Bio-Inspired Molecular Catalysts for Water Oxidation

The catalytic tetranuclear manganese-calcium-oxo cluster in the photosynthetic reaction center, photosystem II, provides an excellent blueprint for light-driven water oxidation in nature. The water oxidation reaction has attracted intense interest due to its potential as a renewable, clean, and environmentally benign source of energy production. Inspired by the oxygen-evolving complex of photosystem II, a large of number of highly innovative synthetic bio-inspired molecular catalysts are being developed that incorporate relatively cheap and abundant metals such as Mn, Fe, Co, Ni, and Cu, as well as Ru and Ir, in their design. In this review, we briefly discuss the historic milestones that have been achieved in the development of transition metal catalysts and focus on a detailed description of recent progress in the field.


Solar Water Oxidation in Nature
Sunlight is an abundant, green, sustainable, and renewable source of energy, however, the photochemical conversion and storage of solar energy has been a challenge [1]. Nature uses photosynthesis to convert solar energy into chemical energy through lightdriven water oxidation and atmospheric CO2 fixation [1][2][3][4][5][6][7][8][9]. Water oxidation, one of the most energetically demanding reactions in nature, takes place in the multi-subunit membrane protein complex, photosystem II (PSII) ( Figure 1A), in plants and cyanobacteria. The structure of PSII has been determined through X-ray diffraction [10][11][12][13] and femtosecond X-ray free electron lasers (XFEL) [14][15][16][17]. Photosystem II is comprised of a core of heterodimeric polypeptides, D1 and D2, surrounded by ~20 polypeptide subunits [14], within which there are more than 1300 water molecules [10]. The waters are mostly located at the stromal and lumenal surfaces, as is commonly encountered in trans-membrane proteins [10].
Photosystem II utilizes visible light to convert water into dioxygen at the catalytic tetranuclear manganese-calcium-oxo (Mn4Ca-oxo) cluster in the oxygen-evolving complex (OEC). The Mn4Ca-oxo cluster in the OEC has a distorted cubane-like structure with three manganese (Mn1-Mn3), one calcium, and four oxygen (O1-O3 and O5) atoms and a 'dangling' manganese atom (Mn4) that is linked to the cubane through di-μ-oxo linkages to the O atoms, O5 and O4 (numbering of the atoms is shown in Figure 1B) [10]. In addition to the μ-oxo-bridged oxygen atoms, the Mn4 and Ca ions are coordinated by the water-derived ligands, W1, W2 and W3, W4 [10,14,18,19], respectively, as well as one imidazole and six carboxylate ligands that are derived from the amino acid residues in the D1 and CP43 polypeptides of PSII ( Figure 1B) [10][11][12]14]. The Mn4Ca-oxo cluster in the OEC is surrounded by additional amino acid residues and structured water molecules that form an extensive network of hydrogen bonds with the ligands of the Mn4Ca-oxo cluster (A) (B) Figure 1. (A) The multi-subunit membrane protein complex, photosystem II (PSII), as observed in the 1.9 Å resolution Xray crystal structure and (B) the tetranuclear manganese-calcium-oxo (Mn4CaO5) catalytic cluster in the oxygen-evolving complex (OEC) of PSII [10]. The manganese (Mn1-Mn4), O 2− (O1-O5), and Ca 2+ ions are depicted as purple, red, and yellow spheres, respectively. The oxygen atoms of the four water-derived ligands that are coordinated to the Mn4CaO5 cluster, W1-W4, are shown in orange, where W1 and W2 are coordinated to the dangling Mn ion, Mn4, and W3, and W4 are ligands to the Ca 2+ ion. The amino acid residues that are coordinated to the metal ions of the Mn4Ca-oxo cluster in the OEC are also shown.

Synthetic Water Oxidation Catalysts
The water oxidation reaction of PSII has been of intense interest as it serves as a blueprint for the development of active and robust synthetic water oxidation catalysts that mimic the OEC of PSII [32][33][34][35]41,42]. The dioxygen produced is released to atmosphere, whereas the electrons and protons from water oxidation may be used to generate economically useful dihydrogen and organic molecules [1,43,44].
Synthetic catalysts for artificial water oxidation (WOC) can be classified into two groups, namely, molecular catalysts and bulk semiconductors. Although semiconductor surfaces (e.g., metals oxides) are easy to synthesize and recover and have displayed reasonable stability for the water oxidation reaction [45][46][47], they are generally vulnerable to harsh reaction conditions (such as low or high pH), often lack structural reproducibility, and are hard to characterize through physical and chemical methods [48]. Since the synthesis of the first well-defined molecular WOC [49], the development of such catalysts has been the subject of extensive research and the strategies that were previously employed in the field can be useful to improve the design of WOC. Amongst the current WOCs, Rubased catalysts have received extensive attention [4,37,[50][51][52], some of which have displayed high catalytic activity [53][54][55]. However, there is also an interest in developing earth-abundant, cost-effective catalysts comprised of first-row (3d) transition metals, such as manganese, iron, cobalt, copper, and nickel. In this review, we describe molecular water oxidation catalysts that are based on 3d (Mn, Fe, Co, Cu, Ni), 4d (Ru), and 5d (Ir) transition metals ( Figure 2), with an emphasis on the progress in the field over the past five years. The examples discussed in this review highlight the role of ligand design in the stability, solubility, and catalytic activity of molecular WOC. We refer the reader to previous and thorough review articles [7,[56][57][58][59][60] for catalytic systems that were developed prior to this period.

Dinuclear Ruthenium Catalysts
The ruthenium-based dinuclear molecular catalyst for water oxidation, known as the "blue dimer", [(H2O)Ru III (bipy)2(μ-O)Ru III (bipy)2(H2O)] 4+ (1) (Figure 3), was first reported by Meyer and coworkers in 1982 [49]. This complex is comprised of a Ru III dimer, where the metal ions are linked by a μ-oxo bridge. Each Ru III ion is coordinated by two bipyridine (bipy) ligands that provide an open coordination site to accommodate a water ligand. The blue dimer displayed a turnover number (TON) and turnover frequency (TOF) of 13.2 and 0.0042 s −1 , respectively, for the chemical oxidation of water using [Ce(NO3)6][(NH4)2] (CAN) as a sacrificial oxidant (Table 1) [61,62]. CAN provides a sufficient oxidation potential to oxidize the investigated catalysts at approximately 1.75 V vs. NHE (pH 0.9) [60,63]. The mechanism of water oxidation by complex 1 was investigated by kinetic measurements in combination with isotope replacement studies to determine the kinetic isotope effect, which suggested that the reaction predominantly proceeded through a water nucleophilic attack (WNA) mechanism with the formation of a high-valent intermediate, 4+ , where uncoordinated water molecules attack the Ru V = O group of the intermediate. However, the results were unable to exclude the possibility of intramolecular or bimolecular pathways in the reaction [64][65][66][67]. The moderately low catalytic performance of complex 1 was attributed to the anation of an active intermediate, in which the [(O)Ru V (μ-O)Ru V (O)] 4+ cation coordinated an anion forming the species, [(bipy)2(H2O)Ru IV (μ-O)Ru III (X)(bipy)2] 4+ (where, X = ClO4ˉ, CF3SO3ˉ and NO3ˉ), which resulted in the deactivation of the system [68].
This was followed by the development of two dinuclear ruthenium complexes [{Ru II (terpy)Cl}2(μ-L)] 2+ (4) and [{Ru II (terpy)(H2O)}2(μ-L)] 4+ (5) (where, L = bis[5-(5′-methyl-2,2′-bipyridinyl)]ethane) by Sakai and coworkers (2009) [75] that displayed activity for chemical water oxidation with TONs of 75 and 106, respectively, using CAN as an oxidant. It was found that the activities of complexes 4 and 5 were much higher than that of the μ-O bridged complex 1. More importantly, an induction time of 2-3 h was observed for complex 4. However, the initial rate of O2 formation for complex 5, the aqua species, did not show an induction period for oxygen formation, implying that the aqua (instead of the chloro) species was active during the catalysis of the water oxidation reaction [75]. Subsequently, Llobet and coworkers (2014) developed a powerful and oxidatively rugged complex, [{Ru II (py-SO3)2(H2O)}2(μ-Mebbp)] − (6) (where, HMebbp = 2,4-bis(bipyridin)-3methyl-pyrazole and py-SO3 = pyridine-3-sulfonate), with a pyrazolate-based equatorial ligand that featured a TON of 22.6 and a TOF of 0.068 s −1 for chemical water oxidation using CAN as an oxidant [70]. In complex 6, Mebbpˉ is a bis(tridentate) monoanionic ligand that was designed to act as a bridging scaffold that placed the two ruthenium ions in close proximity. The Mebbpˉ ligand appeared to induce subtle geometric variations on the relative disposition of the active Ru−OHX groups that regulated the O−O bond formation pathway and influenced the mechanism toward WNA. This was in contrast to the intermolecular bimolecular (I2M) mechanism that was suggested for analogous complexes with 3,5-bis(2-pyridyl)pyrazolato dinucleating, Hbppˉ, ligand due to the strategic disposition and encumbrance of the terpy ligands [57,69,76,77]. This study demonstrated that subtle variations in ligand design could be used to regulate the O−O bond formation pathway of the water oxidation reaction. Additionally, the tridentate dianionic meridional pyridyl-2,6-dicarboxylato (pdc 2− ) ligand was used to generate a dinuclear ruthenium complex 7. Although complex 7 was not a catalyst for water oxidation, it was shown to act as a precursor for a ruthenium-aquo mononuclear complex, [Ru II (pdc-κ 3 -N 1 O 2 )(bipy)(H2O)] (31) (where, pdc = pyridyl-2,6-dicarboxylato and bipy = 2,2′-bipyridine), that was an active water oxidation catalyst with low overpotential of 240 mV at pH 1 and a TOF of 0.2 s −1 [78].

