Hydrogen peroxide as a hydride donor and reductant under biologically relevant conditions

Abstract

Some ruthenium–hydride complexes react with O₂ to yield H₂O₂, therefore the principle of microscopic reversibility dictates that the reverse reaction is also possible, that H₂O₂ could transfer an H⁻ to a Ru complex. Mechanistic evidence is presented, using the Ru-catalyzed ABTS˙⁻ reduction reaction as a probe, which suggests that a Ru–H intermediate is formed via deinsertion of O₂ from H₂O₂ following coordination to Ru. This demonstration that H₂O₂ can function as an H⁻ donor and reductant under biologically-relevant conditions provides the proof-of-concept that H₂O₂ may function as a reductant in living systems, ranging from metalloenzyme-catalyzed reactions to cellular redox homeostasis, and that H₂O₂ may be viable as an environmentally-friendly reductant and H⁻ source in green catalysis.

---

Introduction

Hydrogen peroxide and its descendant reactive oxygen species (ROS) have historically been viewed in biological systems nearly exclusively as oxidants that damage essential biomolecules,^1–3 but recent reports have shown that H₂O₂ can also perform essential signaling functions at low concentrations.^4,5 Due to the damage caused by high ROS concentrations, significant efforts have been devoted to developing antioxidants that catalytically reduce ROS and other oxidizing radicals.⁶ Catalytic antioxidants require other species to serve as terminal reductants (ascorbate, glutathione, NADH, etc.) and, under certain conditions, depletion of endogenous reductants by a catalytic antioxidant can induce, rather than prevent, oxidative stress.^7–9 A catalytic antioxidant that could harness H₂O₂ or other ROS as terminal reductants, akin to catalase and superoxide dismutase,^10,11 would preclude adverse oxidative damage.

Mechanistic studies on the Ru-catalyzed aerobic oxidation of alcohols have provided evidence that O₂ can insert into a Ru–H bond and be subsequently released as H₂O₂ (Scheme 1A, red arrows).^12,13 The principle of microscopic reversibility¹⁴ therefore dictates that it is mechanistically equivalent for H₂O₂ to react with a Ru complex and be subsequently released as O₂ with concomitant formation of a Ru–H intermediate (Scheme 1A, blue arrows). In this reaction, H₂O₂ is oxidized to O₂, and the 2e⁻ liberated in this oxidation will be transferred to Ru in the form of a hydride (H⁻) ligand. The forward and reverse reactions, 1,2-insertion of O₂ into Ru–H (red arrows) and β-hydride elimination (i.e., 1,2-deinsertion of O₂)¹⁵ from Ru–OOH (blue arrows), proceed through a common transition state. The forward and reverse reactions in Ru–H + O₂ ⇌ Ru–OOH could alternatively proceed via 1,1-insertion and 1,1-deinsertion of O₂, respectively, but Ru–OOH would need to rearrange to a higher-energy species for this to be mechanistically feasible (vide infra).

Scheme 1 (A) 1,2-Insertion of O₂ into Ru–H (red arrows), its microscopic reverse β-hydride elimination (i.e., 1,2-deinsertion of O₂) from Ru–OOH (blue arrows), and their common transition state. Structures of (B) Ru1 and (C) ABTS˙⁻. Net reaction (D) and mechanism (E) for Ru1-catalyzed ABTS˙⁻ reduction with an alcohol.

Ruthenium complexes comprising H⁻ and O₂ ligands both bound to the same metal have been previously observed,^16–20 which suggests that the formation of a Ru–H intermediate and O₂ (Scheme 1A, blue arrows) can, under certain circumstances, be thermodynamically and/or kinetically more favorable than the insertion of O₂ into the Ru–H bond (Scheme 1A, red arrows). Alternatively, if direct observation of a Ru–H intermediate produced by deinsertion of O₂ from Ru–OOH is experimentally infeasible, due to existing at too low of a concentration or for too short of a lifetime, then the presence of Ru–H can be demonstrated inferentially via its chemical reactivity.

We recently reported a Ru complex (Ru1, Scheme 1B) that catalyzed the 1e⁻ reduction of ABTS˙⁻ (Scheme 1C) with biologically-relevant alcohols (ascorbate, glucose, NAD⁺, etc., Scheme 1D) as terminal reductants.^21,22 Importantly, ABTS˙⁻ undergoes 1e⁻ reduction at a potential (E_1/2 = +0.68 V vs. NHE) comparable to the ROS generated during oxidative stress,^23–25 therefore the one-electron redox reactivity of ABTS˙⁻ can thermodynamically approximate the corresponding reactivity of oxidizing species in living systems. Mechanistic studies by us²⁶ provided evidence that ABTS˙⁻ was reduced by a Ru–H intermediate formed via β-hydride elimination from a Ru–alkoxide (Scheme 1E). Because the individual steps leading up to Ru–H formation are well-understood and ABTS˙⁻ concentration can be quantified at μM levels,²⁷ we hypothesized that kinetic analysis of Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂ would reveal if any Ru–H intermediate had formed. Herein we demonstrate that H₂O₂ functions as a terminal reductant for Ru1-catalyzed ABTS˙⁻ reduction in aerobic, aqueous solution. Moreover, we provide the first mechanistic evidence that H₂O₂ can function as an H⁻ donor to generate the Ru–H intermediate that reduces ABTS˙⁻, in a manner consistent with β-hydride elimination (i.e., 1,2-deinsertion of O₂) from Ru–OOH (Scheme 1A, blue arrows).

Results and discussion

Peroxide terminal reductant ability is unique to H₂O₂

By itself, Ru1 cannot reduce ABTS˙⁻ to ABTS²⁻ in phosphate buffered saline (PBS, pH 7.4),²⁸ consistent with the fact that a catalyst cannot be consumed or produced by the net reaction (Fig. 1A(i)). Subsequent addition of H₂O₂ caused a decrease in radical absorbance at 734 nm (Fig. 1A(ii)) accompanied by an increase in absorbance at 340 nm, consistent with the 1 : 1 conversion of ABTS˙⁻ to ABTS²⁻ (Fig. S1†). In the absence of Ru1, the addition of H₂O₂ afforded no change in ABTS˙⁻ concentration, which demonstrated that the reactivity of H₂O₂ as a reductant was dependent on the catalyst being present. The oxidation of H₂O₂ to O₂ (E°′ = −0.28 V at pH 7) by ABTS˙⁻ (E_1/2 = +0.68 V) is thermodynamically favorable,^24,29 therefore the lack of reactivity between H₂O₂ and ABTS˙⁻ in the absence of Ru1 demonstrated that the reduction of ABTS˙⁻ with H₂O₂ is under kinetic control. When H₂O₂ was added to a PBS solution containing Ru1 and ABTS²⁻, no ABTS˙⁻ formation was observed, which indicated that Ru1 does not exhibit peroxidase-like reactivity and does not convert H₂O₂ into other ROS capable of oxidizing ABTS²⁻.

