (Fig. 3, right). This result confirms that current densitymeasurements can be directly used to quantify catalyticperformance.

During electrolysis, a slow but constant decrease in current sug-gests catalyst decomposition. After 10 h of continuous electrolysis,the rate reaches about 10% of the original value. Mass spectromet-ric measurements were used to detect CO2 in the gas phase in addi-tion to the dominant O2 content (Fig. S2, Supplementaryinformation). The CO2 fraction is very small but increases withtime, albeit at a much slower rate than oxygen evolution. Thesedata suggest that deactivation of the catalyst is occurring via ace-tate oxidation. Decomposition eventually leads to a black powderthat collects at the bottom of the reactor, which is a ruthenium-containing compound. Our attempts to characterize this decompo-sition product were unsuccessful. Neither X-ray powder diffractionnor infrared spectroscopy experiments delivered significant fea-tures, suggesting it is a mixture of amorphous materials. EDX anal-ysis confirmed that Ru is the only metal ion in this powder. It isimportant to point out, however, that this black solid formedin situ is not participating in the catalysis as supported by severalkey facts: (i) the catalytic activity decreases while at the same timethe solid particles become more abundant; (ii) the electrode is notcoated with any residue as determined by EDX (Fig. S3, Supple-mentary information); and (iii) the ‘‘as used’’ electrode exhibitsno catalytic activity for water electrolysis in a Ru2-free solution.These last two points demonstrate that the catalytic process ishomogeneous, since heterogeneous electrocatalysis would requiredeposition of the active species on the electrode.

Still, we carried out two additional control experiments. Firstwe performed bulk water electrolysis under the same experimen-tal conditions but with RuCl3 (2 mM), instead of Ru2. Negligiblecatalytic activity is observed, confirming that Ru oxide does notform in these conditions (Fig. S4, Supplementary information).Additionally, we performed bulk water electrolysis with an electro-chemical quartz crystal microbalance (EQCM). EQCM allows tomonitor mass changes at the surface of the working electrode

Fig. 3. (left) Evolution of the current density during bulk water electrolysis under an applied 1.5 V potential in a 50 mM NaPi buffer (pH 7) with 1 M NaNO3 containing 1 mMof Ru2; (right) O2 production measured with a fluorescence sensor (red line) and theoretical production calculated for a quantitative Faradaic efficiency (black line). The arrowindicates the termination of electrolysis. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

28 S. Goberna-Ferrón et al. / Journal of Catalysis 315 (2014) 25–32

during the catalytic process (Fig. 4). Significant participation of aheterogeneous catalyst formed during water oxidation wouldrequire its deposition on the electrode surface owing to the lowefficiency of electron transfer between colloidal nanoparticlesand the electrode, especially for multielectron processes. Wedetected a very slow deposition of a solid on the electrode, reach-ing about 2.5 lg after 4 h (Fig. 4a). Such small mass (<2 lg/cm2)cannot account for the catalytic activity observed. Furthermore,the participation of this electrodeposited product in the catalyticprocess can be ruled out. This is confirmed by indirect evidence:current intensity does not increase with time, when the depositbecomes more abundant, and maximum intensity appears at thebeginning of the process before any significant solid depositappears. But also by a direct evidence: when we use this partiallycovered working electrode to perform water electrolysis in a elec-trolyte solution without the presence of Ru2, the current intensitydrops dramatically to intensities even below those found for aclean electrode (Fig. 4b). Thus, the solid deposit is actually decreas-ing the active surface of the electrode and diminishing the oxygenevolution rates. The catalytic current is rapidly recovered whenRu2 is added again to the solution (Fig. 4c and d), and when thesolid deposit does not change in mass. These cycles can be repeated

Fig. 4. Evolution of the current intensity (black line) and of the deposited mass onthe working electrode (red line) during bulk water electrolysis under an applied1.5 V potential in a 50 mM NaPi buffer (pH 7) with 1 M NaNO3. The arrows indicatewhen the working electrode is changed from a solution containing 1 mM of Ru2 (aand c) to a buffer and electrolyte-only solution (b and d), and vice versa. Dotted lineindicates the current found for a clean electrode in the same electrolysis conditionsin the absence of any catalyst. (For interpretation of the references to color in thisfigure legend, the reader is referred to the web version of this article.)

with consistent results supporting the homogenous catalysis, andthe inactivity of any heterogeneous species formed during theturnover conditions. This also suggests that Ru oxide is not actuallypresent in the decomposition product, since some catalytic activityshould be expected if that were the case.