Mononuclear Ruthenium Catalysts
The success of the blue dimer [49] in catalyzing water oxidation and the lower catalytic activity of mononuclear ruthenium complexes in early studies [61] led to the assumption that multinuclear catalysts were required for the successful conversion of water to dioxygen. However, the design of a series of mononuclear ruthenium complexes that were catalytically active for water oxidation challenged this hypothesis [71]. The ease of the chemical syntheses led to a shift towards the design of effective mononuclear ruthenium catalysts for water oxidation. The complex [(ntp)(pic)2Ru(H2O)] 2+ (11) (where, ntp = 2,6-di (1,8-naphthyridin-2-yl)-4-tert-butylpyridine) (Figure 4), reported by Thummel and coworkers, displayed catalytic activity for water oxidation (Table 2) and paved the way for the development of mononuclear ruthenium complexes for water oxidation. Subsequently, a study by Meyer and coworkers on the mononuclear ruthenium complexes [Ru(terpy)(bpm)(OH2)] 2+ (12b) and [Ru(terpy)(bpz)(OH2)] 2+ (12c) (where, bpm = 2,2′-bipyrimidine; bpz = 2,2′-bipyrazine) demonstrated conclusively that a single Ru site is sufficient for catalytic water oxidation [82,83]. A large number of monomeric ruthenium complexes for water oxidation have been reported since the initial findings by Thummel, Meyer and coworkers. The mononuclear ruthenium catalysts that have been reported in the literature can broadly be divided into two classes based on the equatorial and axial ligands that were employed in the respective syntheses.
Berlinguette and coworkers investigated a series of structurally related mononuclear ruthenium catalysts that were formulated as [Ru(terpy)(L)(OH2)] 2+ (where, L = 2,2′-bipyridine (bipy) (12a), 4,4′-dimethoxy-2,2′-bipyridine (bipy-OMe), and 4,4′-dicarboxy-2,2′-bipyridine (bipy-COOH)) [84]. The goal of this study was to determine the effect of the electronic density at the active site on the catalytic performance while holding the balance of the structure at parity. The effects of the systematic modification of the substituent groups on the bipy ligand of the complex indicated that while the presence of electron-withdrawing groups (EWG), such as -Cl and -COOH, suppressed the rate of the reaction, kobs, and enhanced the catalytic TON, the installation of electron-donating groups (EDG), such as -OMe, accelerated the catalytic rate while decreasing the stability of the catalyst [85]. The observation of a reverse relationship between the rate of the reaction and the TON was similar to prior observations by Thummel and coworkers [86]. However, a study by Berlinguette and coworkers suggested that chemical water oxidation driven by Ce IV as an oxidant led to reaction pathways that diverged from the prevailing "acid-base" mechanism for single-site catalysts. The catalysts displayed complicated pathways that involved the incorporation of O atoms from different sources [87,88].
Thummel and coworkers (2008) assessed a series of chloro-coordinated mononuclear ruthenium complexes using terpy, bipy, and related derivatives as ligands that demonstrated high catalytic activity for water oxidation [86]. These studies demonstrated that the presence of an EDG led to an increase in the rate of the reaction with a decrease in the TON, whereas an EWG yielded a higher TON with a decrease in the rate [86]. The parent complex [Ru II (terpy)(bipy)(Cl)] + (13a) in this study was shown to be catalytically active for water oxidation with a TON of 390 using CAN as an oxidant, and it was suggested that the mechanism for complex 13a involved a seven-coordinate intermediate retaining the Ru-Cl bond [86]. However, in contrast to these observations, Sakai and coworkers demonstrated that complex 13a was inactive in the presence of NaCl in solution. This was thought to be due to a dominant shift of the substitution equilibrium, [Ru II (terpy)(bipy)Cl] + + Solv ⇌ [Ru II (terpy)(bipy)(Solv)] 2+ + Cl − , towards the reactant. Moreover, since the oxygen evolution as a function of time suggested that the chloro species was inactive, the real catalyst responsible for dioxygen evolution was inferred to be the aqua species, [Ru II (terpy)(bipy)(Solv)] 2+ , in solution. This suggested that the conversion of the [Ru(terpy)(bipy)Cl] + complex (13a) to the [Ru(terpy)(bipy)(H2O)] 2+ species (12a) may have been involved in the mechanism [75].
Although several studies have proposed a tentative mechanism for water oxidation involving seven-coordinate ruthenium intermediates [82,86,89], it was not possible to isolate and characterize these proposed complexes. This left an open question as to the interaction of water molecules with ruthenium in mononuclear catalysts. The use of negatively charged ligands was thought to be an appropriate means to capture high-valent ruthenium intermediates as they can stabilize higher oxidation states. Given the enhanced catalytic performance of dinuclear ruthenium complexes (9) and (10) with a dicarboxylato ligand where the introduction of negatively charged ligands dramatically lowered the oxidation potential of Ru II to Ru III [72,73], equatorial backbone ligands with terminating carboxylato groups were also introduced in the design of mononuclear ruthenium complexes. The synthesis of a mononuclear ruthenium complex, [Ru(bda)(pic)2] (14a) (where, bda 2− = 2,2′-bipyridine-6,6′-dicarboxylate), was shown to stabilize a possible seven-coordinate Ru IV dimeric intermediate with a proposed [HOHOH] − bridging ligand [90]. This supported the hypothesis that the O-O bond formation could arise from the coupling of two Ru IV =O units, termed as the "interaction between two metal oxo units" or the intermolecular bimolecular (I2M) pathway for water oxidation.
Subsequently, two mononuclear ruthenium complexes, [Ru II (pdc)(pic)3] (15) and [Ru II (pdc)(bipy)(pic)] (16) (where, H2pdc = 2,6-pyridinedicarboxylic acid), were investigated for their catalytic activity in chemical water oxidation [91]. Complex 15 displayed a TON of 553 and a TOF of 0.23 s −1 , which was better than complex 16, which had a TON of 17 and a TOF of 7.2 × 10 −3 s −1 at pH 1.0 for chemical water oxidation in the presence of Ce IV ions [91]. Although both 15 and 16 employed tridentate equatorial backbone ligands containing negatively charged biscarboxylato groups, they were not as catalytically active as [Ru(bda)(pic)2] (14a) [53], which displayed a TON of 2000 and a TOF of 41 s −1 under similar reaction conditions. Upon closer examination, the tetradentate equatorial backbone with two axial picoline ligands in complex 14a formed a highly distorted octahedral configuration with an "open coordination site" (O-Ru-O angle of 123°) that greatly facilitated the access of an aqua ligand [50,90]. The isolation of a Ru IV dimeric intermediate with a [HOHOH]ˉ bridging ligand from water oxidation catalyzed by complex 14a suggested that radical coupling of Ru=O units led to O-O bond formation [90]. As a result of this finding, isoquinolines were employed as axial ligands to facilitate the non-covalent attraction between them and lower the barrier of interaction for the Ru=O units. This strategy succeeded as the complex [Ru(bda)(isq)2] (14b) (where, isq = isoquinoline) and led to a TON of 8369 and a TOF of 303 s −1 [53].
