Mechanisms of Photoisomerization and Water Oxidation Catalysis of Ruthenium(II) Aquo Complexes

Abstract

Polypyridyl ruthenium(II) complexes have been widely researched as promising functional molecules. We have found unique photoisomerization reactions of polypyridyl ruthenium(II) aquo complexes. Recently we have attempted to provide insight into the mechanism of the photoisomerization of the complexes and distinguish between the distal−/proximal-isomers in their physicochemical properties and functions. Moreover, polypyridyl ruthenium(II) aquo complexes have been intensively studied as active water oxidation catalysts (WOCs) which are indispensable for artificial photosynthesis. The catalytic aspect and mechanism of water oxidation by the distal-/proximal-isomers of polypyridyl ruthenium(II) aquo complexes have been investigated to provide the guided thought to develop more efficient molecular catalysts for water oxidation. The recent progress on the photoisomerization and water oxidation of polypyridyl ruthenium(II) aquo complexes in our group are reviewed to understand the properties and functions of ruthenium complexes.

1. Introduction

Polypyridyl ruthenium(II) complexes have been widely researched as promising functional molecules due to appealing photochemical [1, 2, 3] and photophysical [4, 5, 6] properties as well as redox properties [7, 8], which enable them to exhibit a number of functions such as electrochromism [9, 10], proton-coupled electron transfer [11, 12, 13] and photocatalysis [14, 15]. As a result, the polypyridyl ruthenium(II) complexes have been applied to a large variety of devices including sensors [16], photovoltaic cells [17], displays [18] and artificial photosynthesis [19, 20].

We presented irreversible and stoichiometric photoisomerization of distal-[Ru(tpy)(pynp)OH₂]²⁺ (d-1) (tpy = 2,2′;6′,2″-terpyridine, pynp = 2-(2-pyridyl)-1,8-naphthyridine) to proximal-[Ru(tpy)(pynp)OH₂]²⁺ (p-1) as shown in Figure 1 [21, 22, 23], which had not been characterized previously for polypyridyl ruthenium(II) aquo complexes although various photochemical reactions of the ruthenium(II) complexes have been reported [24, 25, 26, 27, 28, 29, 30, 31, 32, 33]. We have attempted to provide insight into the mechanism of the photoisomerization of polypyridyl ruthenium(II) aquo complexes and distinguish between the distal−/proximal-isomers in their physicochemical properties and functions [21, 22, 23, 34, 35, 36]. We have also developed new synthetic strategy to form dinuclear ruthenium(II) complexes utilizing the photoisomerization [37, 38]. Moreover, polypyridyl ruthenium(II) aquo complexes have been intensively studied as active WOCs [21, 22, 34, 35, 37, 39, 40] which are indispensable for artificial photosynthesis. The catalytic aspect and mechanism of water oxidation by the distal-/proximal-isomers of polypyridyl ruthenium(II) aquo complexes have been investigated to provide the guided thought to develop more efficient molecular catalysts for water oxidation. In this chapter, we review the recent progress on the photoisomerization and water oxidation of polypyridyl ruthenium(II) aquo complexes in our group.

Figure 1.

Photoisomerization of distal-[Ru(Rtpy)(pynp)OH₂]²⁺ to proximal-isomers.

2. Photoisomerization of polypyridyl ruthenium(II) aquo complexes

With respect to the photoisomerization of polypyridyl Ru(II) aquo complexes, the photoisomerization of cis-[Ru(bpy)₂(OH₂)₂]²⁺ (bpy = 2,2′-bipyridine) to its trans form in aqueous media was first reported by Meyer [24]. The mechanism of the photoisomerization reaction was later investigated by Planas et al. [41]. In this case, the trans form was present as a photostationary state and slowly went back to the original cis form. To the best of our knowledge, the cis-[Ru(bpy)₂(OH₂)₂]²⁺ had been the only one polypyridyl Ru(II) aquo complex that exhibits photoisomerization behavior before we presented the photoisomerization of d-1 to p-1 [21]. Furthermore, we reported the reversible photoisomerization equilibrium between distal- and proximal-[Ru(tpy)(pyqu)OH₂]²⁺ (d-2 and p-2) isomers with a ligand of 2-(2-pyridyl)quinoline (pyqu) instead of pynp, in contrast to the irreversible photoisomerization of d-1 to p-1. The aspect and mechanism of the irreversible photoisomerization of d-1 are first described, followed by those of the reversible one of d−/p-2 in this section.

2.1 Irreversible photoisomerization of distal-[Ru(Xtpy)(pynp)OH₂]²⁺ to the proximal-isomer

Photoisomerization behavior of d-1 was investigated by ¹H-NMR and UV–Vis spectroscopy. Upon irradiation of visible light to a D₂O solution of d-1, the NMR peak at 9.6 ppm due to d-1 decreased with the irradiation time and disappeared completely after 25 min under the experimental conditions (Figure 2A) [21]. The new peak at 8.9 ppm assigned to p-1 increased with the concomitant decrease of d-1. Figure 2B displays the concentration profile of d-1 and p-1, which indicates that irreversible and stoichiometric photoisomerization from d-1 to p-1 proceeds in water by visible light irradiation (Figure 1). By contrast, no isomerization of d-1 in water was found to occur under thermal treatment. The photoisomerization rate showed a first-order dependence on d-1 concentration, and the kinetic analysis provided the observed rate constant ((k_d-p)_obs/s⁻¹) of photoisomerization (distal to proximal) to be 4.1 x 10⁻³ s⁻¹ under the conditions employed (λ > 420 nm, 180 mW cm⁻²). The (k_d-p)_obs values increased linearly with respect to the light intensity below 255 mW cm⁻², indicating that the photoexcited state participated in photoisomerization under the employed conditions. Arrhenius plots of the photoisomerization gave a straight line in a range of 10 ∼ 35°C, providing 41.7 kJ mol⁻¹ of activation energy (E_a) for the photoisomerization (Table 1). Interestingly, the (k_d-p)_obs value decreased drastically at pD > 7 (pD = −log [D⁺]), while it was unchanged over pD 1-7. The UV–Vis spectrophotometric pH titration of d-1 gave pK_a of 9.7 attributed to the deprotonation of an aquo ligand to form the hydroxo complex, distal-[Ru(tpy)(pynp)OH]⁺. The trend of (k_d-p)_obs change depending on pH corresponds to the fraction of d-1 (aquo form) dissolved in the solution versus pH, suggesting that the hydroxo form of d-1 is inert for the photoisomerization [21, 22].

