Introduction

Solar energy is used to convert CO2 and water into sugars and O2 by a wide variety of photosynthetic organisms, including plants, algae, and cyanobacteria. In this process, the light-driven splitting of water into O2, protons, and “energy rich” electrons is facilitated by photosystem II (PSII), a large cofactor–protein complex found in the thylakoid membranes of chloroplasts. In PSII, the four-electron oxidation of water (2H2O → O2 + 4H+ + 4e−) is effectively catalyzed by a functional unit known as the oxygen-evolving complex (OEC). The OEC consists of an inorganic core composed of four manganese, one calcium, and five oxygen bridges in the form of a Mn4CaO5 cluster surrounded by a functionally important protein matrix [1].

In many ways, natural photosynthesis is the ideal objective when designing direct solar fuel-production systems, because it is driven purely by solar light and uses a catalyst based on inexpensive and abundant metals. The molecular structure of the OEC, in particular, is proof that the reactivity of molecular O2-evolving catalysts can be optimized by molecular design involving a wide range of redox-active metal centers and ligands. Although the development of highly efficient molecular catalysts for water oxidation is a key step in the production of chemical fuels from solar energy, this development is a serious challenge, because the oxidation reaction requires removal of four protons and four electrons [2–6]. Much work has been devoted to study of catalysts which may be applicable to this process, and several materials based on Ru [6], Mn [7–11], Ir [12–14], Co [15–17], and other metals [18–22] have been extensively investigated. Reports on the development of Ru-based O2-evolving catalyst systems, in which cerium(IV) is typically used as oxidant, are especially numerous.

This review begins with a brief summary of historical results concerning the development of Ru-based molecular catalysts active in the O2 evolution reaction. Our own mechanistic studies into O2-evolving catalysts based on Ru complexes are then described. In addition, on the basis of the fundamental properties of the cerium(IV) ion, the crucial function of cerium(IV) as a radical in catalysis of the oxidation of water by Ru complexes is discussed.

General consideration of the mechanisms of O2 evolution

The reaction mechanisms of O2 evolution by molecular catalysts have been widely investigated, because an understanding of such mechanisms is necessary to enable the development of new, highly-active catalysts. In typical reaction mechanisms, a high-valent metal–oxo species has been identified as the crucial intermediate in O2-evolution reactions. The high-valent metal–oxo species (M(n + 2)=O) is typically generated from the corresponding water-coordinated metal ion (Mn–OH2) via two-step proton-coupled electron-transfer (PCET) processes, as summarized in Scheme 1 (L = ancillary ligand).

Scheme 1
scheme 1

Proton-coupled electron-transfer (PCET) reactions of a water-coordinated metal ion

Full size image

In this reaction sequence, higher oxidation states are readily accessible, primarily because of the σ and π-donating character of the oxo group. In addition, proceeding via the PCET process usually avoids high-energy intermediates and/or electrostatic charge buildup during the reactions [23, 24]. The resulting high-valent metal-oxo species is believed to be involved in formation of the O–O bond.

The O–O bond-formation process is the most important step in the reaction mechanism, because it is significantly more difficult than the other reaction steps. Therefore, it is important to have a good understanding of the O–O bond-formation process to enable efficient design of a catalyst suitable for water oxidation. Two possible O–O bond formation pathways have been proposed:

  1. 1

    acid–base coupling between the high-valent metal–oxo species and a water molecule (the OW-AB pathway in Scheme 2); and

  2. 2

    radical coupling between two high-valent metal–oxo species (the OO-RC pathway in Scheme 2).

It is believed the O–O bond-formation process is usually the rate-determining step (RDS) of O2 evolution, and successive reductive elimination affords O2, with regeneration of the water-coordinated catalyst (Scheme 3).

Scheme 2
scheme 2

Proposed reaction pathways for formation of O2 catalyzed by metal complexes

Full size image
Scheme 3
scheme 3

Proposed mechanistic cycles for typical O2 evolution catalyzed by ruthenium complexes via a the OW-AB pathway and b the OO-RC pathway