The catalytic studies involving mononuclear ruthenium catalysts described thus far required a powerful sacrificial oxidant, CAN, for chemical water oxidation. In principle, it should be possible to use a light-absorbing photosensitizer to conduct sustainable lightdriven water oxidation. As described by Åkermark and coworkers [92], a major obstacle that is frequently encountered in light-driven water oxidation is the mismatch between the relatively high redox potential at which a catalyst assumes its active state and the lower potential attainable with a photosensitizer. One way to decrease the redox potential of the active catalyst is to involve PCET, which is a fundamental process that is employed in nature by the OEC of PSII [25,27]. It involves the simultaneous transfer of an electron and a proton, which has a profound effect on the energetics of the water oxidation reaction. As mentioned in the Introduction section, PCET allows for redox leveling at the catalytic site, which is a prerequisite for carrying out the four-electron water oxidation reaction. Additionally, another means of altering the redox potential of the active catalyst is to coordinate electron-donating and redox-active ligands to the metal centers, which would influence the balance between efficiency and stability of the water oxidation catalysts [92]. Thus, Åkermark and coworkers demonstrated that the introduction of imidazole and phenol motifs, in combination with carboxylate groups, facilitated PCET and the formation of high-valent metal-oxo catalytic intermediates at low potentials. This strategy was implemented by the development of two mononuclear ruthenium complexes, [Ru III (L)(pic)3] (where, L = 2-(2-hydroxyphenyl)-1H-benzimidazole-7-carboxylate (17) and L = 2-(2-hydroxyphenyl)-1H-benzimidazol-7-ol (18), which contained negative equatorial backbone ligands comprised of imidazole and phenol motifs with a carboxylate group [92]. By using the imidazole motif, it was possible to introduce a combined redox and proton-transfer mediator, a highly active and essential element, into the mononuclear ruthenium catalysts. Complex 17, with a single carboxylate and phenol moiety, displayed a TON of up to 4000 and a TOF of 7.4 s −1 with [Ru(bipy)3] 3+ as an oxidant for chemical water oxidation, and a postulated [Ru V =O] n+ intermediate of 18 was characterized by high-resolution mass spectrometry [92]. Moreover, to evaluate the possibility of performing lightdriven water oxidation under homogeneous, neutral conditions at pH 7.2, the authors employed a three-component system consisting of complex 17 or 18, a photosensitizer ([Ru(bipy)3] 2+ or [Ru(bipy)2(deeb)] 2+ (where, deeb = 4,4′-di(ethoxycarbonyl)-2,2′-bipyridine)), and a sacrificial electron acceptor (Na2S2O8). Successful evolution of molecular oxygen was detected upon visible-light illumination of this system.  [91,93].
The design of a mononuclear ruthenium complex with a carboxylate-amide motif, [Ru II (HL)(pic)3] (19) (where, L = 6-carbamoylpicolinic acid), was also shown to catalyze water oxidation (TON of 280 and TOF of 1.16 s −1 ) at a neutral pH of 7.2 using [Ru(bipy)3] 3+ as a mild chemical oxidant [94]. This complex was similar to [Ru II (pdc)(pic)3] (15), with the difference that one of the carboxylate ligands was replaced by an amide group. The crystal structure of complex 19 revealed a Ru III ion due to the strong electron-donating ability of the 6-carbamoylpicolinic acid ligand. The presence of the carboxylate-amide ligand in 19 lowered the redox potential of the complex to an extent where catalytic water oxidation could take place under neutral conditions with the mild [Ru(bipy)3] 3+ oxidant [94]. In comparison with the mono-nuclear ruthenium complex, 19, catalysts based on neutral nitrogen containing heterocyclic ligands were generally not compatible with the mild oxidant, [Ru(bipy)3] 3+ . This study once again highlighted the importance of incorporating anionic backbone ligands to decrease the redox potential of ruthenium catalysts.
The above results led to the design of the complex [Ru III (H2pdca)(pic)3] + (20) (where, H4pdca = 2,6-pyridine-dicarboxamide), which was also shown to catalyze water oxidation at a low redox potential using [Ru(bipy)3] 3+ at pH 7.2 with a TON of 400 and a TOF of 1.6 s −1 [95]. The improvement of the catalytic activity in terms of TOF was attributed to the presence of a flexible equatorial backbone ligand. This was followed by the synthesis of a seven-coordinate mononuclear ruthenium complex, [Ru IV (OH)(tda-κ-N 3 O)(py)2] + (21) (where, tda 2− = 2,2′:6′,2″-terpyridine-6,6″-dicarboxylate)) that was found to be an active and robust catalyst with a maximum TOF (TOFmax) of 50,000 s −1 at pH 10.0 using a foot-ofwave analysis (FOWA) [54]. Based on density functional theory (DFT) calculations, it was proposed that the carboxylate moiety in the dianionic ligand, tda 2− , stabilized seven-coordinate intermediates in the high-valent oxidation state of the catalyst. Moreover, the dangling carboxylate group was a putative hydrogen-bonding site that could function as a proton acceptor and hence favor WNA. This could lower the free energy of the activation and lead to O-O bond formation [54]. To our best knowledge, the catalytic activity of complex 21 is the highest that has been reported in literature, albeit it uses FOWA.
The trianionic mononuclear ruthenium complex, [Ru V (O)(t5a-κ-N 2 O)(py)2] (22) (where, t5a 3− = 2,5-bis(6-carboxylatopyridin-2-yl)pyrrol-1-ide and py = pyridine), was demonstrated to be an efficient catalyst with a TOFmax of 9400 s −1 at pH 7 via the FOWA [96]. In this case, the highly anionic nature of the backbone could reduce the redox potential of the Ru IV /Ru V couple, and the flexibility of the carboxylate moiety could facilitate intramolecular proton transfer to facilitate O-O bond formation through WNA. However, unlike complex 21, which required the formation of a seven-coordinate Ru V =O intermediate during water oxidation, complex 22 was not thought to require a seven-coordinate intermediate for O-O formation. This was due to the geometrical distortion of 12° and increased anionic nature of 22 in comparison with 21 (Table 3) [51,96]. The ligands tda 2 ⁻ and t5a 3 ⁻ were shown to be flexible, adaptive, multidentate, and equatorial and were thus termed as "FAME ligands" [51]. We would like to refer the readers to a recent review article that is focused on ruthenium-based molecular catalysts with the ability to achieve seven-coordinate intermediates and unprecedented activity [51].
The modification of the axial ligands to enhance the catalytic performance of mononuclear ruthenium complexes was explored by Sun and coworkers. They designed the complexes [Ru(bda)(Im)2] (36) (where, Im = imidazole) and [Ru(bda)(Im)(DMSO)] (37) (where, DMSO = dimethylsulfoxide) [105,106], which contained both imidazole and DMSO as axial ligands. Complex 36, with two axial imidazole ligands, yielded a TON of 1150 and a TOF of 4.5 s −1 for chemical water oxidation [105]. In contrast, the complex [Ru(bda)(Im)(DMSO)] (37), which contained an imidazole and DMSO axial ligand, exhibited better stability and improved catalytic activity with a TON of 4050 and a TOF of up to 176.5 s −1 [105]. Detailed mechanistic investigations of the catalytic water oxidation reaction using kinetics, electrochemistry, high-resolution mass spectrometry, and density functional theory (DFT) calculations suggested the in situ formation of a Ru II complex with an accessible seventh coordination site. The measured catalytic activity and kinetics revealed the influence of the axial ligands on the catalytic activity, where the increase of catalytic activity for complex 37 with an axial imidazole and DMSO ligands was attributed to the unhindered coupling between terminal oxygen atoms [105]. The catalytic activity of mononuclear ruthenium complexes was shown to be further enhanced with a TON of 6200 and TOF of 506 s −1 by employing two bromo substituted pyrazole-based axial ligands, [Ru II (bda)(L)2] (38) (where bda 2− = 2,2′-bypyridine-6,6′-dicarboxylate and L = 4-Br-3methyl pyrazole). The enhanced catalytic activity of 38 was ascribed to the high hydrophobicity of the complex, which tended to favor dimerization and, hence, facilitate the I2M reaction pathway [48,106]. Complexes 36-38 presented the possibility of simultaneously observing the effects of the axial and equatorial ligand modifications. However, the modification of the equatorial backbone ligand, bda 2− , used in these catalysts has not been fully explored to date. This is most likely due to the challenges that are involved in the synthesis of substituted bda 2− backbone ligands [50]. This is an avenue that could lead to further improvements of the catalytic performance as the introduction of substituents on the bda 2− backbone has been shown to influence the mechanistic pathways of mononuclear catalysts [50].

Iridium Catalysts
Iridium-based homogeneous catalysts were not reported in the literature until 2008. Below, we briefly review the research in this area that has been summarized elsewhere through 2015 [7]. In 2008, Bernhard et al. studied bis-aqua iridium complexes ( Figure 5, 40a-40e) with cyclometalating ligands and a surrogate catalyst, [Ir III (ppy)2(bipy)] + (39) (where, bipy = 2,2′-bipyridine, ppy = 2-phenylpyridine), without coordinated aqua ligands [109]. One of the complexes (40d) achieved a TON of 2760 after a week in Ce IV solution (Table 4), while the investigation of catalytic activity of the surrogate catalyst displayed no oxygen evolution. This finding suggested that open coordination sites (for water ligands) were necessary for catalysis [109]. In 2009, Crabtree and coworkers developed precatalysts based on iridium with pentamethylcyclopentadienyl (Cp*) and 2-phenylpyrimidine (or ppy) ligands [110]. The precursor, [Ir III (Cp*)(ppy)(Cl)] (41a), catalyzed water oxidation driven by Ce IV with a TOF of 54 min −1 , which was considered the most active catalyst at that time [110]. The [Ir V (Cp*)(O)(ppy)] + species was proposed as a possible intermediate in agreement with electronic structure calculations by DFT [110]. The following year, Crabtree and coworkers developed iridium half-sandwich complexes, such as Cp*Ir(N-C)X, [Cp*Ir(N-N)X]X, and [CpIr(N-N)X]X (where, X = Cl, I or NO3ˉ), as precatalysts for water oxidation, which have been reviewed in the literature [7].
