Mononuclear ruthenium polypyridine complexes that catalyze water oxidation

Representative mononuclear Ru polypyridine water oxidation catalysts were discussed by organizing them into four groups according to their ligand environments so as to elucidate the correlation between activity, mechanism, and ligand structure.


Introduction
The successful utilization of solar energy as an alternative to fossil fuels relies on the viable conversion of solar energy into 'solar fuels' that can be stored and distributed in a manner similar to fossil fuels. 1 One approach to achieving this conversion is envisioned as an articial photosynthesis (AP) system that mimics the function of the naturally-occurring photosynthetic system. 2 The AP system includes two half reactions. From an electrochemical point of view, these reactions are the anodic water oxidation reaction (eqn (1)) and the cathodic solar fuel generation, such as proton to hydrogen or CO 2 to methanol reduction. When the overall endothermic redox process is driven by sunlight, solar energy is converted into chemical energy in the form of chemical bonds and dioxygen is liberated concurrently. Exothermic oxidation of the solar fuel by dioxygen releases the energy and closes the energy cycle in a carbonneutral way.
The water oxidation reaction (eqn (1)) is ideally suited to either natural or articial photosynthesis because water and dioxygen are the most abundant electron donor (reductant) and acceptor (oxidant) in the world. This oxidation reaction is energy demanding with a standard redox potential of 1.23 V (all redox potentials presented in this paper are versus standard hydrogen electrode, SHE, unless noted otherwise). In nature, water oxidation is catalyzed by the oxygen evolving complex (OEC) of Photosystem II (PS II). 3,4 In an articial system, a water oxidation catalyst (WOC) would be required to lower the energy barrier (DG ‡ ) of activation for this process. This situation can be illustrated by comparing the schematic energy proles of catalyzed and uncatalyzed water oxidation pathways (Fig. 1). Water oxidation is a complex reaction that involves the removal of four electrons and four protons as well as the formation of the O]O bond. Multiple intermediates are likely to be involved in the catalytic pathway. An ideal WOC should avoid high-energy ('too active') and low-energy ('too stable') intermediates that are likely to require large energy barriers of activation. Thus, the rational design of a WOC becomes a task of manipulating critical intermediates throughout the catalytic cycle. This detailed description, in turn, relies on elucidating the critical intermediates and understanding the inuence of structural factors upon their relative energies. In this regard, the study of molecular ruthenium WOCs during the past decade can provide some clues and inspiration. This review will be restricted to homogeneous Ru-based WOCs whose molecular structures are well dened.

Polypyridine ligand platform
Most molecular ruthenium water oxidation catalysts reported so far are based on polypyridine ligands that include the polypyridine backbone and non-pyridine donors such as imidazole or carboxylate (Scheme 1). These ligands were systematically designed and synthesized with careful concern given to their denticity, rigidity, and conjugation as well as the positioning of substituent groups having different steric and electronic effects. In this manner the inuence of ligand features upon catalytic activity can be compared and illustrated. The suitability of polypyridine ligands is not a coincidence as they meet two basic requirements for catalytic water oxidation. Firstly, the pyridine ring is capable of tolerating harsh oxidation conditions and, secondly, the pyridine ring is stable towards hydrolysis.
When coordinated with Ru(II), the major role of pyridine is to provide its lone pair of electrons as a s-donor to the metal center. The pyridine-Ru coordination bond is quite effective and leads to large ligand eld stabilization energy. As a result, Ru complexes with a pyridine coordinating environment prefer a low-spin electronic conguration. Polypyridines chelate with Ru through multidentate sites resulting in the formation of chelate rings. This multi-binding tethers the ligand and Ru rmly enough to resist ligand displacement by water under acidic or alkaline conditions. The polypyridine ligands are generally believed to be redox insensitive when the Ru(II) complexes are oxidized to higher valences.

Proton-coupled electron transfer (PCET) and high valent ruthenium species
The frontier molecular orbital diagram of an octahedral Ru(II) complex with six identical pyridine ligands is shown in Scheme 2. It can be used as a simplied model to analyze related ruthenium polypyridine systems. Removal of one electron from the highest occupied molecular orbital (HOMO) of such a complex demands considerable energy. Oxidation of [Ru(bpy) 3 ] 2+ (bpy ¼ 2,2 0 -bipyridine), for example, occurs at E 1/2 -(Ru III /Ru II ) ¼ 1.26 V in water. 5 Changing one of the pyridine ligands for a water will stabilize the lowest unoccupied molecular orbital (LUMO) to some extent because the aqua ligand is a weaker s-donor than pyridine. Nevertheless, this change does not signicantly inuence the HOMO orbital of the Ru(II) complex (Scheme 2). As suggested in a study of [Ru II (bpy) 2 -(py)(OH 2 )] 2+ (py ¼ pyridine), the standard redox potential of [Ru III (bpy) 2 (py) (OH 2 )] 3+ /[Ru II (bpy) 2 (py)(OH 2 )] 2+ is 1.04 V, lower than but close to that of [Ru(bpy) 3 ] 2+ . 6,7 This complex in its trivalent state, however, is a much stronger Brønsted acid than in the divalent state. The pK a of [Ru III (bpy) 2 (py)(OH 2 )] 3+ and [Ru II (bpy) 2 (py)(OH 2 )] 2+ are 0.85 and 10.20, respectively. 8,9 In the pH range from 0.85 to 10.20, the PCET redox process of [Ru III (bpy) 2 (py)(OH)] 2+ /[Ru II (bpy) 2 (py)(OH 2 )] 2+ becomes dominant with the redox potential depending on pH, according to the Nernst equation. 10 As a result, the Ru(III) state is thermodynamically easier to access at relatively higher pH (>0.85), for instance, 0.68 V at pH ¼ 7.0.
