The use and burning of fossil fuels have resulted in a series of energy and environmental problems. This has made it necessary to find and use new energy resources that are sustainable, clean, and environmentally benign [1-4]. Solar energy is the most promising energy source for future consumption. Artificial photosynthesis, achieved via water splitting, enables the harvesting and use of solar energy by converting it to chemical energy [5-8]. This process mimics natural photosynthesis, in which solar energy is used to oxidize water, with evolution of O₂ and generation of protons and electrons. The released O₂ is essential for living beings, and the protons and electrons extracted from water participate in the reduction of carbon dioxide to produce carbohydrates, a process by which solar energy is stored in chemical forms. If the protons and electrons combine directly to produce H₂, this process enables light-driven water splitting to O₂ and H₂ gases. The production of H₂ gas by this method is clean and carbon-free [9-11], and reaction of the produced H₂ with O₂ (i.e., in fuel cells) generates electrical energy with water as the only product. Water splitting is therefore important in new energy technologies.

Water oxidation, which is one of the half reactions in water splitting, is challenging both thermodynamically (2H₂O→O₂ + 4H⁺ + 4e^-, △E = 1.23 V vs normal hydrogen electrode (NHE)) and kinetically (extensive rearrangement of atoms) [1, 3]. Water oxidation therefore limits the overall efficiency of water splitting. Catalysts are required for water oxidation. Much effort has been made recently to design and develop efficient and stable catalysts for this process [12-49]. In general, catalysts can be divided into material catalysts and molecular catalysts. Molecular complexes have the following advantages over material catalysts. (1) Their structures are much clearer and are much easier to modify. (2) They are more amenable to mechanistic studies in solution by various spectroscopic methods. (3) The structure–function relationships of molecular catalysts can be systematically studied. (4) They can be easily integrated into molecular assemblies for the construction of practical water oxidation devices. (5) The key O-O bond formation step can be better understood both experimentally and theoretically in molecular systems.

Recently, mononuclear metal complexes have been found to be active and stable catalysts for water oxidation [3, 50, 51]. Although ruthenium and iridium complexes show high activities [52-70], the low natural abundances and high costs of these noble-metal elements limit their widespread applications. The use of cheap and efficient catalysts based on earth-abundant metal elements is therefore desirable. This review article summarizes recent rapid progress in mononuclear WOCs based on first-row transition-metal elements, namely manganese, iron, cobalt, nickel, and copper. Particular attention is paid to catalytic mechanisms and the key O-O bond formation steps. Progress in this area has enabled advances in catalyst design. We believe that such information is critical for designing new catalysts that are more efficient and stable than currently available catalysts.

Nature uses a Mn₄CaO_x cluster to catalyze the oxidation of water to O₂ [71-73]. On the basis of this natural process, many Mn complexes have been synthesized and investigated for use as WOCs. The first well-studied mononuclear Mn catalyst for water oxidation was reported by Sun and co-workers in 2007 [74]. The Mn–corrole complex 1 (Fig. 1) was designed as a bio-inspired model of the oxygen-evolving center of photosystem Ⅱ. Cyclic voltammetry (CV) of 1 in dichloromethane showed two reversible redox events at E_1/2 = -0.32 and 0.61 V (vs Ag/Ag⁺), assigned to Mn^Ⅱ/Mn^Ⅲ and Mn^Ⅲ/Mn^Ⅳ, respectively. This low oxidation potential of Mn^Ⅲ/Mn^Ⅳ indicates that it can be easily oxidized to its high-valent state for water oxidation. On addition of n-Bu₄NOH, the cyclic voltammogram of 1 displayed an additional peak at E_pa = 0.79 V (vs Ag/Ag⁺). This new peak arises from water oxidation and the evolved O₂ was detected by electrochemical reduction at -1.29 V (vs Ag/Ag⁺). Privalov and Sun [75] subsequently reported a detailed computational study of O-O bond formation with the monomeric Mn–corrole 1. The formally Mn^Ⅴ =O species was suggested to be responsible for O-O bond formation. This d² Mn–oxo species has a closed-shell singlet ground state, but is converted to a quintet Mn^Ⅳ–oxyl species in a polar solvent. Two O-O bond formation processes were compared: a concerted pathway and a two-step pathway via coordination of a hydroxide anion on the Mn ion to produce MnOOH. A density functional theory (DFT) study showed that the concerted pathway was more favorable, with an activation energy barrier of 8-10 kcal/mol. However, it should be noted that the energy difference between these two processes is moderate. It is therefore possible that the reaction proceeds via both reaction pathways simultaneously.

This O-O bond formation step was further probed by Sun and co-workers [76], who used a Mn complex with 5, 10, 15-tris(4-nitrophenyl)corrole (2, Fig. 1) to investigate nucleophilic attack by a hydroxide anion on a Mn^Ⅴ=O unit. The oxidation of 2 with t-BuOOH afforded the Mn^Ⅴ=O species, which was characterized using various spectroscopic methods. In a subsequent reaction with n-Bu₄NOH, O₂ was rapidly evolved. The Mn^Ⅳ–OO^- intermediate, which is probably formed by nucleophilic attack of a hydroxide anion on the Mn^Ⅴ=O unit, followed by one-electron oxidation, was examined using electronic absorption spectroscopy, high-resolution mass spectrometry (HRMS), and isotopic ¹⁸O-labeling experiments. The results show that one atom of O₂ came from n-Bu₄NOH and the other came from water. On the basis of these experiments, Sun and co-workers proposed the reaction mechanism shown in Fig. 2. This work is significant because it shows that nucleophilic attack of a hydroxide anion on a Mn^Ⅴ =O unit can form an O–O bond, and this sheds light on water oxidation at the oxygen-evolving center of photosystem Ⅱ.

In 2010, Anxolabéhère-Mallart and co-workers [77] reported activation of a water molecule by electrochemical oxidation of a Mn^Ⅱ–OH₂ complex to a Mn^Ⅳ=O species. The intermediates Mn^Ⅱ–OH₂, Mn^Ⅲ–OH, and Mn^Ⅳ=O were all investigated. Extended X-ray absorption fine structure analysis showed a gradual shortening of the Mn–O bond from Mn^Ⅱ–OH₂ to Mn^Ⅲ–OH and Mn^Ⅳ=O. Experimental results and DFT calculations suggested that the Mn^Ⅳ=O unit is best described as Mn^Ⅲ–oxyl. Although catalytic water oxidation was not observed using this Mn complex, this work elucidates the activation of a water molecule on a Mn center, which is relevant to the natural water oxidation system.