Water electrolysis experiments at different pH values were per-formed at a constant applied potential (1.5 V) with constant stir-ring using a glassy carbon working electrode in a sodiumphosphate NaPi buffer obtained by addition of aqueous solutionsof H3PO4 or NaOH to a neutral 50 mM NaPi buffer solution(Fig. 5). Current density data indicated very little pH dependenceand maintained the same magnitude in the 1 < pH < 10 range.Analogous results were obtained by lowering or increasing thepH value in the 1 < pH < 10 range. This is a unique finding amonghomogeneous WOCs given that pH is typically a key parameterregarding activity and also stability with very strict pH controlrequirements being operative. These data also indicate that Ru2

achieves limiting catalytic rates at low pH values since, althoughoverpotential increases with pH, this effect does not significantlyalter the reaction kinetics.

One final important point is that all experiments described inthis work yield identical results when the mixed-valence Ru2

5+

derivative Ru2(l-O2CCH3)4Cl is used, a result that is in accord withthe fact that the initial oxidation state of the complex does notaffect its catalytic activity or electrochemical behavior.

Fig. 5. Plot of the pH dependence of the current density for bulk water electrolysisunder an applied 1.5 V potential for a solution containing 0.50 mM of Ru2.

Fig. 7. Current density as a function of overpotential for water oxidation under anapplied 1.5 V potential in a 50 mM NaPi buffer (pH 7) with 1 M NaNO3 containing0.5 mM of Ru2.

S. Goberna-Ferrón et al. / Journal of Catalysis 315 (2014) 25–32 29

3.3. Catalysis kinetics

The water oxidation process is pseudo-first order with respectto the catalyst concentration. The intensity of the catalytic wavevs [Ru2] shows a linear behavior up to 1.1 mM (Fig. 6, left) follow-ing Eq. (2):

ic ¼ ncatFA½Ru2�ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffikcatDcat

pð2Þ

where ic is the catalytic peak current, ncat is the number of electronsin the redox reaction (ncat = 4), A is the area of the electrode, kcat isthe catalytic reaction constant and D is the diffusion coefficient ofthe catalyst.

The reversible Ru2+5/+4 couple observed in phosphate buffer

allows for the use of the Randles–Sevcik equation:

ip ¼ ð2:687� 105Þn3=2p m1=2D1=2

cat A½Ru2� ð3Þ

where ip is the peak current for the non-catalytic process, n is thenumber of electrons in the redox reaction (np = 1 for the Ru2

+5/+4 cou-ple), and m is the scan rate. Combining Eqs. (2) and (3), one obtainsan expression independent of D (Eq. (4)) and kcat (TOF) can bedirectly estimated from the slope in the ic/ip vs m�1/2 plot.

ic

ip¼ 0:359

ncat

n3=2p

ffiffiffiffiffiffiffiffikcat

m

rð4Þ

Multiple cyclic voltammograms were recorded in a 50 mM NaPi

buffer (pH 7) with 1 M NaNO3 and 0.85 mM Ru2 as a function ofscan rate between 2 and 20 mV s–1 (Fig. 6, right). The ic/ip vs m–1/2

plot is linear below 10 mV s–1 yielding a limiting TOF = 77 s–1.Current density (j) was measured for 10 min of water electroly-

sis in a 50 mM NaPi buffer (pH 7) with 1 M NaNO3 and 0.5 mM Ru2

under stirring at different potentials using a glassy carbon workingelectrode (Fig. 7). Significant catalytic current appears above 1.3 V,which corresponds to an overpotential (g of 500 mV. The j valueincreases exponentially as a function of g and the logarithmic plotdeviates from linearity above g = 1.0 V reaching a maximumj > 3 mA cm�2 (Fig. 7). Such high current density is not typical forhomogeneous catalysts and confirms the fast kinetics of Ru2. Theslope of the linear fit is greater than 300 mV/decade and such avalue for the Tafel slope suggests that the apparent rate-limitingstep is neither electrochemical nor chemical, and that experimen-tal conditions are affecting the catalytic performance. This situa-tion may be due to the influence of bubble formation, inefficienthomogeneity or local defects as noted in other electrochemical cat-alysts [67,68].

