There is a general consensus among climate scientists that human activities have contributed significantly to shifts in weather patterns (Lynas et al., 2021; Lloyd and Winsberg, 2018; Oreskes, 2004; Goudie, 2019). According to recent estimates and climate model simulations, these activities are closely related to a gradual increase in global temperature on Earth since the Industrial Revolution ( ∼ 1850) (Hansen et al., 2010; Masson-Delmotte et al., 2021). This change in temperatures (atmospheric, oceanic, and land) has been driven mainly by anthropogenic greenhouse gas emissions, causing what is commonly known as climate change (Solomon, 2007; Al-Ghussain, 2018; Fawzy et al., 2020; Khandekar et al., 2005; Ritchie et al., 2023). In this respect, the Intergovernmental Panel on Climate Change’s most recent scientific report concluded concisely that “human influence has warmed the climate at an unprecedented rate in at least the last 2,000 years” (IPPC, 2021; Masson-Delmotte et al., 2021). Moreover, if we look carefully at the first 2 decades of the 20th century, i.e., 2001–2021, the picture is even bleaker since the global surface temperature was almost 0.1 °C higher than from 1850 to 1900.

The main protagonist of these environmental changes is carbon dioxide (CO₂) released into the atmosphere by burning fossil fuels (coal, oil, natural gas), used as non-renewable primary energy sources to drive and maintain modern societies’ development level. Chart 1 shows how these CO₂ emissions have changed from the start of the Industrial Revolution to the present day (Ritchie et al., 2023; Yoro and Daramola, 2020; Soeder, 2021; Letcher, 2019).

Before the COVID-19 pandemic and according to the last IPCC information, atmospheric CO₂ concentrations reached levels not seen in the previous 2 million years, and concentrations of short-lived climate forcers such as methane and nitrous oxide were higher than at any time in at least 80,000 years. These are the three greenhouse gases that have unequivocally contributed to the increase in global average temperature compared to levels observed before the development of industrial societies (Masson-Delmotte et al., 2021). If this scenario continues, we will reach a temperature increase of 4.4 °C by the end of this century, which will be irreversible for centuries and even millennia.

Faced with this challenging environment, finding renewable energy sources and technologies to exploit them to decrease our dependence on fossil fuels and reduce the effects of climate change has become one of the most critical challenges facing the world’s political and scientific agenda in recent years (Goudie, 2019).

Nuclear energy has been considered an important alternative to responding to future energy demand among the various renewable and exploitable energy resources (Fawzy et al., 2020; Prăvălie and Bandoc, 2018). It has begun to be seen as a potential short-term solution to reduce the effects of climate change. However, behind nuclear energy, some controversies remain regarding nuclear accidents and the fact that it has not been able to solve the radioactive waste problem. In addition, its use is linked to military activities. Therefore, efforts must be directed toward promoting and using eco-friendly renewable energy sources (Panwar et al., 2011).

Solar energy is another alternative source that will play a key role in decarbonizing global power systems in the coming decades (Panwar et al., 2011; Kannan and Vakeesan, 2016; Creutzig et al., 2017; Rabaia et al., 2021). Solar energy is plentiful, and as pointed out by the biochemist James Barber: Our Sun is the world’s most powerful energy source, providing the planet with more energy in a single hour than is currently derived from fossil fuels, nuclear power, and all renewable energy sources put together. Its energy supply is inexhaustible in human terms, and its use is harmless to our environment and climate (Barber, 2009; Barber and Tran, 2013).

Besides, the Sun can generate energy using conversion technologies that provide light, heat, electricity, and fuels (Crabtree and Lewis, 2007; Inganas and Sundstrom, 2015). Among these, solar fuels have the potential to be stored for future use and could be a substitute for fossil fuels (Nocera, 2017; Lewis and Nocera, 2006).

Photosynthesis is the example par excellence of absorbing, transferring, and converting solar energy into fuels as energy-rich molecules. Terrestrial plants, green algae, and cyanobacteria have performed this process for over 2 billion years, giving rise to the evolution of aerobic metabolism and complex life (Figure 1) (El-Khouly et al., 2017; Blankenship, 2022; Stirbet et al., 2020; Fischer et al., 2016). Despite oxygenic photosynthesis being the result of a long and challenging process of natural selection, it can be broken down into two fundamental chemical reactions: the oxidation of water to dioxygen (O₂) and the reduction of carbon dioxide into organic matter (see Eqs (1), (2), respectively) (Blankenship, 2022; Krewald et al., 2015a). The first reaction occurs in the heart of photosystem II (PSII) (Brudvig, 2007), an enzyme complex consisting of several protein subunits located in the thylakoid membrane of chloroplasts of green plants and the inner membranes of cyanobacteria.

The X-ray crystallographic resolution of the atomic arrangement of PSII has revealed that the active site responsible for the catalytic transformation of water into oxygen, protons, and electrons is a cluster of manganese, calcium, and oxygen atoms linked by μ-oxo bridges, named because of its activity as the oxygen-evolution center (OEC) (Ferreira, 2004; Umena et al., 2011; Young et al., 2016; Suga et al., 2015). In this way, whereas the oxygen released from the oxidation of water is available for metabolic processes in living systems and to support combustion reactions to sustain our daily lives, the electrons and protons can combine to form H₂ or can eventually be used to chemically convert raw materials such as CO₂ into energy-dense carbohydrates (equations (2) and (3) (Barber and Tran, 2013). The latter processes occur naturally in photosystem I (Blankenship, 2022; Nelson and Ben-Shem, 2005).

For CO₂ fixation, nature has also developed biochemical processes such as the Benson-Bassham-Calvin cycle to reduce carbon dioxide into carbohydrates. Photosynthetic organisms transform about 10¹⁷ g (100 Gt) of CO₂ into biomaterials and organic matter powered by the Sun (Brinkert, 2018). 2 H 2 O + 4 h ν   sunlight → O 2 + 4 e − + 4 H + (1) CO 2 + 4 H + + 4 e − → HCHO + H 2 O (2) 4 H + + 4 e − → 2 H 2 (3)

The oxygen-evolving center is located at the membrane-lumen interface of the PSII and features four coordinated water molecules and several amino acid residues from the protein side chains (Ferreira, 2004; Umena et al., 2011). The water molecules surrounding the OEC and the amino acids that constitute a saturated ligand environment play a significant role in the stabilization and activity of the catalytic site toward the water oxidation process, which is driven by a photo-induced cycle based on five different oxidation states (Lubitz et al., 2019; Dau and Haumann, 2008). Due to the extraordinary catalytic ability of this small cluster to transform clean and abundant sources such as water, CO₂, and solar energy into fuels, the OEC has inspired many bioinorganic chemists to rationally synthesize molecules that emulate not only its structural features but its catalytic properties towards water oxidation (Chen et al., 2021; García-Álvarez et al., 2021; Zhang et al., 2015; Sun, 2015; Chen et al., 2019; Chen et al., 2015; Yagi and Kaneko, 2000; Li et al., 2020; Mullins and Pecoraro, 2008; Yao et al., 2021; Zhang et al., 2021; Chen et al., 2022; Wiechen et al., 2014). Over time, this progress has allowed the integration of artificial photosynthetic models inspired by the chemical principles of nature.

Artificial materials with the potential to catalyze the conversion of water into O₂ and reducing equivalents by mimicking the process of photosynthesis are known as water oxidation catalysts (WOCs). They can be broadly separated into metal oxides and molecular complexes (Yagi and Kaneko, 2000; Smith et al., 2013). Metal oxides were the first materials reported with the electrocatalytic ability for OER and can be traced back to 1903 with the early work of Cohen and Gläser about cobalt oxides (Matheu et al., 2019; Coehn and Gläser, 1902). By contrast, molecular complexes appeared on the scene until 1985, with the famous binuclear Ru-complex or blue-dimer with formula cis-[(H₂O)Ru(bpy)₂(μ-O)Ru(bpy)₂(H₂O)]⁺⁴, a compound featuring two [Ru(bipy)]^3- units linked by an oxo bridge; where each ruthenium(III) ion is coordinated by two solvent molecules involved in the water oxidation process. Electrochemical insights of this complex showed a catalytic process similar to that of the natural OEC (Gilbert et al., 1985). However, two fundamental problems exist in many molecular WOCs: organic systems that interfere with the decrease in catalytic activity due to their oxidative degradation under typical experimental conditions and precious metals that are prohibitively expensive for large-scale artificial photosynthetic systems (Liu et al., 2018; Garrido-Barros et al., 2017; Limburg et al., 2012). Thus, one of the main motivations in the WOC community is the search for fully inorganic catalysts based on earth-abundant metals.

This review article focuses on a fascinating class of molecules that lie at the frontier between metal oxides and molecular complexes called polyoxometalates (POMs) and their role as water oxidation catalysts. In the broadest sense, POMs are discrete anionic clusters of transition metal ions self-assembled readily under one-pot reaction conditions (Long et al., 2007; 2010; Goura et al., 2021; Gumerova and Rompel, 2023). These molecules fall into two categories according to their structural composition: homo- and hetero-POMs, represented by the general formula {M _m O _y } ^n- and {X _x M _m O _y } ^n- , respectively, where M are the addenda atoms such as V, Mo or W in their highest oxidation states, and X represents the heteroatoms, which are intentionally introduced into the framework during the self-assembly to modulate the degree of nuclearity of the polyoxometalate and its properties. The heteroatoms can be almost any element, but commonly found are atoms of Si, P, Al, As, etc.

Hetero-POMs are grouped into six to ten basic structures, from which they derive their huge structural diversity. The Keggin, Wells-Dawson, Anderson, Waugh, Silverton, and Lindqvist-type clusters are among the most representative structures (Wang et al., 2019; Ye et al., 2016; Li and Xu, 2011). Materials derived from Keggin and Wells-Dawson-type POMs have been used as multifunctional catalytic systems due to their high Brønsted acidity (Li et al., 2007; García-López et al., 2019; Mürtz et al., 2024). In general, polyoxometalate-based materials, and especially those incorporating redox-active transition metal ions, possess an enormous potential as water oxidation catalysts due to the following features (Bonchio et al., 2006; Proust et al., 2012; Zhang et al., 2024): ( i ) high stability toward oxidative degradation without altering their parent structure, ( i i ) remarkable redox properties, ( i i i ) behavior as robust molecular ligands, ( i v ) abundant nature of their components, ( v ) structural availability to bind water molecules, and ( v i ) photocatalytic activity.