Fig. 1 Plots of relative [ABTS˙⁻] vs. time which show that Ru1 by itself does not reduce ABTS˙⁻ in aqueous solution (A–F, i). In PBS solutions containing Ru1 and ABTS˙⁻, the addition of H₂O₂ caused the radical absorbance to decrease (A, ii), but no ABTS˙⁻ reduction occurred following the addition of t-Bu₂O₂ (B, ii) or t-BuOOH (C, ii). In contrast, the addition of H₂O₂ to a solution of Ru1 and ABTS˙⁻ in pure water afforded no radical reduction (D, ii). Although neither t-Bu₂O₂ (E, ii) nor t-BuOOH (F, ii) enabled Ru1-catalyzed ABTS˙⁻ reduction, the subsequent addition of H₂O₂ (E and F, iii) did produce decreases in radical absorbance. Conditions: [Ru1]₀ = 5 μM, [ABTS˙⁻]₀ = 50 μM, [ABTS²⁻]₀ = 100 μM, [H₂O₂] or [t-Bu₂O₂] or [t-BuOOH] = 100 μM, PBS (pH 7.4), 25 °C.

To test our hypothesis that H₂O₂ functioned as the terminal reductant for Ru1-catalyzed ABTS˙⁻ reduction via β-hydride elimination from a Ru–OOH species (i.e., Scheme 1A, blue arrows), we explored the reactivity of other peroxides with this system. Di-tert-butyl peroxide (t-Bu₂O₂) cannot react with Ru1 to form a Ru–peroxo species and, if our hypothesis were correct, should therefore be incapable of serving as the terminal reductant. Gratifyingly, no ABTS˙⁻ reduction occurred following the addition of t-Bu₂O₂ to Ru1 and ABTS˙⁻ in PBS (Fig. 1B(ii)), which validated this expectation. Conversely, tert-butyl hydroperoxide (t-BuOOH) can form a Ru–peroxo species, i.e., Ru–OOR, but because the distal oxygen carries a tert-butyl group instead of a hydrogen atom, β-hydride elimination cannot occur, which would prevent oxidation of the peroxide and therefore preclude reduction of ABTS˙⁻. This expectation was validated with the observation (Fig. 1C(ii)) that radical absorbance did not decrease when a PBS solution containing Ru1 and ABTS˙⁻ was treated with t-BuOOH, providing evidence that no Ru–H intermediate was generated. This result was also consistent with our previous findings that primary and secondary alcohols (e.g., EtOH, i-PrOH, etc.) could function as terminal reductants for Ru1-catalyzed ABTS˙⁻ reduction, whereas a tertiary alcohol like t-BuOH could not,^22,26 which reflected the ability (or inability) of the Ru–OR species to undergo β-hydride elimination (i.e., Scheme 1E).

Addition of H₂O₂ to Ru1 and ABTS˙⁻ in pure H₂O (instead of PBS) afforded no change in radical absorbance (Fig. 1D(ii)), which was consistent with our prior observations that a proton acceptor must be present for Ru1-catalyzed ABTS˙⁻ reduction to occur and revealed that deprotonation of H₂O₂ was an essential step of the catalytic cycle leading up to the formation of the radical reducing species. Although neither t-Bu₂O₂ nor t-BuOOH afforded any decrease in radical absorbance (Fig. 1E(ii) and F(ii)), the subsequent addition of H₂O₂ to PBS solutions containing Ru1 and either t-Bu₂O₂ or t-BuOOH did result in ABTS˙⁻ reduction (Fig. 1E(iii) and F(iii)). The absence of ABTS˙⁻ reduction following the addition of t-Bu₂O₂ or t-BuOOH alone was therefore not due to catalyst deactivation, but was instead due to the fact that neither t-Bu₂O₂ nor t-BuOOH could function as terminal reductants. Collectively, these findings provided further evidence that, for ABTS˙⁻ reduction to occur, a Ru–hydroperoxo species must first be formed, which in turn must be capable of undergoing β-hydride elimination to generate the Ru–H intermediate necessary for ABTS˙⁻ reduction.

Reductant ability is not derived from H₂O₂ bond homolysis

Notably, the O–O bond dissociation energy (BDE) values for t-Bu₂O₂ (40.3 kcal mol⁻¹) and t-BuOOH (46.1 kcal mol⁻¹) are both significantly lower than the corresponding value of 49.5 kcal mol⁻¹ for H₂O₂ (Fig. 2).^30–32 Therefore, if the ability of H₂O₂ to reduce ABTS˙⁻ derived from homolytic cleavage of its O–O bond at some point in the mechanism, then the addition of both t-Bu₂O₂ and t-BuOOH should have afforded a decrease in ABTS˙⁻ concentration due to their lower activation barriers of O–O bond homolysis. Likewise, the O–H and O–C BDE values for t-BuOOH (84.2 and 73.6 kcal mol⁻¹, respectively) were both lower than the value of 87.8 kcal mol⁻¹ for the O–H BDE in H₂O₂.^33–35 Consequently, if the mechanism of ABTS˙⁻ reduction by H₂O₂ proceeded through an O–H bond homolysis step, then a decrease in radical absorbance should have occurred following the addition of t-BuOOH, given that homolytic cleavage of its O–H and O–C bonds would both have lower activation energies than O–H bond homolysis in H₂O₂.

Fig. 2 Bond-dissociation energies (in kcal mol⁻¹) for H₂O₂, t-Bu₂O₂, and t-BuOOH.

Furthermore, if homolysis of an O–O, O–H, or O–C bond in H₂O₂, t-Bu₂O₂, or t-BuOOH did occur, it would transiently generate one or more radical species of sufficiently strong oxidizing power (e.g., E°′ = +2.32 V for HO˙, +0.89 V for O₂˙⁻ derived from HO₂˙ at pH 7, etc.) that oxidation of ABTS²⁻ to ABTS˙⁻ (E_1/2 = +0.68 V) would occur.^24,29 Similarly, although there are examples in the literature in which a Ru complex reacts with H₂O₂ or t-BuOOH to generate a Ru(IV)–, Ru(V)–, or Ru(VI)–oxo species,^36–40 these high-valent Ru–oxo species undergo 1e⁻ or 2e⁻ reduction at potentials significantly higher than the ABTS˙⁻/ABTS²⁻ redox couple.^41–43 However, no ABTS˙⁻ formation was observed when PBS solutions containing ABTS²⁻ were treated with H₂O₂, t-Bu₂O₂, or t-BuOOH, either in the presence or absence of Ru1.