Fig. 6. (left) Blank-corrected cyclic voltammograms of a 50 mM NaPi buffer (pH 7) with 1shows the linear relationship between catalytic current and [Ru2]. (right) Blank-correctedof a 50 mM NaPi buffer (pH 7) with 1 M NaNO3 containing 0.85 mM Ru2. The inset show

3.4. Theoretical investigations on the water oxidation mechanism

To address the structural and electronic modifications of[Ru2(l-O2CCH3)4(H2O)2]+ catalyst (S0) during the sequence of theWOC process, we have performed DFT calculations. Some mecha-nistic studies on dinuclear Ru complexes suggest the presence ofintramolecular OAO formation [44] or cooperative effects betweenadjacent ruthenium sites [69]. In this case, however, we canneglect them because geometrical constrains do not allow to alignreacting ligands in parallel; and bimolecular pathways can be dis-carded by the first-order kinetics respect to catalyst concentration.Thus, we foresee a mechanistic scheme resembling that recentlyproposed for different ruthenium single-site catalysts that is illus-trated in Fig. 8 [70–74]. Roughly, the mechanism consist of (i) asequence of proton coupled electron transfer (PCET) events to yieldthe active species RuVAO; (ii) nucleophilic attack of a water mole-cule to form the OAO bond; and (iii) release of oxygen after addi-tional PCET event. In order to investigate the intermediatesinvolved in the above mentioned process, we adopted the compu-tational level by Cramer and Gagliardi [72], and the energeticscheme by Voorhis: [74]

E0 ¼ 1F

DG0ðgÞ þ DGsolv � nHþG0Hþ

ðaqÞ

� �� 4:24V

M NaNO3 and different concentrations of Ru2 (between 0.2 and 1.2 mM). The insetnormalized cyclic voltammograms at different scan rates (between 2 and 20 mV/s)s the linear relationship between ic/ip vs m�1/2.

Fig. 9. LUMO representations for species [RuIIIARuIV@O]+ (S2, top) and[RuIIIARuV@O]2+ (S3, bottom).

30 S. Goberna-Ferrón et al. / Journal of Catalysis 315 (2014) 25–32

where F is the Faraday constant, DG0ðgÞ is the free energy change

associated with oxidation in vacuum, DGsolv is the free energy of sol-vation, G0Hþ

ðaqÞ is the standard free energy of a proton in solution [75]for which we adopted a value of -272.2 kcal mol�1 [76], the nega-tive term references the potentials to the NHE (Eabs = �4.24 V withFermi–Dirac statistics) [77].

Fig. 8 and Table 1 summarize the results. Calculations on[Ru2(l-O2CCH3)4(H2O)2]+ (S0) reproduce the experimental struc-tural data [78] and support a quartet ground state [60–62](Table 1). At our computational level, the electronic configurationis a2p4d2d*p*2 with a small energy gap between the p and d orbitals(0.24 eV); the spin density of the three unpaired electrons beingequally shared between the two Ru sites (Table 1). In terms ofthe electronic structure, [RuIIIARuIIIAOH]+ (S1) has a triplet groundstate and [RuIIIARuIV@O]+ (S2) a quartet ground state. The calcu-lated redox potentials for the two initial consecutive PCET steps(S0 ? S1 and S1 ? S2) are 1.44 and 1.06 V. Experimentally, afterthe peak potentials of the Ru2

+5/+4 couple, we observed a shoulderat � 1.6 V vs NHE once the catalytic process has started. The com-puted value of the first PCET event at the experimental pH 2(1.32 V vs NHE) fits reasonably well to the observed value, indicat-ing that the shoulder in the cyclic voltammetry might be associ-ated with the oxidation [RuIIIARuIIAOH2]+/[RuIIIARuIIIAOH]+.