Further oxidation of [Ru III (bpy) 2 (py)(OH)] 2+ leads to [Ru IV (bpy) 2 (py)O] 2+ (pK a < À6.0). 11 Because both the Ru IV -O/ Ru III -OH and Ru III -OH/Ru II -OH 2 redox events occur as a oneproton coupled one-electron transfers, their redox potentials change in parallel depending on pH (0. 85-10.20). Notably, the potential gap is only 0.11 V between these two redox couples. By comparison, the potential difference between Ru IV /Ru III and Ru III /Ru II couples of cis-[Ru II (bpy) 2 Cl 2 ] 2+ is 1.66 V in MeCN. 12 There are two major factors contributing to the dramatically narrow potential gap for aqua ruthenium complexes such as [Ru II (bpy) 2 (py)(OH 2 )] 2+ . One is the involvement of PCET that avoids charge buildup. 13 The other is the interaction between the ruthenium d orbitals (d xz , d yz ) and oxo p orbitals (p x , p y ), which destabilize the HOMO by combining to form the dp-pp bonding and antibonding molecular orbitals. 14 The p-bonding orbitals (not shown in Scheme 2) are centered on the oxo and are lower in energy than the nonbonding (xy) orbital, while the p* orbitals (HOMO, Scheme 2) are centered at the metal and are higher in energy. In this d 4 electronic scenario of Ru IV ]O, the two lone pairs (p x and p y ) of oxygen are partially delocalized on the ruthenium center aer the interaction. Meanwhile, the d xz and d yz orbitals, which are non-bonding before the interaction, partially delocalize onto the oxo ligand, leading to some electron transfer from ruthenium to oxygen.
For ease of reading, formal oxidation states of a ruthenium center are marked in this paper by assuming that all electron transfer processes of complexes are metal-based. This is in accordance with the conventional view regarding transition metal complexes. It should be noted that spin density in principle distributes over the whole molecule of a metal complex and electron transfer may occur primarily at the ligand such as oxo ligand (see below).

The O-O bond formation
How the O-O bond is formed is a vital aspect of the mechanism of catalytic water oxidation. Due to the requirement for multiple electron transfers in the water to dioxygen oxidation, Ru intermediates with various valence states have to be involved in the catalytic cycle. High valence (Ru IV or Ru V ) ruthenium oxo species are oen postulated as critical intermediates that trigger O]O bond formation.
There are two general mechanisms for O-O bond formation mediated by Ru-oxo species, according to the origin of the oxygen atoms in the generated dioxygen. In the acid-base mechanism (Scheme 3a), water or hydroxide as a Lewis base attacks the terminal oxo group as a Lewis acid. In the radical coupling mechanism (Scheme 3b), two radical-like Ru-oxo species approach and couple with each other. Therefore the favored pathway partly depends on the dominant resonance contributor between Ru N ]O and Ru (NÀ1) -Oc, such as Ru V ]O and Ru IV -Oc, under the reaction conditions. Kinetically, the two pathways may compete with each other. The essential highvalent Ru-oxo species are usually unstable and have only transient lifetimes in the reaction medium. This short lifetime makes direct characterization and observation of these species difficult. In principle, the two pathways for O-O bond formation can be distinguished experimentally by an 18 O-labeled Ru-oxo or water substrate. Moreover, these two O-O bond formation steps show different kinetic orders in the ruthenium-containing intermediates.

Catalytic activity
The activity of molecular Ru WOCs can be described by overpotential (h) and turnover frequency (TOF). The former refers to the difference between the thermodynamic water oxidation potential and the catalytic potential (E cat ) where an appreciable catalytic current is achieved. 15 The latter is straightforwardly dened as the number of catalytic cycles mediated by each catalyst molecule per unit time. These two descriptors are not independent parameters but are linked to each other, because both are related to the activation energy of the rate-determining step ( Fig. 1) in the catalytic cycle. Savéant and coworkers have developed electrochemical models to quantitatively characterize and analyze the TOF-h relationship for a molecular catalyst. [16][17][18] The denition of E cat , however, is somewhat subjective and this parameter has been determined by cyclic voltammetry according to several different criteria. The potential at the onset, the maximum, or half of the maximum of catalytic current have all been designated as E cat . The different methods for estimation of E cat lead to signicant uncertainty concerning this parameter. Hence, caution should be taken in the direct comparison of catalytic potentials.