Collectively, the results from the experiments using H₂O₂, t-Bu₂O₂, and t-BuOOH provided strong evidence that the mechanism for the Ru1-catalyzed reduction of ABTS˙⁻ with H₂O₂ does not involve any strongly oxidizing radicals or high-valent Ru–oxo species, but instead proceeds via heterolytic cleavage of the O–H bonds.

Reduction by H₂O₂ releases O₂ gas

The volume of O₂ gas evolved from these experiments was too small to measure directly because H₂O₂ was consumed from 3.00 mL reaction volumes. For example, reduction of 50 μM ABTS˙⁻ to 50 μM ABTS²⁻ would consume 25 μM H₂O₂ (each H⁻ can reduce 2 ABTS˙⁻) and produce 75 nmol of O₂ (25 μM × 3.00 mL), corresponding to 1.8 μL at 298 K and 1 atm.

When the reaction volume was increased 1000-fold – when 100 μM H₂O₂ was added to a solution of 50 μM ABTS˙⁻ and 5 μM Ru1 in 3.00 L of PBS – 1.8 ± 0.1 mL of O₂ gas (72 ± 2 μmol) were collected, corresponding to a 96 ± 3% theoretical yield (Fig. S2†). UV/vis spectroscopic analysis of 3.0 mL aliquots taken from this reaction before and 30 min after the addition of H₂O₂ (Fig. S3†) revealed that [ABTS˙⁻] had decreased by 44 ± 1 μM (×3.00 L = 132 ± 3 μmol) and [ABTS²⁻] had increased by 42 ± 1 μM (×3.00 L = 126 ± 3 μmol).⁴⁴ Thus, the consumption of 1.0 equiv. of ABTS˙⁻ was accompanied by the formation of 0.95 ± 0.03 equiv. of ABTS²⁻ and the generation of 0.54 ± 0.02 equiv. of O₂ gas.

The evolution of O₂ gas provided additional evidence that no O–O bond cleavage in H₂O₂ was occurring during the Ru1-catalyzed reduction of ABTS˙⁻ with H₂O₂. Furthermore, the release of 1 equiv. of O₂ for the reduction of every 2 equiv. of ABTS˙⁻ demonstrated that each molecule of H₂O₂ functioned as a 2e⁻ reductant, that both O-atoms from H₂O₂ ended up in O₂, and that both H-atoms from H₂O₂ ended up (ultimately) as H⁺.

Ru1 and horseradish peroxidase compete for H₂O₂

Horseradish peroxidase (HRP) catalytically oxidizes ABTS²⁻ to ABTS˙⁻ using H₂O₂ as the terminal oxidant,⁴⁵ which provided us with a convenient method to probe the concentrations of ABTS²⁻ and H₂O₂ indirectly. To a solution of 5 μM Ru1 and 50 μM ABTS˙⁻ in PBS (Fig. 3(i)) was added 20 μM H₂O₂, which produced a gradual decrease in [ABTS˙⁻] over the course of 45 min (Fig. 3(ii)). If our hypothesis were correct, that Ru1-catalyzes the reduction of ABTS˙⁻ to ABTS²⁻ using H₂O₂ as the terminal reductant, then this decrease in [ABTS˙⁻] over time should be accompanied by an increase in [ABTS²⁻] and a decrease in [H₂O₂]. The concentration of 20 μM H₂O₂ was therefore deliberately chosen to be insufficient to achieve quantitative ABTS˙⁻ reduction, such that when the decrease in [ABTS˙⁻] had ceased, the solution would contain ABTS˙⁻, ABTS²⁻, and Ru1, but no H₂O₂. Addition of 10 nM HRP to this solution produced no increase in ABTS˙⁻ absorbance (Fig. 3(iii)), which confirmed that all of the H₂O₂ had been consumed in the previous step. To determine if this lack of HRP-induced ABTS˙⁻ formation was due to enzyme deactivation, a second 20 μM aliquot of H₂O₂ was then added (Fig. 3(iv)). The resulting gradual increase in [ABTS˙⁻] demonstrated that the lack of reactivity in the previous step was due to depletion of the terminal reductant and not enzyme deactivation. Furthermore, the formation of ABTS˙⁻ in Fig. 3(iv) could only occur if there were ABTS²⁻ present at the end of Fig. 3(iii). This, in turn, provided evidence that the decrease in ABTS˙⁻ absorbance observed in Fig. 3(ii) was caused specifically by the one-electron reduction of ABTS˙⁻ to ABTS²⁻.

Fig. 3 Plot of relative [ABTS˙⁻] vs. time following the sequential addition of 5 μM Ru1 (i), 20 μM H₂O₂ (ii), 10 nM HRP (iii), 20 μM H₂O₂ (iv), and then 50 mM EtOH (v). Conditions: [ABTS˙⁻]₀ = 50 μM, PBS (pH 7.4), 25 °C.

The concentration of H₂O₂ added in Fig. 3(iv) was equal to that added in Fig. 3(ii), therefore the HRP had access to a sufficient amount of terminal oxidant to oxidize all of the ABTS²⁻ produced during Fig. 3(ii) and restore the concentration of ABTS˙⁻ to the initial value in Fig. 3(i). However, the [ABTS˙⁻] in Fig. 3(iv) reached a plateau 12 min after the addition of H₂O₂ that was well below this initial value. The fact that the relative [ABTS˙⁻] did not increase back to 100% indicated that some other species was present that was competing with HRP for the H₂O₂. Likewise, the plateau in [ABTS˙⁻] indicated that ABTS˙⁻ formation and reduction had both ceased, which was consistent with the H₂O₂ supply being depleted (i.e., no terminal oxidant available to HRP, no terminal reductant available to Ru1). Moreover, the rate of ABTS˙⁻ formation with Ru1 present (i.e., Fig. 3(iv)) was significantly slower than when no Ru1 was present. Collectively, these results provided strong evidence that HRP-catalyzed ABTS²⁻ oxidation, with H₂O₂ as the terminal oxidant, and Ru1-catalyzed ABTS˙⁻ reduction, with H₂O₂ as the terminal reductant, were both occurring simultaneously and then both ceased when all the H₂O₂ had been consumed. After this plateau was reached, 50 mM EtOH was then added and produced an immediate decrease in radical absorbance that led to quantitative ABTS˙⁻ reduction within 30 min (Fig. 3(v)), which demonstrated that Ru1 was still present and catalytically competent.