Another study of Mn^Ⅴ=O species, by Borovik and co-workers [78], confirmed its participation in water oxidation. In this system, the Mn center rather than the oxygen ligand was oxidized, giving Mn^Ⅴ=O instead of a Mn^Ⅳ–oxyl radical. Isotopic labeling experiments using ¹⁷O-labeled samples showed appreciable spin density on the oxo ligand in Mn^Ⅴ=O; this suggests a possible radical coupling reaction of this species during O–O bond formation.

In 2014, Smith and co-workers [79] reported the catalytic activity of Mn^Ⅱ complexes with pyridinophane (3–5, Fig. 1) in O₂ formation. Complexes 3 and 4 with small substituents on the nitrogen (R = H and Me) catalyzed the disproportionation of H₂O₂, whereas complex 5, which has a bulky t-Bu substituent, was not active in H₂O₂ disproportionation but was active in catalytic water oxidation to evolve O₂. The onset potential for catalytic water oxidation with 5 is about 1.30 V (vs NHE) at pH = 12.2, corresponding to an overpotential of about 0.8 V. Controlled potential electrolysis of 5 at 1.23 V (vs NHE) gave turnover numbers (TONs) of 16–24 with Faradaic efficiencies of (74–81)%. Smith and co-workers [80] used DFT methods to investigate the water oxidation mechanism in the reaction with 5 as the catalyst. A Mn^Ⅴ–bis(oxo) species was proposed as the catalytically active species, from which the O-O bond was formed via an intramolecular radical coupling reaction, with a calculated reaction energy barrier of 14.7 kcal/mol (Fig. 3). It was suggested that the pyridinophane ligand is important for various reasons, e.g., (1) it provides an appropriate electronic environment for accessing multiple oxidation and spin states and (2) it has the right geometry and flexibility for creating a cis divacant octahedral environment for O-O coupling.

Liao and co-workers [81] also used DFT methods to investigate the mechanism of water oxidation catalyzed by 5. Two pathways can lead to the formation of an O-O bond from the Mn^Ⅴ–bis(oxo) species, i.e., either direct intramolecular coupling of the two oxo ligands or nucleophilic attack of the Mn^Ⅴ =O unit by a water molecule. For the direct coupling pathway, the calculated reaction energy barrier was 13.8 kcal/mol, and for the pathway involving nucleophilic attack by water, the calculated energy barrier was 13.1 kcal/mol. In the latter mechanism, one oxo ligand is attacked by a water molecule and the other oxo ligand functions as a base and accepts a proton from the water molecule. Both pathways could therefore contribute to catalysis.

In 2014, D'Eramo and co-workers [82] reported the water oxidation activity of a water-soluble Mn^Ⅲ–chloride complex with 5, 10, 15, 20-tetrakis(4-N-methylpyridyl)porphyrin (6, Fig. 1). Complex 6 catalyzed water oxidation in aqueous phosphate buffer solutions at both pH = 7 and 10. However, the chloride anions of the Mn^Ⅲ–porphyrin were also oxidized during this process, and the catalytic current decreased by approximately 40% in the case of a chloride-free Mn^Ⅲ–porphyrin.

In 2016, Dey and co-workers [83] reported that Mn–corrole 7 (Fig. 1) was active in water oxidation in strong basic aqueous solutions. Two oxidation events appeared, at 0.53 and 0.78 V (vs Ag/AgCl), in the cyclic voltammogram of 7 in acetonitrile, corresponding to Mn^Ⅳ/Mn^Ⅲ and Mn^Ⅴ/Mn^Ⅳ, respectively. An additional irreversible wave appeared at 1.38 V (vs Ag/AgCl). Significantly, the intensity of this wave increased on addition of a NaOH aqueous solution, indicating catalytic O₂ evolution. The function (i_cat/i_p)² (where i_cat is the catalytic current, i_p is the non-catalytic peak current) increased linearly with NaOH concentration, to give a pseudo-first-order rate constant of 11.4 s^-1 with 25 mmol/L NaOH. Catalytic water oxidation was further confirmed by using an edge-plane pyrolytic graphite electrode surface-loaded with complex 7 in strong basic solutions. At pH = 11.0, large catalytic currents were observed at potentials > 1.30 V (vs Ag/AgCl). Controlled potential electrolysis of 7 at 1.40 V (vs Ag/AgCl) gave a TON of 1.90 × 10⁴ over 11.1 h and a turnover frequency (TOF) of 0.47 s^-1. The corresponding Faradaic efficiency for O₂ evolution was calculated to be 82%.

Iron is widely used for O₂ activation in nature [1], and its catalytic properties in water oxidation have attracted much recent attention. In 2010, Collins and co-workers [84] reported the first examples of mononuclear Fe complexes for water oxidation. A series of Fe^Ⅲ complexes with tetraamido macrocyclic ligands (Fe–TAML, 8a–8e, Fig. 4) were synthesized and examined. Complex 8a was inactive, but 8b–8e were active. Complex 8e showed the highest activity, with a TOF of 1.3 s^-1; this shows that electron-withdrawing ligands favor this process. Previous studies by Collins and co-workers [85] suggested that Fe^Ⅴ=O could be generated in the Fe–TAML system. This indicates that a high-valent Fe^Ⅴ=O species (S = 1/2) is probably the key intermediate in water oxidation catalysis. Cramer and co-workers [86] used DFT methods to investigate the reaction mechanism with 8e as the catalyst. Water nucleophilic attack on a Fe^Ⅴ=O unit leads to O–O bond formation. It is worth noting that the supporting TAML ligand must be oxidized by one electron prior to this water nucleophilic attack. Liao and co-workers [87] also studied this process using DFT methods. The calculated activation energy barrier for water nucleophilic attack on TAML^•+–Fe^Ⅴ =O was 15.4 kcal/mol.

In a subsequent study, Kitchin and co-workers [88] immobilized a Fe–TAML complex on glassy carbon or carbon paper for electrocatalytic water oxidation, and obtained a TOF of 0.081 s^-1 with a 45% Faradaic efficiency. Dhar and co-workers [89] reported photochemical water oxidation with a biuret-modified Fe–TAML complex (9, Fig. 4). The maximum TON was 220, and the TOF was 0.76 s^-1. A Fe^Ⅴ=O species was photochemically generated; its formation was supported by electron paramagnetic resonance, ultraviolet-visible (UV-Vis), and HRMS studies. This Fe^Ⅴ=O intermediate is responsible for O–O bond formation. The proposed reaction mechanism with complex 9 is shown in Fig. 5. Irradiation at 440 nm in the presence of Na₂S₂O₈ and [Ru(pyridyl)₃]²⁺ transforms the Fe^Ⅲ–aqua complex into Fe^Ⅴ=O via a two-step proton-coupled electron-transfer (PCET) process. The Fe^Ⅴ=O undergoes water nucleophilic attack to form the O–O bond. The resulting Fe^Ⅲ–OOH is further oxidized by two electrons to evolve O₂ and regenerate the starting Fe^Ⅲ–aqua complex.