DFT calculations indicated that an O-O bond was formed by the intermolecular attack of water to an iridium oxo ligand, and proton transfer to the oxo group was assisted by an additional water molecule [111]. A parallel study on a tris-aqua complex, [Cp*Ir(H2O)3](NO3)2 (42b), was conducted by MacChioni and coworkers, which highlighted that both the synthetic ease to the tris-aqua complex, [Cp*Ir(H2O)3](NO3)2 (42b), as well as the high activity with TOFs up to 15.7 min −1 with long-term activity [112]. Another complex, [Cp*Ir(bzpy)NO3] (44) (where, bzpy = 2-benzoylpyridine), with NO3¯ as a ligand to impart water solubility, was examined with a TON of 1250 and a TOF of 8.46 min −1 , as measured by UV-Vis under CAN [112,113]. The value increased to 12.7 min −1 when the oxygen evolution was measured by a Clark electrode [113] and up to 31 min −1 by probing different catalyst concentrations [114].
Iridium complexes with a carbene-type ligand were developed by Bernhard and coworkers in 2010 and two resulting complexes, 45 and 46, exhibited excellent stability with TONs of 10,000 and 8350 within 5 days, respectively [115]. In 2011, Crabtree and coworkers incorporated an N-heterocyclic carbene (NHC) ligand and Cp* to form the complex [Cp*Ir(NHC)Cl] (47) as a precatalyst that displayed moderate activity with a TOF of 8 min −1 driven by CAN and a TOF of 12-16 min −1 with sodium periodate (at pH 5); deactivation of the precatalyst was ascribed to the low pH in Ce IV solution under the former conditions [116]. As incorporation of a carboxylate moiety into ligands on ruthenium proved to be an effective strategy to improve catalytic activity (vide supra), MacChioni and coworkers developed a series of iridium complexes, 48a-48d and 49, as precatalysts for water oxidation [114]. The complexes 48a and 48b exhibited impressive performances with TOFs of 287 min −1 and 277 min −1 under optimized conditions with Ce IV as sacrificial oxidant, respectively, whereas complex 49, bearing a dicarboxylate moiety, turned out to be the least active [114]. Complex 48d, with a pendant -COOH moiety, featured the lowest potential and exhibited a TOF of 17 min −1 [114]. In 2012, MacChioni and coworkers developed the precatalyst [IrCl(Hedta)]Na (50) with monoprotonated ethylenediaminetetraacetic acid (edta) with a goal to replace Cp* as it slowly degraded in harsh acidic oxidative conditions [117]. The Hedta ligand is easy to synthesize and can stabilize metal ions in high oxidation states [117]. The complex (50) exhibited excellent stability with a TON up to 12,000, whereas it showed only moderate TOFs of 5.4 to 7.3 min −1 [117]. An interesting finding was the neutral complex [Ir(H2O)(Hedta)], which catalyzed water oxidation with a TOF of 2.6 min −1 . It was suggested that the complex [IrCl(Hedta)]¯ underwent ligand exchange with water to generate [Ir(H2O)(Hedta)] for water oxidation [117]. In 2014, MacChioni et al. developed Cp*-based precatalysts 41b, 51a, 51b, 52, 53, 54, and 55, some with strongly electron-donating amido ligands. These complexes showed stabilities through TONs > 500 for all the precatalysts examined. The complexes [Cp*Ir(bimH2)Cl]Cl (54) (where, bimH2 = 2,2′-bismidazole) and [Cp*IrCl(μ 2 -κ 2 -κ 1 -bimH)IrCl2Cp*] (55) displayed high catalytic activities with TOFs of 26 min −1 and 58 min −1 , respectively, as measured by a Clark electrode. Complex 52, bearing an NHC ligand, showed a TOF of 7 min −1 from UV-Vis data [118], which agreed well with analogous complex [Cp*Ir(NHC)Cl] (47) with a TOF of 8 min −1 reported by Crabtree et al. [116]. Although an NHC ligand can effectively stabilize high-valent iridium (IV), as evidenced by the EPR observations [116], Cp*Ir precatalysts incorporating the NHC ligand did not display high catalytic activity.
Molecular iridium catalysts experienced noteworthy progress in 2015 when Brudvig, Crabtree, Schmuttenmaer and coworkers developed an inactive molecular iridium catalyst, [Cp*Ir(pyalc)OH] (56), that dimerized in NaIO4 solution to generate a highly active system once bound to an oxide surface. The so-called het-WOC (56-het) can electrochemically catalyze water oxidation with low overpotential (η = 14 mV) and a high TOF (7.9 s −1 ) and stability, with a TON of 10 6 [119]. This het-WOC is different from a benchmark IrO2 nanomaterial (60-100 nm) that required 580 mV to reach an oxygen evolution rate of 6.6 s −1 per electroactive iridium atom [119].

Catalysts Based on First Row Transition Metals
First-row transition metals have been attracting research attention in the design of efficient molecular catalysts for water oxidation due to their abundance and low cost. However, the development of water oxidation catalysts based on first-row transition metals is relatively limited by virtue of being substitutionally labile, which indicates that ligands coordinated to these metals can exchange with water and/or solvents [4]. In many cases, initial precursor complexes decompose to catalytically active metal oxide particle precipitates. Therefore, studies on water oxidation catalysts based on first-row transition metals should be carefully evaluated to rule out the formation of corresponding metal oxide particles responsible for catalyzing water oxidation. To overcome these challenges, ligands need to be structurally engineered with high metal ion affinity (even in aqueous media) to permit access to and stabilize elevated oxidation states.

Manganese Catalysts
With manganese present in the natural OEC, synthetic complexes based on manganese have attracted significant scientific interest over the years. In early studies, Brudvig, Crabtree and coworkers had proposed that the dimanganese di-μ-oxo complex, [Mn III,IV 2O2(bipy)4][ClO4]3, was active in chemical water oxidation [124,125]. Although numerous synthetic models containing Mn have been constructed in the past 30 years [126], only a few of the complexes have demonstrated catalytic activity for water oxidation [127,128]. The first synthetic functional Mn cubane model, Mn4O4(O2P(Ph)2)6, was synthesized in 1997 [129]. This complex, as well as the Mn4O4((MePh)2PO2)6 catalyst reported by Dismukes and coworkers [130,131], can release dioxygen with UV light absorption in the gas phase as shown in Eq.
These two functional cuboidal clusters release dioxygen (concomitant with ligand dissociation) when excited under photochemical conditions. It has been proposed that an intramolecular two-electron charge transfer weakened Mn-O bonds and considerable distortion of the complex resulted in loss of the phosphinate bridge [130,131]. The distortion of core bonds allowed for the oxygen atoms to move in close proximity to form a peroxy intermediate, thereby lowering the activation barrier for dioxygen formation [130,131]. Subsequently, a synthetic Mn4Ca cluster featuring high structural resemblance to the natural OEC was found to be inactive in water oxidation [132].
In 1994, a chelating bis-porphyrin model was used to form stable catalytic complexes, 62, 63, and 64 ( Figure 6). Complex 64 showed the highest activity at high potentials (1.8-2.0 V vs. Ag/Ag + ) with a TON of up to 9.2 in acetonitrile containing [nBu4N][OH] [133] ( Table 5). The key intermediate involved in water oxidation, a dinuclear Mn V =O (65) (Figure 6), was characterized by spectroscopic methods; however, the mechanism of O-O bond formation was inconclusive as there were two pathways to form an O-O bond: either by the attack of external water on the H2O-Mn V =O group or by the coupling two Mn V =O units in the dimer [134].
In 1999, Brudvig and coworkers synthesized a functional model, [Mn III,IV 2(µ-O)2(terpy)2(H2O)2](NO3)3 (66), for water oxidation with a TOF of 0.67 s −1 [138,139]. The success of the model encouraged the group to develop a series of complexes with the general formula, [Mn III,IV 2(µ-O)2(L)2(X)2] 3+ (L = terpy, terpy derivatives and X = labile ligand), with their catalytic performance assayed in situ and in pure solution with oxone as the oxidant [140]. The parent catalyst with terpy, 66, exhibited the best catalytic performance and the in situ solution showed relatively higher catalytic rate than that of corresponding pure complex, which was ascribed to additional oxone solution [140]. An 18 O isotope labeling experiment was carried out to study the mechanism of water oxidation by the complex with oxone as an oxidant due to the fast exchange of oxygen in hypochlorite. It was pro-

Iron Catalysts
The development of iron-based complexes for water oxidation started in 2010 when Bernhard and Collins and coworkers employed tetra-anionic tetra-amido macrocyclic ligands (TAML), given this ligand type was known to effectively stabilize high oxidation states of transition metals and were robust against oxidation [144]. Fe-TAMLs will catalyze the evolution of dioxygen from water; complex 72e ( Figure 7) exhibited the best catalytic activity with a TOF of 1.3 s −1 and a TON of 16 using CAN as an oxidant at pH 0.7 [144,145]. The catalytic activity of Fe-TAMLs increase from 72a to 72e (Figure 7), with complex 72a the least stable showing no dioxygen evolution [145]. Another Fe-TAML complex (73a) with a methylamino bridging group (-NMe) was reported in 2014 by Dhar and coworkers with an improved TON of 220, whereas the TOF of the complex decreased to 0.67 s −1 with the photosensitizer [Ru(bipy)3] 2+ and Na2S2O8 as the sacrificial oxidant [146]. This complex, 73a, was found to catalyze WO driven under chemical conditions by CAN with a TON of 10 and a TOF of 0.03 s −1 ; the lower performance was ascribed to demetalation at pH 1. To improve the stability of Fe-TAMLs in acidic solution, a nitro group (-NO2) was introduced to further optimize catalytic performance [146]. This modification to 73b showed a slight improvement with a TON of 17 and a TOF of 0.06 s −1 compared to complex 73a under chemical WO with Ce IV as an oxidant. Under photochemical WO conditions, 73a exhibited better activity than 73b with a TON of 60 and a TOF of 0.21 s −1 . Supported by characterization data from techniques such as electron paramagnetic resonance (EPR), ultraviolet-visible spectroscopy (UV-Vis), and high-resolution mass spectrometry (HRMS), it was proposed that a high-valent Fe V =O intermediate was generated during photochemical irradiation. Likewise, O-O bond formation was said to involve a nucleophilic attack by water at Fe V =O to generate a Fe III -hydroperoxo intermediate, which then oxidized to release dioxygen and regenerate the original Fe III -TAML aqua complex [146].