The application of POMs as water oxidation catalysts became a significant issue for designing a purely inorganic catalyst based on a Ru-substituted polyoxotungstate, researched by two pioneering groups in 2008 (Geletii et al., 2008; Sartorel et al., 2008). Since then, a remarkable number of articles have emerged, and it is necessary to organize the most relevant information to provide the reader with a road map for further field development. This review presents a general overview of the water oxidation process performed in natural photosynthesis to analyze the structural and electronic features of the best WOC optimized over billions of years of biological evolution. Furthermore, a comprehensive section on the light-driven water oxidation processes catalyzed by TMSPs places readers on the timeline of these materials, discussing the synthetic strategies, structural features, and catalytic activity. Subsequently, these materials’ likely mechanisms of action are outlined, and finally, we report those artificial photosynthetic devices incorporating POMs.

Photosystem II is a multi-subunit enzyme in photosynthetic plants, algae, and cyanobacteria. It harnesses solar energy to drive a series of electron transfer processes that split water into its primary components: molecular oxygen, protons, and electrons (Figure 2). Although several medium-resolution 3D structures have been obtained, the catalytic site remained hidden for a long time due to structural distortion of the protein during X-ray data acquisition (Umena et al., 2011; Zouni et al., 2001; Kamiya and Shen, 2002). This problem has been gradually resolved by a series of high-resolution structures which, together with theoretical analyses, have provided new insights into the organization of the enzyme and the mechanistic pathway of light-driven water splitting preceding the formation of the O-O bond on the catalytic site.

The X-ray diffraction of isolated photosystem II from the bacterium Thermosynechococcus elongatus at 3.5 Å resolution showed a 650.75 kDa transmembrane protein dimer. The structure comprises two nearly identical monomers divided by a double symmetry axis perpendicular to the lipid layer (Figure 2B) (Ferreira, 2004). According to the structural determination and the complete architecture of PSII, each monomer is composed of 19 protein subunits containing a significant amount of pigment molecules such as α-chlorophylls and beta-carotenes, also including molecules responsible for electron transfer processes such as heme groups, plastoquinones, pheophytin, a non-heme iron, and molecules incorporated during the crystallization process such as carbonate groups. The dimer also comprises two manganese-calcium-based clusters (Ferreira, 2004; Kato et al., 2014). The extrinsic subunits of the inner antenna, or CP43, are associated with the OEC proteins on the luminal side.

The photosynthetic process begins capturing sunlight through the antenna system (the light-harvesting complex or LHC), which is based on several hundred chromophores coupled electronically and integrated into protein molecules, including chlorophylls, carotenoids, and phycobilin molecules (Blankenship, 2022; Ferreira, 2004; Su et al., 2017). The light captured by these systems is funneled to the inner antenna located at the central core of PSII, built by a smaller number of chlorophylls and carotenoids related to the denominated CP43 and CP47 intrinsic proteins (Cardona et al., 2012; Mirkovic et al., 2017). Electronic excitation energy from the peripheral membrane antenna complex absorption is transferred and trapped in a photochemical reaction center, which is converted into electrochemical potential energy. The photosynthetic reaction centers are membrane-bound protein-pigment complexes in which the redox cofactors, i.e., chlorophyll, pheophytin, and quinone, involved in electron transfer are located perpendicular to a two-fold axis (Barber, 2003; Cox et al., 2013; Barber and Archer, 2001). When the excitation reaches a particular Chl a in the reaction center, instead of cascading the energy to a neighboring pigment molecule, an electron is transferred to a pheophytin molecule (Phe_D1), resulting in the first charge separation.

Chlorophyll a is part of a set of four molecules (labeled as PD1, PD2, ChlD1, and ChlD2 in Figure 2) grouped as the P680, or primary electron donor system, associated with the D1 and D2 reaction center proteins. The electron transfer pathways leading to water oxidation begin between the chlorophyll molecule (ChlD1) and pheophytin (PheD1), generating an unstable radical species or [P680^•+ – Phe_D1 ^•-](Vinyard et al., 2013; Barber, 2002; Lubitz et al., 2008; Nelson and Ben-Shem, 2005). The electron received by the pheophytin molecule is immediately delivered to a plastoquinone molecule QA and finally to the plastoquinone QB via a non-heme Fe atom coordinated through a carbonate ligand. When this molecule receives two electrons, it is released from PSII as a reduced plastoquinone (Q_BH₂) and donates its electrons to Photosystem I (PSI) to produce reduced biological hydrogen (Figure 2C). On the other hand, the lower half of the reaction center substitutes the electron to stabilize the P680^•+ system with a low-energy electron from water. The high redox potential of the oxidized P680^•+ species (between 1.3–1.4 vs. SHE) is the key to the unique role of PSII, as it is the potential needed to carry out the oxidation of two water molecules (Barber, 2002).

The oxygen-evolving center enters the scene by oxidizing a water molecule bound to the biological catalyst and the electron transferring it to a redox-active amino acid (TyzZ) of the D1 protein subunit, donating it to the P680 system, and leaving it stable and ready to absorb another photon (Figure 2C) (Vinyard et al., 2013; Barber, 2002; Lubitz et al., 2008; Nelson and Ben-Shem, 2005; Shen, 2015). Four light-driven electron transfer cycles lead to the oxidation of two water molecules with the simultaneous release of O₂ to the atmosphere for the utility of biological systems and the formation of reducing equivalents in the form of energized electrons and protons for synthesizing energized molecules.

The crystal structure of PSII from the cyanobacterium T. elongatus at 3.5 Å resolution resolved the OEC as a distorted cubane-type [Mn₃CaO₄] cluster bonded to an external Mn (Mn4) through a single mono-μ-oxo bridge. In the X-ray crystal structure proposed by Ferreira and co-workers, the Mn─Mn and Mn─Ca distances are 2.7 Å and 3.4 Å, respectively, indicating that the elongated Mn ^… Ca interaction is weak, and this feature seems to play a structural role in the catalytic cycle. Concerning the protein residues, the [Mn₄CaO₄] cubane is wrapped by amino acids from the D1 and CP43 proteins that fulfill the function of biological ligands by binding directly to the cluster metals via the carboxylate groups of the amino acids or stabilizing it through hydrogen bonds or other intermolecular interactions (Barber and Tran, 2013; Crabtree and Lewis, 2007). The residual electron density around the cubane shows a bicarbonate molecule acting as a bridge between the Ca and Mn atoms, temporally replacing the water molecules that bind to the OEC in the active catalytic state.

Later, Umena and co-workers confirmed the distorted cubic geometry of the [Mn₃CaO₄] cluster at a resolution of 1.9 Å in PSII isolated from the thermophilic cyanobacterium Thermosynechococcus vulcanus but adding an oxo group that bridges an Mn atom of the cubane and the external Mn4, giving rise to a [Mn₄CaO₅] system organized in the shape of a distorted chair, where the cubane serves as the base of the seat and the external Mn4 and O4 atoms as the back of this rare asymmetric chair (Figure 2C) (Umena et al., 2011). This crystallographic study confirmed through electron density maps the hypothesis of Ferreira et al. that the Ca and Mn4 atoms are bonded to two water molecules each, suggesting that these solvent molecules may serve as substrates for the formation of the O─O bond on the catalyst. The environment surrounding the [Mn₄CaO₅] cluster does not differ much from that found in previous Ferreira’s structure. Again, some amino acids (glutamic and aspartic acids, for example,) orient their carboxylate groups towards the cubane to add stability, while others are in the second coordination sphere.

Two essential features explaining the unusual activity of the oxygen evolution center, particularly the inorganic [Mn₄CaO₅] cluster towards the water oxidation reaction, are its distorted polyhedral structure and unique redox chemistry. The differences between the bond distances within the cluster define its geometrical distortion. For example, in the structure proposed by Umena and co-workers, the distances around O5 are much longer compared to the other bond distances. This structural detail provides flexibility to the cluster during the catalytic process, and it has even been argued—considering the proximity of this atom to the water molecules on Ca and Mn4—that this O5 can protonate and form part of one of the substrates forming the O-O bond. From an electronic point of view, manganese has a wide range of oxidation states (Zhang and Sun, 2018; Krewald et al., 2015b; Kok et al., 1970; Armstrong, 2008). This feature is one of the reasons why nature chose this metal as the protagonist of photosynthesis since (as will be seen below) Mn rapidly changes its oxidation states during the catalytic cycle.

The steps through which oxygen evolution occurs are known as the Kok cycle or S-state cycle. It comprises four metastable intermediate states (S₀, S₁, S₂, and S₃) and a transition state, S₄, through which O₂ is released, and the catalyst is restored to its resting state (Ferreira, 2004; Armstrong, 2008; Cox et al., 2020; Kern et al., 2018; Pantazis, 2018). Although the PSII community has accepted the Kok cycle, the exact mechanism by which the coupling between the two water molecules in the S4 state occurs remains a mystery, leading to several proposals for the formation of the O-O bond, summarized in the oxo-oxyl coupling and the water nucleophilic attack or WNA (Shamsipur and Pashabadi, 2018; Najafpour et al., 2017; Suga et al., 2019; Vinyard et al., 2015).

In one scenario, a Mn^IV–oxyl radical reacts with an Mn–bridging oxo to generate the O—O bond. Some studies suggest that one of the substrate water molecules binds to the OEC during the S₂ → S₃ transition and is then oxidized to an oxyl radical in the S₄ state to carry out the coupling and formation of the O―O bond. The second scenario aligns more with bioinspired WOCs, where a water molecule (possibly coordinated with a Ca²⁺ ion) attacks a terminal oxo (or oxyl) group in the OEC (Vinyard et al., 2015). While obtaining structural information on the S₄ state has not been forthcoming, understanding the structural motions and reactivity of the oxygen-evolving center during the metastable states has shed light on the pathways leading to the formation and release of molecular oxygen (Pushkar et al., 2018; Capone et al., 2021). Recently, two independent research groups have used femtosecond X-ray crystallography to visualize the metastable states of the Kok cycle (Kern et al., 2018; Suga et al., 2019; Kupitz et al., 2014).

These structural studies appeal for the binding of an oxygen atom during the S2→S3 transition state from a nearby water molecule close to the OEC and agree that this oxygen can either act as one of the substrates for the formation of the O-O bond or replace the O5 position of the [Mn₄CaO₅] cluster during the release of molecular oxygen. In the latter case, the O-O bond formation may occur between the O5 and one of the available water molecules of the Mn4 or Ca atoms. These findings indirectly rule out the mechanism of water nucleophilic attack or the formation of a peroxo intermediate, supporting the possibility that the mechanism occurs via oxo-oxyl coupling (Kern et al., 2018). Finally, when O₂ is released from the [Mn₄CaO₅] system, the S₄ state decays to S₀ with a new water molecule binding, rearranging its structure.

Adding to the brief introduction to POMs, it is important to note that photo-driven water oxidation is more efficient in a very attractive group of POMs denominated transition-metal-substituted polyoxometalates (abbreviated from here as TMSPs) (Han and Ding, 2018; Lauinger et al., 2017; Lv et al., 2012; Geletii et al., 2011). These compounds are obtained by combining lacunary POMs with transition-metal ions.