In 2011, Fillol and co-workers [90] examined a series of mononuclear Fe complexes for use as water oxidation catalysts. These Fe complexes contained various nitrogen-based neutral tetradentate or pentadentate ligands. Fe complexes 10–14, which have two vacant coordination sites in a cis arrangement, were active (Fig. 4), whereas those containing two trans sites or only one vacant site were inactive in water oxidation. A Fe^Ⅳ=O intermediate was suggested on the basis of UV-Vis spectroscopic and ESI-MS results in titration experiments with cerium ammonium nitrate (CAN). Further oxidation of this Fe^Ⅳ=O intermediate with CAN generated a Fe^Ⅴ=O species, which rapidly reacted with a water molecule to form the O–O bond. TONs > 1000 were obtained. On the basis of these results, the authors proposed the reaction mechanism shown in Fig. 6.

In a subsequent study, Fillol and co-workers [91] studied the reaction mechanism and the electronic effects of these Fe-based catalysts on water oxidation. Complex 10 and a family of its derivatives with various substituents on the pyridine ring were examined. Electron-withdrawing groups at the para position of the pyridine ring were favorable. In addition, it was found that a key intermediate was formed between Fe^Ⅳ=O and Ce^Ⅳ prior to formation of a Fe^Ⅴ=O species. UV-Vis titration and kinetic studies suggested that this intermediate was Fe^Ⅳ(O)–(OH)–Ce^Ⅳ, with a bridging hydroxyl group. Fe^Ⅳ(O)–(OH)–Ce^Ⅳ was converted to Fe^Ⅴ(O)–(OH)–Ce^Ⅲ, and a water molecule subsequently attacked the resulting Fe^Ⅴ=O unit to form the O–O bond. This proposal was supported by DFT calculations. The presence of Fe^Ⅳ(O)–(OH)–Ce^Ⅳ was confirmed experimentally by cryospray ionization HRMS and resonance Raman spectroscopy [92]. On the basis of these results, the authors proposed a possible reaction mechanism for the formation of this Fe^Ⅳ–O–Ce^Ⅳ intermediate and its subsequent conversion for O–O bond formation (Fig. 7). The identification of this Fe^Ⅳ–O–Ce^Ⅳ intermediate, which is responsible for the key O–O bond-forming step, provides new insights into the analogous Mn^Ⅴ–O–Ca^Ⅱ unit in the oxygen-evolving center of photosystem Ⅱ.

Bartlett and co-workers [93] reported that anchoring a phosphonate-derivatized Fe complex 12 on WO₃ through covalent bonds improved photoelectrochemical water oxidation. After anchoring the Fe complex on the WO₃ electrode, an increase in the photocurrent density was observed, indicating a fast chemical reaction in the presence of the Fe catalyst. In addition to the increased photocurrent density, a significant increase in the Faradaic efficiency was achieved. The Fe-modified WO₃ electrode gave a Faradaic efficiency of (79 ± 9)% for O₂ evolution, compared with (56 ± 7)% for the unmodified WO₃ electrode.

Similar Fe complexes containing two cis vacant coordination sites were reported to be active in water oxidation [94-96]. In 2013, Yang and co-workers [94] examined the water oxidation activities of a series of Fe complexes with hydrogen-bonding functionalities incorporated into the second coordination sphere. The results showed that complex 15 (Fig. 4) was the most active catalyst in this series, which suggests that intramolecular pendant base groups are favorable for water oxidation. In 2013, Fukuzumi and co-workers [95] reported two Fe complexes, i.e., 16 and 17 (Fig. 4), which can catalyze water oxidation by CAN. The TONs for O₂ evolution were 80 ± 10 and 20 ± 5 for 16 and 17, respectively. Significantly, the authors showed that under acidic conditions these Fe complexes acted as true homogeneous catalysts in water oxidation, whereas under basic conditions Fe(OH)₃ nanoparticles derived from the Fe complexes acted as heterogeneous catalysts. Ligand dissociation and oxidation are responsible for the decomposition of these Fe complexes under catalytic conditions. In 2016, Thapper and co-workers [96] reported that Fe complexes 18 and 19 (Fig. 4) are active WOCs. The addition of CAN to an aqueous solution of 18 or 19 resulted in formation of semi-stable Fe^Ⅳ =O species. However, the catalytic activities of these two complexes are much lower than those of their analogs.

In 2015, Che and co-workers reported a Fe complex with N, N'-dimethyl-2, 11-diaza[3, 3](2, 6)pyridinophane (20, Fig. 4) and its catalytic activity in water oxidation with a variety of chemical oxidants such as CAN, NaIO₄, and oxone [97]. Spectroscopic data and ¹⁸O-labelling experiments suggested that Fe^Ⅳ=O and/or Fe^Ⅴ=O intermediates were involved in the catalytic cycle. The reaction mechanisms differed, depending on the chemical oxidant. With CAN as the oxidant at pH = 1, Fe^Ⅳ=O was identified by ESI-MS and UV-Vis spectroscopy. The O₂ evolution rate had a linear dependence on the concentrations of both 20 and CAN. In contrast, when NaIO₄ or oxone was used as the oxidant at pH = 1, ESI-MS suggested that Fe^Ⅴ=O, with the formula [Fe^Ⅴ(O)₂]⁺, was involved. DFT calculations showed that this [Fe^Ⅴ(O)₂]⁺ intermediate oxidized water to generate O₂, with a calculated activation energy barrier of 15.7 kcal/mol.

In 2015, Thummel and co-workers [98] compared the water oxidation activities of mononuclear and μ-oxo-bridged dinuclear Fe^Ⅲ complexes. The μ-oxo-bridged dinuclear complex, with a TOF of 7920 h^-1, was much more active than the mononuclear analog, with a TOF of only 842 h^-1. Kinetic and spectroscopic studies suggested that the dinuclear Fe^Ⅲ complex remained intact in solution during the catalytic process. A Fe^ⅢFe^Ⅴ=O intermediate was proposed as the catalytically active species, which reacted with water to produce O₂.

Thapper and co-workers [99] reported that Fe–Py5 (Py5 = pyridine-2, 6-diylbis[di(pyridin-2-yl)methanol]) complexes, which contain only one vacant site, were active WOCs. However, the authors proposed that the Py5 ligand underwent de-coordination of one pyridyl group to generate an active species with two open/labile coordination sites available for water binding and activation. These pyridyl-based pentadentate ligands therefore resemble the previously described tetradentate ligands for Fe coordination.