Figure 2.

(A) Time course of ¹H NMR spectral changes during the photoisomerization of d-1 in D₂O with 180 mW cm⁻² visible light irradiation (λ > 420 nm). (B) Kinetic profiles of d-1 (red circle) and of p-1 (blue square). (Reproduced with permission of American Chemical Society from ref. [21]).

Complexes	λmax / nm	pKa	cRu / mM	Photoisomerization parameters				Ref.
	(ε / M^−1 cm^−1)			(kd-p)obs / s^−1b	(kp-d)obs / s^−1b	Φ / % (λ = 520 nm)	Ea / kJ mol^−1
d-1	527 (9,300)	9.7	0.088	n.m.	—	1.5	n.m.	[21]
			3.9	4.1 x 10^−3	—	0.31	41.7	[34]
p-1	524 (9,300)	10.7	3.9	—	n.r	—	—	[21]
d-Cl1	524 (8,600)	9.3	0.088	n.m.	—	2.1	n.m.	[35]
			3.9	4.9 x 10^−3	—	n.m.	n.m.
p-Cl1	518 (8,900)	10.9	3.9	—	n.r	—	—	[35]
d-2	501 (8,300)	9.4	3.9	1.2 x 10^−2	—	1.1	23.3	[34]
p-2	502 (8,700)	10.5	3.9	—	4.3 x 10^−3	0.34	30.6	[34]
d-3	614 (5,700)	n.m.	2.0	n.m.	—	0.05	n.m.	[37]
p-3	612 (6,200)	n.m.	—	n.m.	n.m.	n.m.	n.m.	[37]

Table 1.

Summary of the observed rate constants and internal quantum yields of photoisomerization of d−/p-1, d−/p-Cl1, d−/p-2 and d−/p-3 isomers at 25°C.a

^a

c_Ru is concentrations of ruthenium complexes. The marks of “n.m.” and “n.r.” mean “not measured” and “no reaction”, respectively.

^b

The filtered halogen lamp was used for visible light irradiation (λ > 420 nm, 180 mW cm⁻²).

The internal quantum yield (Φ) for photoisomerization, which is defined as the ratio of the number of the photoisomerized complexes to the number of incident photons of a given energy, was estimated from the UV–vis spectral change in the experiment under monochromatic light irradiation (520 nm, 26.4 mW cm⁻²). The Φ values were calculated according to the following equation:

$$\mathrm{\Phi }=\frac{\mathit{hcN}{{}_{A}k}_{\mathit{pi}}{n}_{\mathrm{int}}}{\mathit{p\lambda A}\left(1‐T\right)}$$E1

where h, c, N_A, k_pi, n_int, p, λ, A and T are Plank’s constant, the speed of light, Avogadro’s number, the rate constant for photoisomerization, initial amount of the complex, photon flux, wavelength, the irradiated area and the transmittance, respectively. The Φ values for photoisomerization from d-1 to p-1 are 1.5 and 0.31% at 0.088 and 3.9 mM, respectively, as shown in Table 1.

In order to investigate influence of chloro substituent on photoisomerization of d-1, distal- and proximal-[Ru(Cl-tpy)(pynp)OH₂]²⁺ (d-Cl1 and p-Cl1; Cl-tpy = 4′-chloro-2,2′;6′,2″-terpyridine) complexes were prepared. When an aqueous solution of d-Cl1 was irradiated with visible light (λ > 420 nm, 180 mWcm⁻²), the stoichiometric photoisomerization of d-Cl1 to p-Cl1 was observed as it is for d-1 (Figure 1). The rate constant for photoisomerization of d-Cl1 was estimated to be (k_d-p)_obs = 4.9 x 10⁻³ s⁻¹ [35], which is higher than that ((k_d-p)_obs = 4.1 x 10⁻³ s⁻¹) [22, 34] observed for d-1 under the same conditions (Table 1). Additionally, the Φ value of d-Cl1 (2.1%) is higher than that (Φ =1.5%) of d-1.

Several groups reported that some polypyridyl complexes undergo photo-substitution reactions via the triplet metal centered (³MC) state from the ³MLCT excited state [25, 26, 27]. According to the reports, a possible mechanism for the photoisomerization of d-1 and p-1 isomers was speculated as follows. The ³MLCT excited state of d-1 is generated by absorption of visible light. The photo-dissociation of the aquo ligand from the exited d-1 proceeds through the thermal accessible ³MC state, leading to the formation of the penta-coordinated [Ru(tpy)(pynp)]²⁺ intermediate. The p-1 isomer is formed by re-coordination of a water molecule to the penta-coordinated intermediate from the opposite direction of a tpy plane. The temperature-dependent transient absorption spectroscopic measurements of d-1 suggest existence of the thermally activated process from the ³MLCT state with an E_a of 49 kJ mol⁻¹ [22], which is close to those from the ³MLCT state to the ³MC state reported for various polypyridyl Ru complexes [42, 43, 44, 45]. The agreement of the E_a value (49 kJ mol⁻¹) with that (41.7 kJ mol⁻¹) calculated from the Arrhenius plot for the photoisomerization (Table 1) also supports the possibility that a main activation process of the photoisomerization reaction is the thermal transition from the ³MLCT state to the ³MC state. However, Density functional theory (DFT) calculations suggested a different activation process, where the distal-penta-coordinated intermediate changes the conformation to proximal-penta-coordinated intermediate for p-1 while maintaining the otherwise octahedral structure of d-1 [22].