Full size image

Catalysis of evolution of O2 by ruthenium complexes

Ruthenium (Ru) is one of the most extensively studied transition metal ions in redox chemistry, catalysis, and photoreaction. On the basis of the well-established fundamental redox chemistry and reactivity of Ru complexes, numerous Ru-based catalysts of O2 evolution have been reported. A typical example of an Ru-based catalyst with applications to water oxidation is cis,cis-[{(H2O)(bpy)2RuIII}(μ-O){RuIII(bpy)2(OH2)}]4+ (1 in Scheme 4; bpy = 2,2′-bipyridine), the so called “blue dimer” [25, 26] reported by Meyer in 1982. This compound, the first successful molecular catalyst of O2 evolution is active in the water-oxidation reaction during either electrochemical or chemical oxidation with Ce4+ as terminal oxidant, although both its catalytic activity and stability are quite low. After discovery, several analogous oxo or chloro-bridged diruthenium complexes were also reported as catalysts of O2 evolution [27–33]. These newer derivatives, however, with the original blue dimer, are poorly stable during the catalysis reaction, because of gradual decomposition of the μ-oxo bridged diruthenium core. Several, more robust, multiruthenium catalysts with rigid bridging ligands have since been developed by Tanaka [34, 35], Llobet [36], Thummel [37] (2–4 in Scheme 4), and others [38–41].

Scheme 4
scheme 4

Structures of dinuclear Ru-based catalysts of O2 evolution

Full size image

Because of the need for multi-electron-transfer processes in the water oxidation reaction mechanism, it has long been assumed that the O2-evolving activity of multinuclear complexes will be substantially higher than that of mononuclear complexes [6]. Indeed, several mononuclear complexes, e.g. Ru(bpy)2Cl2, have been examined as control experiments for 1, and show almost no catalytic activity toward O2 evolution [32, 33]. Before 2008, the basic strategy for development of diruthenium catalysts of O2 evolution was, therefore, to design the structure of the catalyst such that two ruthenium-oxo groups were situated in close proximity, to facilitate oxo–oxo radical coupling, as in the OO-RC pathway in Scheme 2 [6]. Both 18O-isotopic-labeling studies and DFT calculations have in fact revealed that O2-evolution reactions catalyzed by some diruthenium systems, including catalysts 2 and 3, proceed via the intramolecular oxo–oxo radical-coupling pathway [42–45]. It has, however, been suggested that O2 evolution reactions catalyzed by blue dimer and its analogues proceed via the oxo-water acid–base pathway in which O–O bond formation occurs between the oxygen atom bound to the ruthenium center and the oxygen atom of a water molecule in the bulk solution (as shown in the OW-AB pathway in Scheme 2) [46–48]. Because this result contradicts the basic strategy for development of catalysts of O2 evolution, it seems it should be possible to use similar O2 evolution pathways by using monoruthenium complexes as catalysts of O2 evolution.

Recently, several different groups, including our own, have discovered that mononuclear ruthenium complexes with a single labile site have surprisingly high catalytic activity toward evolution of O2 from water in the presence of Ce4+ as oxidant (Table 1) This is a surprising development, because it has been the long-held belief that the mononuclear complexes have almost no activity as catalysts of O2 evolution [33, 49]. Specifically, in 2008, the authors reported that the mononuclear aquaruthenium complex [RuII(terpy)(bpy)(OH2)]2+ (5 in Scheme 5; terpy = 2,2′:6′,2′′-terpyridine) has significant activity on the catalytic evolution of O2 and, in addition, seems robust throughout the course of the reaction [50, 51]. We also reported that other analogous mononuclear aquaruthenium complexes have such O2-evolving activity (6–8 in Scheme 5) [53–56]. Several other researchers have also shown that specific monoruthenium systems (for example 9–16 in Scheme 6) have similarly high catalytic activity towards O2 evolution when Ce4+ is used as the oxidant [57–77]. These findings were a significant breakthrough in this field and led to subsequent reports of a variety of highly efficient catalysts. For example, although turnover frequencies (TOFs) of the most active diruthenium catalysts were ca. 10−2 s−1 [31, 36] they improved dramatically after the discovery of monoruthenium catalysts (TOF = 0.11 s−1 for 11 [65], 3.6 s−1 for 13 [63], and 300 s−1 for 16 [72]).

Table 1 Turnover number (TON) values and associated oxidants for several mononuclear Ru-based catalysts of O2 evolution
Full size table
Scheme 5
scheme 5

Structures of mononuclear Ru-based catalysts of O2 evolution discovered by the authors

Full size image
Scheme 6
scheme 6

Structures of mononuclear Ru-based catalysts of O2 evolution discovered by other researchers

Full size image

Reaction kinetics of O2 evolution catalysis

To further elucidate the O–O bond-formation pathways involved in the catalytic action of mononuclear ruthenium complexes, several researchers, including the authors, attempted to identify the intermediate species involved in the reaction and to confirm the rate-determining step (RDS). The authors initially tried to characterize the active species and other intermediates in O2 evolution catalyzed by 5, by studying mixtures of 5 with 10 equiv. Ce4+ in 0.1 M HClO4 by use of stopped-flow apparatus. In these trials, the intense absorption band of 5 at 475 nm decayed rapidly after mixing, after which new absorption bands at 500 and 700 nm gradually appeared as the reaction proceeded. These spectral changes were analyzed by use of the singular value decomposition (SVD) method in the SPECFIT software package [78]; the results suggested the presence of five detectable Ru-based chemical species during the overall reaction process. These spectral components were successfully identified on the basis of TD-DFT calculations ((U)B3LYP/LanL2DZ) and, on the basis of the results, the RuV=O species was determined to be the primary final product and, as such, is believed to be the dominant higher-valent species during the catalytic O2 evolution reaction [55].