In 2014, Meyer and co-workers [100] reported a six-coordinated Fe^Ⅲ–aqua complex 21 (Fig. 4) as an electrocatalyst for water oxidation in a mixed solution of propylene carbonate and water. The results of electrochemical and kinetic studies suggested that this Fe^Ⅲ–aqua complex was first oxidized to Fe^Ⅴ=O, which then reacted with a water molecule via a reaction that was first order with respect to both the catalyst and water. The determined k_o was 0.035(4) L/(mol·s). This result is consistent with a single-site mechanism instead of a bimolecular mechanism for water oxidation. The maximum TON was 29 over 15 h of electrolysis, giving a Faradaic efficiency for O₂ evolution of 45%. Importantly, a negligible kinetic isotope effect (KIE, 1.08) was observed, suggesting that a concerted oxygen atom–proton transfer mechanism was not involved, i.e., water nucleophilic attack on the Fe^Ⅴ=O unit probably gave the peroxide intermediate Fe^Ⅲ–OOH₂ rather than Fe^Ⅲ–OOH. The reaction mechanism proposed by the authors is shown in Fig. 8.

In 2011, Nocera and co-workers [101] synthesized hangman Co–corrole 22 (Fig. 9) and its β-octafluoro derivative 22-F₈, and showed that they have high water oxidation activities. The TOF of 22-F₈ was 0.81 s^-1 in neutral phosphate buffer with an applied overpotential of 780 mV. Significantly, both complexes are much more active than the non-hangman Co complex with 5, 10, 15-tris(pentafluorophenyl)corrole 23 (Fig. 9) in water oxidation catalysis. A Co^Ⅳ–corrole^•+ moiety was proposed as the catalytically active species. Ertem and Cramer [102] used DFT methods to study the reaction mechanism with 22-F₈. Lai and co-workers [103] also used DFT calculations to investigate the O-O bond formation step with 22-F₈. The formally Co^Ⅴ species has a Co^Ⅳ–corrole^•+ character and is responsible for O-O bond formation with 22-F₈. It was proposed that the pendant carboxyl group serves as an intramolecular base and accepts a proton during water nucleophilic attack on the Co–oxo unit. High-valent Co–oxo units are highly active and their isolation and characterization are challenging. However, Nam and co-workers [104] successfully prepared a stable terminal Co^Ⅳ=O complex. Although these authors did not observe catalytic activity of this Co^Ⅳ=O complex in water oxidation, this work provided experimental evidence for the generation of Co^Ⅳ=O under conditions relevant to water oxidation catalysis.

Cao and co-workers [105] investigated the water oxidation reaction with Co–corrole 23 in detail. The cyclic voltammogram of an indium tin oxide (ITO) electrode surface-loaded with complex 23 showed a pronounced catalytic wave with an onset potential of 1.15 V (vs Ag/AgCl) in phosphate buffer (0.1 mol/L; pH = 7.0). This wave is caused by water oxidation, as shown by the generation of numerous gas bubbles on the ITO electrode surface. The onset overpotential for water oxidation with 23 was 530 mV. The stability of 23 as a WOC was investigated. All the results indicated that 23 was stable and acted as a real molecular catalyst in water oxidation. A TOF of 0.20 s^-1 was obtained at an applied potential of 1.40 V (vs Ag/AgCl). In contrast, the Mn^Ⅲ analog of 23 showed much weaker catalytic activity and decomposed during electrocatalysis in neutral phosphate buffer.

On the basis of the experimental and theoretical results, a catalytic cycle with 23 was proposed (Fig. 10). Starting from Co^Ⅲ–corrole [Cor–Co^Ⅲ–OH₂], two sequential 1H⁺/1e^- oxidation steps lead to [Cor^•+–Co^Ⅲ–OH] and [Cor^•+–Co^Ⅲ–O^•-], respectively. The latter species undergoes water nucleophilic attack to produce [Cor–Co^Ⅲ–OOH]^-. The calculated activation energy barrier is 29.9 kcal/mol with a cluster of four water molecules acting as the proton acceptor, but it is 18.1 kcal/mol with an acetate anion acting as the proton acceptor. The transition states of these two pathways are shown in Fig. 11. Two 1H⁺/1e^- PCET steps then give [Cor–Co^Ⅲ–OO^•-]^- and [Cor^•+–Co^Ⅲ–OO^•-]. The resulting [Cor^•+–Co^Ⅲ–OO^•-] can release O₂ by displacement with a water molecule. This process has a low energy barrier of 3.6 kcal/mol. The O-O bond formation step is therefore the rate-determining step in the catalytic cycle.

In this mechanism, there is a Co-bound pyridine molecule located at the trans position of the Co–oxo unit. The electronic effects of certain ligands can significantly affect the stability and reactivity of their trans ligands. Because the O–O bond formation step is usually the rate-determining step in water oxidation, it is interesting to study the trans ligand effect in water oxidation; such studies have rarely been reported in the literature. Co–corroles are ideal for studying the trans effect because the corrole ligand can provide a stable and rigid square-planar environment for Co with two axial sites trans to each other. Recently, Cao and co-workers [106] synthesized a series of Co–corroles 23 with various axial ligands and examined their catalytic activities in water oxidation. These Co–corroles all showed two oxidation events in acetonitrile, and the oxidation waves shifted in the cathodic direction when electron-donating axial ligands were present. This trend is consistent with increased electron densities on the Co ions. When these Co–corroles were loaded on fluorine-doped tin oxide (FTO) electrodes, they were all active in electrocatalytic water oxidation, with large catalytic waves, in phosphate buffer solutions (0.1 mol/L; pH = 7.0). Importantly, Co–corroles containing electron-donating trans axial ligands were more active. Depending on the electron-donating ability of the trans axial ligands, the onset overpotentials are in the range 510–580 mV, and the catalytic currents at 1.68 V (vs NHE) range from 0.78 to 1.95 mA/cm². Possible reasons for the enhanced catalytic activity with electron-donating trans axial ligands are as follows. The strong trans effect can weaken the Co^Ⅴ =O bond and thus decrease the activation energy barrier for O–O bond formation via water nucleophilic attack on the Co^Ⅴ–O unit. These results show that the trans axial ligands on Co centers can considerably affect the water oxidation activities of Co–corroles.