The TOF of Ru WOCs can also be evaluated by driving the catalyst with a sacricial oxidant in bulk solution. The reduction potential of the oxidant should be positive enough not only to oxidize water thermodynamically but also to enable access to the highest valent intermediate present in the catalytic pathway. A number of sacricial oxidants have been employed in catalytic water oxidation studies. 19 Among them, ceric ammonium nitrate ([(NH 4 ) 2 Ce IV ](NO 3 ) 6 , CAN) and [Ru(bpy) 3 ] 3+ are most oen used under acidic and neutral conditions, respectively. 5,20,21 Both are one-electron oxidants without O-transfer capability. This ensures that water is the only source of oxygen for O 2 evolution. Because [Ru(bpy) 3 ] 3+ easily decomposes, even in the solid state, it is usually generated in situ through the exposure of [Ru(bpy) 3 ] 2+ to sodium peroxydisulfate and light. 22 Alternatively, high purity CAN is commercially available and can be stored for long periods of time by avoiding moisture. Thus the preparation of a CAN solution with a given concentration is convenient and such a solution under acidic conditions (pH ¼ 1.0) is commonly used in mechanistic studies of WOCs.
A high TOF at low overpotential is always desired for an efficient catalyst. The OEC of PSII is able to achieve a maximum TOF of about 500 s À1 under natural conditions and is oen used as a benchmark for this process. 23 For one mononuclear Ru WOC, a striking TOF of 50 000 s À1 was recently reported under electrocatalytic conditions. 24 A Ru WOC may decompose and lose its activity during catalysis. Therefore, turnover number (TON) is also used to assess the catalytic behavior of WOCs. The TON can be dened as the number of oxygen molecules generated per molecule of catalyst before becoming inactivated. The value of the TON is related to both the efficiency and stability of the catalyst. It should be noted that the determination of TOF and TON is inuenced by methodology and experimental conditions. In electrolysis, for example, the reaction rate may be limited by the diffusion of a substrate to the electrode surface, whereas a reaction in bulk solution is governed by the law of mass action. Therefore, how such descriptors of activity are determined should be provided when the catalytic behavior of different WOCs is compared.

Blue dimer
The so-called "blue dimer" (Fig. 2) was initially prepared and investigated by Meyer and coworkers during the early 1980's. [25][26][27] It is the rst ruthenium complex that was shown to be capable of catalyzing water oxidation. Under pH ¼ 1 conditions, the blue dimer was oxidized from Ru III -O-Ru III to Ru V -O-Ru V at a potential >1.5 V via successive proton-coupled one-,  and three-electron transfer processes (through the Ru III -O-Ru IV state). 26 The resulting [(O)Ru V (m-O)Ru V (O)] 4+ intermediate was believed to trigger the O 2 evolution step. 18 O-Labeling studies suggest a complicated mechanism that includes intra-, and inter-molecular coupling and acid-base types of interaction. 28,29 A kinetic study using CAN illustrates the nucleophilic attack of water on the Ru V ]O center and the formation of a peroxo intermediate as the major catalytic pathway. 11,30 This pathway is also supported by DFT calculations. 31 Research involving the blue dimer has inspired the development of both dinuclear and mononuclear ruthenium WOCs using a variety of polypyridine ligands. The discussion of dinuclear Ru WOCs is outside of the scope of this review and we direct interested readers to related references. 32-37

Mononuclear Ru polypyridine WOCs and their catalytic activity
In the past decade, a growing number of mononuclear Ru polypyridine complexes have been reported to catalyze the water oxidation reaction. Compared with multinuclear Ru WOCs, the mononuclear complexes have simpler structures, better-dened spectroscopic properties, and lend themselves more readily to functional group modication. Such mononuclear Ru WOCs thus provided an excellent opportunity for researchers to gain insight into catalytic pathways from both an experimental and theoretical point of view. It is difficult, however, to establish a straightforward correlation between the activity and specic features of these catalysts, because any given structural or electronic feature may simultaneously inuence multiple steps in the catalytic pathway. In order to discuss these WOCs in a systematic manner, we will classify mononuclear ruthenium polypyridine WOCs into four groups according to their ancillary polypyridine scaffolds: 2 ], and (iv) [Ru(LLL) 2 ] types, where L, LL, LLL, and LLLL represent mono-, di-, tri-, and tetra-dentate N/O-polypyridine ligands, respectively, and X represents an aqua or halogen ligand. Instead of listing all reported Ru WOCs, we chose several representative examples from each group shown in Scheme 4 and concentrated on the differences in their catalytic behavior. For each catalyst, we will concern ourselves with the following three questions: (1)  [Ru II (tpy)(bpy)OH 2 ] 2+ (1a, tpy ¼ 2,2 0 ;6 0 ,2 00 -terpyridine) was studied extensively in the early 1980s'. 43 The Pourbaix diagrams for 1 (black solid line in Fig. 3 2+ redox sequence, depending on whether the pH of the medium is greater or less than the pK a1 (1.7) of the [Ru III - is not deprotonated, redox potentials of 1.04 and 1.23 V are determined for the Ru III/II and Ru IV/III couples, respectively. There is still some controversy about the existence of the Ru V state of 1, 40,45 and no absorbance feature for a Ru V species was observed by mixing 1 equiv. of CAN with the [Ru IV ]O] 2+ form of 1a. 44 Nevertheless, the catalytic activity of 1a toward water oxidation has been independently conrmed by several groups. 