To obtain deeper mechanistic insights into the irreversible photoisomerization, the ³MLCT excited states of the d-1 and p-1 were characterized by the time-resolved infrared spectroscopy (TR-IR) [36]. The decay of the photoexcited ³MLCT states for both isomers were investigated by the TR-IR analysis, and the lifetimes of the excited state for the d-1 and p-1 were determined to be 9.7 ns and 6.4 ps, respectively. In general, the decay of the ³MLCT excited states of Ru polypyridyl complexes occurs on nanosecond timescale or above, because the transition from a triplet excited state to a singlet ground state is forbidden by spin selection rules. The very short excited lifetime (6.4 ps) for p-1 imply that a non-radiative process from the ³MLCT state was accelerated, so that p-1 is inert for photoisomerization to d-1.

The large difference in lifetimes of the ³MLCT state between d-1 and p-1 was interpreted by geometry optimization calculations using DFT of both d-1 and p-1 isomers in the singlet ground (S₀) and ³MLCT (T₁) states. While both the structures of d-1 and p-1 isomers in the S₀ states (indicated by lighter color atoms in Figure 3) show no considerable distortion, the aquo ligand was restricted by a hydrogen bond (1.48 Å) between its H atom and an N atom on the pynp ligand for p-1. The transition from S₀ state to T₁ state of d-1 results in a significant change in the dihedral angle of the pynp plane to the tpy plane from 180° to 161°, together with bending at the bond between the naphthyridine and pyridine rings. The distortion for d-1 is likely to originate from the steric hindrance between the extended π*-orbital of the pynp ligand and the π-orbital of the tpy ligand owing to the charge localization on the pynp ligand in the T₁ state. In the case of p-1, on the other hand, no considerable distortion was observed in the transition (Figure 3), but the Ru-O (1.98 Å) and hydrogen bonds (1.06 Å) between the pynp and the water ligands were shorter compared with the Ru-O (2.12 Å) and hydrogen bonds (1.48 Å) in the S₀ state. The shortened hydrogen bond is attributed to the charge localization on the pynp ligand in the T₁ state.

Figure 3.

Overlay of optimized structures of d-1 and p-1 in the S₀ (lighter color) and T₁ (deeper color) states. The top and bottom structures are illustrated from the horizontal directions of the pynp and tpy planes, respectively. The colored labels of atoms are as follows: Green, Ru; gray, carbon; red, oxygen; blue, nitrogen; white, hydrogen. (reprinted with permission from ref. [36] copyright 2015 Elsevier).

Figure 4 shows the DFT-calculated energy diagram of d-1 and p-1 in the S₀, T₁ and putative ³MC states. p-1 in the S₀ state is more stable than d-1 by 65 kJ mol⁻¹ because of hydrogen bond interaction between the pynp and the aquo ligand. The energy difference between the S₀ and T₁ states for d-1 is 176 kJ mol⁻¹, which is remarkably higher than that (101 kJ mol⁻¹) for p-1. The higher energy difference for d-1 is mainly due to the significant distortion of the pynp ligand on the transition from the S₀ to T₁ states (Figure 3). As a result, the energy of d-1 in the T₁ state is higher than that of p-1 in the T₁ state by 140 kJ mol⁻¹. Considering the similar ligand field for both isomers, the energy difference in the ³MC state between d-1 and p-1 is assumed to be not as much as that (140 kJ mol⁻¹) in T₁ states. For p-1, the ³MC state is presumed to be located at much higher energies than that in the T₁ state. This suggests the possibility of the different decay mechanism of p-1 in the T₁ state from the case of d-1 (usual lifetime of 9.7 ns), exhibiting the non-radiative decay through the thermally populated ³MC state. The possible mechanism is direct relaxation of the T₁ to S₀ states according to the energy-gap law: the decay rate decreases exponentially with increasing the energy gap between excited and ground states [46, 47, 48]. Considering almost the same geometries of p-1 between the S₀ and T₁ states, the law can be applied as the weak coupling limit. The lower energy band gap of 101 kJ mol⁻¹ between the S₀ and T₁ states for p-1 could explain the short lifetime, at least qualitatively, based on the energy gap law. The direct relaxation mechanism is consistent with the non-thermal decay process suggested by the very quick decay (unusual lifetime of 6.4 ps) for p-1 in the T₁ state. The direct relaxation process from the T₁ states to the S₀ state (not via the ³MC state) could explain the inert photoisomerization of p-1.

Figure 4.

DFT-calculated energy diagram of d-1 and p-1 in the S₀ and T₁ states. The speculated energy levels of T_n³MC states were described. (see text) (reprinted with permission from ref. [36] copyright 2015 Elsevier).