To further study the hangman effect, Cao and co-workers [107] synthesized a series of Co–corroles with different acid/base pendants 24–27 (Fig. 9) and examined their water oxidation activities in neutral aqueous solutions. The synthesis of these Co–corroles is shown in Fig. 12. The aim of introducing different pendant groups, including –Br, –COOH, –PO(OH)₂, and –CH₂PO(OH)₂, at the 6-position of the dibenzofuran unit, was to increase the proton-accepting ability of the intramolecular base groups, and consequently to enhance the water oxidation performance by assisting the O–O bond formation step. Electrochemical measurements with catalyst-coated FTO working electrodes in phosphate buffers (0.1 mol/L; pH = 7.0) showed that the order of the water oxidation activities was 27 > 26 > 25 > 24. Importantly, a plot of potential (measured at 0.52 mA/cm) vs pK_a of the appended group (for –PO(OH)₂ and –CH₂PO(OH)₂, pK_a2 values were used) showed a linear relationship, with a slope of -55 mV per pK_a unit. This result indicates that the water oxidation mechanism involves a rate-determining PCET step (i.e., O–O bond formation), in which the electron-transfer process becomes more favored when proton transfer is facilitated by a strong basic group.

In 2016, Cao and co-workers [108] reported enhanced water oxidation activity of a pyrene-modified Co–corrole 28 (Fig. 9). A Co complex with 5, 15-bis(pentafluorophenyl)-10-(4)-(1-pyrenyl)phenylcorrole (28), bearing a pyrene substituent, was synthesized (Fig. 13). When 23 and 28 were immobilized on multiwalled carbon nanotubes (MWCNTs), the 28/MWCNT composite outperformed the 23/MWCNT composite, indicating that the pyrene substituent has a significant role in electrocatalysis. The strong non-covalent π-π interactions between the pyrene moiety and MWCNTs enhance the catalytic performance by facilitating fast electron transfer from the electrode to the molecular catalyst and also by increasing adhesion of the molecular catalyst to the carbon support. The presence of strong non-covalent π-π interactions between the pyrene moiety and MWCNTs is supported by spectroscopic data. For example, the electronic absorption spectra of 28 and the 28/MWCNT composite were recorded and compared. The absorption bands at 346 and 381 nm in the spectrum of complex 28 blue-shifted to 334 and 372 nm in the 28/MWCNT composite spectrum. This indicates strong interactions between molecules of 28 and the MWCNT support. The 28/MWCNT composite is also much more efficient than the 23/MWCNT composite in catalyzing oxygen reduction to water [108, 109]. Non-covalent immobilization of molecular catalysts on carbon supports through strong π-π interactions is therefore a simple and straightforward method for the fabrication of hybrid catalyst-carbon materials for efficient electrocatalysis.

Co–porphyrins are another class of Co-based complexes that have been studied as molecular WOCs. Du and co-workers [110] deposited Co–tetraphenylporphyrin or Co–tetrakis(4-bromophenyl)porphyrin on FTO electrodes for electrocatalytic water oxidation. These catalyst films showed good activities in borate buffer solutions (0.5 mol/L; pH = 9.2), with onset overpotentials of 530 and 580 mV, respectively. The molecular nature of these Co-based WOCs was confirmed. TOF values of 0.50 s^-1 for Co–tetraphenylporphyrin and 0.40 s^-1 for Co– tetrakis(4-bromophenyl)porphyrin were obtained in controlled potential electrolysis at an applied potential of 1.3 V (vs Ag/AgCl).

In 2013, Groves and Wang [111] reported electrocatalytic water oxidation with cationic water-soluble Co^Ⅲ–porphyrins (29–31, Fig. 9) in neutral phosphate buffer solutions. The order of the activities of these catalysts was 29 > 30 > 31. Complex 29 gave a TOF of 1.4 × 10³ s^-1 and a Faradaic efficiency of 90%. An inhibitory effect was observed at high phosphate concentrations (i.e., > 60 mmol/L). It was suggested that at low phosphate concentrations, phosphate anions can facilitate O-O bond formation by acting as a base and accepting the proton of the nucleophilic attacking water molecule. However, at high phosphate concentrations, phosphate anions can bind to the Co center and block the binding and activation of water on Co. The onset potentials at pH = 7 decrease linearly in buffers with more basic anions, with a slope of -54 mV per pK_a unit. This result is similar to that reported by Cao and co-workers [107], mentioned above.

In 2013, Sakai and co-workers [112] reported photocatalytic water oxidation with three water-soluble Co^Ⅱ–porphyrins (32–34, Fig. 9). Typically, Co–porphyrins were used as the catalyst, [Ru(bpy)₃]²⁺ was used as the photosensitizer, and Na₂S₂O₈ was used as the sacrificial oxidant. Under the optimum conditions, the TOF for O₂ evolution at pH = 11 was 0.138 s^-1 for 32, 0.118 s^-1 for 33, and 0.170 s^-1 for 34. Kinetic studies showed a second-order dependence of the initial reaction rate on the catalyst concentration; the reaction mechanism shown in Fig. 14 was proposed. However, these Co–porphyrin catalysts decomposed to less active (or inactive) species during catalysis. The decomposition of Co–porphyrins is probably caused by generation of singlet dioxygen (¹O₂) under photocatalytic conditions; this was confirmed by using 9, 10-diphenylanthracene as a chemical probe. Highly active ¹O₂ is a strong oxidant and rapidly reacts with organic substrates. Sakai and co-workers [113] used a new fluorinated Co–porphyrin 35 (Fig. 9) to improve the stability of Co–porphyrin catalysts. These bulky meso substituents were used to increase the tolerance of 35 to ¹O₂ attack through steric shielding. Photocatalytic studies showed that 35 was more efficient than 34 in catalyzing water oxidation: the TON with 35 was approximately twice that with 34. Kinetic studies indicated that the rate-determining step in water oxidation with 35 was water nucleophilic attack on the Co–oxo species.

Several Co complexes with Py5-based ligands have been reported as active WOCs [114-117]. In 2011, Berlinguette and co-workers [114] prepared a mononuclear Co complex with a Py5–OMe ligand (Fig. 15) and showed that it catalyzed water oxidation in phosphate buffer (0.1 mol/L; pH = 9.2). The attack of Co^Ⅳ–hydroxyl/oxo units by a water/hydroxide substrate was proposed to lead to O–O bond formation. In a subsequent study, these authors further confirmed the molecular nature of this Co-based WOC [115]. In 2015, Thapper and co-workers [116] reported a mononuclear Co complex with a Py5–OH ligand (Fig. 15), which was a homogeneous molecular catalyst for photochemical, electrochemical, and chemical water oxidation. Unlike the Py5–OMe ligand, which contains two methoxy groups, the Py5–OH ligand contains two free hydroxyl groups. The overpotentials for water oxidation were 540 and 510 mV at pH = 8 and 9, respectively. Importantly, it was suggested that a Co-bound chloride ligand was critical for this catalytic reaction because it prevented the formation of a dinuclear Co species. Siewert et al. [117] also reported a Co complex with a PyIm4 ligand (Fig. 15) as a molecular catalyst for electrochemical water oxidation. Instead of four pyridine groups, four imidazole groups were attached to the central pyridine unit. The NH moieties in the imidazole groups are thought to act as intramolecular bases for coupling multiple proton and electron transfer steps. The cyclic voltammogram of the Co complex showed a catalytic wave with an onset potential of 1.20 V (vs NHE) in pH = 12.8 aqueous solution, corresponding to an onset overpotential of 730 mV.