In the center of Scheme 2, we show the saturated structure of the Keggin type-anion, commonly referred to as α-[XM₁₂O₄₀] ^n-8; structurally, it consists of a central tetrahedral ion surrounded by eleven octahedra {MO₆}. Depending on the degree of rotation (60°) of the four {M₃O₁₃} triads (each consisting of three edge-shared MO₆ octahedra), the anion can be designated by the labels β-, γ-, δ-, and ε-. Often, these isomers can be surface decorated with metal coordination units (Weinstock et al., 1999; Cao et al., 2019). However, our interest is focused on lacunary species, where one or several polyhedral units of the addenda atoms (V, Mo, W, etc.) have been separated from their primitive structures.

This removal results in cavity systems that can react as multidentate O-donor ligands toward various transition metal ions, replacing these empty sites. Most of the POMs discussed here are based on lacunary species of the Keggin-type anion (Scheme 2). The surprising range of TMSPs lies in the number of vacant sites available in the POM backbone, the type of transition metal that occupies these sites, and how these metals are incorporated into the native structure [Clemente-Juan and Coronado, 1999; Clemente-Juan et al., 2012; Das et al., 2020; Liu et al., 2016].

The intrinsic features of the TMSPs, derived from the combination of their components, allow them to acquire properties applicable in highly relevant research areas such as water splitting, catalysis, environmental science, magnetism, electronic materials, biomedicine, and electrochemistry, among others (Mizuno et al., 2005; Zheng and Yang, 2012; Gao et al., 2014; Wang and Yang, 2015; Bijelic et al., 2019; Horn et al., 2021). As water oxidation catalysts, TMSPs have played a significant role since they meet structural and chemical requirements that a WOC must possess, e.g.,: ( i ) The presence of one or more transition metal ions incorporated in the POM framework with redox capabilities. ( i i ) Accessible sites where water molecules can bind to the transition-metal ions ( i i i ) The ability to carry out proton-coupled electron transfers to neutralize the accumulated charge and ( i v ) the abundance of their elements to make them amenable to large-scale processes (Soriano-López et al., 2013; Gao et al., 2020). Considering that lacunary POMs can incorporate all transition metals, including noble ones, whose catalytic activity toward water oxidation has been well established, it is not surprising that early efforts in the design of POM-WOCs began with ruthenium-substituted POMs (Matheu et al., 2019; Kamdar and Grotjahn, 2019; Concepcion et al., 2010).

At the end of the 20th century, POMs were only used in organic catalysis despite featuring structural motifs in common with some Ru-based WOCs. It was not until 2004 that the POM community addressed the electrocatalytic generation of oxygen with a di-Ru-substituted polyoxometalate, which opened an attractive and multidisciplinary field of research (Neumann and Khenkin, 1995; Xinrong et al., 2000; Neumann and Dahan, 1997; Hill and Prosser-McCartha, 1995; Rüttinger and Dismukes, 1997). Electrochemical oxygen generation with this compound was monitored using a Clark electrode by differential pulse voltammetry in a phosphate buffer solution at pH 8. The synthesis strategy of the complex consisted of reacting Ru[CH₂)₂SO]₄Cl₂ with the polyoxometalate precursor Na₁₂[WZnZn₂(H₂O)₂(ZnW₉O₃₄)₂], allowing the exchange of Zn(II) for Ru(III) ions and obtaining the system Na₁₄[Ru₂Zn₂(H₂O)₂(ZnW₉O₃₄)₂].42H₂O (Ru ₂ -ZnPOM) (Howells et al., 2004).

The X-ray diffraction technique revealed a system with a sandwich-type structure in which two [ZnW₉O₃₄]^12- polyanions are assembled through a four-transition metal belt. This belt is formed by two Zn atoms connected to the two [ZnW₉O₃₄]^12- polyanions through five oxygen atoms. At the same time, a peripherally coordinating water molecule accessible for catalytic processes occupied the sixth position (Figure 3A). On the other hand, the Ru(III) ions are placed at the center of this belt in a saturated octahedral geometry by the O atoms of the POMs scaffolds. In this scheme, the Ru-Ru distance is approximately 3.18 Å.

The authors focused on the structure-activity relationship showing that Ru-free or monosubstituted POMs with a Keggin-type topology [Ru^III(H₂O)PW₉O₃₉]^4- did not exhibit the same activity toward O₂ generation under similar experimental conditions, indicating that the sandwich-type structure allowing the incorporation of redox-active transition metal centers are fundamental to observe efficient catalytic processes (Table 1, entry 1) (Gao et al., 2020; Howells et al., 2004). Although the authors did not present experimentally supported mechanistic proposals, they pointed out that a possible route may involve redox processes during the catalytic cycle and the presence of oxygen species (OH⁻ or H₂O) as substrates for O₂ generation.

Considering the effect of Ru ions on the POM backbone, two research groups published 2008 a tetra-ruthenium POM with promising WOC activity. The complex with structural formula [Ru^IV ₄O₄(OH)₂(H₂O)₄( γ –SiW₁₀O₃₆)₂] (Ru ₄ -SiPOM) crystallized using two synthetic approaches (Figure 3B). First, Hill and co-workers obtained Ru4-SiPOM in a one-pot synthesis by reacting the polyoxometalate precursor salt K₈[ γ -SiW₁₀O₃₆]·12H₂O with ruthenium chloride in an acidic aqueous medium and then adding rubidium chloride to this reaction system to introduce alkali cations to stabilize the polyanionic system (Geletii et al., 2008). On the other hand, Bonchio and co-workers isolated the complex through a Lego-type strategy by adding an excess of CsCl to an aqueous mixture containing the cation complex [Ru₄O₆(H₂O)n]⁴⁺, obtained by reaction of K₄Ru₂OCl₁₀, and K₈[ γ -SiW₁₀O₃₆]·12H₂O at pH 1.8. The only difference between the two resulting crystal structures was the nature of the alkali-metal counterions that stabilize the overall charge of the Ru ₄ -SiPOM polyanion. Both crystal structures showed a tetra-ruthenium(IV)-oxo core (Figure 3B, inset) sandwiched by two divacant [ γ –SiW₁₀O₃₆] systems. In the central core, the Ru···Ru distances fall in the range of 3.47–3.66 Å.

The Hill group characterized the redox processes of this polyanion by analytical and electrochemical techniques. Cyclic voltammetry at acidic pH (0.1 M HCl) showed oxidation and reduction weaves corresponding to the {Ru₄O₆} cluster. In contrast, at pH 7 in a phosphate buffer and 0.6 mM Ru ₄ -SiPOM, there was a significant increase in current at E = 950 ─ 1,050 mV, which was assigned to electrocatalytic water oxidation. With these findings, the analysis was extended using the [Ru(bpy)₃]³⁺ system as the oxidizing agent, observing that the addition of Ru ₄ -SiPOM accelerates the water oxidation reaction significantly more than when using only the [Ru(bpy)₃]³⁺species in the absence of the catalyst. The catalytic performance was monitored by quantifying the formation of the reduced Ru(bpy)₃]²⁺ compound and the O₂ production versus time by gas chromatography (Table 1 , entry 2)^[71] To rule out any impurities, the group compared the catalytic effect using RuO₂ as a possible decomposition product of the Ru ₄ -SiPOM system at the same concentration and under the same experimental conditions, thus corroborating that the catalysis came from the polyoxometalate.

On the other hand, the Bonchio group characterized the protonation states of the Ru ₄ -SiPOM system by UV-Vis and acid-base spectrophotometric titration, confirming the observations of Hill’s group that the Ru(IV) centers in the central core undergo reversible protonation equilibria, which, added to the redox properties of ruthenium, explains the catalytic efficiency of this tetra-ruthenium POM. Water oxidation was evaluated by employing an excess of Ce(IV) salt as an oxidizing agent and Ru ₄ -SiPOM at a concentration of 4.3 µmol in water at pH 6 (Table 1, entry 3). The O₂ generation was monitored by gas chromatography, obtaining a yield of about 90%. IR and Raman spectroscopy confirmed the catalyst’s stability after the process.

In subsequent work, Geletti et al. demonstrated that the Ru ₄ -SiPOM system integrated into an artificial photosynthetic scheme at physiological pH catalyzes the reaction 4) efficiently concerning the calculated quantum yield (Geletii et al., 2009). This system consisted of four primary components, i.e., a typical oxidizing agent ([Ru(bpy)₃]³⁺), an electron acceptor (S₂O₈ ^2-), the ruthenium(IV) POM-based catalyst, and a Xe lamp as a visible light source (Table 1 , entry 4). According to the authors, the quantum yield of the system (defined as O₂ produced/quantum absorbed by the photosensitizer) was relatively high (almost 10%) compared to other photocatalytic systems based on molecular catalysts. 4 Ru bpy 3 3 + + 2 H 2 O → POM 4 Ru bpy 3 2 + + 4 H + + O 2 (4)

In this photosynthetic system, the photosensitizer [Ru(bpy)₃]²⁺ acts as a visible light harvester, generating the [Ru(bpy)₃]²⁺/[Ru(bpy)₃]^2+* system, which produces the oxidized [Ru(bpy)₃]³⁺ species through photooxidation employing S₂O₈ ^2- as the oxidizing agent. The [Ru(bpy)₃]³⁺ species oxidize the redox core of the polyoxometalate in a four-electron process, preparing it to oxidize two water molecules to generate O₂ and reestablish the [Ru(bpy)₃]²⁺ species (Geletii et al., 2009; Süss-Fink, 2008). Eq. 5 summarizes this process. 4 Ru bpy 3 2 + + 2 S 2   O 8 2 − + 2 h ν → 4 Ru bpy 3 3 + + 4 SO 4 2 − (5)

Bonchio and others performed kinetic studies on this type of complex photocatalytic process using time-resolved techniques (such as flash photolysis) to help clarify two physicochemical problems related to this light-driven scheme: ( i ) the interactions between the Ru ₄ -SiPOM catalyst and the [Ru(bpy)₃]²⁺ species and ( i i ) the kinetics of the hole transfer from oxidized [Ru(bpy)₃]³⁺ molecules to the Ru ₄ -SiPOM system, leading to the final stages of the process that determines O₂ evolution. The study established a global electron transfer system based on forming ionic pairs between the Ru ₄ -SiPOM catalyst and the oxidized sensitizer in a stoichiometric ratio of 1:4, respectively (Natali et al., 2012a). The oxygen evolution process in the photocatalytic cycle is strongly determined by the quenching of the [Ru(bpy)₃]²⁺* species, favored by strong electrostatic POM/sensitizer interactions. On the other hand, studies using flash photolysis at low concentrations of the catalyst ( ∼ 0.5 µM) allowed determination of the rate of the final processes related to O₂ evolution, which is in the millisecond range with exceptional values of TON and TOF, as can be seen in Table 1, entry 5.