β-Hydride elimination from Ru–OOH is proposed mechanism

We propose the mechanism for Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂ is conserved with the previously reported mechanism in which EtOH is the terminal reductant.²⁶ Addition of Ru1 to a solution of ABTS˙⁻ and ABTS²⁻ in PBS will result in rapid exchange⁴⁶ of the Cl ligand with ABTS²⁻, ABTS˙⁻, and H₂O (Scheme 2, orange arrows) to afford [L_nRu-A_red]¹⁻, [L_nRu-A_ox], and [L_nRu-OH₂]¹⁺, respectively. Because ABTS²⁻ inhibits Ru1-catalyzed ABTS˙⁻ reduction by binding to Ru,²⁶ kinetic experiments were performed with an excess of ABTS²⁻ present to ensure reproducible data. Substitution of ABTS²⁻ in [L_nRu-A_red]¹⁻ (step 1) and ABTS˙⁻ in [L_nRu-A_ox] (step 2) by H₂O will also form [L_nRu-OH₂]¹⁺. Exchange of H₂O for H₂O₂ will afford [L_nRu-(H₂O₂)]¹⁺ (step 3), which will be converted into [L_nRu-OOH] upon H⁺ dissociation to buffer (step 4). The OOH ligand will then undergo β-hydride elimination (viaTS5) to release O₂ and generate [L_nRu-H] (step 5). Although β-hydride elimination from a Ru–OOH species is unknown, the reverse reaction, insertion of O₂ into a Ru–H bond, is known^12,13 and proceeds via the same transition state structure (i.e., TS5). Alternatively, 1,1-deinsertion of O₂ would also afford a Ru–H intermediate, but [L_nRu-OOH] would first need to rearrange to a higher-energy species (vide infra). Computational studies of other [RuCl(L₂)(η⁶-cymene)] complexes (L₂ = a bidentate ligand) have shown that decreases in cymene hapticity to accommodate additional ligand binding have activation barriers below 19 kcal mol⁻¹,⁴⁷ which suggests that a similar hapticity decrease in TS5 would be thermally accessible. Once it has formed, [L_nRu-H] will then be oxidized to [L_nRu-H]¹⁺ by ABTS˙⁻, affording ABTS²⁻ (step 6). Dissociation of H⁺ from [L_nRu-H]¹⁺, 1e⁻ oxidation of [L_nRu] by ABTS˙⁻, and subsequent coordination of ABTS²⁻ to [L_nRu]¹⁺ to restart the catalytic cycle (dashed arrow) will not influence the reaction rate or appear in the rate equation because they occur after the rate determining step. Although these transformations cannot be directly observed, literature precedents suggest that they are feasible under these reaction conditions.²⁶

Scheme 2 Proposed mechanism for Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂. Forward (clockwise) and reverse (counter-clockwise) reactions are colored in green and blue, respectively. Each K_n or k_n corresponds to the equilibrium or rate constant, respectively, for the forward reaction in step “n” (i.e., step 1 has equilibrium constant K₁, etc.). TS5 is the structure of the transition state for step 5. The dashed arrow includes multiple transformations that occur after the rate-determining steps (see ref. 26). All complexes shown above are in the Ru²⁺ oxidation state except for [L_nRu-H]⁺, which is Ru³⁺. The spectator ligand set “L_n” comprises the η⁶-cymene and κ²-(C,O)-benzimidazolylidene–carboxylate ligands, but their hapticity/denticity may decrease to accommodate binding of additional ligands.

If the proposed mechanism is valid, then ABTS˙⁻ reduction can only occur if the Ru–H intermediate has formed, and this Ru–hydride intermediate, in turn, can only form if the OOH ligand on Ru has undergone β-hydride elimination. Within these constraints, an observation that [ABTS˙⁻] has decreased thus serves as an indirect indication that a Ru–OOH species has undergone β-hydride elimination and generated a Ru–H intermediate. Although β-hydride elimination from a Ru–OOH species (i.e., step 5) is unknown, the reverse reaction, insertion of O₂ into a Ru–H bond, is known^12,13 and proceeds via the same transition state (i.e., TS5). Furthermore, there are literature examples of Ru complexes that have both hydride and O₂ ligands bound to the same metal center,^16–20 which suggests that the formation of a Ru–H intermediate and O₂ (i.e., step 5) can be more thermodynamically favorable and/or faster than the reverse reaction (insertion of O₂ into the Ru–H bond to form a Ru–OOH species) under the appropriate experimental conditions.

Rate law evidence for proposed mechanism

To test the validity of our proposed mechanism for Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂, we derived the general rate law equation for the catalytic cycle presented in Scheme 2 as a function of the initial rate of ABTS˙⁻ reduction (v₀). If the proposed mechanism is valid, v₀ should be equal to the product of the rate constant for step 6 (k₆) times the concentrations of ABTS˙⁻ and [L_nRu-H] (eqn (1)). Utilizing the pre-equilibrium approximation allowed the initial rate of ABTS˙⁻ reduction (v₀) to be expressed as functions of [ABTS˙⁻]₀, [ABTS²⁻]₀, [H⁺]₀, [H₂O₂]₀, and [Ru1]₀ (eqn (2)–(4); y = v₀; x = concentration of independent variable; a, b, and c = constants; see eqn (S1)–(S8)† for full derivation). If the concentrations of all other species are held constant, the relationship between v₀ and [ABTS˙⁻]₀ will follow eqn (2), the relationship between v₀ and [ABTS²⁻]₀ as well as v₀ and [H⁺]₀ will follow eqn (3), and the relationship between v₀vs. [H₂O₂]₀ will follow eqn (4).

The v₀ values for Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂ increased non-linearly as the initial ABTS˙⁻ concentration increased, with v₀ tapering off at higher values of [ABTS˙⁻]₀ (Fig. 4A), and could be successfully fit using eqn (2). Although substrate binding saturation kinetics would produce a similar curve, this phenomenon could be ruled out because previous studies with EtOH as the terminal reductant instead revealed a linear relationship with [ABTS˙⁻]₀.²⁶ The non-linearity observed with H₂O₂ as the terminal reductant could be attributed to competitive binding between ABTS˙⁻ and H₂O₂ (steps 2 and 3, Scheme 2), due to the significantly lower concentrations of terminal reductant employed in the current study (i.e., 100 μM for H₂O₂vs. 50 mM for EtOH) and its poorer Lewis basicity (pK_a = 11.6 for H₂O₂vs. 15.7 for EtOH).⁴⁸ With a sufficiently high terminal reductant concentration or metal binding ability, the contribution of step 2 to the mechanism becomes negligible and the overall rate equation simplifies to a linear form. Because the v₀vs. [ABTS˙⁻] data could be fit using eqn (2), wherein the [ABTS˙⁻] term occurs in both the numerator and denominator, the mechanism of ABTS˙⁻ reduction with H₂O₂ must involve (i) ABTS˙⁻ dissociation from Ru before [L_nRu-H] can form (consistent with step 2) and in a subsequent process (ii) bimolecular electron-transfer reaction to ABTS˙⁻ from [L_nRu-H] (consistent with step 6).