In 2015, Ding and co-workers [118] reported the mononuclear Co complex 36 (Fig. 9) as a catalyst for visible-light-driven water oxidation. When a typical three-component system with complex 36 as the catalyst, [Ru(bpy)₃]²⁺ as the photosensitizer, and Na₂S₂O₈ as the oxidant was used, the maximum TON for O₂ evolution was 1610. Verpoort and co-workers [119] reported another mononuclear Co complex, i.e., 37 (Fig. 9), as a catalyst for both electrochemical and photochemical water oxidation. In electrochemical studies, the catalytic water oxidation wave increased during CV scans, indicating that 37 was converted to another more active form. The current crossover observed in the first CV cycle indicated catalyst film deposition. However, characterization of this film suggested that the film was a molecular Co species rather than cobalt oxide/hydroxide. In light-driven water oxidation, a TON of 90 was obtained in 1 h of irradiation. A reaction mechanism based on DFT calculations was proposed.

In 2014, Ding and co-workers [120] reported a mononuclear Co-substituted silicotungstate as a catalyst for visible-light-induced water oxidation. All-inorganic silicotungstate ligands are robust in oxidizing environments [121]. The cyclic voltammogram of this Co-substituted silicotungstate showed an oxidation peak assigned to Co^Ⅲ/Co^Ⅱ, at 0.82 V (vs Ag/AgCl), and a catalytic wave with an onset potential of 0.96 V (vs Ag/AgCl) in borate buffer solution (pH = 9.0). This result indicates that this Co complex is active in water oxidation. In photocatalytic studies, a TON of 313, TOF of 3.2 s^-1, and quantum yield of 27% for O₂ evolution were obtained in solutions (pH = 9.0) under the optimum conditions. Several lines of evidence from experimental and control studies suggested that this Co-substituted silicotungstate, rather than its decomposition product, was the real molecular catalyst.

In 2015, Cao and co-workers [122] reported the first Ni–porphyrin catalyst for water oxidation. A cationic water-soluble Ni^Ⅱ complex with meso-tetrakis(4-N-methylpyridyl) porphyrin 38 (Fig. 16) was investigated as a homogeneous WOC. The cyclic voltammogram of 38 showed a large catalytic wave with an onset potential of 1.0 V (vs NHE) in neutral phosphate buffer (0.10 mol/L). This value is much lower than that for the Co–porphyrin analog 30 (Fig. 9). The normalized cyclic voltammograms (i/ν_1/2, i = current and ν = scan rate) showed that the catalytic current decreased with increasing scan rate. This result suggests a catalytic process with a chemical rate-determining step. In water oxidation, this chemical rate-determining step is probably O-O bond formation. The KIE value of 1.55 (KIE = k_{cat, H2O}/k_{cat, D2O} = (i_{cat, H2O}/i_{cat, D2O})²) supports the suggestion that the O-O bond formation step is the rate-determining step in the catalytic cycle. The molecular nature of 36 in water oxidation was carefully investigated, and a TOF of 0.67 s^-1 was determined.

Experimental and theoretical mechanistic studies suggested that the two-electron-oxidized species formed from 38 was the catalytically active species, and a reaction mechanism was proposed for water oxidation (Fig. 17). Two-electron oxidation gives Por–Ni^Ⅲ–O^•, which reacts with a water molecule via nucleophilic attack of water on the oxo unit. This O-O bond formation step is the rate-determining step and has a calculated activation energy barrier of 24.9 kcal/mol when a cluster of four water molecules accepts the proton, or 18.0 kcal/mol when an acetate anion is the proton acceptor. The transition-state structures (Fig. 18) show an increased O-O bond length (2.15 vs 1.96 Å, 1 Å = 0.1 nm). This change indicates that the transition state with an acetate anion as the proton acceptor is formed at an earlier stage than the transition state with four water molecules acting as the proton acceptor.

Lu and co-workers [123] reported that the macrocyclic Ni complex 39 (Fig. 16) acted as a WOC. The cyclic voltammogram of 39 showed an irreversible Ni^Ⅲ/Ni^Ⅱ oxidation wave at 0.87 V (vs NHE) and a large catalytic wave at 1.41 V (vs NHE) in phosphate buffer (0.1 mol/L; pH = 7.0). Both waves had linear pH dependence, with slopes of -59 mV per pH unit. The two-electron-oxidized form is therefore probably the catalytically active species in water oxidation. During controlled potential electrolysis at 1.55 V (vs NHE), the current increased gradually from 0.15 to 0.90 mA/cm² in the first hour, without stirring, and then stabilized at ca. 0.9 mA/cm² during further electrolysis. The authors proposed that a new active species was generated during electrolysis. Theoretical studies suggested that the Ni complex with two trans vacant sites was slowly converted to the cis isomer, and the O-O bond was formed through HO-OH coupling on the Ni ion. Lu and co-workers [124] subsequently reported Ni complex 40 (Fig. 16), containing two cis vacant sites, for water oxidation. DFT calculations suggested that the coupling of two hydroxyl groups on the formally Ni^Ⅳ ion of complex 40 was possible. This intramolecular O-O bond formation had a calculated activation energy barrier of 18.9 kcal/mol. These authors also reported that an analog of complex 40, which contained two cis vacant sites, was an active WOC [125]. A similar O-O bond formation mechanism, with a 7.9 kcal/mol activation energy barrier, was proposed.

