Hydrogen Peroxide as a Sustainable Energy Carrier: Electrocatalytic Production of Hydrogen Peroxide and the Fuel Cell

Abstract

This review describes homogeneous and heterogeneous catalytic reduction of dioxygen with metal complexes focusing on the catalytic two-electron reduction of dioxygen to produce hydrogen peroxide. Whether two-electron reduction of dioxygen to produce hydrogen peroxide or four-electron O₂-reduction to produce water occurs depends on the types of metals and ligands that are utilized. Those factors controlling the two processes are discussed in terms of metal-oxygen intermediates involved in the catalysis. Metal complexes acting as catalysts for selective two-electron reduction of oxygen can be utilized as metal complex-modified electrodes in the electrocatalytic reduction to produce hydrogen peroxide. Hydrogen peroxide thus produced can be used as a fuel in a hydrogen peroxide fuel cell. A hydrogen peroxide fuel cell can be operated with a one-compartment structure without a membrane, which is certainly more promising for the development of low-cost fuel cells as compared with two compartment hydrogen fuel cells that require membranes. Hydrogen peroxide is regarded as an environmentally benign energy carrier because it can be produced by the electrocatalytic two-electron reduction of O₂, which is abundant in air, using solar cells; the hydrogen peroxide thus produced could then be readily stored and then used as needed to generate electricity through the use of hydrogen peroxide fuel cells.

1. Introduction

The rapid consumption of fossil fuel is expected to cause unacceptable environmental problems such as the greenhouse effect by CO₂ emission, which may lead to disastrous climatic consequences in the near future [1]. Even if climate change, such as that due to global warming, turns out to be a less than expected important problem, we are certainly on the verge of running out of fossil fuels by the end of 21^st century, because the consumption rate of fossil fuel is expected to increase further by worldwide rapid population and economic growth, particularly in the developing countries [2]. Thus, renewable and clean energy resources are urgently required in order to solve global energy and environmental issues. Among renewable energy resources, solar energy is by far the largest exploitable resource [3–8]. Of course, solar energy has been utilized for ages in photosynthesis, leading to accumulated fossil fuel which we have been using so rapidly. It is therefore quite important for us to obtain sustainable solar fuels such as hydrogen or others [3–10].

Hydrogen is a clean energy source for the future and it can be used to reduce the dependence on fossil fuels and the emissions of greenhouse gases in the long-term [11–14]. The important advantage of hydrogen is that carbon dioxide is not produced when hydrogen is burned to produce only water. Hydrogen should be ideally produced by splitting water using solar energy. However, the storage of hydrogen has been a difficult issue, because hydrogen is a gas having a low volumetric energy density. Tank systems have been employed, either for gaseous pressurized hydrogen or liquid hydrogen. However, high-pressure equipment and a large demand for energy for cryogenic purposes are involved. Other approaches, such as in the use of metal hydrides, carbon nanotubes, and metal–organic frameworks can store or liberate only low amounts of hydrogen and unfavorable high temperatures are required to release the stored hydrogen [15–19]. Thus, none of the existing processes for storage and carriage of hydrogen are environmentally benign.

On the other hand, hydrogen peroxide has merited significant attention, because H₂O₂ can oxidize various chemicals selectively to produce no waste chemicals but water [20–22]. Hydrogen peroxide can be an ideal energy carrier alternative to oil or hydrogen, because it can be used in a fuel cell leading to the generation of electricity [23]. Thus, a combination of hydrogen peroxide production by the electrocatalytic reduction of dioxygen in air with electrical power generated by a photovoltaic solar cell and power generation with a hydrogen peroxide fuel cell provides a sustainable solar fuel [24]. Currently H₂O₂ is mainly produced by the anthraquinone process, in which the hydroquinone in an organic solvent is oxidized by molecular oxygen to produce H₂O₂ and quinone. The quinone formed can then be reduced by hydrogen using Ni or Pd catalysts. Thus, H₂O₂ is produced by the reduction of oxygen with hydrogen. In recent years, more than 3.5 million metric tons of H₂O₂ are produced all over the world annually in recent years [25]. In this review, first we describe homogeneous vs. heterogeneous catalytic reduction of dioxygen with a variety of metal complexes and then we introduce recent development in the electrocatalytic production of H₂O₂ and hydrogen peroxide fuel cells.