21,40,[45][46][47][48] A TOF of 6.1 Â 10 À5 s À1 was observed for 1a in the presence of excess CAN (200 equiv.) under pH 1.0 conditions. 44 The rate of CAN consumption depended on the concentration of 1a but not the concentration of CAN, inferring either k O-O or k O 2 as the ratedetermining step which does not involve the CAN oxidant. Berlinguette and coworkers found that introduction of electrondonating methoxy groups at the 4,4 0 positions of the bpy ligands enhances the catalytic efficiency of 1a. 44 For the modied complex (1b), a TOF of 1.5 Â 10 À4 s À1 was obtained under the same conditions as were applied for 1a, and the rate constants k e and k O-O were measured as 3.7 M À1 s À1 and 3.0 Â 10 À5 s À1 (Table 1) respectively by using stopped-ow techniques. 44 Unlike 1a, the rate of CAN consumption for 1b is rst order relative to both the catalyst and CAN with a rate constant smaller than k e . Therefore, the oxidation of [Ru IV -OO] to [Ru V -OO] was proposed to be the rate-limiting step in the catalytic cycle of 1b. Yagi et al. reported that electron-donating groups on the tpy moiety of 1a also remarkably improve the catalytic performance. 49 Llobet and coworkers found that the uoride substituents at the 6,6 0 positions of bpy ligand of 1a not only perturb the electronic feature but also act as internal base. 50 Complexes 2a-2c are composed of the same polypyridine ligands as 1a but with halogens instead of the aquo ligand in 1a. They show catalytic water oxidation activity in aqueous medium. It is believed that they convert to 1a by dissociation of the halogen ligand in the aqueous environment and the resulting aqua complex 1a plays the role of an authentic catalyst. 45,46 In the model complex [Ru II (tpy)(pynp)OH 2 ] 2+ Scheme 5 Generalized mechanism for water oxidation by Type I Ru catalysts in pH ¼ 1.0 aqueous medium.
Compared to the bpy ligand in 1a, the 2,2 0 -bipyrimidine (bpm) ligand of 3a elevates the redox potential of the Ru III/II couple and reduces the potential of the Ru IV/III couple to such an extent that the former is more positive than the latter. 38 Besides various bidentate ligands, 53 several tridentate ligands in place of tpy have been incorporated into complexes of the [Ru(LLL)(LL)OH 2 ] motif, which are able to catalyze water oxidation. WOC 4, for example, possesses a 2,6-bis(1-methylbenzimidazol-2 0 -yl)pyridine (Mebimpy) ligand that has a stronger s-donating ability than tpy. 40,41 Unlike 1a, complex 4 tentatively undergoes the oxidation of either the [Ru IV ]O] or [Ru IV -OO] intermediate as the rate-determining step. Complex 5 contains a negatively charged 2,2 0 -bipyridine-6-carboxylate (bpc) ligand that can donate lone pair electrons of oxygen to stabilize the high-valent Ru center via pp-dp interaction. 42 The advantage of introducing the anionic carboxylate donor can be understood by comparison of 1a, 3a and 5 (see Fig. 3 and Table  1). For 5 we observed a slight decrease in the potential (1.57 V) but a dramatic increase in the kinetics (k e ¼ 1.

Type II: [Ru(LLL)(L) 2 X] n+ WOCs
Unlike the Type I WOCs discussed in the previous section, Type II complexes bind three monodentate ligands in addition to a tridentate ligand. The aqua ligand, if there is one, always occupies the fourth binding site in the equatorial plane dened by the ruthenium and the tridentate ligand. One of the earliest examples of this group is [Ru II (npm)(pic) 2 OH 2 ] 2+ (6, npm ¼ 4-tbutyl-2,6-di-(1 0 ,8 0 -naphthyrid-2 0 -yl)-pyridine) prepared by Thummel and coworkers in 2005. 33 The single-crystal X-ray structure of 6 shows that the two external 1,8-naphthyridyl nitrogens do not coordinate with Ru II but one of them does form an H-bond with the coordinated water. The higher pK a1 of 6 (>13.5) as compared to 7 (11.2) indicates that the intramolecular H-bond inhibits proton dissociation from the bound water. The pH dependence of the redox potentials of 6 in aqueous solution is summarized in a recent mechanistic study. 55 The pH slope of À59 mV pH À1 in the pH > 2.9 region of the Pourbaix diagram (Fig. 4)  loss of an electron and a proton. While a DFT simulation predicted a similar thermodynamic energy change for these two pathways under standard conditions (pH ¼ 0), the latter pathway is more favored at higher pH since it is a protoncoupled process and the former one is not. 56 Complex 7 has a coordination geometry very similar to 6. However, it does not possess any vacant nitrogen site that can form an H-bond with a bound water. The electrochemical behavior of 7 as displayed in the Pourbaix diagram (Fig. 4) is quite different from that of 6. 57 PCET couples corresponding to [Ru III -OH]/[Ru II -OH 2 ] appeared in a broad region. The pK a1 values of [Ru III -OH 2 ] and [Ru II -OH 2 ] were deduced from the potential/pH relationship as 1.2 and 11.2, respectively. Under acidic and neutral conditions, a prominent catalytic current was observed in the cyclic voltammogram of 7 with an onset that was clearly separated from the Ru III /Ru II redox wave. While it is determined that a ruthenium species of higher oxidation state than Ru III is needed to trigger water oxidation, no redox wave can be distinguished unambiguously for further oxidation of the Ru III intermediate at pH < 10. Thus one can speculate that the [Ru IV ]O] form of 7 is responsible for O-O bond formation in the same fashion as 6. The mechanistic details of 7 have not been elucidated, however, the TOF (1.3-3.7 Â 10 À2 s À1 ) 57,58 of 7 is very close to the TOF (3.2 Â 10 À4 s À1 ) 56 of 6, measured in CANdriven O 2 evolution experiments.