A systematic study of iron complexes for water oxidation was reported by Lloret-Fillol and Costas et al. in 2011 [147]. Neutral tetradentate organic ligands were employed to form complexes 74-80 (Figure 7), and their catalytic performances are listed in Table 6. To compare other metals with the same Me2Pytacn ligand (where, Me2Pytacn = 1-(2′-pyridylmethyl)-4,7-dimethyl-1,4,7-triazacyclononane), complexes [M(OTf)2(Me2Pytacn)] (M = Ni, Co, Mn) were synthesized and it was found that these metal complexes were inactive for water oxidation, whereas complex 74, featuring Fe, was highly active with a TON of 70 and a TOF of 0.12 s −1 using CAN as the oxidant [147]. A structural feature shared among complexes 74-78 is that they exhibit two cis labile coordination sites, which could serve as a rationale in the design of iron-based catalysts for WO. In support of this, amongst complexes 74-80, the latter two (79 and 80) are inactive; 79 features two trans labile ligands, and complex 80 has only one available (solvated) coordination site. An isotope labeling experiment involving complex 74 confirmed water (as opposed to NaIO4) as the oxygen source for dioxygen and a kinetic study showed Fe IV =O to be an intermediate (resting state) as characterized by UV-Vis and electrospray ionization mass spectrometry (ESI-MS). The mechanism was suggested to involve the oxidation of Fe IV =O to form Fe V =O, which undergoes attack by water to form Fe III -hydroperoxide required for O-O bond formation. Fe III -hydroperoxide will then undergo further oxidation by Ce IV to form Fe IV -hydroperoxide, which catalyzes the release of dioxygen [147]. DFT and time-dependent DFT calculations [148] were employed to study the catalytic cycle driven by Ce IV . A more energetically favorable pathway was suggested by evoking the +IV oxidation state for all oncycle intermediates. In 2014, Sun and coworkers explored 11 iron-based complexes, all of which featured 2 cis labile coordination sites. In addition, carboxylate or ethanolate moieties were incorporated into the design of iron-based catalysts although these modifications were not found to be as effective as similar designs for ruthenium-based catalysts for WO [149]. Dinuclear complexes structurally mimicking the 'blue dimer' (vide supra) were found to be inactive toward WO [149]. Only 2 complexes out of the 11 that were examined, 81 and 82, catalyzed WO to generate dioxygen with a moderate catalytic performance, with a TOF of 0.1 s -1 for complex 82 [149]. This study also ruled out the presence of nanoparticulate iron oxide in the system, which can be responsible for WO [149].
In 2015, two other aminopyridyl iron complexes, 83 and 84, were reported by Costas and Lloret-Fillol and coworkers to be topological isomers that achieved TONs of 160 and 380, respectively, and TOFs of 0.16 s −1 and 0.28 s −1 , respectively [150]. The mechanism of dioxygen evolution was proposed to involve an iron-oxo-cerium intermediate, O=Fe IV -O-Ce IV , which was characterized by UV-Vis, cryospray high-resolution mass spectrometry (CSI-HRMS), and resonance Raman spectroscopy [150]. A different mechanism based on the formation of an iron-oxo-cerium adduct was proposed for complex 83 in water oxidation, and this heteronuclear dinuclear species was experimentally characterized [150]. Two iron-based complexes, 85 and 86, with different tetraazadentate ligands were reported by Thapper and coworkers in 2016 [151]. These two complexes leveraged a design wherein labile ligands are oriented in a cis conformation. Complexes 85 and 86 catalyze WO with TONs of 14 and 2 and TOFs of 0.18 s −1 and 2 × 10 −3 s −1 , respectively, using Ce IV as oxidant [151].
Thummel et al. reported an interesting Fe III (dpa) complex, 87, with a square planar tetradentate polypyridyl ligand featuring two labile trans ligands that evolves dioxygen with a TOF of 0.23 s −1 driven by Ce IV [152]. Another two mononuclear polypyridyl ironbased complexes with the pyridine-2,6-diylbis[di(pyridin-2-yl)methanol] (Py5OH) ligand were developed by Thapper and coworkers in 2016 [153]. The only difference between complexes 88 and 89 is that the axial chloride ligand in 88 was replaced by a methanol ligand in 89; both are catalytically active under acidic Ce IV /[Ru(bipy)3] 3+ -driven chemistry, however, their activity depends on specific conditions for WO. For instance, complex 88 exhibited better catalytic performance than complex 89 under neutral pH with [Ru(bipy)3] 3+ as an oxidant for WO (TOF of 2.2 s −1 vs. 0.9s −1 ); however, the opposite results were observed when acidic conditions were employed. Additionally, the axial chloride ligand played a role in stabilizing catalyst 88 relative to 89, which degraded at a higher pH (10.5 vs. 9.5) [153]. Dynamic light scattering experiments were employed to examine the integrity of the complexes; no iron oxide (nano)particles were found in these systems, yet these particles were detected when the pH achieved ≥9 [153].
Complexes 79, 81, 90, and 91 are representatives of iron catalysts containing secondary or tertiary amino ligands. The first two complexes were discussed earlier in this section. In 2015, Lau et al. reported that complex 90 with a cyclam ligand exhibited catalytic activity toward WO when NaIO4 was used as oxidant, with TONs up to 1030 [154]. In the same year, Zhan and coworkers reported the first water-soluble iron-cyclen complex (91) for electrochemical WO with a TOF up to 65 s −1 [155]. However, the result from this study was proven to be incorrect due to the formation of iron oxide particulates, which were shown by Najafpour et al. to be responsible for WO [156].
Impressive progress in iron-based catalysts for water oxidation was made in 2016 when Kawata and Masaoka and coworkers reported that a penta-nuclear iron catalyst achieved a TON of 10 7 and a TOF of 1900 s −1 under electrochemical water oxidation [157]. Intramolecular O-O bond formation was proposed as the mechanism that involved two iron-oxo moieties [157]. The dinuclear iron complexes, 92-95, have also been developed and examined. Complex 92, reported in 2014 by Najafpour and coworkers, can catalyze water oxidation with a TOF of 0.72 s −1 in the presence of Ce IV , whereas the catalytic activity of complex 93, reported by Ding and Ma and coworkers, exhibited a TON of 2380 and a TOF of 2.2 s −1 using oxone as an oxidant [158,159]. Related bis(μ-O) complexes, 94 and 95, were investigated for water oxidation. In the presence of Ce IV , 94 showed reasonable activity with a TOF of 2.2 s −1 [152], while electrocatalytic water oxidation was achieved with complex 95 with a TOF of 1.2 s −1 [160].

Cobalt Catalysts
The study of cobalt-based catalysts for water oxidation started approximately a decade ago. The design of such catalysts has been challenging given that the (pre-)catalyst complexes are prone to degradation under harsh (oxidative) conditions and the substitutional lability of cobalt, particularly as Co 3+ . When a cobalt-based complex degrades, the Co 2+/3+ ion forms as the corresponding hydroxide (and/or oxide), which can efficiently catalyze water oxidation to release dioxygen. Numerous studies have claimed new cobaltbased molecular catalysts with impressive catalytic performances [161,162], but upon further investigation, cobalt oxide nanoparticles were shown to be the catalytically active species responsible for WO [163,164].