The authors also calculated the electronic structure of the RuV=O species by use of the DFT method (UB3LYP/LanL2DZ) and determined that the O(oxo) atom of the species has substantial radical character (Fig. 1), which can also be understood as Mulliken spin density located at the O(oxo) atom [55]. This clearly indicates that the RuV=O species has the characteristics of a highly reactive oxo radical which can be reasonably represented by the two resonance structures RuV=O ↔ RuIV–O·. The oxygen atom bound to the Ru ion thus simultaneously exhibits characteristics of not only a Lewis acid but also a radical.

Fig. 1
figure 1

Spin density distributions of [RuV(terpy)(bpy)(O)]3+. The values of the spin density of the oxygen atom and ruthenium atoms are depicted with α spins having positive signs

Full size image

Several of groups, including the authors’ group, also attempted to determine the RDS of O2 evolution as catalyzed by mononuclear Ru complexes. The rate of O2 formation catalyzed by 6, as estimated from the initial slope, is described by a quadratic equation based on catalyst concentration and which is independent of Ce4+ concentration (i.e., is zero order) [54]. The rate law for this reaction can therefore be described as follows:

$$v_{{{\text{O}}2}} = {\text{d}}\left[ {{\text{O}}_{2} } \right]/{\text{dt}} = k[{\text{Catalyst}}]^{2}$$

Because the rate-determining step for catalysis by 6 is second-order with regard to catalyst concentration and zero order with regard to Ce4+ concentration, the OO-RC pathway (Scheme 2) is likely to be the predominant mechanism of O2 evolution as catalyzed by 6. The kinetics of Ce(IV)-driven catalytic oxidation of water in aqueous solution was also determined to be second-order, depending on the concentration of 15 [68], which suggests that the O2 evolution reaction catalyzed by 15 also proceeds via the OO-RC path [70].

In contrast, electrochemical studies performed by our group using an ITO electrode modified with a derivative of 5 revealed that O–O bond formation can also occur by direct attack of a water molecule, as in the OW-AB pathway [53]. The Ru–Ru distance in this system, 35 Å, does not allow two or more ruthenium centers to engage in the same O2 evolution event, hence the electrocatalytic evolution of O2 from water which occurs in this process is due solely to events at a single aquaruthenium site. This same pathway has also been described by Thummel [57–59], Meyer [60–63], van Voorhis [79], and Llobet [73, 74]. In Meyer’s work, the rate of evolution of O2 catalyzed by 10 in aqueous 0.1 M HNO3 was first-order with regard to catalyst concentration and zero-order with regard to Ce4+ concentration, so the reaction catalyzed by 10 must proceed via the OW-AB pathway, in accordance with the rate law:

$$v_{{{\text{O}}2}} = {\text{d[O}}_{2} ]/{\text{dt}} = k[{\text{Catalyst}}]$$

Meyer explains these results by theorizing that the nucleophilic attack of water on RuV=O is the RDS of the O2 evolution reaction as it proceeds via the OW-AB pathway, which progresses via the putative intermediate species RuIII–OOH. This species is subsequently oxidized to RuIV(OO), followed by reductive elimination of O2.

In contrast, the rate of evolution of O2 during catalysis by 5 is linearly dependent on both catalyst concentration and Ce4+ concentration [51, 54]; the rate law for this reaction can therefore be summarized as:

$$v_{{{\text{O}}2}} = {\text{d}}\left[ {{\text{O}}_{2} } \right]/{\text{dt}} = k[{\text{Catalyst}}][{\text{Ce}}^{4 + } ]$$

This rate law suggests that the RuV=O species and cerium(IV) form an encounter complex during the transition state in the RDS. This is problematic, however, because cerium(IV) has long been regarded as a simple, one-electron oxidant and, if this is the case, the observed rate law cannot be explained. In addition, several researchers have reported that the nitrate ion attached to cerium(IV) is also involved in the catalytic cycle of O2 evolution as a minor pathway [76, 80]. These results indicate the anomalous function of cerium(IV) in O2 evolution. Consequently, to better describe the reactivity it is necessary to obtain a better understanding of the evidently complicated properties of cerium(IV) as a metal ion.