The catalytic activity has been investigated for complexes 8a-8c, in which halogen ligands instead of aqua occupy the equatorial coordination site. 58 A 10-12 min induction period (the concentration of catalyst is 0.04 mM) was observed before 8a and 8b began to catalyze O 2 evolution in the presence of excess CAN. Their TOFs are lower than that of [Ru II (tpy)(pic) 2 OH 2 ] 2+ (7). These observations are consistent with the suggestion that halogen/water exchange is required to generate the authentic WOC 7. On the contrary, the iodide complex [Ru II (tpy)(pic) 2 I] + (8c) catalyzed CAN-driven O 2 evolution without any induction period and achieved a TOF of 0.16 s À1 that is greater than its aqua analog 7. This unusual catalytic performance of 8c suggests a mechanism that involves the iodide group and differs from what is proposed for 6 or 7. Thus far no insights regarding this concern have been revealed.
Complexes 9 and 10 preserve the same coordination geometry as other Type II complexes. 47 Otherwise, there is no aqua or 'labile' halogen monodentate ligand. By comparing 9 and 10, it is found that a dianionic carboxylate ligand, rather than the neutral tpy, facilitates picoline/water exchange at the Ru III state. 59 DFT model studies estimate a lower energy barrier for 10 than 9 by about 10 kcal mol À1 , corresponding to a remarkably faster picoline/water exchange rate for 10. This accelerated exchange rate is attributed to destabilization of the ruthenium d z 2 orbital by carboxylate, resulting in a large energy gap between the binding orbitals of Ru and picoline. We suggest that the aqua complex [Ru III (pdc)(pic) 2 OH 2 ] (pdc ¼ 2,6-pyridinedicarboxylate) derived from 10 is the actual WOC initiating catalytic O 2 evolution. The TOF (0.23 s À1 ) of 10 is signicantly greater than that of 7. The introduction of an amide group in place of one carboxylate group of pdc further lowers the oxidation potential and enhances the catalytic activity of the complex. 60

Type III: [Ru(LLLL)(L) 2 ] n+ WOCs
The complex cis-[Ru II (bpy) 2 (OH 2 ) 2 ] 2+ (11) has the same coordination environment as either ruthenium site of the blue dimer where an aqua ligand replaces the oxo-bridge. Thus 11 represents a monomeric analog of the blue dimer. An electrochemical study by Meyer et al. showed that complex 11 can lose 4H + /4e À in a stepwise fashion within a narrow potential range 0.8-1.5 V vs. NHE and form a Ru bis-oxo complex cis-[Ru(bpy) 2 (O) 2 ] 2+ with a formally VI ruthenium center. 61 Mixing 11 and CAN in 0.1 M CF 3 SO 3 H resulted in O 2 evolution and RuO 2 precipitation simultaneously. 61,62 Therefore, there is some uncertainty about whether 11 or RuO 2 actually catalyzes water oxidation. In a more recent study, Llobet and coworkers revisited complex 11. 63 They found that cis-[Ru II (bpy) 2 (OH 2 ) 2 ] 2+ is capable of catalyzing dioxygen production at a much faster rate than either its trans-isomer or RuO 2 , although the catalytic performance of 11 is limited to several turnovers. An 18 Olabeling experiment demonstrated that the dioxygen evolved from the rst catalytic cycle originated from both the complex aqua ligand and the solvent water molecules. This result supports an 'acid-base' pathway, as depicted in Scheme 7, and rules out intramolecular O-O bond formation. A DFT simulation computed the activation free energy of the water nucleophilic attack and O 2 release steps to be 24.5 and 25.1 kcal mol À1 , respectively. Both values are greater than the activation energy of the tautomerization step. Scheme 6 Proposed mechanism for water oxidation by complex 6 in aqueous medium.
Scheme 7 Proposed mechanism for water oxidation by complex 11 in aqueous medium.