In 2010, Hill and coworkers reported a cobalt-based molecular catalyst, [Co4(H2O)2(PW9O34)2] 10− , with a polyoxometalate (POM) ligand (polytungstophosphate) for water oxidation with a TOF of >5 using [Ru(bipy)3] 3+ as an oxidant in sodium phosphate buffer at pH 8 [161]. In 2011, Hill and coworkers reported that the cobalt-POM, [Co4(H2O)2(PW9O34)2] 10− , was an efficient catalyst for photon-driven water oxidation with a TON of 224 using the photosensitizer [Ru(bipy)3] 2+ , Na2S2O8 as the sacrificial electron acceptor, and light [165]. However, a study on the cobalt-POM, [Co4(H2O)2(PW9O34)2] 10− , under electrochemical conditions revealed that degradation of the catalyst resulted in the formation of CoOX, which was determined to be the real catalyst for WO [163]. The absorbance at 580 nm measured by UV-Vis spectroscopy decreased over a period of 3 h, which corresponded to the degradation of the catalyst and formation of CoOx films (identified by UV-Vis, scanning electron microscopy (SEM), and energy-dispersive X-ray (EDX)). The activity of CoOX was confirmed by observing similar catalytic activity from an electrode with deposited CoOX films as compared to cobalt-POM. The film deposited on an electrode was rinsed prior to insertion into a solution of sodium phosphate buffer at pH 8 (without the cobalt-POM), and the catalytic performance by the controlled-potential electrolysis was similar to the cobalt-POM system [163].
Inspired by the topology of the OEC of PSII (i.e., a tetranuclear manganese-calciumoxo cluster, Mn4CaO5), several tetranuclear cobalt complexes for WO have been developed [143,[166][167][168]. In 2011, Dismukes and coworkers reported a tetra-cobalt complex, [Co4O4(μ-OAc)4(py)4] (96), featuring a cuboidal Co4O4 core that was found to catalyze water oxidation by a light-driven process using [Ru(bipy)3] 2+ as the photosensitizer and Na2S2O8 as the sacrificial electron acceptor (TON >40 and TOF of 0.02 s −1 ) [143] (Table 7). Scandola and coworkers also studied the photocatalytic activity of [Co4O4(μ-OAc)4(py)4] (96) and reported a quantum yield of 0.3 [167]. Following this work (in 2012), Sartorel and coworkers developed a series of isostructural tetra-cobalt complexes with para-substituted pyridine ligands [166]. Six complexes were generated (97a-97f) and screened under photochemical conditions, as described above, e.g., components of light, [Ru(bipy)3] 2+ , and Na2S2O8 [166]. These complexes have TONs of approximately 140, and amongst them, 97a has the largest electron transfer rate (2.51 × 10 8 M −1 s −1 ) and the highest quantum efficiency (80%) [166]. The overpotential of these complexes varies in a narrow range of 0.50-0.57 V, indicating no apparent effect from the substitution on the pyridine [166]. Nocera and coworkers questioned the catalytic activity of these complexes with a report identifying a Co(II) impurity in Co4O4(OAc)4(Py−X)4 as being responsible for the activity in water oxidation [169]. The study qualitatively and quantitively suggested the presence of Co(II) by EPR, nuclear magnetic resonance (NMR) line broadening analysis, and electrochemical titration. The EPR spectrum of Co4O4(OAc)4(Py−COOMe)4 that was purified by column chromatography does not show a broad line feature over the g = 10 to 2 range, whereas the spectrum of the crude material showed a broad signal over the same range [169] that was suggestive of Co(II) [170]. To quantitatively confirm the presence of a Co(II) species, a 31 P NMR line broadening analysis was employed to determine the concentration of Co 2+ ions, which was on average 0.16 mM for a crude sample of 852 ppm [169]. Moreover, EDTA was used to titrate solutions of crude and purified Co4O4(OAc)4(Py−COOMe)4; CV showed almost complete catalytic suppression of crude sample with EDTA, whereas the purified sample with EDTA exhibited CV wave reversibility [169].
In 2016, Fukuzumi et al. reported a dinuclear cobalt complex, [Co III (TPA)(μ-OH)2Co III (TPA)] 4+ (100) (where, TPA = tris(2-pyridylmethyl)amine), as a molecular catalyst for light-driven water oxidation using [Ru(bipy)3] 2+ and Na2S2O8 [171]. The complex achieved a high quantum yield (44%) and a TON of 742 without forming CoOx particles; the study employed NMR and dynamic light scattering to rule out the formation of nanoparticles [171]. During the period of photocatalytic activity under the same condition in a deuterated borate buffer, 1 H NMR spectra of the solution showed that the catalyst remained intact over 2 h, and dynamic light scattering did not support the formation or presence of nanoparticles. Chemical water oxidation with [Ru(bipy)3] 3+ as an oxidant was also carried out and dioxygen evolved with a TON of 4.3.
Another dinuclear cobalt complex was published recently with bridging μ-OH and μ-O2 structural features [172]. Under photochemical water oxidation conditions with [Ru(bipy)3] 2+ and Na2S2O8, the complex afforded dioxygen yield of 46.7%, however, this value plummeted to 9.6% when bipy was added as a chelating agent, and complete deactivation was observed with an elevated bipy concentration [172]. Upon the addition the chelating agent, it was observed that these cobalt polyridy complexes possessing a μ-OH/μ-O2 structural motif (98, 99, and 101 in Figure 8) do not catalyze water oxidation [172].
An unusual cobalt-based water oxidation catalyst, 102, with oligopyridine ligands was developed by Lau and coworkers in 2014 [173]. The complex is highly distorted from octahedral geometry and exhibits a double helical type structure by X-ray crystallography [173]. It was found to be active in chemical water oxidation with [Ru(bipy)3] 3+ as an oxidant (TON of 56 and a TOF of 1.9 s −1 ) and photochemically active catalyst using [Ru(bipy)3] 2+ photosensitizer and Na2S2O8 as a sacrificial electron acceptor (TON of 442) [173]. Dynamic light scattering was used during photocatalysis and no (nano)particles were observed. In addition, inductively coupled plasma atomic emission spectroscopy (ICP-AES) was used to measure supernatant cobalt content of a solution that was centrifuged post-photocatalysis, and the result showed that 98.1% of the total cobalt remained in solution. The titration of terpy (to scavenge free Co 2+ ions) reduced oxygen evolution by around 10%, which was ascribed to the oxidation of the free terpy ligand [173]. Nocera and coworkers reported a cobalt complex (103) with aqua ligands positioned on each cobalt so as to investigate the mechanism of oxygen evolution reaction (OER) at the edge sites of cobaltate clusters; 18 O-labeling experiments demonstrated the mechanism of the water oxidation by the complex involved intra-molecular coupling between edge-site oxygen atoms [174].
Berlinguette and Thapper and coworkers reported similar mononuclear Co(II) polypyridyl complexes in 2011 and 2015, respectively [175,176]. The difference between these two complexes (104 and 105) is primarily the axial ligand, which is an aqua ligand in 104 (a dicationic species) and chloride one in complex 105 (a monocationic complex). Complex 104, [Co(PY5)(OH2)] 2+ (where, PY5 = 2,6-(bis(bis-2-pyridyl)-methoxymethane)pyridine), was found to electrochemically catalyze water oxidation over the pH range of 7.6 to 10.3 [175]. The CV of 104 showed two oxidation processes at 0.75 V (pH 2.2) and 1.43 V over the pH range of 7.6-10.3, respectively. The first oxidation process was related to the [Co III -OH] 2+ /[Co II -OH2] 2+ redox couple, while the second oxidation proposed the formation of a [Co IV -OH] 3+ intermediate [175]. The formation of an O-O bond was suggested through the nucleophilic water attack on a Co(IV)-oxo species, which purportedly formed upon the deprotonation of a [Co IV -OH] 3+ species [175]. The mechanism of water oxidation involving complex 104 was studied by DFT and was suggested to involve a [ − ⃛ ] 2+ species which can be interpreted as a cobalt-oxene moiety [Co II -(•O•)] 2+ [177]. This cobalt-oxene moiety was thought to have enhanced the radical character of oxygen, thereby promoting nucleophilic attack of hydroxide substrate on oxygen to form the O-O bond [177]. In consideration of the 'oxo wall' concept, established for latetransition-metal oxo complexes (Groups 9-11), high-valent metal oxo species are rare, unstable, and require careful examination for Co, Ni, and Cu when these species are claimed as intermediates in catalytic reactions [178]. Complex 105, [Co(PY5OH)(Cl)] + , was inspired by studies of complex 104 and features free hydroxy groups (in lieu of methoxy), which allows for further functionalization [175,176]. It was found to be active in chemical water oxidation (with [Ru(bipy)3] 3+ as an oxidant) with a TON of 15 and photochemically active (using [Ru(bipy)3] 2+ and Na2S2O8) exhibiting a TON of 51 and a TOF of 1.3 s −1 . Incorporation of a chloride ligand in complex 105 is important considering improved observed catalytic activity relative to [Co(PY5OH)(OH2)] 2+ (axial aqua ligand replacing axial chloro in 105), which exhibited lower catalytic activity (TON of 6.5 using [Ru III (bipy)3] 3+ as chemical oxidant).