Fig. 4 Dependence of the initial rate (v₀) of Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂ on (A) [ABTS˙⁻]₀ = 5, 10, 25, 50, 75, or 100 μM; (B) [ABTS²⁻]₀ = 50, 100, 150, or 200 μM; (C) pH = 7.4, 7.9, 8.4 or 8.9; (D) [H₂O₂]₀ = 25, 50, 100, 200 or 400 μM; (E) [Ru1]₀ = 1, 2, 3, 4 or 5 μM; (F) T = 15, 25, 35 or 45 °C. The data points (◇) and error bars were determined from the average and standard deviation values obtained from 4 independent experiments performed on 4 different days, and the grey traces represent the model fits generated by eqn (2) (for ABTS˙⁻), eqn (3) (for ABTS²⁻ and H⁺), and eqn (4) (for H₂O₂). Conditions: [Ru1]₀ = 5 μM, [ABTS˙⁻]₀ = 50 μM, [ABTS²⁻]₀ = 100 μM, [H₂O₂]₀ = 100 μM, PBS (pH 7.4), 25 °C.

Plots of v₀vs. [ABTS²⁻]₀ and v₀vs. [H⁺]₀ (Fig. 4B and C) revealed that v₀ decreased non-linearly with [ABTS²⁻]₀ and [H⁺]₀, respectively, and could each be fit by eqn (3). These results demonstrated that ABTS˙⁻ reduction can only occur after ABTS²⁻ dissociation from Ru (consistent with step 1) and H⁺ dissociation from H₂O₂ (consistent with step 4). The lack of ABTS˙⁻ reduction in pure H₂O (vide supra) provided additional support that H⁺ dissociation to solution is essential for reactivity. The plot of v₀vs. [H₂O₂]₀ (Fig. 4D) revealed a positive correlation and could be fit using eqn (4), whereby the deviation from linearity at higher concentrations indicated that H₂O₂ must bind to Ru at some point prior to ABTS˙⁻ reduction (consistent with step 3). The linear relationship between v₀ and [Ru1]₀ (Fig. 4E) suggested that the observed reactivity was predominantly produced by a mononuclear species, consistent with our previous mechanistic studies.²⁶

The Eyring–Polanyi plot (Fig. 4F) revealed a positive entropy of activation (ΔS^‡ = 25.5 ± 1.9 cal mol⁻¹ K⁻¹), which indicated disorder was increasing during the rate determining step. This result was consistent with the proposed mechanism, in which a single intermediate, [L_nRu-OOH], fragments into two separate molecules, [L_nRu-H] and O₂, via β-hydride elimination (step 5). Moreover, the ΔS^‡ value observed for ABTS˙⁻ reduction by Ru1 with H₂O₂ fell within the range of values observed for Ru1 with other terminal reductants (methanol, ethanol, isopropanol, and ethylene glycol, ΔS^‡ = 11.4–32.8 cal mol⁻¹ K⁻¹) which have previously been shown to generate Ru–H intermediates via β-hydride elimination.²⁶

Kinetic isotope effect evidence for proposed mechanism

Our prior mechanistic studies revealed a solvent kinetic isotope effect (KIE) of 1.74 for Ru1-catalyzed ABTS˙⁻ reduction, reflecting the role of the solvent, H₂O, as H⁺ acceptor (or D⁺ acceptor in the case of D₂O) in step 4.²⁶ When this solvent KIE was factored out, the ratio of observed rate constants (k_obs) for Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂ in protio PBS vs. D₂O₂ in deutero PBS was determined to be 2.10 ± 0.24, which indicated significant O–H/O–D bond breakage was occurring during the rate determining step. Unfortunately, the individual contributions of step 4 (H⁺ dissociation) and step 5 (β-hydride elimination) could not be deconvoluted because D–O–O–H cannot be obtained in pure form and the pertinent Ru intermediates – L_nRu–O(D)OH and L_nRu–O(H)OD – will undergo H/D exchange in protio and deutero PBS on a faster timescale than Ru1-catalyzed ABTS˙⁻ reduction. Therefore, the observed ratio of 2.10 ± 0.24 was treated as the product of the individual O–H/O–D KIE values for step 4 and step 5, and the square root of this ratio was calculated (1.45 ± 0.08) as an averaged approximation of how each step contributed to the overall mechanism (i.e., 1.45 for step 4 and 1.45 for step 5).

Interestingly, this value was nearly identical to the O–H/D KIE of 1.45 measured with NAD⁺.²¹ Although H₂O₂ and NAD⁺ are structurally dissimilar, their O–H pK_a values (11.6 for H₂O₂ and 11.8 for NAD⁺) are nearly identical,^48,49 therefore their O–H bond polarizations will be highly conserved. As a result, H/D isotopic substitution in H₂O₂ should impact the transition state structure leading to O–H bond breakage to a similar extent as in NAD⁺. The fact that H₂O₂ and NAD⁺ afforded nearly identical O–H/O–D KIE values that aligned with their highly similar pK_a values, despite their significant structural differences, provided further support for the proposed mechanism shown in Scheme 2.

Evidence for Ru–H intermediate

The range of suitable experimental conditions under which Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂ could still occur precluded direct detection of a Ru–H intermediate by ¹H NMR or IR spectroscopy, or by ESI-MS. No radical reduction occurs in PBS solutions containing more than 20% CH₃CN, and the maximum concentration attainable for stock solutions of Ru1 in CH₃CN is 1 mM, therefore the highest concentration attainable for Ru1 in the ABTS˙⁻ reduction experiments is 200 μM, which is well below the detection limit of IR spectroscopy.

In addition, the proposed Ru–H intermediate is formed in the rate determining step, and there are 5 other possible Ru-containing species, and both of these factors would cause the concentration of any Ru–H intermediate to be significantly less than 200 μM. Furthermore, ABTS˙⁻ is a paramagnetic species that can broaden or even suppress ¹H NMR peaks via relaxation. It is therefore unsurprising that no Ru–H intermediate could be detected when Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂ was monitored by ¹H NMR or IR spectroscopy.