These two initial PCET are found as the most energeticallydemanding steps of the catalytic cycle (see Fig. 8), and comparedto those of mononuclear Ru at analogous computational level(ranging from 0.60 to 1.10 V) [70,72], they are significantly higher.If we take a closer look to the electronic structures of S0 and S1, weobserve a significant electron spin density delocalization throughthe RuARu bond (Table 1). Thus, the oxidation state of the activeRu atoms of S0 and S1 should not be regarded as 2+ and 3+ butsomewhere between 2+ and 3+, and 3+ and 4+, respectively. Con-sequently, removing one electron from a Ru site is more energeti-cally costly because the electron is stabilized by the adjacent Rusite through the metal–metal bonding. On the contrary, reachingthe active species RuVAO (0.88 V) is lesser energetically demand-ing than in mononuclear Ru complexes (ranging from 1.32 to1.48 V) [70,72]. As we describe in detail below when the RuV-oxospecies forms, the RuARu bond can elongate, thus releasing thetrans effect on the oxo group.

Fig. 8. Schematic representation of the water oxidation by Ru2(l-O2CCH3)4(H2O)2]+ cata

The double bond between the RuV and the O atom in S3 impedesthe formation of a sigma RuARu bond leading to a ground statewith four unpaired electrons, three in the RuIII and one in theRuV. For this structure, we have computed the quintet and the trip-let states, the latter being 2.2 kcal�mol�1 lower in energy. In bothstates, the RuARu distance has been computed rather long,2.60 Å for the triplet, and 2.79 Å for the quintet; values that suggest

lyst based on DFT calculations. The potentials reported are relative to NHE at pH 0.

Table 1Electronic and structural data for species involved in the WOC mechanism.

Species Conf Spin q dRuARu dRu2AOa q(Ru1)a q(Ru2)a

S0 d5–d6 3/2 +1 2.29 (2.25) c 2.34 (2.27)c 1.48 1.48S1 d5–d5 1 +1 2.32 1.92 1.27 0.65S2 d5–d4 3/2 +1 2.42 1.70 1.66 0.94S3

b d5–d3 2 +2 2.78 1.70 2.58 0.76S3

b d5–d3 1 +2 2.60 1.69 2.30 0.66S4 d5–d5 1 +1 2.32 2.04 1.69 0.68S5 d5–d4 3/2 +1 2.33 2.19 1.34 0.53

a Ru1 and Ru2 refer to non-active and active catalytic side, respectively.b For structure S3, the triplet ground state was computed to be 2.2 kcal mol�1

below the lowest quintet.c Values in parenthesis correspond to X-ray bond distances.

S. Goberna-Ferrón et al. / Journal of Catalysis 315 (2014) 25–32 31

the absence of a direct metal–metal bond (or with a very weakinteraction).

Fig. 9 shows the LUMO in the two formal oxo species,[RuIIIARuIV@O]+ (S2) and [RuIIIARuV@O]2+ (S3). These are theacceptor orbitals for the nucleophilic attack of water [27]. In S3,the LUMOs show a significant contribution of the RuV-oxop-antibonding interaction, in which the weights of the p(O) anddxz(Ru) atomic orbitals are, respectively, 62% and 25% for thealpha component. In the corresponding orbital for S2, thecontribution of p(O) is low (11%) and the MO orbital being furtherpolarized toward the metal with 69% of dxz(Ru). In addition, thebeta is mainly a non-bonding dxy Ru-type orbital perpendicular tothe Ru-aqua axis, whereas the p�RuAO orbital has became theLUMO+3. Therefore, S2 is expected to be less reactive towardnucleophiles as previous DFT calculations have already shown formononuclear Ru complexes [73]. Similarly, we have recentlycorrelated the electrophilicity of nitrides with the electronicstructure of the frontier orbitals [79]. The water attack results in atriplet hydroperoxo intermediate [RuIIIARuIIIAOOH]+ (S4). Next, aPCET leads to the formation of a quartet [RuIIIARuIVAOO]+ (S5) thatreleases molecular oxygen upon water displacement to regenerateS0, closing the catalytic cycle. These last steps are less energeticallydemanding indicating that once the OAO bond is formed, thereaction proceeds smoothly.