With the results obtained with the system Ru ₄ -SiPOM and interested in exploring potential structure-activity correlations in water oxidation, Besson and co-workers prepared the compound Cs₉[Ru^IV ₄O₅(OH)(H₂O)₄(γ-PW₁₀O₃₆)₂]·17H₂O (Ru ₄ -PPOM), changing only the nature of the heteroatom (from Si to P) in the POM framework (Besson et al., 2010). The complex was obtained by reacting a cesium polyoxometalate salt, Cs₇[γ-PW₁₀O₃₆]·H₂O, with ruthenium trichloride in an acidic aqueous solution (pH 0.6) under ambient conditions. The X-ray crystal analysis showed comparable structural features with Ru ₄ -SiPOM, i.e., a tetrahedral tetra-ruthenium(IV) oxo core sandwiched by two [γ-PW₁₀O₃₆]^7- staggered Keggin-type anions (Figure 3C). However, an essential structure difference in charge and redox properties is that in Ru ₄ -PPOM, the central [Ru^IV ₄O₆] core was found to be monoprotonated (In Ru ₄ -SiPOM, the central core is diprotonated). Nevertheless, the redox potentials of Ru ₄ -PPOM and its analogous Si-POM were very similar.

The results of the photocatalytic water oxidation are shown in Table 1, entry 6. An interesting fact from this work was that the oxygen yield calculated from the sacrificial electron acceptor (S₂O₈ ^2-) consumption was 25% lower than that calculated for the Ru ₄ -SiPOM compound under similar experimental conditions. Considering that the heteroatom is the only structural difference between the two catalysts, the authors addressed these contrasting results by analyzing the equilibrium constants of both systems in the limiting step leading to O₂ evolution, i.e., in the final reaction Ru^V ₄ + 2H₂O → Ru^IV ₄ + O₂ + 4H⁺. This entire process is determined by four consecutive series of oxidations, including the gradual oxidation of the [Ru^IV ₄O₆] system to its oxidized form [Ru^V ₄O₆] by the photogenerated sensitizer [Ru(bpy)₃]³⁺ (Han and Ding, 2018; Besson et al., 2010). For the silicon-centered system (pH 7), k is 0.25 s^-1, whereas for the phosphorus analog (pH 5.8), the water oxidation is ca. 0.07 V thermodynamically unfavorable. From this, it follows that k (Ru ₄ -SiPOM) < k (Ru ₄ -PPOM), and this difference could explain the lower catalytic activity of Ru ₄ -PPOM compared to the Si-centered system.

In addition to tungsten-based polynuclear catalysts, some polyoxomolybdates have been electrostatically stabilized by Ru-based photosensitizers, allowing their catalytic activity for water oxidation to be studied. Incorporating a coordination complex in the POM scaffolds is a suitable strategy to obtain hybrid molecular systems in which the coordination units are grafted on the surface of POMs or act as counterions to stabilize the charge of the anionic POMs. In these systems, the organic ligands generally serve as neutral donors (Sánchez-Lara et al., 2021; Sánchez-Lara et al., 2024). A study by Gao and co-workers 2013 explored the correlation between oxygen evolution and the degree of nuclearity of three POM-ruthenium photosensitizer hybrids (Gao et al., 2013). The compounds were prepared by solvothermal conditions at 110 °C using (TBA)₂[Mo₆O₁₉] and α -(TBA)₄[Mo₈O₂₆] as polyoxometalate sources and [Ru^II(1,10-phen)₃Cl₂] as counterions and photosensitizers. For one of the reported compounds, sulfur is required as an additional starting material. The centrosymmetric anions [Mo₆O₁₉]^2-, [Mo₅S₂O₂₃]^4- and α -[Mo₈O₂₆]^4- obtained were electrostatically stabilized by the cationic [Ru^II(1,10-phen)₃]²⁺ complexes with a stoichiometry according to the anionic charge of the polyoxometalate ion.

The catalytic performance was investigated using the classical system, which employs a sacrificial agent, which, in this case, was S₂O₈ ^2-. An external source was not used as a photosensitizer because the compounds already include ruthenium-based II) light harvester systems in their crystal structures. The results concerning O₂ generation appeared to depend on the size of the POM cluster, i.e., the catalytic efficiency obeyed the following order: α-[Mo₈O₂₆]^4- > [Mo₅S₂O₂₃]^4- > [Mo₆O₁₉]^2-. Although the authors related this activity with the number of terminal M = O bonds involved in water oxidation, an additional answer for this behavior may be due to the number of [Ru(phen)₃]²⁺ cations incorporated in each system. The compound with the best catalytic efficiency toward O₂ formation contains two molecules of the sensitizer [Ru(phen)₃]²⁺ in its molecular formula, whereas the [Mo₅S₂O₂₃]^4- and [Mo₆O₁₉]^2- anions crystallized with a single molecule of [Ru(phen)₃]²⁺.

The authors propose that the mechanism of action is related to a radical cation process that takes place through the transfer of an electron from the peripheral bonds Mo=O to the [Ru(phen)₃]²⁺ system, generating radicals on the surface of polyoxometalate Mo-O^+• and oxidized species [Ru(phen)₃]³⁺ (Gao et al., 2013). The oxidizing potential of Mo-O^+• is high enough to produce O₂ from water. This hypothesis was supported using free radical scavengers such as hydroquinone.

In 2012, Patzke and others conducted a complex comparative analysis between three complexes in which the redox-active core is exposed or intercalated by robust Keggin-type ligands (Car et al., 2012). The system analyzed as a possible WOC was a trivacant triruthenium-substituted POM [{Ru₃O₃(H₂O)Cl₂}(SiW₉O₃₄)] (Ru ₃ -SiPOM) and its catalytic activity was compared with two cobalt and nickel sandwiched POMs with formulae [Co₄(H₂O)₂(SiW₉O₃₄)₂] (Co ₄ -SiW ₉ ) and [Ni₄(H₂O)₂(SiW₉O₃₄)₂] (Ni ₄ -SiW ₉ ), respectively. Ru ₃ -SiPOM was obtained from an acidic aqueous solution (pH 2.5) of RuCl₃·nH₂O with the sodium salt α-Na₁₀[SiW₉O₃₄]·15H₂O. Precipitation of Ru ₃ -SiPOM as a black solid was favored upon adding an excess of KCl to the solution. Due to the low tendency that presents the open systems to crystallize, the structure was proposed based on computational methods as a small [{Ru₃O₃(H₂O)Cl₂}(SiW₉O₃₄)]^7- anion where the three Ru atoms are arranged in a Ru₃O₃Cl cluster and fused to the trivacant [SiW₉O₃₄]^10- scaffold. As seen in Figure 3D, the catalytically active Ru ions are exposed on the surface of the POM, replacing some tungsten atoms of the native Keggin structure.

As in the previous cases, the light-driven catalytic activity of varying amounts of the three compounds was studied with the system containing [Ru(bpy)₃]²⁺ and S₂O₈ ^2- as photosensitizer and sacrificial electron acceptors at pH 5.8 using Na₂SiF₆ instead of phosphate buffer to avoid decomposition of the sensitizer. Some general data, along with TON and TOF values, are summarized in Table 1 (entry 7). With this photocatalytic system, the best results were observed at a concentration of 50 µM catalyst (Ru ₃ -SiPOM), reaching a TON value of 23 after a 2-h exposure. The authors noted that this value is lower than the tetra-ruthenium(IV)-oxo system. Interestingly, the polyoxometalate Co ₄ -SiW ₉ with a sandwich-type structure showed similar activity since the TON and TOF values were very close to those observed for the Ru ₃ -SiPOM system; the isostructural Ni ₄ -SiW ₉ compound showed no activity under the same experimental conditions, indicating that the open structure and the nature of the transition metal favors catalysis.

The authors performed reference experiments by modifying the catalytic conditions to investigate whether the Ru ₃ -SiPOM system is present as the species responsible for water oxidation throughout various photocatalytic processes. Reference experiments in buffer solutions (Na₂SiF₆ pH 5.8) supported by DLS and TEM confirmed the integrity of Ru ₃ -SiPOM, as no heterogeneous particles or by-products were identified under the experimental conditions. Furthermore, additional experiments with the RuCl₃·nH₂O salt as a potential decomposition material of POM did not lead to O₂-evolving under the experimental conditions under which the catalysis was carried out.

On the other hand, a brown precipitate was identified under dark conditions while purging the headspace vial after sensitizer addition. According to other works (Natali et al., 2012a), this pointed to the formation of an adduct between the anionic Ru ₃ -SiPOM complex and the cationic photosensitizer (PS), Ru ₃ -SiPOM/PS. The formation of Ru ₃ -SiPOM/PS was corroborated by adding a solution of Ru ₃ -SiPOM to a [Ru(bpy)₃]Cl₂ buffered solution (Car et al., 2012). The catalytic studies indicated that the Ru ₃ -SiPOM/PS adduct was efficient under illumination and dark conditions, carrying out a redox process via four consecutive cycles that lead to the formation of the {Ru^V ₃SiPOM-PS} oxidizing species, which in turn oxidizes water to restore the POM/PS adduct.

Over time, research on these materials, particularly the Ru ₄ -SiPOM system, has focused on their immobilization on the electrodes or carbon nanotube surfaces to develop modified electrochemical systems capable of performing heterogeneous catalytic processes, including water splitting (Toma et al., 2011; Anwar et al., 2014).

In the following sections of this review, we will address the cases where Earth-abundant transition metals (Co, Ni, Mn, Cu, and Fe) have been used as redox metal clusters self-assembled by robust POM ligands. Considering that Co-based systems have been appearing significantly in the specialized literature, we have decided to start with this metal ion and continue with the other transition metals according to their status in the research on this attractive field.

A good start to addressing the field of Earth-abundant transition-metal substituted POMs with catalytic capabilities for water oxidation is the famous cobalt system Na₁₀[Co₄(H₂O)₂(PW₉O₃₄)₂] (Co ₄ -PPOM) synthesized and structurally characterized for the first time in 1973 (Weakley et al., 1973), and re-investigated by Hill and others in 2010 as a promising WOC (Yin et al., 2010). Before this work, some Co-substituted POMs with catalytic activity for organic oxidations had already been reported, while in the water oxidation field, some cobalt oxides CoO_x had shown good activity towards water oxidation under heterogeneous conditions (Kholdeeva, 2004; Tang et al., 2010; Kanan and Nocera, 2008; Jiao and Frei, 2009). Therefore, the Co ₄ -PPOM system filled a gap in homogeneous catalysis and was a breakthrough in synthesizing schemes based on 3 days elements. Co ₄ -PPOM was synthesized under mild conditions in a one-pot process by mixing the corresponding precursor salts based on W, P, and Co in water. The resulting suspension was heated to reflux before adjusting the pH to 7. The solution was then saturated with sodium chloride to shift the equilibrium toward product formation.