Fundamental properties of cerium(IV)

Over the last three decades, cerium(IV) ammonium nitrate (CAN) has been most commonly used as a so-called simple, one-electron oxidant, and numerous catalysts of O2 evolution based on metal complexes have been developed in conjunction with use of cerium(IV) as an oxidant. Despite this significant body of research, however, several anomalies have been observed in the function of cerium(IV) in the catalysis of O2 evolution.

To further clarify the Ce(IV)-based evolution of O2 from water, it is vital to understand the complicated properties of cerium(IV). Cerium(IV) is well known not only as a strong oxidant, but also as a strong Lewis acid [81–83]. The first and second hydrolysis equilibrium constants (pK h1, pK h2) for cerium(IV) were determined by Yu and O’Keefe as follows [84, 85]:

$$\begin{gathered} {\text{Ce}}^{\text{IV}} \left( {{\text{OH}}_{2} } \right)_{n} \to {\text{Ce}}^{\text{IV}} \left( {\text{OH}} \right)\left( {{\text{OH}}_{2} } \right)_{n - 1} + {\text{H}}^{ + } \quad {\text{p}}K_{\text{h1}} = - 0.7 \hfill \\ {\text{Ce}}^{\text{IV}} \left( {\text{OH}} \right)\left( {{\text{OH}}_{2} } \right)_{n - 1} \to {\text{Ce}}^{\text{IV}} \left( {\text{OH}} \right)_{2} \left( {{\text{OH}}_{2} } \right)_{n - 2} + {\text{H}}^{ + } \quad {\text{p}}K_{{{\text{h}}2}} = 0.717 \hfill \\ \end{gathered}$$

Owing to this hydrolysis equilibrium, the formal potential of the Ce(IV)/Ce(III) system changes on varying the acid concentration; for example, E 0 = +1.26 V versus NHE in 0.1 M HNO3 yet +1.37 V versus NHE in 1.0 M HNO3 [86]. In addition, compared with CeIV(OH), CeIV(OH)2 readily forms less-reactive [87–90] μ-oxo cerium(IV) dimers (for example CeIV(μ-O)CeIV) [91, 92] or oligomers [93], which leads to additional complications. At pH > 2–3, cerium(IV) also forms stable precipitates, for example CeO2 [84, 85]. Notwithstanding these complexities, CeIV(OH) is the dominant chemical species under the standard reaction conditions used during O2 evolution (pH ≈ 0).

Importantly, it has been suggested the CeIV(OH) species exists in an unusual electronic state. The authors computed the electronic structure of one of the possible hydroxocerium(IV) ions, [CeIV(OH)(NO3)5]2−, by use of the DFT method (UM05-2X/CEP-4G) [54] and found that Mulliken atomic spin densities of +0.89 and −1.02 are present on the O(hydroxide) and Ce atoms, respectively (Fig. 2). These spin density distributions show that one spin occupies the 4f orbital of the cerium ion whereas the other is primarily located on the O(hydroxide) atom, indicating that this ion can be described more accurately as [CeIII(·OH)(NO3)5]2−, and so the spins localized on the CeIII and O(hydroxide) atoms are antiferromagnetically coupled. The spin density is also somewhat localized on the O(nitrate) atom. These results reveal the singlet biradical character of the hydroxocerium(IV) ion [54].

Fig. 2
figure 2

Spin density distributions of [CeIV(OH)(NO3)5]2− in the open-shell singlet state. The values of the spin density of the oxygen atom and cerium atoms are depicted with α and β spins having positive and negative signs, respectively

Full size image

Although it has been widely assumed that compounds of Ce4+ have a 4f0 configuration (e.g., [Xe](6 s)0(4f)0 for Ce(IV)) and simple diamagnetism, some cerium(IV) complexes also have singlet biradical character [94–96]. Neumann and Fulde first investigated the detailed electronic structure of Ce(C8H8)2 (cerocene, in which C8H8 2− is the cyclooctatetraene dianion), and revealed that the compound is better described as CeIII(C8H8 2−)(C8H −·8 ), in which one spin is localized on the cerium ion (4f1) and the other on one of the ligands (Scheme 7) [94]. This is now understood as an open-shell singlet state in which two spins within the molecule are antiferromagnetically coupled. In other words, cerocene also has a singlet biradical ground state.