Complex 11 is prone to lose bpy ligands when it is oxidized to a high oxidation state. 61 This loss is attributed to the rapid decomposition of 11 under water oxidation conditions and thus the low catalytic turnover. The trans-isomer of 11, however, is more stable with respect to ligand dissociation. Thummel and coworkers incorporated a rigid phenanthroline moiety to replace the central bpy of qpy (2,2 0 :6 0 ,2 00 :6 00 ,2 000 -quaterpyridine) thus preparing the tetradentate ligand 2,9-di-(pyrid-2 0 -yl)-1,10phenanthroline (dpp), in which rotation about the central bpybpy bond has been restricted. 66,67 Complex 12 involving the equatorial tetradentate dpp ligand and two axial pic ligands is the earliest example in the category of [Ru(LLLL)(L) 2 ] WOCs. 47,67 In the presence of CAN, 12 was reported to catalyze O 2 evolution with a TOF of 1.2 Â 10 À2 s À1 . 68 Although no water is coordinated with the Ru(II) center of 12, its Pourbaix diagram (Fig. 5) clearly demonstrates features of PCET processes. Theoretical studies corroborated that, in the medium and high pH regions, complex 12 accommodates a water molecule during the 2e À / 2H + PCET oxidation resulting in a seven-coordinate 18 2+ ET and follows 2e À /2H + PCET redox steps to produce a sevencoordinate [Ru V (O)] 3+ species. Water association to the ruthenium center is presumed to occur concurrent with the redox process. DFT simulation proposes an 'acid-base' mechanism for the O-O bond formation between seven-coordinate [Ru V (O)] 3+ intermediate and water. 64 It requires a calculated thermodynamic potential of 1.94 V that is the highest in the predicted catalytic cycle (Scheme 8) for 12. It should be noted that the X-ray structure of 12 shows a considerably large 125 external N-Ru-N (dpp) angle. This feature might facilitate water insertion in the primary coordination sphere of the complex. A recent study on several analogs of 12 indicates that both electronic and steric modication affects the catalytic performance. 68 It is difficult, however, to establish a straightforward structure-activity correlation.
The Pourbaix diagram (Fig. 5) of complex 13 shows quite different features from that of 12. 65 Firstly, the redox potentials of Ru III/II , Ru IV/III , and Ru V/IV are well separated over the whole pH range from 0 to 12. Secondly, the Ru III /Ru II oxidation process is coupled with proton transfer when the pH is higher than 5.5, indicating water molecule association in the redox step because complex 13 in its divalent state does not bind an aqua ligand. Thirdly, the IV oxidation state of 13 can be reached at a lower potential than that of 12. At pH ¼ 1.0, for example, the Ru IV/III redox potential is about +1.05 V. Sun's group successfully isolated the Ru IV species from pH ¼ 1.0 aqueous solution as a dimeric {m-(HOHOH)[Ru IV (bdc)(pic) 2 ] 2 } 3+ (bdc ¼ 2,2 0 -bipyridine-6,6 0 -dicarboxylate) complex, of which each Ru IV center is seven-coordinated incorporating one hydroxyl ligand in the equatorial plane. 69 The structure of the Ru IV species might be stabilized by a hydrogen bonding network including a solvated water molecule, hydroxyl ligand, and carboxylate groups. It implies a possible proton-shuttling path from the hydroxy ligand to the bulk solvent during water oxidation.
A cyclic voltammogram of 13 under acidic conditions showed the onset of a catalytic current at a more positive potential than the Ru V /Ru IV redox potential. A kinetic study at pH ¼ 1.0 using a stopped-ow technique suggests a catalytic cycle for 13 as displayed in Scheme 9. 65 The O-O bond was proposed to form via coupling of two [Ru V ]O] + species which can be regarded as a resonance form of the Ru IV oxyl radical [Ru IV -Oc] + . Dioxygen release from the resulting [Ru IV -OO-Ru IV ] 2+ intermediate was believed to be the rate-determining step under stoichiometric CAN conditions. In the presence of excess CAN, however, [Ru IV -OO-Ru IV ] 2+ can be rapidly oxidized to a superoxo [Ru IV -OcO-Ru IV ] 3+ intermediate which liberates O 2 at a fast rate. As a result, the radical coupling step becomes rate-determining. This hypothesis is supported by the experimental observation that the rate of water oxidation by 13 was second order with respect to the catalyst when a large excess of CAN was used. 65 The complex was reported to be capable of Scheme 8 Proposed mechanism for water oxidation by complex 12 in aqueous medium.

Scheme 9
Proposed mechanism for water oxidation by complex 13 in aqueous medium.
