The hydrophobic effect is often introduced by using lipophilic substituents. Generally, the hydrophobicity modification of ligands can pre-organize complexes and boost the association of two metal centers, thus improving water splitting catalytic activity. Complex 1a, complex 12, and complex 13 WO catalysts are good examples of this phenomenon. The hydrophobic modification on ligands improves the WO catalytic activity of Ru-bda from 22 s⁻¹ to 146 s⁻¹ (Liu et al., 2021). The octyl substituent in the carbene ligand of triazolylidene Cp*Ir-complexes induces the association of the iridium species, leading to a ∼10-fold increase of TOF (101 min⁻¹) comparing with their methyl counterparts (TOF = 9.9 min⁻¹), shown in Figure 5A (Corbucci et al., 2015). By exploiting the extreme hydrophobicity of semi-fluorinated side chains (Figure 5B), Chen et al. (2016) reported an enhanced efficient WOC, Co-(BimC₃F₈), with a TOF of 1.83 s⁻¹, a 15-fold increase, at neutral pH without soluble cobalt, salts.

π—π interactions were extensively used in multi-components photocatalytic system, especially in promoting the electron transfer between photosensitizers and catalysts (Wang et al., 2022). Neel et al. (2017) reviewed how to exploit non-covalent π interactions for catalyst design. In chemical-driven water splitting process, π—π stacking is employed to facilitate the intermolecular interactions and accelerate bimolecular coupling. For example, complex 14 with isoquinoline (isoq) displays an order of magnitude higher TOF of 303 s⁻¹ than that of 32 s⁻¹ for complex 1a with picoline (pic), as shown in Figure 5C (Duan et al., 2012b). A faster catalysis was observed by introducing MeO-isoq, which causes a more favorable π—π stacking effects in water (Richmond et al., 2014). Correlated ab initio calculations demonstrated that modulation of π—π stacking dispersion interactions can lower activation barrier, therefore a smaller driving force for the catalysis is obtained (Johansson et al., 2021).

Attempts to improve the performance of WSCs also include the introduction of halogen substituents. Xie et al. (2018) studied in detail the influence of halogen substituents on the performance of complex 1b (Figure 5D). A 10-fold enhancement of TOF (330 s⁻¹) was found for R = I compare to R = H, which revealed that iodine can accelerate the O-O bond formation by facilitating the intermolecular interactions i.e. bimolecular coupling, due to its easy-polarization.

Electrostatic interactions were found to be another effective strategy to regulate binuclear catalysis. While attractive electrostatic interactions can facilitate inter-catalyst coupling and promote the catalytic performance, repulsive electrostatic interactions have a negative effect on the catalytic activity (Sampson et al., 2014; Pye and Mankad, 2017). In their recent study, Yi et al. (2021) prepared a family of Ru-bda catalysts (complex 15, Figure 6), functionalized with positively charged Me-bpy⁺ (Nmethyl-4,4′-bipyridinium) and/or negatively charged p-SO₃-py^- (pyridine-4-sulfonate) group, and identified the intermolecular electrostatic interactions by various methods. Complex 15c and the mixture M ([15a]: [15b] = 1:1) present 8–20 times higher TOF than complex 15a and 15b with repulsive effects. It was proved that electrostatic interactions are benefit to the formation of pre-reactive dimers, which were key intermediates in improving the catalytic activities. In fact, Richmond et al. (2014) had discovered the electrostatic effect in their earlier report, but they attributed the lowered catalytic activity to the high steric hindrance between the Me-bpy⁺ ligands.

In addition to aforementioned intermolecular interactions, Timmer et al. (2021) discovered that off-set interactions that introduced by de-symmetrization of the axial ligands in complex 16 and 17 (Figure 7A) can provide enough space for the O-O bond formation and reduce reaction barrier. DFT calculations suggested that reduced kinetic barrier of the second-order O-O bond formation ensures high catalytic performance especially at low catalyst concentrations. For Ru-bda catalysts with isoq in the axis, the position of pyridine substituents is crucial for stacking. Instead of direct π—π interactions, the off-set interaction brought by bromide shorten the distance of Ru-Ru in the pre-reactive dimer.