2.2 Reversible photoisomerization between distal- and proximal-[Ru(tpy)(pyqu)OH₂]²⁺ isomers

The photoisomerization between d-2 and p-2 was examined to reveal the influence of structures of bidentate ligands on the photoisomerization of d-1 (Figure 5) [34]. On irradiation of visible light to a D₂O solution of p-2, the 8.9 ppm peak characteristic of p-2 in the ¹H NMR spectrum (Figure 6A) decreased with increasing the 9.5 ppm peak assigned to the d-2 isomer. The ¹H NMR spectral change reached saturation in 5 min, while the 8.9 ppm peak did not completely disappear. This result suggests that the photoisomerization of p-2 to d-2 reaches a photo-stationary state (Figure 5). This shows that irreversible photoisomerization alters to reversible one by the replacement of the pynp ligand to pyqu in mononuclear Ru aquo complexes [21, 22]. From the integrated peak areas at 8.9 and 9.5 ppm, the equilibrated concentration ratio of p-2:d-2 was calculated to be 76:24 (Figure 6B). The kinetic profiles of the photoisomerization of p-2 to d-2 were well fitted with a reversible reaction model, giving the observed rate constants of the forward reaction (proximal to distal, (k_p-d)_obs/s⁻¹) of 4.3 ± 0.1× 10⁻³ s⁻¹ and the back reaction (distal to proximal, (k_d-p)_obs/s⁻¹) of 1.2 ± 0.03 × 10⁻² s⁻¹ under the conditions employed (λ > 420 nm, 180 mW cm⁻²), respectively (Table 1) [34]. The observed equilibrium constant were given to K_obs (= (k_p-d)_obs/(k_d-p)_obs) = 0.34 ± 0.01 under the employed conditions. Both (k_p-d)_obs and (k_d-p)_obs increased linearly with respect to light intensity up to 180 mW cm⁻², suggesting that the photoexcited states of p-2 and d-2 isomers are also involved in the forward and back reactions, respectively [34]. As shown in Figure 7, (k_p-d)_obs and (k_d-p)_obs decreased from pD = 8 to 12, and the photoisomerization did not take place at all above pD =12 [34]. This result implies that both hydroxo isomers, proximal- and distal-[Ru(tpy)(pyqu)OD]⁺ are inert for the forward and back photoisomerization reactions, respectively, as observed in the case of distal-[Ru(tpy)(pynp)OH]⁺ [21, 22]. The pD-dependent (k_p-d)_obs and (k_d-p)_obs analysis demonstrated inflection points at pD = 10.5 and 9.8 for the forward and back reactions, respectively. These values are close to the pK_a of p-2 (pK_a = 10.5) and d-2 (pK_a = 9.4) (Table 1). The difference in pK_a values (1.1) between p-2 and d-2 results in the markedly pD-dependent K_obs values (inset of Figure 7). K_obs increased above pH = 9, and reached its maximum value (K_obs = 2.5) at pH = 12. This trend is consistent with the observation that the yield of d-2 generated in the photoisomerization raised from 26–65% on increasing from pH 5.7 to 12.

Figure 5.

Photoisomerization equilibrium between proximal- and distal-[Ru(tpy)(pyqu)OH₂]²⁺ (p-2 and d-2) isomers.

Figure 6.

(A) Time course of ¹H NMR spectral changes during the photoisomerization of p-2 (3.9 mM, pD = 8.4) with 180 mW/cm² visiblelight irradiation (λ > 420 nm) in D₂O. (B) Kinetic profiles of p-2 (blue square) and d-2 (red circles). (reproduced from ref. [34] with permission of John Wiley & Sons, Inc.).

Figure 7.

Plots of observed rate constants ((k_p-d)_obs and (k_d-p)_obs) vs. pD for the photoisomerization reactions shown in Figure 5. Visible light (λ > 420 nm, 180 mW/cm²) was irradiated to the p-2 (3.9 mM) in D₂O at 25°C. Inset shows diagram of observed equilibrium constants K_obs vs. pD. (reproduced from ref. [34] with permission of John Wiley & Sons, Inc.).

The thermal dependent kinetics for (k_p-d)_obs and (k_d-p)_obs in a temperature range of 25 to 70°C gives E_a = 30.6 ± 2.9 and 23.3 ± 2.1 kJ mol⁻¹ for the forward and back reactions, respectively. The higher E_a of the forward reaction ((k_p-d)_obs) compared to the back one ((k_d-p)_obs) is attributable to the unfavorable loss of the hydrogen bond (C-H···O) between the H atom bonded to 8-C of the quinoline moiety of the pyqu ligand and the O atom of the aquo ligand upon water dissociation for photoisomerization. The van’t Hoff plots for K_obs gave ΔH = 7.7 ± 2.7 kJ mol⁻¹, that is close to enthalpy for the hydrogen bond (5.4 kJ mol⁻¹) of C-H···O in the benzene-water complex [49].

The Φ values for the forward and back photoisomerization reactions were estimated to be 0.34% and 1.1% from the experiments using monochromatic light (520 nm, 26.4 mW cm⁻²) (Table 1). The higher Φ (1.1%) of d-2 compared to that (0.34%) of p-2 could be attributed to the enthalpy of hydrogen bond (C-H···O) of p-2 (decreased Φ for p-2) and the steric repulsion between the H atom bonded to 8-C of the quinoline moiety of pyqu and the tpy plane for d-2 (increased Φ for d-2). The Φ value of 1.1% for photoisomerization from d-2 to p-2 is 3.5 times higher than that (0.31%) for the photoisomerization from d-1 to p-1 under the same concentration [34]. This could also be ascribed to the dominant steric repulsion between the tpy ligand and the H atom bonded to 8-C of the quinoline ring of pyqu for d-2, compared to the repulsion between the tpy ligand and the uncoordinated nitrogen of the naphthyridine ring of pynp for d-1.

2.3 New synthetic strategy for dinuclear ruthenium(II) complexes utilizing the photoisomerization

The photoisomerization of Ru(II) aqua complexes was applied to the strategic synthesis of dinuclear Ru complexes that are difficult to be synthesized by conventional thermal reactions so far. Herein, we succeeded to newly synthesize several dinuclear Ru(II) complexes, proximal, proximal-[Ru₂(tpy)₂LXY]³⁺ (L = 5-phenyl-2,8-di(2-pyridyl)-1,9,10- anthyridine, X and Y = other coordination sites; denoted as p,p-Ru₂XY) utilizing the photoisomerization of a mononuclear Ru aquo complex [37]. The tetradentate backbone ligand L was used to form a proximal, proximal -dinuclear Ru(II) structure (Figure 8). The reaction of [Ru(tpy)Cl₃] with L gave distal-[Ru(tpy)LCl]⁺ (d-3Cl) selectively in ethanol in a 59% isolated yield. This selective formation of d-3Cl was possibly caused by the steric hindrance of L with a chloro ligand. d-3Cl was then converted to distal-[Ru(tpy)LOH₂]²⁺ (d-3) by chloro subtraction with a silver salt in water in a 90% isolated yield. The thermal reaction of d-3 with a second ruthenium center [Ru(tpy)Cl₃] for dimerization failed to give the proximal, proximal-dinuclear Ru(II) complex owing to the steric hindrance located between the tpy ligand on the d-3 and the one on [Ru(tpy)Cl₃]. However, if the photoisomerization of d-3 to proximal-[Ru(tpy)LOH₂]²⁺ (p-3) is utilized (Figure 8), the steric constraint of d-3 for formation of the proximal,proximal-dinuclear Ru(II) species could be avoided. The photoisomerization of d-3 stoichiometrically progressed to form p-3 in water/ethanol mixture under visible light irradiation (λ > 420 nm). The subsequent thermal reaction of p-3 with [Ru(tpy)Cl₃] in water/ethanol mixture successfully generated proximal,proximal-[Ru₂(tpy)₂L(μ-Cl)]³⁺ (p,p-Ru₂(μ-Cl) in a 67% isolated yield that was unambiguously characterized by X-ray diffraction [37]. The p,p-Ru₂(μ-Cl) was converted to the proximal,proximal-[Ru₂(tpy)₂L(μ-OH)]³⁺ (p,p-Ru₂(μ-OH)) in neutral or slightly basic aqueous medium via substitution of the μ-Cl bridge with an OH⁻ ion as shown in Figure 9 [37, 38]. The p,p-Ru₂(μ-OH) was then converted to proximal,proximal-[Ru₂(tpy)₂L(OH)(OH₂)]³⁺ (p,p-Ru₂(OH)(OH₂)) via insertion of a water molecule to the central core to reach equilibrium between p,p-Ru₂(μ-OH) (∼10%) and p,p-Ru₂(OH)(OH₂).