In 2016, Nath and co-workers [126] reported a selenide-coordinated Ni complex for electrocatalytic water oxidation. This selenide-coordinated Ni complex can efficiently catalyze water oxidation by reaching a current density of 10 mA/cm² at a low overpotential of 200 mV in KOH solution (1.0 mol/L). This small overpotential is caused by the low electronegativity of selenide, which leads to increased covalency of the Ni–Se bonds. As a consequence, the Ni center can be oxidized at much lower potentials to generate the catalytically active species for water oxidation. In 2016, Sun and co-workers [127] reported a water-soluble Ni–Py5 complex for electrocatalytic water oxidation. The cyclic voltammogram of this Ni–Py5 showed a Ni^Ⅲ/Ni^Ⅱ oxidation peak at 1.41 V (vs NHE) in a phosphate buffer (pH = 7). This Ni^Ⅲ/Ni^Ⅱ oxidation had a linear pH dependence, with a slop of -61 mV per pH unit. After this metal-based oxidation, a catalytic wave for water oxidation was observed. The molecular nature of this catalyst was confirmed, and a reaction mechanism that involved a water nucleophilic attack pathway was proposed. In this mechanism, the phosphate anions were important because they greatly enhanced water oxidation catalysis, and a rate constant of 1820 L/(mol·s) was achieved.

In 2012, Mayer and co-workers [128] reported the first example of a homogeneous Cu catalyst for electrochemical water oxidation. Mixing various simple Cu salts with bipyridine at high pH led to in situ formation of Cu–bipyridine–hydroxo complex 41 (Fig. 19). The cyclic voltammogram of this Cu complex in aqueous solutions (pH = 11.8–13.3) showed large and irreversible catalytic currents for O₂ evolution. Electrochemical, electron paramagnetic resonance, and other studies indicated that the dominant catalytic species was a monomeric [(bpy)Cu(OH)₂] moiety, rather than heterogeneous nanoparticles formed during electrocatalysis. This Cu complex is among the most active homogeneous WOCs, with a TOF of 100 s^-1. In 2014, Lin and co-workers [129] used 6, 6'-dihydroxy-2, 2'-bipyridine to synthesize a similar Cu–bipyridine–hydroxo complex, i.e., 42 (Fig. 19), for electrocatalytic water oxidation. The introduction of two pendant hydroxyl groups enabled complex 42 to catalyze water oxidation at much lower potentials than those needed by its analog 41. Experimental and computational studies suggested that this 6, 6'-dihydroxy-2, 2'-bipyridine ligand is redox non-innocent and participates in electron-transfer processes. This can facilitate oxidation of the Cu center and can provide an active WOC with a low overpotential.

In 2013, Meyer and co-workers [130] reported a Cu complex 43 with a self-assembled triglycylglycine macrocyclic ligand (Fig. 19), as an electrocatalyst for water oxidation in phosphate buffer (pH = 11). The cyclic voltammogram of 43 showed a well-defined, reversible oxidation wave at E_1/2 = 0.58 V (vs NHE) and an additional, irreversible catalytic wave at E_{p, a} = 1.32 V (vs NHE). The first oxidation was assigned to the Cu^Ⅲ/Cu^Ⅱ redox couple, was pH dependent, and decreased at 59 mV per pH unit. The catalytic wave had an onset potential of 1.10 V (vs NHE), corresponding to an overpotential of 520 mV. Mechanistic studies suggested that a formally Cu^Ⅳ species was the catalytically active species in water oxidation; a proposed mechanism is shown in Fig. 20. In this mechanism, the Cu^Ⅳ=O intermediate reacts with a water molecule to form the O-O bond; this is the rate-determining step.

In 2015, Pap and co-workers [131] reported two similar mononuclear Cu–tetrapeptide complexes as WOCs. These Cu complexes, i.e., 44 and 45 (Fig. 19), were synthesized with two different l-2, 3-diaminopropionic acid-based peptides, which form stable 1:1 complexes with Cu^Ⅱ in basic solutions. The Cu^Ⅲ/Cu^Ⅱ redox potentials were 0.72 and 0.76 V (vs Ag/AgCl) for 44 and 45, respectively. The cyclic voltammograms of these Cu complexes in phosphate buffers (pH = 11) showed electrocatalytic waves for water oxidation, with onset potentials above 1.0 V (vs Ag/AgCl). TOFs of 24 and 53 s^-1 were calculated for 44 and 45, respectively.

Llobet and co-workers [132] reported a series of Cu complexes with a new family of tetraanionic tetradentate amidate ligands as WOCs. The Cu complexes 46a–d (Fig. 19) were active electrocatalysts for water oxidation in basic aqueous solutions. Importantly, the overpotential for water oxidation decreased greatly, down to 170 mV, with increasing electron-donating capacity of the aromatic ring. The two-electron-oxidized form L^•–Cu^Ⅲ–OH was suggested as the catalytically active species in water oxidation (Fig. 21). This species reacts with an additional hydroxide anion to produce a peroxo intermediate, L–Cu^Ⅱ–(HO-OH). However, in a typical water nucleophilic attack step, the hydroxide anion directly attacks the metal–oxo unit to form the O-O bond in a single step. A DFT calculation suggested that one electron from the incoming hydroxide anion transfers to the L^•–Cu^Ⅲ–OH moiety to form a (HO---OH)^•- radical anion fragment with a partial O-O bond. This (HO---OH)^•- radical anion can hydrogen bond to the L–Cu^Ⅲ complex. Subsequent electron transfer converts the (HO---OH)^•- radical anion to a singlet HO-OH peroxo species and an L–Cu^Ⅱ unit. These authors labeled this type of mechanism as single-electron-transfer water nucleophilic attack, which differs from the traditional two-electron-transfer step. In this mechanism, O-O bond formation has a low activation energy barrier of 5.5 kcal/mol.

In 2017, Crabtree and co-workers [133] reported electrocatalytic water oxidation by Cu complex 47, which has an oxidation-resistant ligand (Fig. 19). Catalytic water oxidation with 47 occurs at relatively low overpotentials of 520–580 mV in basic solutions, with a TOF of 0.7 s^-1. In 2014, Meyer and co-workers [134] reported electrocatalytic water oxidation with Cu complex 48 (Fig. 19). Electrochemical kinetic studies indicated a single-site water oxidation mechanism. It was suggested that phosphate buffer anions assist electron transfer to the electrode and proton transfer to the phosphate anions, and facilitate the rate-limiting O-O bond formation step through an atom–proton transfer process. The two-electron oxidized form with either a d⁷ Cu^Ⅳ–OH or a d⁸ L^•–Cu^Ⅲ–OH electronic configuration reacts with a water molecule to form the O-O bond, during which a phosphate anion can act as a base and assist proton transfer. In 2016, Cao and co-workers [135] reported a water-soluble Cu complex with tris(2-fluoro-6-pyridylmethyl)amine for both photocatalytic and electrocatalytic water oxidation. The strong binding affinity of the polypyridine ligand and its robust and oxidation-resistant features are critical for the observed activity of this Cu complex.