4. Summary

The results of this study reveal that metal–metal bonded diru-thenium complexes are viable water oxidation catalysts. The Ru2

paddlewheel compound catalyzes water electrolysis over a widepH range (1–10) and at high catalytic rates with quantitative effi-ciencies, reaching current densities over 1 mA cm–2 in dilute solu-tions. The advantage of this approach is that the metal–metal bondplatform allows for the use of common ligands such as acetate todevelop new WOCs without the need for complex organic chelat-ing agents. Among Ru-containing WOCs, the Ru-paddlewheel com-plex is one of the fastest reported with an estimated TOF = 77 s–1.Most Ru catalysts achieve rates well below 1 s–1, with the notableexception being the family of complexes containing the 6,60-dicar-boxylate-2,20-dipyridyl ligand (TOF > 300 s–1).12c,f Moreover, fewtransition metal WOCs are capable of reaching these rates apartfrom some copper12l,p and cobalt examples.12b,m However, in thelatter cases, unlike the present one, the catalysts require strictpH conditions. Ultimately, slow oxidative deactivation due to theinstability of the acetate ligands were observed at high potentialsbut such decomposition can be avoided by improving metal–metalbond stability by employing different bridging ligands as well asother transition metals to tune the performance. This simpleproof-of concept example paves the way for further developmentof an attractive new family of fast water oxidation electrocatalysts.

DFT calculations at M06-L/SMD level were carried to investigatethe intermediates involved in the plausible mechanism for wateroxidation catalyzed by [Ru2(l-O2CCH3)4(H2O)2]+. We have foundthat the initial PCET event emerges as the most energetic demand-ing step, which has been estimated to happen at a redox potential<1.5 V (vs NHE), in very good agreement with experimental results.In this step, the presence of a direct RuARu bond likely induces theobserved initial overpotential. Nevertheless, it is also responsiblefor the lower potentials needed in the next steps that become moreand more favored in the water oxidation cycle, what supports thevery fast kinetics found. The RuV-oxo group of S3 is prone to nucle-ophilic attack of an external water molecule yielding the formationof the OAO bond; the oxidation of RuIV- to RuV-oxo being energet-ically simple. It is worth mentioning that the diruthenium complexcan act as single-site catalyst with the adjacent Ru behaving as anauxiliary ligand that tunes reactivity providing a remarkably activecatalyst.

Acknowledgments

We thank the financial support of the EU (ERC Stg Grant CHEM-COMP), Spanish Ministerio de Economía y Competitividad (GrantsCTQ2012-34088 and CTQ2011-29054-C02-01), the Generalitat deCatalunya (2009SGR462), and the ICIQ Foundation. JSL thanks ICIQand URV for a pre-doctoral fellowship. KRD is grateful to the WelchFoundation of Texas (A-1449) for support of this work in the formof summer financial support for BP.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jcat.2014.04.010.

References

[1] F.A. Cotton, C.A. Murillo, R.A. Walton (Eds.), Multiple Bonds Between MetalAtoms, third ed., Springer, US, New York, 2005.

[2] M.P. Doyle, R. Duffy, M. Ratnikov, L. Zhou, Chem. Rev. 110 (2010) 704–724.[3] J. Hansen, H.M.L. Davies, Coord. Chem. Rev. 252 (2008) 545–555.[4] D.J. Timmons, M.P. Doyle, Chiral Dirhodium(II) Catalysts and Their

Applications (Chapter 13 in ref. 1), 2005[5] F.A. Cotton, C. Lin, C.A. Murillo, Acc. Chem. Res. 34 (2001) 759–771.[6] H. Masaka, R. Clérac, C.S. Campos-Fernández, K.R. Dunbar, Inorg. Chem. 40

(2001) 1663–1671.[7] H. Miyasaka, Y. Asai, N. Motokawa, K. Kubo, M. Yamashita, Inorg. Chem. 49

(2010) 9116–9118.[8] N. Motokawa, S. Matsunaga, S. Takaishi, H. Miyasaka, M. Yamashita, K.R.

Dunbar, J. Am. Chem. Soc. 132 (2010) (1951) 11943–11951.[9] H. Miyasaka, T. Izawa, N. Takahashi, M. Yamashita, K.R. Dunbar, J. Am. Chem.