Figure 8.

Synthetic scheme of a μ-Cl bridged dinuclear ruthenium complex, p,p-Ru₂(μ-Cl) using photoisomerization of d-3. (reproduced from ref. [37] with permission of American Chemical Society).

Figure 9.

Reversible bridging-ligand substitution reactions among p,p-Ru₂(μ-Cl), p,p-Ru₂(μ-OH)p,p-Ru₂(OH)(OH₂).

3. Water oxidation catalysis by ruthenium(II) aquo complexes

The mononuclear ruthenium polypyridyl aquo complexes are promising WOCs to understand water oxidation chemistry and provide the guided thought for efficient WOCs because of their simple structures, high catalytic activities, ease of chemical modification, and informative knowledge including catalytic mechanisms [13, 21, 22, 23, 34, 35, 40, 50, 51, 52]. Controlling the electron density on the metal center as an active site by alternating the substituent groups on their ligands is a common approach to reveal the mechanism and influencing factors for water oxidation catalysis.

The molecular WOCs take part in the process of the four electrons and four protons removal from two water molecules, either consecutively or concerted, to form the O-O bond. Understanding the mechanism of the O-O bond formation is vital in improvement of molecular WOCs. Two main classes of proposed reaction mechanisms at metal centers are shown in Figure 10; water nucleophilic attack (WNA) on metal-oxo centers (M = O) and interaction of two M-O centers (I2M) [53, 54, 55, 56]. In the WNA mechanism (Figure 10a), an electrophilic high-valent metal-oxo (Mⁿ⁺ = O) species is formed via multiple consecutive oxidation steps. A nucleophilic attack of a water molecule on the Mⁿ⁺ = O species occurs, that leads to formation of a hydroperoxide (M⁽ⁿ⁻²⁾⁺-OOH) species. Further oxidation and deprotonation steps generate a M⁽ⁿ⁻¹⁾⁺-OO⁻ intermediate, which releases O₂ and converts to M⁽ⁿ⁻²⁾⁺-OH₂ with incorporation of OH₂. A lot of mononuclear complexes have been proposed to undergo the WNA mechanism for water oxidation [21, 22, 23, 34, 35, 40, 51, 52, 57, 58]. In the I2M mechanism (Figure 10b), the coupling of either two Mⁿ⁺-O oxyl radicals or coupling of one Mⁿ⁺-O oxyl radical with another Mⁿ⁺-O unit of a non-radical character affords a peroxo Mⁿ⁺-O-O-Mⁿ⁺ intermediate, which releases O₂ and returns to 2 M⁽ⁿ⁻²⁾⁺-OH₂ with incorporation of OH₂. It involves intramolecular [37, 59, 60] and intermolecular [61, 62] pathways.

Figure 10.

Schematic representation of the two mechanistic pathways for O-O bond formation. (reproduced with permission from ref. [20]. Copyright 2019 Wiley).

In this section, we introduce our recent progress on chemical and electrochemical water oxidation catalyses by mono- and dinuclear ruthenium(II) aquo complexes in homogeneous systems. Firstly, substitution effects on the catalytic activity and mechanism of mononuclear ruthenium(II) aquo complexes for chemical water oxidation are described in Section 3-1. Secondly, the difference in the catalytic properties between distal−/proximal-isomers is explicated in Section 3-2. Finally, the electrocatalytic activities of a series of dinuclear ruthenium(II) complexes are discussed in Section 3-3.

3.1 Substitution effect on the catalytic activities of mononuclear ruthenium complexes for water oxidation

To investigate the catalytic aspects of [Ru(Rtpy)(bpy)(H₂O)]²⁺ (Rtpy = 2,2′:6′,2″-terpyridine derivatives) complexes (4R) having a variety of 4′-substituent groups on Rtpy ligand (Figure 11), chemical water oxidation experiments were performed in a homogeneous aqueous solution using a Ce^IV as a sacrificial oxidant [39, 40]. For 4EtO (R = EtO), O₂ was significantly evolved from the catalytic solution, and the amount (n_O2/μmol) of O₂ evolved increased linearly until 100 min and then saturated to 35 μmol at 4 h, which is 5 times higher than that for 4H (Figure 12) [39]. The initial O₂ evolution rates (v_O2/mol s⁻¹) were calculated from the initial slopes. The v_O2 for 4EtO increased with the Ce^IV concentration (c_Ce/M) and reached a saturation at c_Ce = 1.0 M (Figure 13), indicating that O₂ evolution is zero order with respect to Ce^IV under the large excess Ce^IV conditions. This behavior was well-analyzed by Michaelis–Menten-like kinetics to give the maximum catalytic rate (v_max) of 1.5 (±0.08) x 10⁻¹ mol s⁻¹ (4EtO mol)⁻¹ and the constant (K_m = 1.2 (±0.06) mmol) in terms of the Ce^IV concentration for the half value of v_max. The oxidation reaction from Ru^IV=O to Ru^V=O by Ce^IV could involve the redox equilibrium prior to oxygen evolution because the redox potential (1.45 V vs. SCE) of Ru^IV=O/Ru^V=O for 4EtO is close to standard potential (E° = 1.47 V vs. SCE (1.71 V vs. NHE) [63]) of Ce^III/IV. The redox equilibrium presumably leads to saturation of the v_O2 with increase of the Ce^IV concentration (Figure 13). The v_O2 increased linearly with the 4EtO concentration (c_Ru/M) under the large excess Ce^IV conditions (c_Ce = 0.1 M, c_Ru = ∼ 0.2 mM, 5.0 ml water) (Figure 14), showing that the O₂ evolution is a first order process with respect to 4EtO. This result is consistent with the case of earlier-reported mononuculear Ru(II) aquo complexes [50, 51].