2. Catalytic reduction of dioxygen with metal complexes

2.1. Cobalt porphyrins

No reduction of O₂ occurred by ferrocene derivatives such as 1,1′-dimethylferrocene [Fe(C₅H₄Me)₂] in acetonitrile (MeCN) at 298 K [26], because electron transfer from Fe(C₅H₄Me)₂ (E_ox = 0.26 V vs. SCE) [27] to O₂ (E_red = −0.86 V vs. SCE) [28] is highly endergonic. In the presence of HClO₄ in MeCN solvent, however, O₂ is slowly reduced by Fe(C₅H₄Me)₂ to produce hydrogen peroxide (H₂O₂) via proton-coupled electron transfer from Fe(C₅H₄Me)₂ to O₂ [27]. Mononuclear cobalt complexes with macrobicyclic hexamine cage ligands have been reported to act as oxygen reduction catalysts for H₂O₂ production [29,30]. The addition of cobalt porphyrin monomers or dimers and HClO₄ to air-saturated MeCN solutions containing ferrocene derivatives results in significantly accelerated O₂-reduction. (Scheme 1) [31,32]. The stoichiometry of the oxidation of ferrocene derivatives by O₂ in the presence of HClO₄ (two- vs four-electron reduction of O₂) was determined by the concentration of ferrocenium cations compared with the concentration of O₂, when the concentration of O₂ made to be much smaller [32]. When a monomer cobalt porphyrin such as 2,3,7,8,12,13,17,18-octaethyl-porphinato cobalt(II) [Co(OEP)] was employed as a catalyst, two equiv. of the ferrocenim cation ([Fe(C₅H₄Me)₂]⁺) were produced from O₂, using Fe(C₅H₄Me)₂ in the presence of excess HClO₄ in benzonitrile (PhCN) at 298 K as shown in Fig. 1 [32]. Thus, only two-electron reduction of O₂ occurs and no further reduction occurs to produce more than two equiv. of Fe(C₅H₄Me)₂⁺ [Eq (1)].

Scheme 1.

Fig. 1.

Time profiles of formation of Fe(C₅H₄Me)₂⁺ monitored at 650 nm in electron transfer oxidation of Fe(C₅H₄Me)₂ (1.0 × 10⁻¹ mol L⁻¹) by O₂ (1.7 × 10⁻³ mol L⁻¹), catalyzed by Co₂(DPX) (2.0 × 10⁻⁵ mol L⁻¹), Co₂(DPA) (2.0 × 10⁻⁵ mol L⁻¹), Co₂(DPB) (2.0 × 10⁻⁵ mol L⁻¹), Co₂(DPD) (2.0 × 10⁻⁵ mol L⁻¹), and Co(OEP) (3.0 × 10⁻⁵ mol L⁻¹) in the presence of HClO₄ (2.0 × 10⁻² mol L⁻¹) in PhCN at 298 K.

$$2\text{Fe}{({\text{C}}_5{\text{H}}_4\text{Me})}_2+{\text{O}}_2+2{\text{H}}^{+}\underset{\text{Co}(\text{OEP})}{\to }2\text{Fe}{{({\text{C}}_5{\text{H}}_4\text{Me})}_2}^{+}+{\text{H}}_2{\text{O}}_2$$ (1)

Confirmation that a stoichiometric amount of H₂O₂ was formed was carried out using iodometric measurements [31].