Another attractive way to facilitate the catalytic activity is to govern proton-coupled electron transfer (PCET) process by introducing proton acceptors/donators on ligands (Young et al., 2009; Wang et al., 2019; Zhang et al., 2021). Zhang et al. (2014) proposed ligand assisted PCET process for complex 18, where H on the 6 and 6′positions of bpy is replaced with hydroxyl groups, an internal base for proton transfer (Figure 7B). This directly lower the potential of complex about 200 mV, consequently resulting in its capability of driving WO to peroxide at a relatively low potential. Gao et al. (2007) designed bio-inspired manganese complex 19 with corrole ligands. The electrochemical data show that Mn corrole complexes 19 can catalyze oxidation of water to produce oxygen at quite low oxidation potentials, as indicated by its easy oxidation to high-valent states. Nocera et al. systematically studied the electrocatalytic behavior of Co/Ni hangman porphyrins complexes. They reported that owing to the pre-organization of water within the hangman cleft, the catalytic performance of these complexes (20, 21) can be dramatically boosted by employing carboxyl acid as a proton acceptor (Dogutan et al., 2011; Bediako et al., 2014; Neuman et al., 2020). Later, Sun et al. (2017) synthesized complex 22 and proved that the pendant hangman carboxyl moiety can act as intramolecular base to accelerate the APT process during the O-O bond formation. In the following up study, Nocera et al. found that the rate of catalysis of hangman iron porphyrins complexes 23 can be affected by nearly 3 orders of magnitude by improving the hanging group’s proton-donating ability (Graham and Nocera, 2014). Recently, the impact of carboxylate unites on electrocatalyzed WO process were deeply discussed by Das et al. (2021) Same as previous studies, the free carboxylic acid/carboxylate units can provide proton donor/acceptor sites through a chemically non-innocent way, hence can improve the overpotential and activity of the WO reaction dramatically.

Clearly, these intermolecular interactions offer inspirations for future design of WSCs. The comparison between modified and parent complexes have been summarized in Table 2.

Pyridine, a [N] type monodentate ligand, is one of the earliest developed monodentate chelating ligands where a ruthenium WOC was developed and is still widely used today (Carlin, 1961; Gersten et al., 1982). It can form coordinating bond with many transition metals such as Ru (Daniel et al., 2018; Kamdar and Grotjahn, 2019), Co, (Khademi et al., 2022; Navarro et al., 2022), Ni (Wang et al., 2019), and Cu (Lee et al., 2020), etc. And be used to prepare WSCs. Pal (2018) has well explained the coordinating mechanisms and emphasized its broad application in the fifth chapter of book “Pyridine”. Recently, pyridine was noticed again by Li et al. (2021), Zhu et al. (2022) as additives retarding the back-electron transfer or electron transfer between TiO₂ photoanode and the oxidized dye in solar water oxidation, or between an organic light absorber and a molecular WOC on a photoanode.

Based on the coordinating mechanisms of pyridine, Duan et al. (2009) reported the first case of Ru-bda WO catalyst. DFT prediction implied that complex with higher HOMO energy has higher durability, i.e., stability and lifetime (Duan et al., 2012a). When change the pyridine ligand to phthalazine (pzt) ligands, an extraordinary TON of 55400 and high TOF of 286 s⁻¹ were obtained. On the basis of computational analysis, a series of [N] ligands (Figure 12) with two benzene methine groups are occupied by “N” in ring, such as pyrimidine (pmd), pyrazine (prz), pyridazine (pdz), cinnoline, and phthalazine (ptz). This family of monodentate ligands shed light on the development of other types of WSCs.