Figure 11.

Chemical structures of mononuclear ruthenium(II) aquo complexes.

Figure 12.

Time courses of the amount (n_O2/Mol) of O₂ evolved in chemical water oxidation experiments in an aqueous solution at 25°C using a Ce^IV oxidant. c_Ce = 0.1 M (0.5 mmol); complexes, 20 μM (0.1 μmol); pH = 1.0; liquid volume, 5.0 ml. (○) 4EtO, (●) 4H and (□) without complex. (reproduced with permission from ref. [39]. Copyright 2011 the Royal Society of Chemistry).

Figure 13.

Plots of v_O2 versus the Ce^IV concentration (c_Ce/M) in chemical water oxidation experiments using 0.2 mM 4EtO (5.0 ml water) at 25°C. the solid line is the simulated curve based on a Michaelis–Menten-like kinetic equation. (reproduced with permission from ref. [39]. Copyright 2011 the Royal Society of Chemistry).

Figure 14.

Plots of initial O₂ evolution rate (v_O2/Mol s⁻¹) versus the amount of 4R complexes. c_Ce = 0.1 M, 5.0 ml water, pH = 1.0. (reproduced with permission from ref. [40]. Copyright 2019 American Chemical Society).

The mechanism of water oxidation by 4R is shown in Figure 15 based on the WNA mechanism [50, 51, 56]. 4R (abbreviated to Ru^II-OH₂ as the oxidation state) is oxidized by Ce^IV to Ru^III-OH₂ by a 1-electron process, and subsequently oxidized to Ru^IV=O by two-proton/one-electron process at pH = 1.0 [39, 40]. Ru^IV=O is further oxidized to Ru^V=O, involving the above-mentioned redox equilibrium (step I, Figure 15). Ru^V=O undergoes water nucleophilic attack (step II, Figure 15) to form an O-O bond in the Ru^III-OOH intermediate. Finally, O₂ is produced via further oxidation to Ru^IV-OO⁻, with Ru^II-OH₂ regenerated by incorporation of water.

Figure 15.

Mechanism of water oxidation by 4R complexes at pH = 1.0. (reprinted with permission from ref. [40]. Copyright 2019 American Chemical Society).

The v_O2 showed a first-order dependence on the complex amount for all the 4R derivatives (Figure 14) [40], suggesting that O₂ is produced by a unimolecular reaction of 4R. Turnover frequencies (k_O2/s⁻¹) were calculated from the slopes of the linear relationships. The k_O2 values were largely affected by the Rtpy ligands, being variable from 0.05 to 44 x 10⁻² s⁻¹ (as the highest k_O2 for 4EtO) by a factor of 880 (Table 2). The k_O2 values were plotted with respect to Hammett constant σ_p of the 4′-substituents of Rtpy ligands in Figure 16A. However, the Me₂N group is protonated to give Me₂NH⁺ under the catalytic conditions (pH = 1.0), and the σ_p value of Me₂NH⁺ was assumed as σ_p = 0.71 of the midpoint between σ_p = 0.60 for protonated amino group and σ_p = 0.82 for protonated trimethylamino group [64]. The k_O2 values were almost constant (3.4 ∼ 6.1 x 10⁻² s⁻¹) in a range of σ_p = −0.17 (R = Me) ∼ 0.23 (R = Cl), although the k_O2 value (5 x 10⁻⁴ s⁻¹) of 4Me₂N is lower than these values. The k_O2 value increased sharply at σ_p = −0.24 (R = EtO) to 4.4 x10⁻¹ s⁻¹ as the maximum, and thereafter decreased with the σ_p decrease to −0.37 (R = OH). This result demonstrates that very critical Hammett constant (σ_p = −0.27 ∼ −0.24) exists for the high k_O2 values. The very low k_O2 values for 4OH and 4Me₂N arise from the difficulty of Ru^V=O formation. Most likely, 4OH and 4Me₂N are considered to decompose through their ligand oxidation during water oxidation catalysis, as reported recently [65, 66].

Complexes	Abbreviation	kO2/10^−3 s^−1	Ref
[Ru(tpy)(bpy)OH2]^2+	4H	34	[39]
[Ru(Cltpy)(bpy)OH2]^2+	4Cl	43	[40]
[Ru(Metpy)(bpy)OH2]^2+	4Me	61	[40]
[Ru(MeOtpy)(bpy)OH2]^2+	4MeO	240	[40]
[Ru(EtOtpy)(bpy)OH2]^2+	4EtO	440	[40]
[Ru(PrOtpy)(bpy)OH2]^2+	4PrO	331	[40]
[Ru(OHtpy)(bpy)OH2]^2+	4OH	0.9	[40]
[Ru(Me2Ntpy)(bpy)OH2]^2+	4Me2N	0.5	[40]
distal-[Ru(tpy)(pynp)OH2]^2+	d-1	3.8	[21]
proximal-[Ru(tpy)(pynp)OH2]^2+	p-1	0.48	[21]
distal-[Ru(Cltpy)(pynp)OH2]^2+	d-1Cl	6.3	[35]
proximal-[Ru(Cltpy)(pynp)OH2]^2+	p-1Cl	0.39	[35]
distal-[Ru(tpy)(pyqu)OH2]^2+	d-2	1.0	[34]
proximal-[Ru(tpy)(pyqu)OH2]^2+	p-2	1.7	[34]

Table 2.