Catalytic organic transformation reactions which proceed through Ru–H intermediates have been studied by mass spectrometry, however these reactions typically employ millimolar catalyst concentrations.⁵⁰ Nonetheless, we sought to detect the Ru–H intermediate proposed for the Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂ using high-resolution Fourier transform mass spectrometry with electrospray ionization (ESI-MS). Because [L_nRu-H] is neutral, it would need to acquire a charge to be detectable, such as by being converted to [{L_nRu-H} + H]¹⁺ (e.g., m/z = 503.1267) via protonation. Unfortunately, no peaks corresponding to this exact species were observed. One possibility is that the H⁻ ligand of [L_nRu-H] reacts with H⁺ to release H₂ gas and thereby afford [L_nRu]¹⁺ (e.g., m/z = 501.1102), a species that was, in fact, observed in positive mode (Fig. S4†). However, this same species could also be produced by ligand dissociation from other intermediates, such as dissociation of ABTS²⁻ from [L_nRu-A_red]¹⁻, ABTS˙⁻ from [L_nRu-A_ox], H₂O from [L_nRu-OH₂]¹⁺, and H₂O₂ from [L_nRu-(H₂O₂)]¹⁺. Moreover, the ability of ESI-MS to detect the formation of [{L_nRu-H} + H]¹⁺ was severely hampered by the fact that the Ru1-catalyzed ABTS˙⁻ reduction reaction solutions already contained both positively and negatively charged species that would not require ionization, and at much higher concentrations than any un-ionized [L_nRu-H]. The presence of these charged species can cause severe ionization suppression in ESI-MS, significantly reducing the ability of this technique to ionize and detect low-concentration neutral molecules, such as the reaction intermediate [L_nRu-H].

Although the ¹H NMR, IR, and ESI-MS experiments did not lead to direct observation of [L_nRu-H], the results did not exclude its formation and they were not inconsistent with the indirect evidence for the formation of [L_nRu-H] provided by the UV/vis spectroscopic kinetic experiments. By itself, H₂O₂ was incapable of reducing ABTS˙⁻ to ABTS²⁻, which could only occur if Ru1 was also present. In addition, the ABTS˙⁻ reduction rate was linear with Ru1 concentration. Collectively, these results indicated that (1) a Ru-containing species functioned as the catalyst and (2) one or more mononuclear Ru-containing species were intermediates in the catalytic cycle.

No ABTS˙⁻ reduction occurred with Ru1 by itself unless H₂O₂ was also present, which demonstrated that H₂O₂ was the terminal reductant. The inability of t-Bu₂O₂ and t-BuOOH to reduce ABTS˙⁻ or to oxidize ABTS²⁻ in the presence of Ru1 provided evidence that the terminal reductant ability of H₂O₂ did not involve O–O or O–H bond homolysis and that no oxidizing radicals or high-valent Ru–oxo or Ru–hydroxo species were generated during the catalytic cycle. The non-linear relationship between the ABTS˙⁻ reduction rate and [H₂O₂] that gradually approached saturation at higher concentrations was consistent with H₂O₂ coordinating to Ru before the rate determining step. Likewise, the inverse relationship between the ABTS˙⁻ reduction rate and [H⁺] indicated that one H-atom was lost from H₂O₂ as H⁺. Collectively, these results demonstrated that (3) a Ru(H₂O₂) species must be formed before the rate determining step, (4) H₂O₂ only underwent O–H bond breakage and only via heterolysis, and (5) that one O–H bond heterolyzed as O⁻ and H⁺. The most likely product of this combination of processes would be a Ru–OOH species.

Others have demonstrated that H₂O₂ formation is responsible for the aerobic oxidation of alcohols^12,13 by and chemotherapeutic activity⁵¹ of Ru-based catalysts, and H₂O₂ could only be produced by a Ru–H intermediate if it underwent insertion of O₂ into the Ru–H bond, which would yield a Ru–OOH species. The microscopic reverse of this reaction is de-insertion of O₂ from Ru–OOH, in which the H-atom is transferred to Ru as H⁻ and the O–O single bond is converted to a double bond. The observation that 1 equiv. of O₂ gas was released for every 2 equiv. of ABTS˙⁻ reduced to 2 equiv. of ABTS²⁻ provided evidence that deinsertion of O₂ was indeed occurring. The large positive ΔS^‡ value observed for Ru1-catalyzed ABTS˙⁻ reduction with H₂O₂ demonstrated an increase in disorder during the rate determining step, which would be consistent with the fragmentation of one ligand into multiple ligands (regardless of whether these ligands remained bound to or subsequently dissociated from Ru).

It is important to note that de-insertion of O₂ from Ru–OOH could proceed via 1,1-deinsertion of O₂, rather than β-hydride elimination (i.e., 1,2-deinsertion of O₂), and still yield a positive ΔS^‡ value. For 1,1-deinsertion to occur, however, the atom connectivity in [L_nRu-OOH] would need to rearrange to [L_nRu-O(O)H]. The possible structures for [L_nRu-O(O)H] (IVa and IVb) would be expected to be higher in energy than [L_nRu-OOH]: both are formally charge-separated species, and IVb would contain a protonated O₂ ligand (Fig. 5). As a result, any equilibrium between [L_nRu-OOH] and IVa or IVb would be expected to favor [L_nRu-OOH]. Similarly, dissociation of H⁺ from [L_nRu(H₂O₂)] would be expected to occur preferentially at the OH group directly bound to Ru, which would yield [L_nRu-OOH], rather than at the other OH group, which would yield IVa. Although 1,1-deinsertion and β-hydride elimination (i.e., 1,2-deinsertion) would both afford large, positive values for ΔS^‡, we believe that β-hydride elimination is more probable than 1,1-deinsertion of O₂ because it represents a lower-energy pathway.

Fig. 5 β-Hydride elimination (i.e., 1,2-deinsertion of O₂) could proceed from [L_nRu-OOH], whereas 1,1-deinsertion would first require rearrangement to a different form, such as IVa or IVb.