Wang and coworkers synthesized mononuclear cobalt complex 106 with an esterfunctionalized (-COOMe) TPA ligand (TPA = tris(2-pyridylmethyl)amine); photo-induced catalysis showed a TON of 127.7 and a TOF of 3.8 s −1 using [Ru(bipy)3] 2+ and Na2S2O8 [179]. In comparison, complex 107 with non-functionalized TPA showed reduced activity by roughly an order of magnitude (TON of 12.6 and TOF 0.3 s −1 ); the high efficiency of complex 106 was attributed to the moderately more electron-withdrawing nature of COOMe-functionalized TPA (or steric considerations), thought to increase the stability of 106 by precluding self-oxidation and dimerization [179]. Interestingly, both complexes 106 and 107 have chloride ligands, which enhanced the catalytic activity relative to complex 104.
Lau and coworkers reported the cobalt complex, 108, with a quaterpyridyl backbone and axial aqua ligands. This complex was shown to be both chemically and photochemically active for WO displaying a TON of 160 and initial TOF of 4 s −1 with [Ru(bipy)3] 3+ as the oxidant, and a TON of 335 after irradiation at 457 nm for 1.5 h in the presence of photosensitizer [Ru(bipy)3] 2+ and sacrificial electron acceptor Na2S2O8 [180]. The study by Lau et al. suggested that the actual mechanism is likely dependent on the stability of ligandmetal-oxo species, LM=O, vs. metal oxide (MO), and that a strongly chelating and oxidation-resistant ligand is more likely to support molecular LM=O for water oxidation. In the absence of such a ligand, LM=O is unstable with respect to the decomposition to MO, which is active as the real catalyst for water oxidation [180]. In 2012, Fukuzumi and coworkers reported a series of polyamino-based mononuclear cobalt complexes (109)(110)(111) and one Cp*-based cobalt complex with bipyridine shown as complex 112 [181]. Evidence has been presented to indicate that both 109 and 112 converted to nanoparticles under photocatalytic water oxidation conditions and that Co(OH)X nanoparticles were responsible for the activity [181]. The catalytic activity of Co(NO3)2 as pre-catalyst was examined as a reference material to demonstrate the catalytic ability of complexes 109 to 112 and the yield of dioxygen evolved from water oxidation with these complexes increases in the order of 111 < 112 < 110 < Co(NO3)2 < 109 [181].
In 2013, Sakai and coworkers reported a series of water-soluble porphyrin-based cobalt complexes 113a-113c that photochemically catalyzed water oxidation using the photosensitizer, [Ru(bipy)3] 2+ , and sacrificial electron acceptor Na2S2O8 [182]. The TOF of 113c was found to be dependent on the pH and the highest activity was achieved at pH 11 with a TOF of 0.17 s −1 [182], and the TOF of 113c can reach up to 0.62 s −1 with pH 9 [183]. Due to the reported observation of a second order dependence on catalyst concentration, the mechanism for water oxidation catalyzed by complexes 113a-113c was proposed to involve radical coupling of two Co(III) or Co(IV) oxyl species [182]. Oxyl radical involvement was supported via DFT calculations, however, a Co(IV) or oxyl radical species was not observed in the EPR spectroscopy studies [182]. Modification to the porphyrin ligand at the meso position of complex 113c yielded a fluorinated Co-porphyrin. Moreover, 113d not only exhibited enhanced catalytic activity for water oxidation with a TON of 571 and a TOF of 1.1 s −1 , but also fundamentally changed the mechanism for water oxidation from oxyl-oxyl coupling for 113c [182] to nucleophilic attack by water on Co(IV)=O species due to the observed first-order dependence on catalyst concentration for 113d [184]. A similar enhancement of catalytic activity was observed for 113e by the replacement of fluoride atoms with larger (blocking) chloride atoms.
Groves and coworkers also investigated porphyrin-based cobalt complexes for water oxidation and a representative of the type is cationic complex 114, which electrochemically catalyzed water oxidation to dioxygen under neutral conditions (pH 7). Complex 114 was confirmed as the active molecular catalyst with much evidence; EDTA titration showed that cobalt ions were not released from the porphyrin during catalysis, and the analysis of the working electrode surface by energy dispersive X-ray spectroscopy showed no evidence of deposition [185]. In 2018, Najafpour and coworkers reported Co II (phthalocyanine) (115) and Co II (salen) (116). These complexes were screened for electrochemical catalytic water oxidation, and it was found that 116 was active at pH 11, whereas no dioxygen was observed for complex 115 [186]. High resolution visible spectra showed peaks at 410-430 nm and 680-740 nm that are characteristic of nanoparticles, which is indicative of the nanoparticulate nature of the real catalyst. Gross and coworkers reported a cobalt corrole complex, 117, as electrochemically active for water oxidation [187].
In 2014, Lin and coworkers reported a copper complex, [(dhbp)Cu(OH2)2] (121) (where, dhbp = 6,6′-dihydroxy-2,2′-bipyridine) that mimicked the function of the redoxactive tyrosine residue, YZ, of PSII as a mediator in the electron transfer process [192]. The electrocatalytic complex, 121, leveraged a non-innocent redox active ligand with pendant hydroxyl groups that likely participated in PCET so that the overpotential was 510 to 560 mV. Due to difficulties in the characterization of intermediates, the proposed intramolecular PCET process could not be determined experimentally [192]. In 2015, Llobet and coworkers reported a new family of copper catalysts, 122a-122d, using derivatives of a tetra-anionic tetradentate amidate ligand that reduced the overpotential to 170 mV by the tuning of the electronic structure in this series of Cu complexes [193].
Wei et al. reported a new type of copper-azo catalyst, 124, for chemical water oxidation driven by Ce IV with an initial TOF of 4.0 kPa h −1 . The proposed mechanism for O-O bond formation was the coupling of two Cu III =O units [194]. Najafpour and coworkers reinvestigated the catalytic performance of 124, under both chemical and electrochemical conditions for water oxidation, and found no oxygen evolution in the presence of Ce IV at pH ~ 1. The EDX results of a film formed on a fluorine-doped tin oxide (FTO) electrode after bulk electrolysis indicated low amounts of Cu that suggested decomposition of the catalyst. Combined with cyclic voltammetry (CV), which displayed an oxidation peak comparable to water oxidation mediated by CuO, it suggested that the true catalyst was a copper oxide film that was formed during the measurements [195].
In 2017, Sun and coworkers investigated the difference in electrocatalytic performance of two similar copper complexes; the diamine-dipyridine (129) and the diiminedipyridine (130) complexes [199]. Complex 130 showed a lower overpotential of 700 mV and a higher TOF of 50.4 s −1 relative to complex 129, with an overpotential of 1070 mV and a TOF of 13.5 s −1 [199]. Moreover, CPE results showed that 130 displayed better stability than 129, which was possibly due to the fact that the degradation of the C=N bonds in the ligand of 130 was avoided [199]. The author attributed the difference of catalytic activity between 129 and 130 to the conjugate effect and the type of the coordinating N atom of the ligand in molecular catalysts [199]. The study highlighted a possibility of the simultaneous enhancement of the activity, decrease in the overpotential, and improvement of the stability of molecular catalysts through fine tuning the ligand structures. Sun and coworkers also reported two copper complexes (131 and 132) bearing amine-pyridine ligands for electrochemical water oxidation, which displayed TOFs of 13.1 s −1 and 18.7 s −1 with overpotentials of 440 mV and 570 mV, respectively, at pH 11.5 [200].
Cao and Lai et al. reported a copper catalyst, 133, that incorporated a dianionic tridentate ligand, with a carbonate group in the equatorial plane, that electrochemically catalyzed water oxidation at pH 10 with an observed overpotential of 650 mV [201]. Computational studies indicated that the HCO3ˉ group functioned as an intra-molecular base (assisted in proton removal from water) and facilitated O-O bond formation when the purported oxo radical intermediate [LCu III -O •¯] experienced water nucleophilic attack [201]. In 2018, inspired by the cuboidal Mn4CaO5 cluster in the OEC of PSII, Wu and coworkers reported the first set of copper-cubane complexes (134a, 134b) as effective electrocatalysts for water oxidation with TOFs of 267 s −1 and 105 s −1 at 1.70 V and 1.56 V, respectively. The EDX, XPS, and electrochemical studies excluded the formation of a film during catalysis and in situ EPR and Raman spectroscopy measurements indicated highvalent Cu III and Cu III O • as intermediates in the catalytic cycle [202].
Catalyst 135 was reported by Sakai and coworkers and was found to follow secondorder kinetics that suggested a bimolecular radical coupling mechanism was at play [203]. Meanwhile, Najafpour and coworkers reinvestigated stabilities and activities of copper complexes 135-138 comparing them to a copper(II) salt, i.e., copper(II) sulfate pentahydrate. The copper(II) salt was found to have the highest activity for water oxidation among 135-138 from linear sweep voltammetry (LSV) and CV [204]. Catalyst 135, 136 and 138 are stable at pH 11, except catalyst 137, which decomposed to form Cu 2+ ions during longterm amperometry suggesting copper ions are the true catalyst for water oxidation [204]. This explained why catalyst 137 displayed better activity than the rest of molecular catalysts examined (135,136,138) [204]. In 2019, Cao and coworkers reported a copper-porphyrin complex, 139, which catalyzed water oxidation electrochemically with a TOF of 30 s −1 and a low overpotential of 310 mV at neutral pH conditions [205]. Moreover, it can catalyze water oxidation to hydrogen peroxide in acidic conditions at pH 3 [205]. In 2020, Zhang, Verpoort and coworkers reported a copper-based electrocatalyst, [Cu(TCA)2] (140) (where, HTCA = 1-mesityl-1H-1,2,3-triazole-4-carboxylic acid), that displayed first-order kinetics and a single-site mechanism was implied [206].