Summary of k_O2 by mononuclear Ru(II) aquo complexes.

Figure 16.

Plots of the turnover frequency (k_O2/s⁻¹) versus Hammett constant (σ_p) of 4′-substituent groups (A) and versus the redox potentials of a Ru^IV=O/Ru^V=O pair (B). The Me₂N group is protonated to a Me₂NH⁺ group under the experimental conditions for water oxidation (pH = 1.0), and the σ_p value of Me₂NH⁺ was assumed as σ_p = 0.71 of the midpoint between σ_p = 0.60 for protonated amino group and σ_p = 0.82 for protonated trimethylamino group [64]. The redox potentials were estimated from E_ap values, where E_ap is an anodic peak potential corresponding to the Ru^IV=O/Ru^V=O couple in a cyclic voltammogram for each of 4R complexes. The red dashed line indicates the standard redox potential of Ce^III/IV [63]. (Reproduced with permission from ref. [40]. Copyright 2019 American Chemical Society).

The k_O2 values were plotted with respect to redox potentials of Ru^IV=O/Ru^V=O in Figure 16B. The k_O2 values increased with the decrease of the redox potentials in an order of 4H, 4Cl, 4Me < 4MeO < 4PrO < 4EtO. This indicates that the potential for formation of Ru^V=O species is essential in the water oxidation catalysis under the employed conditions. This k_O2 profile can be explained by the increased fraction of Ru^V=O in the equilibrium. In this case, the oxidation rate (step I in Figure 15) from Ru^IV=O to Ru^V=O by Ce^IV could be comparable with the rate of the nucleophilic attack of water on Ru^V=O (step II in Figure 15) to be involved in a rate-determining step. The O-O bond formation process via water nucleophilic attack could be no longer a rate-determining step singularly. One might deservedly expect that the k_O2 values of 4H, 4Cl, and 4Me with the higher redox potentials of Ru^IV=O/Ru^V=O are higher than those of 4MeO, 4PrO and 4EtO with the lower redox potentials since water nucleophilic attack on Ru^V=O (step II, Figure 15) is assumed to accelerate because of the higher electrophilicity of the oxo of Ru^V=O for the formers. By contrast, the k_O2 values of 4H, 4Cl, and 4Me were indeed lower than those of 4MeO, 4PrO and 4EtO, which could be explained by the slow oxidation rate (step I in Figure 15) from Ru^IV=O to Ru^V=O involved in a rate-determining step.

3.2 Comparison in catalytic activities between distal and proximal isomers of mononuclear ruthenium complexes

To explore the catalytic aspects of a series of distal and proximal isomers for the mononuclear Ru(II) aquo complexes, chemical water oxidation experiments were conducted in a homogeneous aqueous solution using a Ce^IV oxidant [21, 22]. O₂ was significantly evolved from a mixed solution of d-1 and Ce^IV, and n_O2 increased linearly with time until 50 min (Figure 17). v_O2 increased linearly with respect to the amount of the complex, and the slope of the linear relationship provides k_O2 = 3.8 x 10⁻³ s⁻¹ (Figure 18). The same chemical water oxidation experiments were carried out after visible light irradiation to the solution of d-1 for 1 h to generate p-1 completely (Figure 1). n_O2 dramatically decreased compared with the case before light irradiation (Figure 17). The linear plot of v_O2 with the catalyst amount provided k_O2 = 4.8 x 10⁻⁴ s⁻¹ for p-1, indicating that the observed k_O2 value is decreased due to photoisomerization of d-1 to p-1 by nearly an order of magnitude. This result tells us that we have to pay attention to the observed catalytic activity decrease due to photoisomerization of d-1 to p-1 when d-1 is applied to photocatalytic systems.

Figure 17.

Time courses of the amount (n_O2/μmol) of O₂ evolved in chemical water oxidation experiments in an aqueous solution at 25°C using a Ce^IV as a sacrificial oxidant. c_Ce = 0.1 M (0.5 mmol); Ru complex, 1.0 μmol; pH = 1.0; liquid volume, 5.0 ml. (●) d-1, (■) p-1. (Reproduced with permission from ref. [21]. Copyright 2011 American Chemical Society).

Figure 18.

Plots of initial rate (v_O2 / mol s⁻¹) of O₂ evolved vs. the amount (μmol) of d-1Cl (■), p-1Cl (●), d-1 (□), p-1 (○). Conditions: c_Ce = 0.1 M (0.5 mmol); liquid volume, 5.0 mL; pH = 1.0. (reproduced with permission from ref. [35]. Copyright 2015 Elsevier).

The chemical water oxidation catalyzed by d-1Cl and p-1Cl was investigated under the same conditions as the d−/p-1 isomer system to understand the effect of 4′-chloro-substitusion on tpy ligand on the catalytic activity of the distal−/proximal-isomer complexes for water oxidation [35]. v_O2 increased linearly with respect to the complex amount for d-1Cl and p-1Cl, as is the case of the d−/p-1 isomer system (Figure 18). k_O2 (6.3 × 10⁻³ s⁻¹) for d-1Cl was 15 times higher than that (3.9 × 10⁻⁴ s⁻¹) for p-1Cl. From a perspective of the effect of the chloro-substitution on tpy, k_O2 for d-1Cl was also 1.6 times higher that (3.8 × 10⁻³ s⁻¹) for d-1, while k_O2 for p-1Cl was 1.2 times lower than that (4.8 × 10⁻⁴ s⁻¹) for p-1. If assuming that the O-O bond formation via the nucleophilic attack of water on the Ru^V=O intermediate is the rate-determining step, k_O2 could increase by the chloro substitution because electrophilicity of Ru^V=O increased, as is the case for d-1Cl relative to d-1. However, another explanation is needed for the k_O2 decrease for p-1Cl relative to p-1. For instance, there are cases that the rate for oxidation of Ru^IV=O to Ru^V=O could be involved in a rate-determining step, or that the stabilities of the complexes are different, as pointed out in the Section 3-1.