Similar to pyridine-based [N] ligands in terms of metal-coordination properties and adjustable structures, a more electron-donating ligand, imidazole drew much attention. Imidazole has also been applied in preparing Ru-bda catalysts. The imidazole ligand can in situ form an active complex with the Ru^II center under the catalytic conditions (Wang et al., 2012). The key factor of catalytic performance in this case is that bulky ligand changes coupling between terminal oxygen atoms, and electronic properties.

In addition to [N] ligand, [C] ligand is another usual type of monodentate ligands such as carbene. Hetterscheid and Reek, 2011 studied [IrCp*(Me₂NHC)(OH)₂] (Me₂NHC = N-dimethylimidazolin-2-ylidene) complex with a large TON of 2000. DFT calculations show that the oxidant potential of this WOC can heavily influence its catalytic water oxidation via various competing channels (Diaz-Morales et al., 2014; Venturini et al., 2014). Iglesias and Oro (2018) reviewed iridium–NHC catalysts, and more carbene-containing complexes will be discussed in detail in the following sections.

Other monodentate ligands such as halogen ions (F, Cl, Br, I) and aqua (OH₂) were also extensively explored. When comparing the performance of complexes with different halogen ligands, two factors are usually considered: 1) steric effect coming from the different size of halogen atoms (I > Br > Cl), 2) the O–H∙∙∙halogen hydrogen bond intensity that may affect the water/proton exchange rate (Brammer et al., 2001). Besides these two factors, however, (Kaveevivitchai et al., 2012) proposed the third possibility: halogen atoms difference may change the water splitting pathway even if the complexes have the same chemical structure but different halogen atoms. Cyclic voltammetric data of Ru catalysts 29 d and 31 show that the aqua complexes are more difficult to oxidize than the analogous halide complexes. The WO data show that the aqua complexes performed better than the chloride and bromide complexes because the later ones require initial exchange of water for the Cl or Br ligand to produce the active intermediate species (Tseng et al., 2008; Masaoka and Sakai, 2009). However, the iodo-complexe presents unusual behavior where it catalyze considerably accelerate production of oxygen than the aqua-complex, as shown in Figures 13A,B. The unusually high initial rate for I-containing complex indicates the seven-coordinate rather than the six-coordinate intermediate pathway (Tseng et al., 2008).

[NN] ligand is the most studied type of chelating ligands and hundreds of derivatives have been developed so far. It is difficult to get an absolute conclusion of which ligand is better than the other since the catalytic performance is usually mechanism-dependent, and the same factor may have different effects but there are some general rules that can be followed, such as changing the steric geometry, electronic density around metal center, pKa, its ability of accepting/donating protons or transferring electrons, or building up PCET pathway, etc.

2,2′-bipyrimidine (bpm) and 2,2′-bipyrazine (bpz) are stronger π-acceptors and less electron-donating groups with respect to 2,2′-bipyridine (bpy), hence usually result in higher oxidation potentials but lower TON values. For example, E^o (Ru^III/II) of [Ru(tpy)(bpm)(OH₂)]²⁺ and [Ru(tpy)(bpz)(OH₂)]²⁺ catalysts are increased relative to that of [Ru(tpy)(bpy)(OH₂)]²⁺ by metal-ligand back-bonding to bpm or bpz in Ru(II). However, rate constants for the rate limiting step of catalysts with bpm or bpz is larger than that with bpy, demonstrating that a less energy is required for the O-O bond formation for WOCs with bpm or bpz (Concepcion et al., 2008; Concepcion et al., 2009b; Concepcion et al., 2010). Similar observations were also reported for the family of [Cp*Ir(NN)Cl] catalysts (36) where Cp* is pentamethylcyclopentadienyl and [M(NNN)(NN)(OH₂)]²⁺ catalysts (37) where M is Ir, Ru, or Os (Mcdaniel et al., 2008; Concepcion et al., 2009b; Blakemore et al., 2010). DFT calculation indicates that more nitrogen atoms in ligand result in less electron density at the reactant Ru-O bond, further explicating the slightly smaller TOF of catalyst with bpy than that with bpm or bpz (Jarvis et al., 2013). Moreover, the nitrogen atoms in the chelating ligands can also change the pKa and proton-electron transfer pathway of catalytic reactions. The potential-pH diagram for bpm complex and its comparison with bpy complex revealed that PCET avoiding charge buildup leads to the thermodynamical instability of Ru(III) and leads to its poor TON performance (Concepcion et al., 2009a).