As shown in Figure 4A, the main feature of the [Co₄(H₂O)₂(PW₉O₃₄)₂]^10- polyanion is a flat central cluster featuring four cobalt atoms in an octahedral environment, resembling a fragment of a composite layer in Co(OH)₂, separating two moieties of the anionic {PW₉O₃₄}^9- scaffolds. The reader may notice that the POM anion in the empirical formula is the same as that used in Ru ₄ -PPOM, as described in the previous section. An essential structural characteristic for catalytic water oxidation processes is that the sixth coordination position on the two external Co atoms of the central cluster is completed by a water molecule shielded by the POM anion that, as will be seen later, plays a pivotal role in the water oxidation mechanism. In the crystal structure, ten Na⁺ ions placed in available positions in the crystal lattice stabilize the high anionic charge of Co ₄ -PPOM.

To gain insight into the O₂ generation, the authors compared the catalytic behavior of the Co ₄ -PPOM complex with seven different Co-substituted POMs, featuring a variation of the Co core structure and the type of POM framework. In the cyclic voltammogram (CV), the complex showed a significant increase in the catalytic current at low overpotential (1.2 V vs Ag/AgCl) in phosphate buffer (pH 8) at 1 mM Co ₄ -PPOM. This current was attributed to water oxidation. Interestingly, this catalytic phenomenon was not observed for the other cobalt POMs because they were not stable under the experimental conditions for water oxidation. Under specific requirements given in Table 2 (entry 1), a high TON value was reached (>1,000 in ca. 3 min), which is one of the highest values reported for a molecular WOC.

The stability of Co ₄ -PPOM was confirmed through several spectroscopic and electrochemical techniques such as UV-Vis, ¹³P NMR, IR spectroscopies, and CV before and after catalysis and over time. Poisoning experiments were also carried out using bipyridine as a chelating agent of potential free Co²⁺ released if catalyst decomposition occurs. Since no significant changes in the catalytic activity of water oxidation were observed, it was postulated that polyoxotungstate ligands prevent the formation of Co particles. Some theoretical calculations supported the thesis that POM is almost inert during catalysis and stabilizes the central Co core. This last argument justifies using polyhedral structures based on POMs to stabilize active sites where water oxidation occurs.

An intriguing experimental debate that began by questioning the role of Co ₄ -PPOM as a true molecular catalyst highlighted the importance of distinguishing the catalytic activity of the parent compound and its potential decomposition products formed under the strongly oxidizing conditions required, especially when these products are active species with water oxidation capabilities. In this sense, Stracke and Finke proposed, through an electrochemical approach supported by several spectroscopic and microscopic tools such as UV-Vis, SEM, and EDX, that the catalytic activity observed for the Co ₄ -PPOM compound was due to the in situ deposition of cobalt particles (oxide or hydroxide solids) on the working electrode surface under oxidizing conditions ( ∼ 1 V vs. Ag/AgCl) (Stracke and Finke, 2011). With the techniques outlined above, the authors argued that the [Co₄(H₂O)₂(PW₉O₃₄)₂]^10- system is not the primary catalyst and may act as a precatalyst that, under hydrolysis, releases active cobalt species previously known as competent WOCs. In the same line, Bonchio and others studied Co ₄ -PPOM by nanosecond laser flash photolysis, suggesting that the catalytic activity could correspond to soluble molecular species but not Co ₄ -PPOM (Natali et al., 2012b).

In response to these concerns, Hill, Geletti, et al. proposed the following experimental approaches: i selective extraction of the Co4-PPOM system from the stock solution in which the catalysis was carried out employing THpANO3 to leave any remaining Co_xO _y or Co⁺²(aq) species released from the polyoxometalate cluster in the aqueous phase, and i i quantification of these species with two methods, cathodic adsorption voltammetry, and inductively coupled plasma mass spectrometry (Vickers et al., 2013). The results showed that extracting POM from the aqueous phase significantly decreased catalysis (almost 98%). At the same time, the remaining heterogeneous or soluble cobalt species were found at concentrations below 0.1 µM, which cannot be attributed to the overall catalytic process observed, confirming that Co ₄ -PPOM is the dominant catalyst As a final point, the authors stressed the importance that “catalytic studies of molecular species, especially POM-WOCs, under one set of experimental conditions be compared only with extreme caution, if at all, with those under different conditions.”

A set of studies reported between 2012 and 2014 confirmed that including a polynuclear cobalt, core is not essential to observe water oxidation catalysis by showing that POMs containing mono- and di-cobalt cores can exhibit water oxidation. First, Sakai and co-workers evaluated in 2012 the WOC activity of two Co-containing Anderson- and Evans-Showell-type anions, i.e., [CoMo₆O₂₄H₆]^3- (Co-POM-Mo) and [Co₂Mo₁₀O₃₈H₄]^6- (Co ₂ -POM-Mo), respectively (Tanaka et al., 2012). Both anions are used extensively as starting materials to produce a variety of salts and functionalized metal complexes (An et al., 2017; Sang et al., 2018). In the [CoMo₆O₂₄H₆]^3- anion, six assembled {MoO₆} octahedral units are arranged around a central CoO₆ system to form the Anderson-type polyoxometalate in Figure 4B, while the [Co₂Mo₁₀O₃₈H₄]^6- anion can be seen as a dimer of two monolacunary Anderson [CoMo₅H₂O₁₉]^3- clusters, assembled via two hexacoordinate CoO₆ units with shared edges, producing an acute angle of nearly 45° between them (Figure 4C) (An et al., 2017).

Catalytic studies showed that these small Mo-based POMs have capabilities for O₂ generation at basic pH under the typical visible light-driven system based in the [Ru(bpy)2]²⁺/Na₂S₂O₈ pair. The control experiments carried out with Co-free polyanions, [PMo₁₂O₄₀]^3- and [Mo₇O₂₄]^6-, did not show catalytic activity, indicating that the mononuclear cobalt sites are responsible for O₂ formation (Tanaka et al., 2012). The highest TON values for these two polyoxometalate systems at µ M concentrations are shown in Table 2, entries 9 and 10, respectively. The authors performed stability studies on Co4-PPOM to characterize the catalyst’s possible decomposition products, concluding that the compound is stable under the experimental conditions of the applied photocatalytic system. The stability of these platforms is likely related to the type of structure of the POMs. A sandwich-type design may be more vulnerable to decomposition processes because the central metal ions are more exposed. In contrast, the metal ions are firmly coordinated with the O atoms from the POM backbone in Anderson-type networks. Another reason may be the total charge of the molecule, which may accelerate proton-coupled electron transfer processes.

Subsequently, a dinuclear Keggin-type POM doped with two mixed-valence cobalt atoms (Co^III/Co^II) was reported in 2013 as an efficient molecular WOC using a typical photocatalytic setup (Song et al., 2013). As noted in Figure 4D, the complex is compatible with the structural formula K₇[Co^IIICo^II(H₂O)W₁₁O₃₉]·15H₂O (Co ₂ POM), where one Co^III ion was incorporated within the POM backbone acting as a central heteroatom, and the Co^II ion replaces one of the addenda W atoms in a peripheral position. The compound was synthesized in several steps using a methodology previously reported by Baker and McCutcheon (Baker and McCutcheon, 1956).

The catalytic activity of Co ₂ -POM was investigated and compared under the same conditions (Table 2, entry 11) with four Co-substituted Keggin anions, two of which exhibited the formula [Xⁿ⁺Co²⁺W₁₁O₃₉]^(12-n)- (XW₁₁; Xⁿ⁺ = Si⁴⁺, and P⁵⁺), and the other two were analogs of Co ₂ -POM, i.e., [Co²⁺W₁₂O₄₀] and [Co³⁺W₁₂O₄₀]. Of all these complexes, only the disubstituted Co ₂ -POM system showed catalytic activity (Table 2, entry 11), suggesting that the two Co atoms’ redox properties in the POM network favored the water oxidation processes. However, considering the other POMs established as control groups did not exhibit any catalytic activity, the experimental conditions set for a particular compound may not be suitable for other cobalt/POM systems (Song et al., 2013). Five stability studies using various analytical techniques (e.g., DLS, cyclic voltammetry, catalyst recycling studies, etc.) confirmed that the Co ₂ -POM compound is responsible for the activity observed under the typical photocatalytic scheme.

Finally, the complex with formula K₁₀[Co-(H₂O)₂(γ-SiW₁₀O₃₅)₂]23·H₂O (Co-SiPOM) reported in 2006 by Ulrich Kortz’s group, who studied it from a structural and electrochemical point of view (Bassil et al., 2006), was re-investigated by Ding’s group in 2014 as a potential WOC (Xiang et al., 2014). The X-ray crystal structure of this polyanion is shown in Figure 4E; an accessible octahedral Co^II ion was incorporated into the framework of the lacunary polyoxometalate [Si₂W₂₀O₇₀]^12- through four W-O-Co μ₂-oxo bridges. Interestingly, the inner coordination sphere of this central core is saturated with two possibly active water molecules located axial to each other. The water oxidation catalyzed by this compound was characterized by CV in borate buffer (pH 8), observing a catalytic current at ∼ 0.1 V vs. Ag/AgCl, which was not observed in the control experiment using the [Ru(bpy)₃]²⁺ agent in the absence of Co-SiPOM system.

The O₂ generation was performed in the ordinary photocatalytic system using a visible light source, the Ru(bpy)₃]²⁺ complex, and the electron acceptor Na₂S₂O₈. The activity was compared with two structural analogs based on Ni and Mn and six mono, di, or tetra cobalt-substituted POMs, including the Co ₄ -PPOM. In summary, the Co-SiPOM catalyst was active toward O₂ evolution with results comparable to the Co ₄ -PPOM compound (Table 2, entry 12). At the same time, the Ni and Mn analogs and the other systems evaluated did not show any catalytic activity under these experimental conditions. As in the previous cases, the stability of the catalyst against hydrolysis was studied in detail using various practical approaches.

The water oxidation activity of these four POMs featuring mono and dicobalt cores was influenced by the Co^II/III system’s intrinsic redox properties and the coordination environments imposed by the polyoxometalate-based ligands. In Co-SiPOM, the Si heteroatom does not appear to be essential for water oxidation; however, in Co-POM-Mo, Co ₂ -POM, and Co-SiPOM, where external cobalt ions are accessible to water molecules, the catalytic activity seems to be higher. On the other hand, the water oxidation activity of Co ₂ -POM-Mo and Co ₂ -POM seemed to improve through an electron transfer between the two Co ions incorporated in the POM scaffold.