Scheme 7
scheme 7

Singlet biradical character of cerocene

Full size image

The singlet biradical character of the hydroxocerium(IV) ion indicates that a highly reactive hydroxyl radical (HO·) is, formally, localized throughout the ion (Scheme 8). If this is so, cerium(IV) functions not only as an oxidant but also a radical. Indeed, cerium(IV) oxidizes aldehydes, ketones, and alcohols at the α-carbon atom via a radical-like H-atom abstraction mechanism, suggesting radical-like reactivity of the hydroxocerium(IV) species [97–99]. This is further supported by the fact that cerium(IV) readily generates free radicals in response to photo-excitation [100–102]. Cerium(IV) complexes with an O-donor ligand also have a broad absorption band in the near-UV to visible region, corresponding to the charge transfer band from the O atom to the metal [103]. Evans and Uri have reported that photo-excitation of this absorption band induces generation of a free hydroxyl radical, leading to O2 evolution [100].

Scheme 8
scheme 8

Possible resonance structures of the CeIV(OH)(NO3) species

Full size image

Mechanism of cerium(IV)-driven water oxidation

The radical-like character of cerium(IV) clearly explains the mechanism of O2 evolution during catalysis with 5. As noted above, the rate of evolution of O2 catalyzed by 5 is first-order with regard to both catalyst and Ce4+ concentrations. In addition, the O(oxo) atoms of both the RuV=O species and the CeIV(OH) species have substantial radical characteristics. On the basis of these findings, we proposed that the pathway of O2 evolution during catalysis by 5 proceeds primarily via the oxo-hydroxocerium(IV) radical-coupling pathway (the OH-RC pathway). In this process, cerium(IV) functions both as an oxidant and as a radical and provides an efficient pathway enabling radical–radical coupling between the catalyst and the cerium(IV) [54, 55]. Because the spin density is also partially located on the O(nitrate) atom of the CeIV(OH)(NO3) species, the NO3 ligand might also have radical-like reactivity. Our interpretation of the reaction mechanisms seems to take into account the specific characteristics of cerium(IV). It is also noteworthy that most Ru-based catalysts of O2 evolution reported to date have used CAN as oxidant [25–80], which implies that use of CAN is crucial.

In summary, it is reasonable to conclude that O2-evolving reactions catalyzed by 5 proceed preferentially via the oxo-hydroxocerium(IV) radical coupling path (the OH-RC pathway; Scheme 9), even though this reaction may also proceed via coupling with oxygen in the water bulk (the OW-AB pathway) in the absence of the hydroxocerium(IV) species. A possible interpretation of these findings is that O–O bond formation occurs as the RDS via coupling of the RuV=O (↔RuIV–O·) and hydroxocerium(IV) species, leading to formation of either the RuIV–O–O(H)–CeIII or the RuIV–O–O–CeIII species, followed by elimination of Ce3+ to give RuIII–(OO·) [55]. The RuII–OH2 species is then regenerated by reductive elimination of O2 from RuIII–(OO·).

Scheme 9
scheme 9

O–O bond formation step in the oxidation of water promoted by the radical-like character of the hydroxocerium(IV) species

Full size image

The nitrate ion attached to cerium(IV) is also involved in the catalytic cycle of O2 evolution, as follows:

  1. 1

    As reported by Berlinguette and co-workers, a small number of oxygen atoms derived from [Ce(NO3)6]2− are incorporated into the final dioxygen product during reactions catalyzed by 5 [76].

  2. 2

    Stull et al. identified new intermediate species by analysis of distinct features in EPR, optical, and resonance Raman (RR) spectra when the cerium(IV)-nitrato complex was used as an oxidant [80].

The radical-like character of cerium(IV) also clearly explains the specific reactivity of the nitrate ion attached to cerium(IV), because the spin density is also partially located on the O(nitrate) atom of the CeIV(OH)(NO3) species.

Concluding remarks

In this review we have summarized recent notable developments in investigations of the mechanisms of O2-evolving reactions catalyzed by ruthenium complexes with cerium(IV) as an oxidant. Although mononuclear complexes are believed to have almost no activity compared with multinuclear complexes, in recent years several mononuclear ruthenium complexes with a single labile site have been developed as efficient catalysts for oxidation of water, and have been described herein. These findings led to dramatic improvement of O2-evolving activity, (e.g., the rate of O2 evolution). Three distinct O–O bond-formation pathways, the OO-RC, OW-AB, and OH-RC pathways, involved in evolution of O2 catalyzed by mononuclear ruthenium complexes were considered. This review also affirmed the crucial function of cerium(IV) in the water-oxidation process. The discovery of the utility of this species was so important that most Ru-based catalysts of O2 evolution reported to date have incorporated use of cerium(IV) as oxidant. These novel concepts of catalysis will provide new ideas with regard to the design of more highly active catalysts of O2 evolution.