The lack of reactivity with t-BuOOH provided evidence that the reduction of ABTS˙⁻ with H₂O₂ did not involve either heterolytic or homolytic cleavage of the O–O bond, because the activation barriers for these transformations would be lower for t-BuOOH than for H₂O₂ and they would generate intermediates capable of oxidizing ABTS²⁻. Furthermore, the evolution of O₂ gas provided additional evidence that the O–O bond in H₂O₂ is not broken heterolytically or homolytically. With these constraints, the only way an HOO⁻ ligand on Ru could fragment into multiple species and be accompanied by the release of O₂ gas would be via breakage of the O–H bond. Moreover, the only way this O–H bond could break that would be consistent with the fact that H₂O₂ must be oxidized to be able to serve as terminal reductant for Ru1-catalyzed ABTS˙⁻ reduction would be if this O–H bond broke heterolytically to afford O₂. Consequently, the fragmentation of an HOO⁻ ligand to generate O₂ could only occur if H⁻ was also generated. Indeed, for every 2 equiv. of ABTS˙⁻ reduced to 2 equiv. of ABTS²⁻ by H₂O₂, 1 equiv. of O₂ gas was released, indicating that H₂O₂ functions as a 2e⁻ reductant in this reaction through its H-atoms. Because the two O-atoms are lost from H₂O₂ as O₂ and one H-atom is lost as H⁺, the other H-atom must carry the 2e⁻ for reducing 2 equiv. of ABTS˙⁻, which would most likely occur in the form of a hydride (H⁻).

Because Ru complexes comprising both H⁻ and O₂ ligands have been sufficiently stable to be characterizable by single-crystal X-ray diffraction,^16–20 β-hydride elimination from an HOO⁻ ligand can be sufficiently thermodynamically favorable to drive the conversion of Ru–OOH to Ru–H. Moreover, the ΔS^‡ value observed with H₂O₂ fell within the range of values observed with non-tertiary alcohol-based terminal reductants that were also shown to proceed through β-hydride elimination transition states. Collectively, these findings provide evidence that (6) the oxidation of H₂O₂ to O₂, which supplies the electrons necessary for ABTS˙⁻ reduction, must occur via elimination of H⁻ from an HOO⁻ ligand. The most probable destination of any H⁻ eliminated from an HOO⁻ ligand bound to Ru would be that same metal center, the product of which would be a Ru–H intermediate.

Summary and conclusions

Our findings demonstrate that (i) a mononuclear Ru-containing species is the catalyst for ABTS˙⁻ reduction, (ii) H₂O₂ is the terminal reductant, (iii) H₂O₂ is a 2e⁻ reductant, (iv) H₂O₂ coordinates to Ru before the rate determining step, (v) the two O-atoms from H₂O₂ depart as O₂ gas, and (vi) the two H-atoms from H₂O₂ depart as H⁺ and H⁻. Although the experimental constraints of the ABTS˙⁻ reduction reaction were ultimately incompatible with direct observation of any Ru–H intermediate by ¹H NMR, IR, or ESI-MS, the mechanism presented in Scheme 2 (with the caveat that the conversion of Ru–OOH to Ru–H could proceed via either β-hydride elimination or 1,1-deinsertion of O₂) properly accounts for all of the aforementioned findings and provides a general rate law that accurately models all of the UV/vis spectroscopy kinetic data.

This report constitutes, to the best of our knowledge, both (i) the first instance of H₂O₂ functioning as a terminal reductant under biologically-relevant conditions and (ii) the first instance of H₂O₂ functioning as a hydride donor. However, the ability of H₂O₂ to function as both an oxidant and reductant is not unprecedented and, in fact, serves as the basis for H₂O₂ fuel cells.^52,53 Given the impressive advances using H₂O₂ as an oxidant in green catalysis,^54–57 the newfound ability of H₂O₂ to function as an H⁻ donor and reductant (the byproduct of which is O₂) will lead to complementary advances using H₂O₂ as a green H⁻ donor and reductant. Furthermore, establishing the proof-of-principle that H₂O₂ can act as an H⁻ donor in a chemical reaction (i.e., reduction of ABTS˙⁻ to ABTS²⁻) provides the foundation for future discoveries of biological reactions in which H₂O₂ acts as an H⁻ donor in living systems. The ability of Ru1 to reduce oxidizing species using H₂O₂ as the terminal reductant under biologically-relevant conditions provides a strong impetus to investigate the therapeutic efficacy of Ru1 in maintaining cellular redox homeostasis or modulating essential cellular redox processes. The findings of our efforts in these areas will be detailed in a future report.

Conflicts of interest

The authors declare no competing financial interest.

Acknowledgements

This work was supported by the National Science Foundation (DMR-1555224). We thank A. Mangalum for prior work with Ru1 and helpful discussions.