In 2017, Cao and co-workers [136] reported electrocatalytic water oxidation by a water-soluble Cu complex with the dianionic tridentate pincer ligand N, N'-2, 6-dimethylphenyl-2, 6-pyridinedicarboxamidate 49 (Fig. 19) in carbonate buffer solutions. This tridentate pincer ligand binds a Cu^Ⅱ ion through three nitrogen atoms, which define an equatorial plane. The fourth coordination site in the equatorial plane is occupied by a carbonate group, which can act as a potential proton shuttle during water activation, including steps involved in formation of the catalytically active species and subsequent O-O bond formation. The cyclic voltammogram of 49 in carbonate buffer (0.10 mol/L; pH = 10) showed a reversible Cu^Ⅲ/Cu^Ⅱ oxidation wave at 0.94 V (vs NHE) and a pronounced catalytic water oxidation wave with an onset potential of 1.29 V (vs NHE). Differential pulse voltammetry of 49 gave two oxidation waves, at E_{p, a} = 0.93 and 1.51 V (vs NHE), in carbonate buffer (pH = 10). These results indicate that the two-electron oxidized form of 49 is the catalytically active species in water oxidation.

On the basis of experimental and theoretical studies, the catalytic cycle shown in Fig. 22 was proposed. One-electron oxidation of 49 gives a Cu^Ⅲ species [L–Cu^Ⅲ–CO₃H], which then binds a water molecule. Subsequent intramolecular proton transfer from the Cu-bound water molecule to the -CO₃H group, accompanied by structural rearrangement, generates [L–Cu^Ⅲ(H₂CO₃)–OH], with an activation energy barrier of 5.3 kcal/mol. In the resulting structure, the carbonic acid molecule moves to the axial position relative to the Cu atom, with a long Cu-O bond length of 2.54 Å (1 Å = 0.1 nm), and participates in hydrogen-bonding interactions with the Cu-bound hydroxyl group (Fig. 23(a)). After release of the carbonic acid molecule, the resulting species [L–Cu^Ⅲ–OH] undergoes a one-electron oxidation to give an oxo radical species, [L–Cu^Ⅲ–O^•-], which reacts with a water molecule to form the O-O bond. This water nucleophilic attack step is assisted by a Cu-bound HCO₃^- anion, which acts as a proton accepter. As shown in Fig. 23(b), in the transition state for O-O bond formation, the HCO₃^- anion is weakly bound to the Cu atom as an axial ligand, with a Cu-O bond distance of 2.38 Å (1 Å = 0.1 nm), and also hydrogen bonds to the attacking water molecule to assist this proton-transfer process. Computational evidence shows that the Cu-bound carbonate group can act as an intramolecular base and assist the removal of protons for water activation, including steps involved in formation of the catalytically active species and subsequent O-O bond formation.

This review article summarizes recent advances in the development of new and efficient WOCs based on first-row transition metals, namely Mn, Fe, Co, Ni, and Cu. Considerable insights have been gained in terms of catalyst design and water oxidation mechanisms at single-metal sites. Despite these achievements, catalysts that can function at low overpotentials with high catalytic rates (i.e., catalytic currents) are still lacking. To further improve the efficiency of molecular WOCs, it is crucial to better understand the details of the reaction mechanism, in particular, the catalytically active species and the rate-determining step. Generation of the catalytically active species at low potentials is the key to decreasing the overpotentials required for water oxidation, and accelerating the rate-determining step can significantly improve the reaction rate. A better understanding of the reaction mechanism will therefore enable rational design of ligands and molecular catalysts for highly efficient water oxidation. Several ligand design strategies can be considered for lowering overpotentials and increasing reaction rates.

The following strategies can be used to lower the overpotentials for water oxidation, i.e., to generate the catalytically active species at lower potentials. (1) High-valent metal–oxo units are usually the catalytically active species. Nucleophilic attack of the metal–oxo unit by water or coupling between two metal–oxo units through a bimetallic pathway leads to O–O bond formation. As a consequence, negatively charged ligands or multidentate ligands with strong σ-donating abilities are favored. (2) Redox non-innocent ligands can participate in electron transfer and therefore help to accumulate oxidizing equivalents. This feature is particularly valuable for water oxidation at single-metal sites. (3) Metal complexes containing ligands with strongly electron-donating substituents catalyze water oxidation at low overpotentials. However, the increased electron density of the ligand backbone decreases the stability of molecular catalysts. Fine-tuning of the ligand electronic structure is therefore the key to producing robust WOCs with low overpotentials. On the basis of these factors, negatively charged multidentate amidate ligands, and porphyrin and corrole ligands are excellent candidates for synthesizing WOCs that function at low overpotentials.

The following strategies can be used to improve the reaction rate, i.e., to accelerate the rate-determining step. (1) Introduced intramolecular basic groups facilitate water nucleophilic attack by accepting protons. As mentioned above, nucleophilic attack of the metal–oxo unit by water is commonly suggested as the reaction pathway in the rate-limiting O–O bond formation step. During this process, one proton from the attacking water molecule is lost, with generation of a peroxo intermediate. Intramolecular basic groups can act as proton acceptors and assist this process. (2) In some cases, the O–O bond is formed through a two-step, single-electron-transfer water nucleophilic attack instead of the traditional two-electron-transfer step. In this mechanism, it is suggested that one electron from the incoming hydroxide anion transfers to the metal–oxo/hydroxo species to form a partial O-O bond with radical anion features; subsequent electron transfer converts the radical anion to a singlet peroxo state. DFT calculations suggest that the activation energy barriers to O-O bond formation via this mechanism are low. Ligands that can stabilize the partially O-O bonded radical anions are therefore favorable. (3) Metal complexes with two cis labile/vacant coordination sites have the potential to form the O-O bond through HO-OH coupling on the same metal ion. Ligands that can provide a pre-organized geometry that affords this coordination environment are likely to improve the water oxidation efficiency.

In addition to these factors, PCET is a fundamental aspect of water oxidation that needs to be considered. PCET can prevent charge accumulation and therefore can lower overpotentials by avoiding the generation of high-energy intermediates. The coupling of protons and electrons is therefore generally thermodynamically favored in the water oxidation reaction, although the removal/addition of protons to the reaction can impede the kinetics. Stability is another issue for molecular WOCs. Organic ligand backbones can potentially be destroyed under the harsh oxidative conditions required for water oxidation. Ligands with sufficient tolerance of oxidative decomposition are therefore mandatory, and these generally do not contain active C-C and C-H bonds. Although the mononuclear first-row transition-metal complexes reported so far still cannot fulfill the requirements for large-scale water oxidation in practical applications, the results summarized here will certainly contribute to our understanding of the water oxidation process. This information is therefore valuable for achieving the ultimate goal of artificial photosynthesis through water splitting.