catalyzing water oxidation with a TOF of 12 s À1 . Electron withdrawing and hydrophobic substituent groups on the axial ligands boost the catalytic activity. 70 In one case, where isoquinoline was introduced as the axial ligand, an astonishing TOF of 303 s À1 was obtained. 65 This elevated rate is attributed to the noncovalent intermolecular attraction between isoquinolines which lowers the energy barrier for the radical coupling step. The systematic study of Ru WOCs with bdc ligands, including 13 and its analogues, has recently been reviewed by Sun et al. 71 Concepcion and coworkers prepared the complex [Ru II (bdp)(pic) 2 ] (H 2 -bdp ¼ 2,2 0 -bipyridine-6,6 0 -diphosphonic acid) as a phosphonate analog of 13. 12 Using CAN as an oxidant at pH ¼ 1.0, [Ru II (bdp)(pic) 2 ] is found to catalyze water oxidation via an acid-base pathway involving a seven-coordinate [Ru IV -OH] À intermediate and a rate-limiting oxidation step. The TOF (0.3 s À1 , assuming 100% CAN efficiency) of [Ru II (bdp)(pic) 2 ], however, is almost two orders of magnitude smaller than the TOF of 13, highlighting the favoured radical coupling rather than acid-base mechanism for a highly efficient catalyst. Llobet et al. investigated the catalytic water oxidation behavior of [Ru II (tda)(py) 2 ] (H 2 -tda ¼ 2,2 0 :6 0 ,2 00 -terpyridine]-6,6 00dicarboxylic acid), in which the pentadentate tda ligand contains one pyridine moiety more than bdc. 24 The authors proposed a seven-coordinate Ru V ]O state of the complex with a dangling carboxylate group that can form H-bond with incoming water molecule and thus facilitate the electrophilic attack of the oxo to the water molecule. An impressive TOF of 8000 s À1 at pH 7.0, assessed by electrochemical method, makes [Ru II (tda)(py) 2 ] the most efficient mononuclear WOC ever reported. Chemical-driven water oxidation catalysis for the complex was not revealed in the study.

Type IV: [Ru(LLL) 2 ] n+ WOCs
The primary coordination sphere of [Ru II (tpy) 2 ] 2+ is saturated by six rigid Ru-N(tpy) coordination bonds. The complex does not possess a vacant coordination site to accommodate a water molecule and replacement of one of the bound pyridines by water has never been observed. To behave as a WOC therefore, [Ru II (tpy) 2 ] 2+ must expand its coordination sphere to seven by the addition of a water molecule, much like complex 12. Such hepta-coordination demands a pentagonal bipyramid geometry which would dictate the impossible situation of a single tpy ligand spanning both axial sites with the Ru-tpy coordination (N-Ru-N) arranged in an approximate linear fashion. When the tpy ligand binds with a single metal center it forms two adjacent ve-membered chelate rings that dene an exterior N-Ru-N angle of only about 158 . If the size of one of these chelate rings is increased from ve to six, however, the resulting ligand could span both axial sites. The tridentate ligand 2-(quinol-8 0 -yl)-1,10phenanthroline (phenq) binds Ru(II) as a 6-5 chelator and thus can accommodate 7-coordinate pentagonal bipyramid geometry. The [Ru(phenq)(tpy)] 2+ complex (15) thus shows modest WOC activity (TON ¼ 334). 72 Several other Ru(II) complexes involving tridentate 6-5 chelators have likewise been shown to be active as WOCs, pointing to the importance of conformational effects in designing active catalyst systems. It is possible that the ruthenium center coordinates with a water molecule at high valent state, such as Ru(IV), which is more electron-decient than the divalent state. Meanwhile, the complex has to reorganize its structure to provide space in the coordination sphere for the association of an oxygen (water) ligand. This reorganization might be accomplished by weakening and elongation of certain N-Ru coordination bond. We expect that the substitution of quinoline for pyridine in some of the ligand systems shown in Scheme 1 will provide an interesting and useful new family of metal binders.

Auxiliary pathways contribute to dioxygen evolution
Chemical-promoted catalytic water oxidation is usually performed in the presence of a large excess of a sacricial oxidant, hundreds to thousands of equivalents relative to the amount of the Ru catalyst. Partly due to such harsh conditions competing pathways have been observed and proposed to contribute to O 2 evolution concomitant with the primary catalytic pathways described above. Berlinguette and co-workers found that not all oxygen atoms of dioxygen were derived from water when they studied water oxidation catalyzed by 1a. 44 44 Llobet and co-workers reported that the mononuclear catalyst 1 could lose its bpy ligand and convert to an oxo-bridged dinuclear [Ru IV -O-Ru IV ]O] 4+ species (Scheme 10) in situ during CAN-promoted water oxidation. 73,74 They managed to isolate the dinuclear complex and characterized its structure by single-crystal X-ray diffraction. This conversion is slow but irreversible through a self-assembly type process. While the dinuclear complex exhibited catalytic activity similar to 1 towards water oxidation, it is a more robust WOC than 1. A DFT calculation supported a catalytic cycle for the dinuclear species which coexisted in parallel with the catalytic cycle of the mononuclear catalyst 1. Very recently, Sakai and co-workers found that catalyst 13 could lose monodentate pyridine ligands and assembled to a trimeric ruthenium species upon oxidation in a very similar manner as 1. 75 The isolated trinuclear ruthenium complex has a 'Ru III -O-Ru IV -O-Ru III ' motif with m-oxobridges. In a light-driven [Ru(bpy) 3 ] 2+ /S 2 O 8 2À photochemical Scheme 10 Generation of oxo-bridged dinuclear catalyst during the CAN-promoted catalytic water oxidation process.
system (pH ¼ 8.0), it is capable of catalysing O 2 evolution with a TOF of about 0.9 s À1 . Mechanistic details of the trinuclear Ru complex are under investigation. Investigation of complex 14 by Lau and co-workers revealed that the qpy ligand was oxidized to qpy-N,N 000 -dioxide in a pH ¼ 1.0 aqueous solution of CAN (Scheme 11). 76 The resulting Ru(III) complex was isolated and structurally characterized by singlecrystal X-ray diffraction. It showed a considerably shorter induction period (about 1 min) as compared to 14 (about 5 min) in CAN-promoted O 2 evolution experiments. Aer the induction period, the O 2 evolution rates for 14 and its di-N-oxide counterpart are comparable. 18 O-Labeling experiments indicated that the oxygen atoms of the di-N-oxide are not found in the catalytically generated dioxygen. These observations imply oxidative conversion of the qpy ligand as part of the Ru complex with the di-N-oxide 16 being the authentic catalyst for water oxidation. The kinetics of qpy to qpy-N,N 000 -dioxide, however, have not been disclosed in detail. It is not clear yet if there are competing catalytic pathways that might include both 14 and 16.