Inspired by the high performance of Mn-ligating His332 of PS II, where deprotonation process improves the catalytic activity, imidazole-containing ligands such as 2,2′-diimidazole (H₂bim), 2-(2′-pyridyl)-imidazole (pimH), and 2-(2′-pyridyl)-benzoimidazole (pybim) were designed. An imidazole ring can not only provide more nitrogen atoms to tune electron density at metal center but also serve as a proton donor to tune proton-electron transfer pathways. In addition, the deprotonation of imidazole moiety on ligand could lower catalytic onset potential. Stott et al. (2017) studied Cu-based catalysts Cu(NN)(OH₂)₂ with introducing an imidazole ring into the bpy ligand. The experimental results showed that deprotonation of an ionizable imidazole ring can lower the metal reduction potentials and catalytic overpotentials of catalysts. Similarly, Okamura et al. (2012) modified [Ru(tpy)(bpy)(OH₂)]²⁺ by replacing bpy with H₂bim, a lower water oxidation potential was observed compare to analogous Ru WOCs due to the high donor power and multi-electron storage ability.

In addition to changing nitrogen atoms and proton donors, steric hindrance can also influence the catalytic performance. As discussed in the preceding text, the steric impact on the activity and stability of complexes depends on water splitting conditions and mechanisms hence we can’t conclude its positive or negative effect. However, the as-needed enhanced steric effect of backbone can be generally designed either by bond fixation (to hinder rotation) such as replacing bpy with phen or by increasing the size of backbone such as modifying an extra benzene ring on ligand, as shown in Figure 14. For instance, the catalytic activity of [Ru(tpy)(NN)Cl]²⁺ decreased with fusing an extra ring onto the bpy ligand (pyqn) and the activity is fully suppressed when there are two phenyl rings (bqn) (Tseng et al., 2008; Zeng et al., 2015).

Besides the commonly used bipyridine type bidentate [NN] ligands, amine-contained ligands were also developed in recent years especially in designing earth-abundant transition-metal complexes for molecular WSCs. Amine-type ligands are advantageous as 1) their simple characterizing conditions (Lu et al., 2016), 2) the easy active conformation during OER (Lee et al., 2020) in comparison with bpy ligands. For example, by mixing a Cu(II) salt and 1,2-ethylenediamine (en), Lu et al. (2016) synthesized Cu(en) catalyst with high WO activity over a wide pH range. WO catalysis occurs in solutions from pH 7 to 10 for [Cu^II(en)₂(OH₂)₂]²⁺ complexes. At higher pH, a catalytically active layer of CuO/Cu(OH)₂ formed on the electrode surface, producing O₂ in a high Faradaic yield and at an overpotential superior to other Cu-based surface catalysts.

Notably, the [NN] ligands generally determine the electrochemical tunability by virtue of the substituents on ligands or the intrinsic structure changes. Despite their decent catalytic activity, however, the complexity of ligands’ designs and synthesis upgrade the difficulties of obtaining homogeneous WSCs that are simultaneously simple, robust, and effective. 2-Phenylpyridine (ppy) ligand (Figure 14), where a nitrogen in bpy is replaced by carbon, is presented as an efficient ligand for WSCs owing to the formation of strong carbon-metal bond that extremely robust under typical conditions. Since it was first reported by Schmid et al. (1994) in preparing a WO catalyst [Ir (ppy)₂(H₂O)₂]⁺ (40), ppy was proven versatile enough to generate aquo complexes with various types of cyclometalating ligands by Mcdaniel et al. (2008). In the following work, Vilella et al. (2011) reported that [Ir(O)(X)(ppy)₂]ⁿ (X = OH₂, OH⁻ or O²⁻, depending on the pH) was the active catalytic species and X = O specie has the most basic internal base, hence demonstrated the lowest energy barrier for O–O bond formation.