In parallel with these efforts, some works were conducted to study the effect of replacing the PO₄ ^3- group in Co ₄ -PPOM on its electronic and catalytic properties. Hill and co-workers first attempted such a direction by keeping the same tetranuclear cobalt core but using the Si-centered POM used in Ru4-SiPOM. However, the authors failed to obtain crystals of the desired isostructural complex ([Co₄(H₂O)₂[ α -SiW₉O₃₄)₂]^10-), even following a procedure already described (Zhang et al., 2007). Instead, they obtained a compound with an interesting structural arrangement, formulated as [{Co₄(μ-OH)(H₂O)₃}[Si₂W₁₉O₇₀]^11- (Co ₄ -SiPOM) [Zhu et al., 2021]. This compound crystallized from an aqueous solution at pH ∼ 6 containing the POM precursor salt Na₁₀[SiW₉O₃₄]·nH₂O and CoCl₂.6H₂O. The critical factor in obtaining this product was the addition of high concentrations of KCl to stabilize the negative charge of Co ₄ -SiPOM with K⁺ ions. X-ray analysis showed that Co ₄ -SiPOM is disordered over Co1 and W8 positions with refined occupancies of 0.5 for each polyanion (Figures 4F, G). Therefore, the molecule exhibits an irregular structure consisting of trivacant [A- α -SiW₉O₃₄]^10- and [ α -SiW₁₀O₃₇]^10- moieties, stabilizing a central {Co₄(μ-OH)(H₂O)₃} core. The {Co₄} cluster contains three Co atoms in octahedral geometry and one in a tetragonal pyramidal environment.

The activity of Co ₄ -SiPOM was examined for light-driven water oxidation at pH seven to nine using the typical photocatalytic platform. From the kinetics of O₂ formation, its yield increases at higher pH, with a maximum TON value of ∼ 80 and an apparent TOF (TOF_ap = TON/time) of ∼ 0.1 s^-1. The O₂ yield is based on persulfate consumption at 24%, according to Eq. (7) (Table 2, entry 13). The hydrolytic stability of Co ₄ -SiPOM under turnover conditions was studied vs. time at relatively high concentrations by UV-Vis, indicating that Co ₄ -SiPOM slowly undergoes hydrolysis at different pH values, possibly due to the instability of the central {Co₄} core enclosed within the POM framework (Zhu et al., 2012b). The main hydrolysis products were isolated as crystalline materials from aged Co ₄ -SiPOM solutions and identified by X-ray crystallography as potassium salts of [{Co₄(μ-OH)(H₂O)₃}](Si₂W₁₉O₇₀)]^11- and [Co(H₂O)SiW₁₁O₃₉]^6-. Although these by-products are more stable due to the additional stabilization of the K⁺ ions, their catalytic activity was lower than that of Co ₄ -SiPOM. 2 S 2 O 8 2 −   2 H 2 O + 2 h ν → 4 SO 4 2 − + O 2 + 4 H + (7)

Subsequently, a POM isostructural to the previously discussed Co ₄ -PPOM was reported, which showed remarkable water oxidation activity (Table 2, entry 14) (Lv et al., 2014). The structure of this compound retains the same central {Co₄} core as its analog phosphate-centered Co ₄ -PPOM. However, the phosphorus heteroatom was replaced by the tetrahedral vanadate(V) ion, resulting in a different structural formula Na₁₀[Co₄(H₂O)₂(VW₉O₃₄)₂]·35H₂O (Co ₄ -VPOM; Figure 4H).

Although both compounds exhibit the same structural arrangement, they display different electronic structures due to the redox properties of vanadium, influencing the catalytic performance. DFT calculations on the crystal structure of the Co₄-VPOM system supported this hypothesis, which showed that the HOMO level consists mainly of d orbitals from Co²⁺ ions, with some mixing of the oxygen orbitals of the inorganic tungsten fragments [VW₉O₃₄]. In contrast, the LUMO orbital consists mainly of the 3 d orbitals of vanadium with some mixing of the cobalt central core. The fact that vanadium works with a d⁰ electronic configuration allows for electronic transfers and redox processes that were not observed for the Co ₄ -PPOM system. Magnetic studies established the differences in the electronic features, revealing the ferromagnetic nature of Co ₄ -VPOM. A more complex analysis of the charge transfer phenomena in this compound was studied using two X-ray spectroscopic techniques and published elsewhere (Liu et al., 2018).

Although we focus this review on light-driven schemes, it is important to note that heterogeneous water oxidation catalysis, a system where the POMs appear stable and robust, has been reported. This methodology is helpful in poorly soluble systems or where it is necessary to avoid the formation of metal particles (e.g., CoOx) due to the inherent decomposition of the catalyst under oxidizing conditions. Carbon pastes modified with cobalt-substituted POM have shown a strong electrochemical oxidation wave, which is absent in pure carbon paste electrodes under the same catalytic conditions, suggesting the participation of catalytic processes promoted by the Co-POM components (Haider et al., 2019; Arens et al., 2020; Soriano-López et al., 2023). This versatility of exploring catalytic water oxidation using various methodologies favors exploring artificial photosynthetic systems, a topic we study below.

Hill’s group made the first report on Ni-substituted POMs in 2012. The complex was synthesized by reacting the POM precursor salt Na₁₀[SiW₉O₃₄] with NiCl₂-6H₂O and KCl in an aqueous medium at pH 6.8; again, the addition of the alkaline salt to favors the stability of the anionic polyoxometalate and its subsequent crystallization from the mother solution (Zhu et al., 2012a). Green needle crystals obtained by slow evaporation under ambient conditions were studied by X-ray diffraction, revealing a Wells-Dawson-type open structure [Si₂W₁₈O₆₆]^16- composed by the assembly of two [A-β-SiW₉O₃₄]^10- systems communicated through a pair of W-O-W bridging bonds. The resulting open structure allowed the incorporation of a central pentanuclear core {Ni₅(OH)₆(OH₂)₃}⁴⁺, which is of particular interest for water oxidation processes because the metal ions have water molecules in their coordination sphere, which are accessible for catalytic processes.

During catalytic experiments, the precipitation of a poorly soluble adduct formed between Ni ₅ -SiPOM and the photosensitizer was observed in dark and light-driven reactions. The Ni ₅ -SiPOM–[Ru(bpy)₃] ⁿ⁺ ion-pair was identified as the catalytically active species since the filtered solutions of these systems before the photo-illumination showed no O₂ evolution. The kinetic conditions of catalytic O₂ evolution in the presence of 2 μM of Ni ₅ -SiPOM are presented in Table 3 (entry 24) (Zhu et al., 2012b). Although Ni ₅ -SiPOM appears to be an effective catalyst, no detailed studies were conducted to interrogate the electronic or photophysical characteristics of the ion pair generated during the catalytic cycle. The association of the photosensitizer with the POM anion positively affected the stability of Ni ₅ -SiPOM since studies showed no nickel hydroxide/oxide particle formation under the conditions employed.

On the other hand, Ding and others designed a dimeric POM enclosing the Ni cluster [{β-SiNi₂W₁₀O₃₆(OH)₂(H₂O)}₂]^12- (Ni ₂ -SiPOM) that efficiently catalyzes water oxidation under visible light (Yu et al., 2016). Ni ₂ -SiPOM was prepared in a one-pot scheme by combining the salt K₈[β-SiW₁₁O₃₉] with NiSO₄.6H₂O in a potassium acetate solution under gentle warming and was separated in the solid-state as the hydrate potassium salt, K₁₂[β-SiNi₂W₁₀O₃₆(OH)₂(H₂O)}₂]·20H₂O (Figure 6C). The centrosymmetric polyanion Ni ₂ -SiPOM is compatible with two divacant lacunary [β-SiW₁₀O₃₆]^10- anions, which complete their Keggin-type structures incorporating two independent octahedral Ni^II ions. The corresponding asymmetric unit [{β-SiNi₂W₁₀O₃₆(OH)₂(H₂O)}₂]^12- grows and is linked with another symmetry-related fragment through two μ₃-oxygen atoms (O1 and O1’), resulting in a perfect parallelogram with Ni‒O bond distances in the range of 2.0 Å as shown in Figure 6D.

The catalytic activity was evaluated using the typical water oxidation system. The optimal conditions shown in Table 3, entry 25, were established by modifying different experimental conditions, such as the pH value of the buffer solutions and the concentrations of the catalyst, photosensitizer, and oxidizing agent. At concentrations above 15 μM, the catalytic activity decreases due to the appearance of heterogeneous species formed between the POM and the ruthenium-based photosensitizer (Yu et al., 2016). Several experiments confirmed that Ni ₂ -SiPOM is the dominant catalyst, ruling out the formation of soluble ions or nickel particles during water oxidation.

Following the structural rules of the [Mn₄CaO₅] cluster in plants, three POMs ligands stabilizing nickel centers with a cubane-type assembly, Na₂₄[Ni₁₂(OH)₉(CO₃)₃(PO₄)(SiW₉O₃₄)₃]·56H₂O (Ni ₁₂ -SiPOM), Na₂₅[Ni₁₃(H₂O)₃(OH)₉(PO₄)₄(SiW₉O₃₄)₃]·50H₂O (N ₁₃ -SiPOM), and Na₅₀[Ni₂₅(H₂O)₂(H₂O)₂(OH)₁₈(CO₃)₂(PO₄)₆(SiW₉O₃₄)₆]·85H₂O (Ni ₂₅ -SiPOM) were prepared, characterized by single crystal X-ray diffraction, and studied as molecular water oxidation catalysts (Figures 6E, G, I) (Han et al., 2015). The systems were prepared in water at basic pH using N₁₀[A-α-SiW₉O₃₄].18H₂O, NiCl₂.6H₂O, and Na₃PO₄.12H₂O. Furthermore, to obtain Ni ₁₂ -SiPOM and Ni ₂₅ -SiPOM as crystalline materials, it was necessary to add Na₂CO₃ and heat the reaction medium below 90 °C to activate the PO₄ ^3- groups.

Ni ₁₂ -SiPOM and N ₂₅ -SiPOM contain the partial cubane {Ni₃O₄} cluster encapsulated by PO₄ ^3- and CO₃ ^2- ligands (Figures 6F–J), with central units formulated as {Ni₃(CO₃)₃(PO₄)} and {Ni₃(CO₃)₂(PO₄)}, respectively. In Ni ₁₂ -SiPOM, the system {Ni₃(CO₃)₃(PO₄)} is covered by the carbon-free [A-α-SiW₉O₃₄]-(NiOH)₃] scaffolds. However, in N ₂₅ -SiPOM, two central {Ni₃(CO₃)₂(PO₄)} cores are connected through a {Ni(H₂O)₂} bridge, resulting in a complex entity {Ni₂₅(H₂O)₂(OH)₁₈(CO₃)₂(PO₄)₆}, stabilized by several [A-α-SiW₉O₃₄]-(NiOH)₃] units. On the other hand, in Ni ₁₃ -SiPOM the {Ni₄(H₂O)₃(PO₄)₄} cubane is wrapped by three {[A-α-SiW₉O₃₄]-(NiOH)₃} fragments. It is noteworthy that the fully inorganic {Ni₄(H₂O)₃(PO₄)₄)} moiety shown in Figure 6H encloses a central {Ni₄O₄} cubane, which structurally mimics the assembly present in the biological catalyst responsible for water oxidation in photosynthetic organisms.