Conclusions and outlook
A series of 15 mononuclear Ru(II) polypyridine complexes have been selected as representative WOCs to review and divided into four groups according to the disposition of the pyridine ligands around the metal center. Type I complexes have a tridentate, bidentate, and monodentate ligand occupying the 6 coordination sites of Ru(II). Type II have a tridentate and three monodentates. Type III are (mostly) tetradentate in the equatorial plane plus two axial monodentates. Finally, Type IV is unique with two tridentate ligands binding in a meridional fashion. While all multi-dentate ligands are polypyridine based, the monodentate ligand could be a water, halogen, or a substituted pyridine.
The aqua ligand can release protons upon oxidation of the Ru(II) complex. Such a PCET process is essential to the formation of a high valent, Ru(IV) or Ru(V), ruthenium oxo species at a moderate potential. There are Ru(II) examples in every group that do not possess an aqua ligand. They coordinate with a water molecule by ligand exchange or reorganization of structure to provide a vacant binding space.
The catalytic activity of the Ru(II) complexes is evaluated by analysis of the proles of oxygen evolution vs. time. Since only the oxidative half of the water splitting reaction is under scrutiny, a sacricial oxidant must be used in a stoichiometric fashion. Typically we have used ceric ammonium nitrate under acidic conditions as this sacricial reagent. The catalytic activity of these Ru(II) complexes towards water oxidation has been discussed in light of two fundamentally different mechanisms: one involving attack of a water molecule on the oxygen of a high valent Ru]O species and the other involving the formation of a Ru-Oc radical that could then dimerize to give a Ru-O-O-Ru species. The former mechanism appears to be the most prevalent for the systems under discussion. Only complex 13, among all the candidates, prefers the latter mechanism. Meanwhile 13 is the most active WOC in terms of TOF in CAN-driven O 2 evolution experiments. Its superior activity evokes a putative favoring of the radical coupling pathway for a highly efficient WOC. It appears that the fundamental differences in mechanism among the four types of complexes involve the chemistry of the critical higher valent ruthenium oxo intermediates. On one hand, the Ru]O intermediates trigger O-O formation; on the other hand, they represent the highest formal valence of the ruthenium center in the catalytic cycle.
Isolation of high valent Ru]O intermediates in situ is quite challenging due to their thermal instability and the strong solvation effect of the prerequisite aqueous medium. The structures of the Ru]O intermediates are related to the arrangement of the coordinating ligands. It is reasonable to envision the location of oxo ligand outside and inside the plane of tridentate polypyridine ligand for Type I and II WOCs, respectively. For Type III WOCs, a seven-coordinate structure seems favored for the Ru oxo intermediate. Transient spectroscopic techniques are able to probe kinetics in the catalytic cycle. Thus far, O-O bond formation, electron transfer, or O 2 liberation have been proposed as rate-determining for different WOCs.
The complicated mechanistic details, especially the various rate-limiting steps, of diverse WOCs make it almost impossible to establish a universal correlation between the structure and activity of WOCs. Nevertheless, there are some basic principles that can be applied to individual steps regardless of catalyst group. For example, the anionic carboxylate ligand has been found to enhance the rate of O 2 liberation step for both Type I and Type II WOCs. The introduction of electron-donating substituents, in general, facilitates the electron transfer process.
The design of homogeneous transition metal WOCs should meet some basic requirements: access to the metal-aqua and metal]O states, validity of O-O bond formation, and stability and solubility in aqueous solution. It is important to target the rate-determining step of the catalytic cycle. The goal is to reduce the activation energy of this step by modication of the ligands. Specic ligand features can be considered involving both the inner and outer coordination spheres. The former includes ligand properties such as rigidity, conjugation, s/p-donating ability, coordination vacancy, and interaction of ligand donors. The latter includes the electronic effect of substituents, hydrogen bonding properties, steric repulsion and hydrophilicity.
Looking to the future, there is a mounting effort to extend redox catalysis to include more earth abundant metals, especially the rst row transition metals. Both Co and Ni have been widely investigated as proton reduction catalysts to produce hydrogen and Fe, Co, Mn, and others have been used in systems active towards water oxidation. As with Ru(II), it is the ligand environment that will ultimately control the redox activity. Lessons learned from ruthenium-based catalysts should inspire and motivate the development of catalysts based on other transition metals. The future promises considerable new development directed towards the realization of a practical system for articial photosynthesis.