Interestingly, complex 141 ( Figure 10) also served as an electrocatalyst for hydrogen production by water reduction [209]. Multiple characterization techniques, such as, SEM, EDX and dynamic light scattering (DLS), were employed to verify that complex 141 was a stable active homogenous catalyst. Additionally, X-ray analysis indicated that the catalyst remained intact after being recovered from controlled electrolysis.  [209]. Therefore, an HO-OH coupling mechanism from the triplet cis Ni intermediate was the preferred pathway for O-O bond formation.
Complex 141 displayed relatively low catalytic performance with a TON of 15 (Table  9), which may have been due to the conformational change during catalysis from an inactive trans-isomer to the active cis-isomer. Recall, in a study involving an iron-based molecular catalyst for water oxidation, Costa and coworkers indicated the importance of two cis labile coordination sites is a key structural feature for catalysis [147]. Thus, Lu et al. hypothesized that design of a six-coordinate Ni II complex with two cis labile sites to facilitate the HO-OH coupling, would improve catalytic performance. In 2016, a Ni II complex, 143, possessing two cis labile sites for electrocatalytic water oxidation was reported by Lu et al., and the catalyst was found to be less efficient than complex 141, which suggested that more Ni II complexes need to be investigated to understand the structure-performance relationship [210]. This study also revealed that base (acetate buffer) played a role in facilitating the formation of a key intermediate and aided in atom-proton transfer (APT) for O-O bond formation. No catalytic activity was observed when complex 143 was in phosphate buffer due to occupation of two cis labile sites by bidentate HPO4 2− , whereas complex 143 displayed catalytic activity in acetate buffer [210]. The acetate buffer not only aided in generation of a peroxide intermediate, [Ni II L(OOH)] 2+ , formed through intramolecular O-O coupling, but it also acted as a proton acceptor to lower the barrier to O-O bond formation [210]. Meyer and coworkers had reported that added base could enhance the rate in a catalytic water oxidation with a Ru complex, [Ru(bda)(isq)2] (where, bda = 2,2′-bipyridine-6,6′-dicarboxylate and isq = isoquinoline) [211]. Similar findings were observed with the complex, Co-porphyrin, where the base played multiple roles in the catalytic cycle [185]. In this case, the sodium phosphate buffer (Na-Pi) was reported to: (1) act as a proton acceptor for the conversion of Co III -OH to Co IV -O intermediate, (2) act as a base to deprotonate water coupled with the rate-limiting O-O bond formation step, and (3) inhibit catalytic water oxidation activity at high buffer concentrations [185]. In addition, Meyerstein et al. used carbonate and phosphate buffer to investigate their role in electrochemical water oxidation catalyzed by Ni IV (meso-L)X2 (L = 1,4,8,11-tetraazacyclotetradecane; X= CO3 2− or PO4 3− ) [212]. The study highlighted the role of carbonate in stabilizing Ni III L (L = 1,4,8,11-tetraazacyclotetradecane) and engaging in the redox process as a noninnocent ligand [212].
Lu and coworkers investigated the role of buffer in a homogeneous electrocatalytic system involving complex 144 that exhibited a moderate overpotential of ~480 mV in phosphate buffer at neutral pH. The electrocatalysis results indicated a dual role of HPO4 2− as a proton acceptor to facilitate PCET and as an inhibitor for electrolysis by replacing cis labile sites with HPO4 2− [213]. Based on the catalytic activity of copper complexes with polypeptide ligands [190,196], a nickel-glycine electrocatalyst was reported by Allen and coworkers with an overpotential of 475 mV in phosphate buffer at pH 11. However, further electrochemical studies suggested that the formation of nickel oxide was responsible for catalysis, and XPS confirmed the presence of a nickel hydroxide film [214]. Ni-POM was reported as a homogeneous catalyst for photochemical water oxidation in 2012 by Hill and coworkers [215] and by Wang and coworkers in 2015 [216]. The pentanickel silicotungstate complex, K10H2[Ni5(OH)6(OH2)3(Si2W18O66)]·34H2O, with photosensitizer [Ru(bipy)3] 2+ and Na2S2O8 as a sacrificial electron acceptor in sodium borate buffer at pH 8, revealed that the complex can catalyze photo-driven water oxidation with a TON of 60 and a TOF of ~0.18 s −1 . Various techniques, such as UV-Vis, infrared spectroscopy (IR), and DLS, were employed to demonstrate the stability of the complex in buffered solution; the integrity of the complex remained for one month [215].
In 2017, Lu, Ke and coworkers investigated the steric effect on nickel complexes with a macrocylic ligand reminiscent of complex 141 by studying two nickel complexes, 147 and 148, with four or six methyl groups, respectively [219]. The results showed that [Ni(Me8L)](ClO4)2 (148) was more active in electrocatalysis than [Ni(Me6L)](ClO4)2 (141) which outperformed [Ni(Me4L)](ClO4)2 (147), suggesting that the steric repulsion between methyl groups and phosphate anions bound axially to the in situ formed Ni III can affect catalytic activity. Bruner et al. reported that a nickel phenolate complex, 149, can catalyze water oxidation electrochemically under neutral conditions with a TOF of 0.15 s −1 [220]. Ding and coworkers reported a homogeneous nickel complex (150a) bearing an o-phenylenebis(N'-methyloxamidate) ligand for electrochemical water oxidation with a TOF of 0.4 s −1 in phosphate buffer at pH 11 [221]. Two other complexes, investigated by Ding and coworkers, that share a similar structure to 150a but have the functional groups with nitrogen atoms replaced by one or two carboxyl groups displayed instability and decomposed to form NiOX [221]. This suggested that nitrogen functional groups are more beneficial than carboxyl groups to stabilize nickel-based homogeneous catalysts. Llobet and coworkers further investigated the electronic effects on the catalytic activity by modification of the o-phenylene moiety of complex 150a to generate complexes 150b and 150c. The introduction of electro-donating groups on the phenyl ring in 150b and 150c decreased the overpotential to ~ 170 and 220 mV, respectively.
The study also demonstrated that 150b shared similar reversible behavior in the first redox wave as complex 150a, however, 150c displayed a prominent decrease in the reduction wave, suggesting the presence of deactivation [222]. In 2019, Verpoort and coworkers reported the first Ni-NHC complex (151) for homogeneous electrochemical water oxidation catalysis in sodium acetate buffer (NaOAc) and sodium phosphate buffer (Na-Pi) at pH 7 and pH 9 [223]. The study also demonstrated improved catalyst activity in the phosphate buffer with an overpotential of 800 mV at pH 9, and the complete loss of activity in the acetate buffer, which highlighted the recent finding on the role of phosphate ions as a proton acceptor to enhance catalytic activity in PCET [223]. In 2020, Sun and coworkers reported a homogeneous Ni-TAML complex (152) (where TAML is a redox-active ligand) for electrochemical water oxidation with a TOF of 0.32 s −1 in neutral phosphate buffer with an overpotential of 680 mV [224]. The mechanistic study proposed the generation of [Ni III (TAML 2− )-O • ] attacked by a water molecule for O-O bond formation indicating a baseassisted APT process [224].

Conclusions
The development of molecular catalysts for water oxidation has experienced rapid progress over the past two decades, especially ruthenium (4d metal-based) catalysts. The design principles that were employed for ruthenium-based catalysts are being applied in the development of molecular catalysts based on first-row (3d) metals, including Mn, Fe, Co, Cu, and Ni, as well as iridium (a 5d) metal. The development of efficient and robust molecular catalysts based on the first-row metals is challenging. This is due, in part, to the fact that 4d metals (ruthenium) can bind a ligand backbone better as compared to firstrow (3d) metals due to larger atomic radii, which allows for a stronger interaction (overlap) between the central metal ion and its ligands and a reduction in the repulsion between all coordinated ligands. In contrast, the first-row metals, Mn, Fe, Co, Cu, and Ni, are generally substitutionally labile, which indicates that the ligands coordinated to the metals can undergo ligand exchange (with solvent), thus limiting the development of water oxidation catalysts based on these metals. Additionally, an added complication is the decomposition of the first-row metal catalysts to corresponding oxides and/or hydroxides that are also catalytically active for water oxidation. This can interfere with quantitative measurements of the catalytic activity of molecular catalysts of the first-row transition metals. Although the current catalytic performance of molecular catalysts based on the first-row transition metals is moderate, it remains promising and is a worthwhile endeavor given their abundance and economic feasibility.

Conflicts of Interest:
The authors declare no conflict of interest.