Water oxidation was performed under the photocatalytic system outlined above. The catalysts showed the following order of efficiency Ni ₁₂ SiPOM < Ni ₁₃ -SiPOM < Ni ₂₅ -SiPOM with TON_max = 204.5 (see Table 3, entries 26–28), demonstrating that the nuclearity of the internal clusters influenced the water oxidation processes. The authors tested the activity of these compounds with various polyoxotungstates stabilizing different arrangements of Ni ions, which did not show any photocatalytic activity under controlled experimental conditions (pH 9) (Han et al., 2015). The ability of the Ni ₁₂ SiPOM, Ni ₁₃ -SiPOM, and Ni ₂₅ -SiPOM to efficiently sustain the water oxidation processes was explained by the ability of the photosensitizer to oxidize the POMs systems, by the nuclearity of the Ni centers affecting the number of possible active sites for the water molecules, and finally, by the cube-like arrangement of these nickel clusters. Stability studies of the catalysts ruled out the formation of nickel hydroxides, oxides, or other species during the catalytic process and the structural decomposition of the POMs as a function of time under oxidizing conditions.

In 2018, Das and others studied the electrocatalytic water oxidation process of a POM-supporting nickel(II) complex with interesting structural features, which was resolved by X-ray crystallography and other techniques as [Ni(bpy)₃]₃][{Ni(bpy)₂(H₂O)}{HCoW₁₂O₄₀}]·3H₂O (Ni-CoPOM) (Singh et al., 2018). The complex was obtained through solvothermal synthesis at 160 °C from a suspension formed by small quantities of K₆[CoW₁₂O₄₀]·6H₂O, Ni(OAc)₂·4H₂O, and 2,2′-bipyridine in an acidic solution. The hydrothermal reaction conditions allowed the covalent grafting of the [HCoW₁₂O₄₀]^5- anion by {Ni^II(bpy)₂(H₂O)} (Figure 6K). Additionally, the asymmetric unit contains 1.5 discrete [Ni(bpy)₃]²⁺ complexes stabilizing the POM anionic charge and is complemented by three lattice water molecules in general positions.

Several electrochemical experiments were conducted to determine the water oxidation activity of Ni-CoPOM and identify the true catalyst since several metal ions in the crystal structure can support WOC activity. The experiments showed that the [Ni(bpy)₂(H₂O)]²⁺ unit linked to the POM structure is the catalytically active species and that its activity is related to both the redox nature of the Ni²⁺ ions and its coordination environment. However, the Co²⁺ ion enclosed in [HCoW₁₂O₄₀] may also have an active role in the global catalytic performance. On the other hand, the nickel ions within [Ni(bpy)₃]²⁺ were discarded as potential WOC centers since they are coordinatively saturated.

[Ni(bpy)₂(H₂O)}{HCoW₁₂O₄₀] catalyzed the oxygen evolution reaction at neutral pH in phosphate buffer with an overpotential of 476 mV and a high TOF of 18.5 s^-1 O₂ evolved per mol of Ni ion at a current density of 1 mA cm^-2 (Singh et al., 2018). The same research group recently reported a similar system with a copper(II)-bipyridine complex attached to the [HCoW₁₂O₄₀]^5- anion; the active unit [{Cu(bpy)(H₂O)}₂[CoW₁₂O₄₀]] is an efficient electrocatalyst for the hydrogen evolving reaction in a near-neutral medium (Singh et al., 2021). These results introduce a new way of developing POM-based electrocatalysts for water splitting.

Mimicking the photosynthetic oxygen-evolving center with artificial systems to overcome the demanding water oxidation processes has been one of the most challenging issues bioinorganic chemists have faced over the past 2 decades. Nevertheless, after several efforts, a significant breakthrough in this direction was achieved when Zhang and co-workers synthesized 2015 the first artificial [Mn₄CaO₅] cluster through a two-step procedure using low-cost commercial chemicals (Zhang et al., 2015; Chen et al., 2017; Li et al., 2020). The artificial complex contains an asymmetric Mn₃CaO₄ unit attached to the fourth external manganese and is stabilized by eight (CH₃)₃CCO₂ ⁻ pendant units, forming a complex with the structural formula [Mn₄CaO₄(Bu^tCO₂)₈(Bu^tCO₂H)₂(py)] (Bu^t = tert-butyl; py = pyridine) (Zhang et al., 2015). In the molecule, the synthetic Mn₄CaO₄ cluster is similar to the catalytic structure in photosystem II (Figure 7) (Umena et al., 2011). Despite the remarkable structural similarity, the synthetic OEC was not stable under the established experimental conditions, and the catalytic performance could not be determined due to the rapid decomposition of cubane in the presence of water in the solution studied, which is essential for determining the WOC activity.

Using POMs as multidentate ligands to direct the assembly of cubane-type clusters may be a good design strategy to overcome the structural issues observed in artificial Mn₄CaO₄ models; this will lead to a molecule without the exact structural features of the biological catalyst (such as its organic environment). However, the incorporation of polynuclear Mn clusters into POM-based frameworks can not only solve the stability problem but also improve catalytic activity through charge transfer, redox processes, and the rapid removal of reducing equivalents generated by the water-splitting reaction, thus providing modern platforms for photosynthetic water oxidation.

An example of the above is the compound reported by Cronin’s group in 2011 formulated as K₁₈[Mn₂Mn₄(μ₃-O₂)-(H₂O)(β-SiW₈O₃₁)(β-SiW₉O₃₄)(γ-SiW₁₀O₃₆)]·40H₂O (Mn ₅ -SiPOM) (Mitchell et al., 2011). According to the authors, Mn ₅ -SiPOM represented the first model of an Mn-cubane wrapped by POM scaffolds (Figure 8A). The synthetic strategy consisted of adequate monitoring of both the temperature and pH, allowing the isomerization of the starting material {γ-SiW₁₀O₃₆} into the {β-SiW₈O₃₁} and {β-SiW₉O₃₄} polyanions. During the self-assembly process, the three lacunary species encapsulated a mixed-valence {Mn^III ₂Mn^II ₄O₄}²⁺ cubane, which closely resembles that found in the structure of the PSII (Figure 8B). This Mn cluster may be described as a {Mn₅O₄} cubane-type cluster linked to an external Mn^II atom via one of the μ₃–O vertices of the cube.

From an electrochemical point of view, Mn ₅ -SiPOM exhibited an interesting and similar redox activity both in solution and solid state, showing that the compound is stable under electrochemical conditions. Its stability was confirmed by ESI-MS experiments, revealing that Mn ₅ -SiPOM is exceptionally stable in H₂O/CH₃CN mixture (Mitchell et al., 2011). Moreover, the complex was also able to form microtubular architectures through morphogenesis when a crystal was exposed to an aqueous solution containing bulky cations. Although the WOC activity of the Mn ₅ -SiPOM compound has not been studied in detail, the design of materials inspired by this approach may open a window for the development of electronically interesting POM-based catalytic devices.

Scancola, Kortz Bonchio, and coworkers prepared the first Mn-substituted POM with WOC activity in 2014 (Al-Oweini et al., 2014). The complex shown in Figure 8C with formula Na_3.5K_2.5[Mn^III ₃Mn^IVO₃(CH₃COO)₃(A-α-SiW₉O₃₄].20H₂O was synthesized under mild conditions by reacting the polyoxometalate sodium salt [α-SiW₉O₃₄]^10-, prepared previously with a mixed-valence manganese complex [Mn^III ₈Mn^IV ₄O₁₂(CH₃COOH)₁₆(H₂O)₄]·2CH₃COOH.4H₂O; the reaction was carried out at pH 6 using a sodium acetate buffer solution; KCl was also used to stabilize the anionic charge of the resulting complex. The system crystallized in the centrosymmetric space group P 1 ¯ , and the molecule consists of an asymmetric manganese oxo system {Mn^III ₃Mn^IVO₃}, enclosed by a trivacant [α-SiW₉O₃₄]^10- anion and three acetate ligands. More importantly, the structural features and oxidation states of the central {Mn^III ₃Mn^IVO₃} core are almost exactly those in the reduced S₀ state of the natural [Mn₄CaO₄] cluster in PSII (Figure 8D).

The experiment to corroborate water oxidation activity was conducted electrochemically in a Na₂SiF₆/NaHCO₃ buffer solution (pH 5.2) using a 0.5 µM concentration of Mn ₄ -SiPOM. In the cyclic voltammogram, a strong catalytic current was observed at an onset potential of 1.25 V (vs. Ag/AgCl), which was attributed to water oxidation; this behavior was not observed in the buffer solution or the unpolished electrode, suggesting that the catalytic process was homogeneous.

Once the electrochemical conditions under which the catalyst oxidizes water were established, oxygen production in the photo-induced system was tested using [Ru(bpy)₃]²⁺ and S₂O₈ ^2-, and a wide range of catalyst Mn ₄ -SiPOM concentrations (from 6.3 to 50 µM) at pH 5.2 using the buffer noted above (Table 4, entry 29). Under illumination, O₂ evolved immediately (although with low yields), reaching a plateau after 30 min of exposure. The oxygen formation depended on catalyst concentration and persulfate consumption, while the low O₂ yield between 1.2 and 3.7% was attributed to the gradual decomposition of the photosensitizer under illumination. Control experiments with manganese(II) ions introduced as MnSO₄.H₂O led to insignificant O₂ production even at high concentrations, indicating that the action mechanism leading to water oxidation involves the four Mn ions of the redox active {Mn^III ₃Mn^IVO₃} core.