The k_O2 value (1.7 × 10⁻³ s⁻¹) of p-2 was higher than that (1.0 × 10⁻³ s⁻¹) of d-2 by a factor of 1.7 (Table 2) [34]. This result is in contrast to the d−/p-1 and d−/p-1Cl isomer systems, in which k_O2 of distal-isomers are higher than those of the proximal-ones by one order of magnitude. The k_O2 value of p-2 is 3.5 times higher than that (4.8 × 10⁻⁴ s⁻¹) of p-1, though the value of d-2 is 3.8 times lower than that (3.8 x 10⁻³ s⁻¹) of p-2 under the same conditions.

3.3 Electrocatalytic activities of a series of dinuclear ruthenium complexes for water oxidation

We investigated catalytic activities of p,p-Ru₂(OH)(OH₂) and the related mono- and dinuclear Ru(II) complexes (Figures 8 and 9) for electrochemical water oxidation in homogeneous solution [37]. The cyclic voltammogram (CV) of p,p-Ru₂(OH)(OH₂) displayed a higher anodic current after 1.2 V vs. SCE attributed to water oxidation (Figure 19). The catalytic current density increased to 3.5 mA cm⁻² (the blank without the complex generates 0.31 mA cm⁻²) at a potential of 1.4 V and pH 6.0, that is 4.8 and 9.2 times higher compared to those of d-3 and p,p-Ru₂(μ-Cl). Importantly, the current density value (3.5 mA cm⁻² at pH 6.0) obtained from p,p-Ru₂(OH)(OH₂) was much higher than that (1.4 mA cm⁻² at pH 9.0) obtained for p,p-Ru₂(μ-OH), under even thermodynamically unfavorable pH conditions. These results suggest that the proximal,proximal-dinuclear Ru(II) core structure with vicinal aquo and hydroxo groups is inevitably essential for efficient electrocatalytic water oxidation. Bulk electrolysis was performed in a nearly neutral phosphate buffer solution (pH 6.0) for p,p-Ru₂(OH)(OH₂) at 1.3 V vs. SCE. A higher charge amount of 2.1 C compared to the that (0.55 C) of the blank without the complex was obtained, and a 4.2 μmol (Faradaic efficiency 76–80%) of O₂ was produced after 1 h electrolysis. This evolved O₂ amount obtained by p,p-Ru₂(OH)(OH₂) corresponds to 5.3 equivalent of the total p,p-Ru₂(OH)(OH₂) amount (0.6 μmol) in the electrolyte solution, ensuring the catalytic water oxidation. The UV–Visible absorption spectrum of the electrolyte solution after the electrolysis displayed an intense band at 694 nm, assigned to proximal,proximal-[Ru^III₂(tpy)₂L(OH)₂]⁴⁺ (abbreviated to Ru^III-OH:Ru^III-OH as a oxidation state). This observation suggests that Ru^III-OH:Ru^III-OH is involved in the catalytic cycle. For the proposed electrocatalytic cycle for water oxidation under neutral conditions, p,p-Ru₂(OH)(OH₂) was electrochemically oxidized by the proton-coupled electron transfer reactions via Ru^III-OH:Ru^III-OH and Ru^IV=O:Ru^IV–OH states most possibly to the Ru^V=O:Ru^V=O state, which could oxidize water to O₂ together with the regeneration of Ru^III-OH:Ru^III-OH.

Figure 19.

CVs of 1 mM p,p-Ru₂(OH)(OH₂) (red), d-3 (blue), p,p-Ru₂(μ-Cl) (green), and blank (black dots) in a 0.1 M phosphate buffer (pH 6.0) at a scan rate of 50 mV s⁻¹. CV of p,p-Ru₂(μ-OH) (black) was measured at pH 9.0 because it gradually converts to p,p-Ru₂(μ-Cl) at acidic conditions. Inset shows CV of 0.5 mM p,p-Ru₂(OH)(OH₂) in a 0.1 M phosphate buffer (pH 7.0) at a scan rate of 20 mV s⁻¹. (Reprinted with permission from ref. [37] with permission of American Chemical Society).

In order to provide mechanistic insights into O–O bond formation for O₂ production, the H/D isotope effect on electrocatalytic water oxidation by p,p-Ru₂(OH)(OH₂) and d-3 were examined in H₂O and D₂O media. A large H/D isotope effect (1.7) on electrocatalytic water oxidation by d-3 was observed relative to the blank experiment (1.1). This result is consistent with the proton transfer-concerted O–O bond formation by the WNA mechanism. On the other hand, the isotopic effect (1.1) on the electrocatalysis by p,p-Ru₂(OH)(OH₂) was comparable with the blank (1.1). The lower isotope effect indicates proton-non-concerted chemical reaction process in the electrocatalytic cycle, and p,p-Ru₂(OH)(OH₂) is likely to produce O₂via the I2M mechanism.

4. Conclusions

Recent progress on the aspects and mechanistic insights of photoisomerization of Ru(II) aquo complexes in our group was reviewed to unveil the photoisomerization reactions and its mechanism comprehensively. The controls of properties and functions of mononuclear Ru(II) aquo complexes by the photoisomerization were exemplified in terms of their water oxidation catalyses. An example of application of the photoisomerization is demonstrated; the employment of the photoisomerization enabled to synthesize the dinuclear Ru(II) complexes that have been difficult to be synthesized by conventional thermochemical processes. The synthesized dinuclear complexes also serve as efficient catalysts for water oxidation. New design and development of variety types of Ru complexes are desired to explore unique reactions, functions and application based on the photoisomerization for future molecular systems such as artificial photosynthetic devices.