Crabtree et al. reported a series of highly active and robust cyclopentadiene (Cp*)-containing Ir complexes and compared their water splitting behaviors (Hull et al., 2009; Blakemore et al., 2010). Apparently, the robust carbon-metal bond contributes to an increase of TON for [(Cp*)Ir(ppy)Cl]⁺ (41), but a slight decrease of TOF in comparison with [(Cp*)Ir(bpy)Cl]⁺ (36), where a [NN] ligand was used. Subsequently, mechanism studies showed that unlike ppy-contained Cp* iridium complex, where water molecules directly interacted with the Ir^V = O and formed O-O bond, in the bpy analogue, [Ce^IV(NO₃)₃(OH)] complex bridged the Ir^IV-O• species and a water molecule, as shown in Figure 15 (Bucci et al., 2016). Furthermore, (Savini et al., 2010) studied a water soluble complex [(Cp*)Ir(bzpy)Cl]⁺ (bzpy = 2-benzoylpyridine) with long-term activity 2 times higher than [(Cp*)Ir(ppy)Cl]⁺ by creatively introducing a –C(O)– bridging between the two aryl rings. Obviously, the presence of an electron withdrawing group stabilizes the complex by enhancing the π-back donation from the metal. However, adding ketone group increased the flexibility of ligand and made them less robust than their ppy counterpart (Savini et al., 2011).

Another type of [CN] ligand is N-heterocyclic carbene (NHC)-pyridine ligands, as shown in Figure 14. Vaquer et al. (2013) synthesized aqua-Ru complexes with different number of carbene ligands and revealed a linear relationship between the number of carbene ligands and Δ E_1/2, where Δ E_1/2 = E_1/2 (Ru^IV/III) - E_1/2 (Ru^III/II). The enhanced Δ E_1/2 increased the stability of Ru(III) oxidation state and therefore electrocatalytic driving force for WO. Vivancos et al. (2018) have recently reviewed the N-heterocyclic carbenes complexes in details in terms of their synthesis, catalysis, and other applications.

The strong coordination bonds between NHC ligands and transition metal centers can also significantly increase the stability of complexes containing [CC] ligands, such as bmmptraz. The influence of NHC on the catalytic water splitting activity was obvious. Albrecht et al. compared the catalytic performance of [(Cp*)Ir (bmmptraz)(MeCN)]²⁺ containing [CC] ligand (43) and its [CN] type counterpart (41) (Lalrempuia et al., 2010). Both the TON and TOF of 43 were larger than 41. Moreover, by comparing the electrochemical behavior of [(Cp*)Ir (ppy)Cl]⁺ (41) and [(Cp*)Ir(PhIm-ph)Cl]⁺ (43), Crabtree et al. demonstrated that the NHC ligand on high-valent iridium has a stabilizing effect (Brewster et al., 2011).

Some unusual bidentate ligands were also developed in the past decade. An oxidation- and dissociation-resistant [NO] ligand, pyalkH, has been proved useful in stabilizing unusually high oxidation states, such as Rh (Sinha et al., 2015), Ir (Shopov et al., 2015; Shopov et al., 2017), and Mn (Michaelos et al., 2016). Fisher et al. (2017) prepared Cu(pyalk)₂ WO electrocatalyst with high activity and stability. The oxhydryl group in pyalkH provided a deprotonated site for producing alkoxide form. The complex showed a decent TOF of 0.7 s⁻¹ under basic conditions (pH > 10.4). Mechanism studies suggested that only the cis form of Cu(pyalk)₂ could convert H₂O to O₂. Based on the same principle, various [OO] ligands were reported (Figure 16) (Concepcion et al., 2010; Lippert et al., 2010), as reviewed by Limburg et al. (2012).