On the same basis, Streb’s group synthesized 2016 a molecular vanadium-manganese compound using one-pot self-assembly from readily available precursors, formulated as (n-Bu₄N)₃[Mn₄V₄O₁₇(OAc)₃]·3H₂O (Mn ₄ V ₄ ) (Schwarz et al., 2016). Crystallization of Mn ₄ V ₄ was achieved by solvent diffusion of diethyl ether into an acetonitrile solution containing (n-Bu₄N)₄[V₄O₁₄]4H₂O, Mn(OAc)₃, and (n-Bu₄N)MnO₄. X-ray studies showed a central {Mn₄O₄} oxo core stabilized by two different kinds of ligands, i.e., a vanadate species [V₄O₁₃]^6- and three acetate groups (Figure 8E). While [V₄O₁₃]^6- coordinated through O atoms from two tetrahedral VO₄ ^3- units, the acetate ions were coordinated through carboxylate groups, resulting in the anionic entity [Mn₄V₄O₁₇(OAc)₃]^3- with idealized C _3v symmetry. Three (n-Bu₄N)⁺ cations stabilize the anion in the crystal structure. The authors indicate that the Mn‒O distances and Mn···Mn interactions in this artificial cubane are similar to those in the cubic [Mn₄CaO₄] assembly from the oxygen evolution center. In addition, the oxidation states of the manganese ions, calculated using bond valence sum (BVS) from the crystal structure, are compatible with a central {Mn^III ₂Mn^IV ₂O₄}⁶⁺ unit (Figure 8F), analogous to the metastable S₁ state of the natural catalytic cycle.

Electrochemical analyses in anhydrous MeCN revealed that Mn ₄ V ₄ undergoes successive one-electron oxidation processes leading to the metastable electronic states S₂ (M^IIIMn^IV) and S₃ (Mn^IV) of the water oxidation cycle; these redox events were confirmed by EPR through comparison of the spectrum of the native compound {Mn^III ₂Mn^IV ₂O₄}⁶⁺ with the one- and two-electron oxidized systems obtained by bulk electrolysis, i.e., {M^IIIMn^IV ₃O₄}⁷⁺ and {Mn^IV ₄O₄}⁸⁺. The water oxidation activity of Mn ₄ V ₄ in the standard system [Ru(bpy)₃]²⁺/S₂O₈ ^2- was studied in a non-buffered MeCN/H₂O solution. The catalytic results support the electrochemical findings by showing that Mn ₄ V ₄ is a very active WOC with a maximum TON of 1,150 and TOF of 1.75 s^-1, using a concentration of 0.3 μM catalyst (Table 4, entry 30) (Schwarz et al., 2016). The TON values decreased considerably at concentrations above 0.6 μM, possibly due to the formation of adducts between the photosensitizer and the polyanion. Finally, deactivation of the catalytic activity was observed in recovering/recycling experiments, indicating that the experimental conditions could limit WOC performance.

Subsequently, Ding and others studied the O₂ production by a manganese antimoniotungstate, [Mn₃(H₂O)₃(SbW₉O₃₃)₂]^12- (Mn ₃ -SbPOM) and compared its catalytic activity with that of two structural analogs with different numbers of manganese ions and other heteroatoms in their structure: [Mn(H₂O)₅(PW₉O₃₄)₂]^9- and [Mn₃(H₂O)₃(AsW₉O₃₃)₂]^12-(Yu et al., 2017). The abbreviated compound Mn ₃ -SbPOM was synthesized following a process previously described by Bernt Krebs in 1998, which consisted of reacting the Na₉[SbW₉O₃₃] salt with MnCl₂.4H₂O in an aqueous medium (pH 6–7) (Bösing et al., 1998). After the addition of NH₄NO₃ solution, the complex crystallized by slow evaporation as a mixed ammonium-sodium salt Na₁₁(NH₄)[Mn₃(H₂O)₃(SbW₉O₃₃)₂]. The resulting crystal structure corresponds to two kegging-type [SbW₉O₃₃]^9- networks associated with three Mn²⁺ ions (Figure 8G). Each manganese atom has a distorted square pyramidal geometry with the apical position filled by a water molecule. Under a photochemical setup, the catalytic performance of Mn ₃ -SbPOM afforded TON and TOF values of 103 and 0.4 s^-1 (Table 4, entry 31). In contrast, its analogs showed low O₂ photocatalytic activity. No soluble ions or Mn oxides were detected during the water oxidation process.

In 2019, the compound [Mn(H₂O)₃)₂(H₂W₁₂O₄₂)]^6-, which consists of a polymeric network extending in 2D through manganese and tungsten centers, was prepared and characterized. The basic unit can be formulated as [Mn(H₂O)₃)₂(H₂W₁₂O₄₂)] _n ^6n- , which is stabilized by hydrated alkali metal ions such as sodium and potassium. An essential feature of this system is that the manganese centers are solvated by three water molecules, which can be used for catalytic processes. This complex undergoes electron transfer processes leading to the formation of manganese species in high oxidation states that promote water oxidation (Teillout et al., 2019). On the other hand, six Mn-substituted POMs selected from literature with different structural features were recently studied as electrocatalytic WOCs in solution and composite films. The controlled experiments indicated that the Mn ions’ content, oxidation state, and central core affected the electrocatalytic water oxidation activity (Wu et al., 2022).

It has been exemplified that an effective strategy in the search for new POM-WOCs is to test previously reported compounds with interesting structural features whose water oxidation activity has yet to be studied. This is the case of the first copper-substituted polyoxometalate studied as WOC, [Cu₅(OH)₄(H₂O)₂(α-SiW₉O₃₃)₂]^10- (Cu ₅ -SiPOM). This compound had already been studied in 2005 as a material with interesting magnetic properties but was not explored as a catalyst for water oxidation until 2015 by Ding and others (Figure 9A) (Nellutla et al., 2005; Yu et al., 2015). Cu ₅ -SiPOM was prepared by reacting the potassium salt K₁₀[α-SiW₉O₃₄] with CuCl₂.2H₂O in a hot acidic aqueous solution (pH 4.8). Crystals of the corresponding hydrated potassium salt, K₁₀[Cu₅(OH)₄(H₂O)₂(α-SiW₉O₃₃)₂].18H₂O (Figure 9B) were isolated by slow evaporation. The structure was described as a dimeric polyanion formed of two [α-SiW₉O₃₄]^10- frameworks fused through two bridging W–O bonds, which enclosed a central [Cu₅(OH)₄(H₂O)₂]⁶⁺ core, in which the Cu²⁺ ions have a Jahn-Teller distorted octahedral geometry. The Cu—O distances in the copper cluster are in the range of 1.907–2.387 Å.

The light-driven water oxidation of the Cu ₅ -SiPOM system was compared under the same conditions with that of four copper POMs, which differ in the number of copper centers and polyoxometalate networks: [{Cu(H₂O)SiW₈O₃₁}₂]^12-, [Cu₄(H₂O)₂(OH)₄(Si₂W₁₆O₅₈]^8-, [Cu₆Cl(SbW₉O₃₃)₂]^7-, and [Cu₂₀Cl(OH)₂₄(H₂O)₁₂(P₈W₄₈O₁₈₄)]^25-. Of these, only Cu ₅ -SiPOM showed oxygen evolution, indicating that the number of copper centers is unrelated to catalytic activity. A possible answer to the activity of this complex can be found in the arrangement of the [Cu₅(OH)₄(H₂O)₂]⁶⁺ cluster, in its protonation state, and in its stabilization with the robust tungsten-based ligands [SbW₉O₃₃]^9- (Yu et al., 2015). In the best experimental conditions shown in Table 4, entry 32, and in a concentration range of 5–30 μM, the copper system showed O₂ evolution under illumination in the typical photocatalytic system, reaching a plateau after 5 min, as seen in Figure 9C. Above 20 μM, the catalytic activity decreased due to interactions between the polyoxometalate and the photosensitizer. Several experiments confirmed that Cu ₅ -SiPOM is the active and dominant catalyst.

Subsequently, one Cu(II)-substituted POM was reported to exhibit water oxidation capabilities at neutral pH, mimicking the conditions in which the OEC performs water oxidation. The complex was reported by Ding’s group in 2018 as a catalyst for O₂ formation through homogeneous electrocatalytic water oxidation (Yu et al., 2018). The antimony-containing polyoxometalate [Cu₃(H₂O)₃[(SbW₉O₃₃)₂]^12- (Cu ₃ -SbPOM) was prepared following a synthetic strategy reported by Kortz, which consisted of reacting CuCl₂ ^.2H₂O with Na₉[SbW₉O₃₃]19.5H₂O in an aqueous solution at pH 6.2. Single crystals of the complex Na₁₂[(SbW₉O₃₃)₂Cu₃(H₂O)₃] were obtained by slow evaporation. The crystal structure features two [α-SbW₉O₃₃]^9- polyanions related by three equivalent Cu²⁺ ions, resulting in a sandwich-type design (Figure 9D). The compound shows structural features similar to the Mn ₃ -SbPOM manganese system described in the previous section. Different approaches confirmed electrocatalytic water oxidation and the system’s stability under electrochemical conditions.

Due to its abundance on Earth, iron is a very cheap metal and less toxic than other transition metals; therefore, iron compounds can play a leading role as WOCs (Liu et al., 2019). Nevertheless, water oxidation catalysis of Fe-substituted POMs began to be reported only recently. In 2005, Ding and co-workers informed a Fe-POM complex formulated through X-ray crystallography as Na₂₇[Fe^III ₁₁(H₂O)₁₄-(OH)₂(W₂O₁₀)₂(α-SbW₉O₃₃)₆]·103H₂O (Fe ₁₁ -SbPOM) obtained via a facile one-pot synthesis by combining FeCl₃ and Na₉[α-SbW₉O₃₃] in an acidic aqueous medium (Du et al., 2015a). Compound Fe ₁₁ -SbPOM was crystallized by slow evaporation at room temperature. The solid-state structure consisted of a large entity with a nanoscale size of 1.75 × 2.48 × 2.50 nm³ built by six robust [α-SbW₉O₃₃]^9- anions linked by a central [Fe₁₁(H₂O)₁₄(OH)₂(W₂O₁₀)₂]²⁷⁺ core (Figures 10A, B). According to BVS calculations, the oxidation states for all Fe atoms are +3. Under photochemical conditions, the TON reached 1815 with an initial TOF 6.3 s^-1 (Figure 10C). Parallel studies on Fe ₁₁ -SbPOM also showed remarkable photocatalytic activity for hydrogen evolution at neutral pH without adding cocatalysts or photosensitizers (Du et al., 2015b).

Subsequently, the same research group described a water oxidation system using bismuth vanadate BiVO₄ instead of the typical ruthenium photosensitizer, NaIO₃ as the sacrificial agent, and the molecular Fe ₁₁ -SbPOM complex as the catalyst (Hu et al., 2019). Under optimal conditions, the system evolved O₂ in a yield of 19.1%, sustaining an initial TOF of 173 h^-1 in an acetate buffer at pH 3, under which Fe ₁₁ -SbPOM is stable (Figure 10 d). Another spherical giant POM of formula, Mo₇₂Fe₃₀O₂₅₂(CH₃COO)_12-[Mo₂O₇(H₂O)]₂[H₂Mo₂O₈(H₂O)](H₂O)₉₁]·nH₂O, was studied as water oxidation catalysts in the [Ru(bpy)₃]²⁺/persulfate system at pH 9, showing a TON of 13.99 with a TOF of 3 min^-1 (Kaushik et al., 2020).