Soc. 128 (2006) 11358–11359.[10] H.T. Chifotides, K.R. Dunbar, Rhodium Compounds (Chapter 12 in ref. 1), 2005.[11] K. Kadish, L. Wang, A. Thuriere, E. Van Caemelbecke, J. Bear, Inorg. Chem. 42

(2003) 834–843.[12] D. Baxter, M. Chisholm, M. Lynn, E. Putilina, S. Trzaska, T. Swager, Chem. Mater.

10 (1998) 1758–1763.[13] K. Mashima, M. Tanaka, Y. Kaneda, A. Fukumoto, H. Mizomoto, K. Tani, H.

Nakano, A. Nakamura, T. Sakaguchi, K. Kamada, K. Ohta, Chem. Lett. 26 (1997)411–415.

[14] J.D. Aguirre, A.M. Angeles-Boza, A. Chouai, J.-P. Pellois, C. Turro, K.R. Dunbar, J.Am. Chem. Soc. 131 (2009) 11353–11360.

[15] H.T. Chifotides, K.R. Dunbar, Acc. Chem. Res. 38 (2005) 146–156.[16] M.P.J. Doyle, Org. Chem. 71 (2006) 9253–9260.[17] A.F. Hayduk, A.M. Macintosh, D.G. Nocera, J. Am. Chem. Soc. 121 (1999) 5023–

5032.[18] T.R. Cook, A.J. Esswein, D.G. Nocera, J. Am. Chem. Soc. 129 (2007) 10094–

10095.[19] T.S. Teets, D.A. Lutterman, D.G. Nocera, Inorg. Chem. 49 (2010) 3035–3043.[20] A.J. Esswein, A.S. Veige, P.M. Piccoli, A.J. Schultz, D.G. Nocera, Organometallics

27 (2008) 1073–1083.[21] A.J. Esswein, A.S. Veige, D.G. Nocera, J. Am. Chem. Soc. 127 (2005) 16641–

16651.[22] A.F. Heyduk, D.G. Nocera, Science 293 (2001) 1639–1641.[23] T.S. Teets, D.G. Nocera, J. Am. Chem. Soc. 133 (2011) 17796–17806.[24] T.S. Teets, T.R. Cook, B. McCarthy, D.G. Nocera, J. Am. Chem. Soc. 133 (2011)

8114–8117.

32 S. Goberna-Ferrón et al. / Journal of Catalysis 315 (2014) 25–32

[25] X. Liu, F.Y. Wang, Coord. Chem. Rev. 256 (2012) 1115–1136.[26] D.J. Wasylenko, R.D. Palmer, C.P. Berlinguette, Chem. Commun. 49 (2013) 218–

227.[27] Y. Jiang, F. Li, B. Zhang, X. Li, X. Wang, F. Huang, L. Sun, Angew. Chem. Int. Ed.

52 (2013) 3398–3401.[28] J.J. Concepcion, J.W. Jurss, M.K. Brennaman, P.G. Hoertz, A.O.T. Patrocinio, N.Y.

Murakami, J.L. Iha, J. Templeton, T.J. Meyer, Acc. Chem. Res. 42 (2009) 1954–1965.

[29] J.D. Blakemore, N.D. Schley, D. Balcells, J.F. Hull, G.W. Olack, C.D. Incarvito, O.Eisenstein, G.W. Brudvig, R.H. Crabtree, J. Am. Chem. Soc. 132 (2010) 16017–16029.

[30] L. Duan, F. Bozoglian, S. Mandal, B. Stewart, T. Privalov, A. Llobet, L. Sun, Nat.Chem. 4 (2012) 418–423.

[31] R. Lairempuia, N.D. McDaniel, H. Muller-Bunz, S. Bernhard, M. Albrecht,Angew. Chem. Int. Ed. 49 (2010) 9765–9768.

[32] J. Lloret-Fillol, Z. Codolé, I. García-Bosch, L. Gómez, J.J. Pla, M. Costas, Nat.Chem. 3 (2011) 807–813.

[33] W.C. Ellis, N.D. McDaniel, S. Bernhard, T.J. Collins, J. Am. Chem. Soc. 132 (2010)10990–10991.

[34] N.S. McCool, D.M. Robinson, J.E. Sheats, G.C. Dismukes, J. Am. Chem. Soc. 133(2011) 11446–11449.