Notes and references

1. J. A. Imlay, Annu. Rev. Biochem., 2008, 77, 755–776.
2. J. A. Imlay, Annu. Rev. Microbiol., 2003, 57, 395–418.
3. R. Kohen and A. Nyska, Toxicol. Pathol., 2002, 30, 620–650.
4. H. J. Forman, M. Maiorino and F. Ursini, Biochemistry, 2010, 49, 835–842.
5. E. A. Veal, A. M. Day and B. A. Morgan, Mol. Cell, 2007, 26, 1–14.
6. S. Miriyala, I. Spasojevic, A. Tovmasyan, D. Salvemini, Z. Vujaskovic, D. St. Clair and I. Batinic-Haberle, Biochim. Biophys. Acta, 2012, 1822, 794–814.
7. A. Tovmasyan, R. S. Sampaio, M.-K. Boss, J. C. Bueno-Janice, B. H. Bader, M. Thomas, J. S. Reboucas, M. Orr, J. D. Chandler, Y.-M. Go, D. P. Jones, T. N. Venkatraman, S. Haberle, N. Kyui, C. D. Lascola, M. W. Dewhirst, I. Spasojevic, L. Benov and I. Batinic-Haberle, Free Radical Biol. Med., 2015, 89, 1231–1247.
8. M. K. Evans, A. Tovmasyan, I. Batinic-Haberle and G. R. Devi, Free Radical Biol. Med., 2014, 68, 302–314.
9. S. D. Amaral and B. P. Espósito, BioMetals, 2008, 21, 425–432.
10. P. Chelikani, I. Fita and P. C. Loewen, Cell. Mol. Life Sci., 2004, 61, 192–208.
11. T. Fukai and M. Ushio-Fukai, Antioxid. Redox Signaling, 2011, 15, 1583–1606.
12. F. Nikaidou, H. Ushiyama, K. Yamaguchi, K. Yamashita and N. Mizuno, J. Phys. Chem. C, 2010, 114, 10873–10880.
13. K. Yamaguchi and N. Mizuno, Angew. Chem., Int. Ed., 2002, 41, 4538–4542.
14. D. G. Blackmond, Angew. Chem., Int. Ed., 2009, 48, 2648–2654.
15. “β-Hydride elimination from Ru–OOH” and “1,2-deinsertion of O₂ from Ru–OOH” refer to the same process and can be used interchangeably in this context, but we primarily use the former in this manuscript to emphasize the reactivity of H₂O₂ as a hydride donor.
16. A. E. W. Ledger, A. Moreno, C. E. Ellul, M. F. Mahon, P. S. Pregosin, M. K. Whittlesey and J. M. J. Williams, Inorg. Chem., 2010, 49, 7244–7256.
17. L. J. L. Häller, E. Mas-Marzá, A. Moreno, J. P. Lowe, S. A. Macgregor, M. F. Mahon, P. S. Pregosin and M. K. Whittlesey, J. Am. Chem. Soc., 2009, 131, 9618–9619.
18. J. Matthes, S. Gründemann, A. Toner, Y. Guari, B. Donnadieu, J. Spandl, S. Sabo-Etienne, E. Clot, H.-H. Limbach and B. Chaudret, Organometallics, 2004, 23, 1424–1433.
19. M. Jiménez-Tenorio, M. C. Puerta and P. Valerga, Inorg. Chem., 1994, 33, 3515–3520.
20. M. Jiménez-Tenorio, M. C. Puerta and P. Valerga, J. Am. Chem. Soc., 1993, 115, 9794–9795.
21. Y. Htet and A. G. Tennyson, J. Am. Chem. Soc., 2016, 138, 15833–15836.
22. Y. Htet and A. G. Tennyson, Chem. Sci., 2016, 7, 4052–4058.
23. Unless specified otherwise, all potentials are reported relative to NHE and have been adjusted to account for the effects of H⁺ concentration at pH 7.
24. S. L. Scott, W.-J. Chen, A. Bakac and J. H. Espenson, J. Phys. Chem., 1993, 97, 6710–6714.
25. U. Jungwirth, C. R. Kowol, B. K. Keppler, C. G. Hartinger, W. Berger and P. Heffeter, Antioxid. Redox Signaling, 2011, 15, 1085–1127.
26. Y. Htet and A. G. Tennyson, Angew. Chem., Int. Ed., 2016, 55, 8556–8560.
27. R. Re, N. Pellegrini, A. Proteggente, A. Pannala, M. Yang and C. Rice-Evans, Free Radical Biol. Med., 1999, 26, 1231–1237.
28. All experiments were performed in PBS at pH 7.4 unless specified otherwise.
29. P. M. Wood, Biochem. J., 1988, 253, 287–289.
30. F. Agapito, B. J. Costa Cabral and J. A. Martinho Simões, J. Mol. Struct.: THEOCHEM, 2005, 729, 223–227.
31. C. D. Wijaya, R. Sumathi and W. H. Green Jr, J. Phys. Chem. A, 2003, 107, 4908–4920.
32. R. D. Bach, J. Phys. Chem. A, 2016, 120, 840–850.
33. S. J. Blanksby, T. M. Ramond, G. E. Davico, M. R. Nimlos, S. Kato, V. M. Bierbaum, W. C. Lineberger, G. B. Ellison and M. Okumura, J. Am. Chem. Soc., 2001, 123, 9585–9596.
34. J. M. Simmie, G. Black, H. J. Curran and J. P. Hinde, J. Phys. Chem. A, 2008, 112, 5010–5016.
35. T. M. Ramond, S. J. Blanksby, S. Kato, V. M. Bierbaum, G. E. Davico, R. L. Schwartz, W. C. Lineberger and G. B. Ellison, J. Phys. Chem. A, 2002, 106, 9641–9647.
36. M.-Z. Wang, C.-Y. Zhou, M.-K. Wong and C.-M. Che, Chem.–Eur. J., 2010, 16, 5723–5735.
37. C. S. Yi, K.-H. Kwon and D. W. Lee, Org. Lett., 2009, 11, 1567–1569.
38. D. Chatterjee, A. Mitra and R. E. Shepherd, Inorg. Chim. Acta, 2004, 357, 980–990.
39. A. S. Goldstein, R. H. Beer and R. S. Drago, J. Am. Chem. Soc., 1994, 116, 2424–2429.
40. C.-M. Che, T.-F. Lai and K.-Y. Wong, Inorg. Chem., 1987, 26, 2289–2299.
41. A. Gerli, J. Reedijk, M. T. Lakin and A. L. Spek, Inorg. Chem., 1995, 34, 1836–1843.
42. C.-M. Che, V. W.-W. Yam and T. C. W. Mak, J. Am. Chem. Soc., 1990, 112, 2284–2291.
43. C.-M. Che, W.-T. Tang, W.-T. Wong and T.-F. Lai, J. Am. Chem. Soc., 1989, 111, 9048–9056.
44. ABTS˙⁻ reduction was not quantitative after 30 min because the temperature of the 3.00 L solution was 19 °C, but the reduction was quantitative after 30 min at 25 °C (see Fig. S1†).
45. L. Pitulice, I. Pastor, E. Vilaseca, S. Madurga, A. Isvoran, M. Cascante and F. Mas, Biocatal. Biotransform., 2013, 2, 1–5.
46. A. F. A. Peacock, M. Melchart, R. J. Deeth, A. Habtemariam, S. Parsons and P. J. Sadler, Chem.–Eur. J., 2007, 13, 2601–2613.
47. S. Dinda, K. L. Sebastian and A. G. Samuelson, Organometallics, 2010, 29, 6209–6218.
48. K. O. Christe, W. W. Wilson and E. C. Curtis, Inorg. Chem., 1979, 18, 2578–2586.
49. S. Sen, U. Pal and N. C. Maiti, J. Phys. Chem. B, 2014, 118, 909–914.
50. X. Wu, J. Liu, D. D. Tommaso, J. A. Iggo, C. R. A. Catlow, J. Bacsa and J. Xiao, Chem.–Eur. J., 2008, 14, 7699–7715.
51. Y. Fu, M. J. Romero, A. Habtemariam, M. E. Snowden, L. Song, G. J. Clarkson, B. Qamar, A. M. Pizarro, P. R. Unwin and P. J. Sadler, Chem. Sci., 2012, 3, 2485–2494.
52. S. Fukuzumi, Joule, 2017, 1, 689–738.
53. S. Fukuzumi and Y. Yamada, ChemElectroChem, 2016, 3, 1978–1989.
54. M. S. Chen and M. C. White, Science, 2007, 318, 783–787.
55. S.-I. Murahashi, N. Komiya and H. Terai, Angew. Chem., Int. Ed., 2005, 44, 6931–6933.
56. R. Noyori, M. Aoki and K. Sato, Chem. Commun., 2003, 1977–1986.
57. B. S. Lane and K. Burgess, Chem. Rev., 2003, 103, 2457–2474.

---

Footnote

† Electronic supplementary information (ESI) available: Detailed experimental procedures, rate law equation derivations, additional spectra, and digital images. See DOI: 10.1039/c8sc05418e

---