[35] D.J. Wasylenko, C. Ganesamoorthy, J. Borau-Garcia, C.P. Berlinguette, Chem.Commun. 47 (2011) 4249–4251.

[36] S.M. Barnett, K.I. Goldberg, J.M. Mayer, Nat. Chem. 4 (2012) 498–502.[37] Q. Yin, J.M. Tan, C. Besson, Y.V. Geletii, D.G. Musaev, A.E. Kuznetsov, Z. Luo, K.I.

Hardcastle, C.L. Hill, Science 328 (2010) 342–345.[38] H. Lv, Y.V. Geletii, C. Zhao, J.W. Vickers, G. Zhu, Z. Luo, J. Song, T. Lian, D.G.

Musaev, C.L. Hill, Chem. Soc. Rev. 41 (2012) 7572–7589.[39] J.W. Vickers, H. Lv, J.M. Sumliner, G. Zhu, Z. Luo, D.G. Musaev, Y.V. Geletii, C.L.

Hill, J. Am. Chem. Soc. 135 (2013) 14110–14118.[40] A. Sartorel, M. Carraro, G. Scorrano, R. De Zorzi, S. Geremia, N.D. McDaniel, S.

Bernhard, M. Bonchio, J. Am. Chem. Soc. 130 (2008) 5006–5007.[41] S. Goberna-Ferron, L. Vigara, J. Soriano-López, J.R. Galan-Mascaros, Inorg.

Chem. 51 (2012) 11707–11715.[42] M.-T. Zhang, Z. Chen, P. Kang, T.J. Meyer, J. Am. Chem. Soc. 135 (2013) 2048–

2051.[43] J. Soriano-Lopez, S. Goberna-Ferron, L. Vigara, J.J. Carbo, J.M. Poblet, J.R. Galan-

Mascaros, Inorg. Chem. 52 (2013) 4753–4755.[44] C. Sens, I. Romero, M. Rodriguez, A. Llobet, T. Parella, J. Benet-Buchholz, J. Am.

Chem. Soc. 126 (2004) 7798–7799.[45] F. Bozoglian, S. Romain, M.Z. Ertem, T.K. Todorova, C. Sens, J. Mola, M.

Rodriguez, I. Romero, J. Benet-Buchholz, X. Fontrodona, C.J. Cramer, L.Gagliardi, A. Llobet, J. Am. Chem. Soc. 131 (2009) 15176–15187.

[46] S. Romain, F. Bozoglian, X. Sala, A. Llobet, J. Am. Chem. Soc. 131 (2009) 2768–2769.

[47] J.J. Concepcion, J.W. Jurss, J.L. Templeton, T.J. Meyer, J. Am. Chem. Soc. 130(2008) 16462–16463.

[48] Y. Xu, A. Fischer, L. Duan, L. Tong, E. Gabrielsson, B. Akermark, L. Sun, Angew.Chem. Int. Ed. 49 (2010) 8934–8937.

[49] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato,X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M.Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y.

Honda, O. Kitao, H. Nakai, T. Vreven, J.A. Montgomery Jr., J.E. Peralta, F. Ogliaro,M. Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, R. Kobayashi, J.Normand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M.Cossi, N. Rega, J.M. Millam, M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo,J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C.Pomelli, J.W. Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth,P. Salvador, J.J. Dannenberg, S. Dapprich, A.D. Daniels, O. Farkas, J.B. Foresman,J.V. Ortiz, J. Cioslowski, D.J. Fox, Gaussian09 (Revision A.02), Gaussian, Inc.,Wallingford, CT, 2009.

[50] Y. Zhao, D.G. Truhlar, J. Chem. Phys. 125 (2006) 194101.[51] Y. Zhao, D.G. Truhlar, Acc. Chem. Res. 41 (2008) 157–167.[52] Y. Zhao, D.G. Truhlar, Theor. Chem. Acc. 120 (2008) 215–241.[53] D. rae, U. Haussermann, M. Dolg, H. Stoll, H. Preuss, Theor. Chim. Acta 77

(1990) 123–141.[54] W.J. Hehre, L. Radom, P.v.R. Schleyer, J.A. Pople, Ab Initio Molecular Orbital

Theory, Wiley, New York, 1986.[55] M.M. Francl, W.J. Pietro, W.J. Hehre, J.S. Binkley, M.S. Gordon, D.J. Defrees, J.A.

Pople, J. Chem. Phys. 77 (1982) 3654–3665.[56] W.J. Hehre, R. Ditchfield, J.A. Pople, J. Chem. Phys. 56 (1972) 2257–2261.[57] P.C. Hariharan, J.A. Pople, Theor. Chim. Acta 28 (1973) 213–222.[58] A.V. Marenich, C.J. Cramer, D.G. Truhlar, J. Phys. Chem. B 113 (2009) 6378–

6396.[59] M.A.S. Aquino, Coord. Chem. Rev. 170 (1998) 141–202.[60] F.A. Cotton, E. Pedersen, Inorg. Chem. 14 (1975) 388–391.[61] F.A. Cotton, M. Matusz, B. Zhong, Inorg. Chem. 27 (1988) 4368–4372.[62] A.J. Lindsay, G. Wilkinson, M. Motevalli, M.B. Hursthouse, J. Chem. Soc. Dalton

Trans. (1985) 2321–2326.[63] C.R. Wilson, H. Taube, Inorg. Chem. 14 (1975) 2276–2279.[64] E. Yeager, Electrochim. Acta 29 (1984) 1527–1537.[65] J.P. Collin, J. Abdelaziz, S. Jean Pierre, C.K. William, A.M. Margaret, L.K. Nancy,

T.A.H. William, D.S. Galen, Inorg. Chem. 29 (1990) 2238–2241.[66] F.A. Cotton, T. Datta, L. Labella, M. Shang, Inorg. Chim. Acta 203 (1993) 55–60.[67] N.D. Schley, J.D. Blakemore, N.K. Subbaiyan, C.D. Incarvito, F. D’Souza, R.H.

Crabtree, G.W. Brudvig, J. Am. Chem. Soc. 133 (2011) 10473–10481.[68] J.D. Blakemore, N.D. Schley, G.W. Olack, C.D. Incarvito, G.W. Brudvig, R.H.

Crabtree, Chem. Sci. 2 (2011) 94–98.[69] F. Liu, J.J. Concepcion, J.W. Jurss, T. Cardolaccia, J.L. Templeton, T.J. Meyer,

Inorg. Chem. 47 (2008) 1727–1752.[70] M. Hirahara, M.Z. Ertem, M. Komi, H. Yamazaki, C.J. Cramer, M. Yagi, Inorg.

Chem. 52 (2013) 6354–6364.[71] Z.-L. Lang, G.-C. Yang, N.-N. Ma, S.-Z. Wen, L.-K. Yan, W. Guan, Z.-M. Su, Dalton

Trans. 42 (2013) 10617–10625.[72] L. Vigara, M.Z. Ertem, N. Planas, F. Bozoglian, N. Leidel, H. Dau, M. Haumann, L.

Gagliardi, C.J. Cramer, A. Llobet, Chem. Sci. 3 (2012) 2576–2586.[73] J.J. Concepcion, M.-K. Tsai, J.T. Muckerman, T.J. Meyer, J. Am. Chem. Soc. 132

(2010) 1545–1557.[74] L.-P. Wang, Q. Wu, T.V. Voorhis, Inorg. Chem. 49 (2010) 4543–4553.[75] J.A. Mejias, S. Hamad, S. Lago, J. Phys. Chem. B 105 (2001) 9872–9878.[76] M.D. Tissandier, K.A. Cowen, W.Y. Feng, E. Gunddlach, M.H. Cohen, A.D.

Earhart, J.V. Coe, T.R. Tuttle, J. Phys. Chem. A 102 (1998) 7787–7794.[77] C.P. Kelly, C.J. Cramer, D.G. Truhlar, J. Phys. Chem. B 111 (2007) 408–422.[78] A. Bino, F.A. Cotton, T.R. Felthouse, Inorg. Chem. 18 (1979) 2599–2603.[79] S. Romo, N.S. Antonova, J.J. Carbó, J.M. Poblet, Dalton Trans. (2008) 5166–5172.