In contrast to the monomeric cobalt porphyrin, when a cofacial dicobalt porphyrin (Co₂(DPX) in right column in Scheme 1) was employed as a catalyst, four equiv. of [Fe(C₅H₄Me)₂]⁺ was produced in the catalytic reduction of O₂ by Fe(C₅H₄Me)₂ in the presence of HClO₄ in PhCN at 298 K (Fig. 1) [Eq. (2)] [32]. Thus, the four-electron reduction of O₂ occurs here efficiently. It was separately confirmed that no H₂O₂ was formed in during this process [32].

$$4\text{Fe}{({\text{C}}_5{\text{H}}_4\text{Me})}_2+{\text{O}}_2+4{\text{H}}^{+}\underset{{\text{Co}}_2(\text{DPX})}{\to }4\text{Fe}{{({\text{C}}_5{\text{H}}_4\text{Me})}_2}^{+}+2{\text{H}}_2\text{O}$$ (2)

The other cofacial dicobalt porphyrins [Co₂(DPA), Co₂(DPB) and Co₂(DPD) in the left column of Scheme 1] also catalyze the reduction of O₂ by Fe(C₅H₄Me)₂, but the amount of Fe(C₅H₄Me)₂⁺ formed is less than four equiv. based on the amount of O₂ (Fig. 1) [32]. This indicates that the clean four-electron reduction of O₂ by Fe(C₅H₄Me)₂ occurs only in the case of Co₂(DPX) and other dicobalt porphyrins as catalysts lead to a mixture of processes, i.e., involving both two- and four-electron stoichiometries.

The cyclic voltammograms of Co(OEP) and [Co₂(DPX)] are compared in Fig. 2, where the one-electron reduction potential of Co(OEP) is determined to be 0.31 V corresponding to the Co(II)/Co(III) couple, whereas the Co(II)/Co(III) couple for [Co₂(DPX)] is split into two one-electron redox waves at 0.53, 0.39 V (vs. SCE) [32]. This indicates that two Co ions are interacting with each other in [Co₂(DPX)].

Fig. 2.

Cyclic voltammograms of (a) Co₂(DPX) and (b) Co(OEP) (1.0 × 10⁻³ mol L⁻¹) in PhCN containing 0.1 mol L⁻¹ TBAP; scan rate 100 mV s⁻¹.

The catalytic cycle of the two-electron reduction of O₂ by Fe(C₅HMe)₂ in the presence an acid in PhCN is shown in Scheme 2 [32]. The initial electron transfer from Fe(C₅H₄Me)₂ to Co(III)OEP⁺ is thermodynamically feasible judging from the one-electron oxidation potentials known for the compounds in question: Fe(C₅H₄Me)₂ (E_ox = 0.26 V vs. SCE) and Co(OEP)⁺ (E_red = 0.31 V vs. SCE in Fig. 2). Indeed, electron transfer from Fe(C₅HMe)₂ to Co(OEP)⁺ occurs and this is followed by the subsequent fast electron transfer from Co(II)OEP to O₂ in the presence of an acid to produce the hydroperoxyl species Co(III)(OEP)O₂H⁺, which is further oxidized by Fe(C₅H₅)₂ in the presence of an acid to produce H₂O₂, accompanied by regeneration of Co(III)OEP⁺. The initial electron transfer is the rate-determining step in the catalytic cycle, when the catalytic rate is given by Eq. (3) [32]. In such a case the catalytic rate

Scheme 2.

$$\text{d}[\text{Fe}{{({\text{C}}_5{\text{H}}_4\text{Me})}_2}^{+}]/\text{d}t=2k_{\text{et}}[\text{Fe}{({\text{C}}_5{\text{H}}_4\text{Me})}_2][\text{Co}{(\text{OEP})}^{+}]\text{in}\text{the}\text{presence}\text{of}{\text{O}}_2$$ (3)

does not depend on the concentration of O₂ or an acid. In addition, the observed second-order rate constant is twice that of the rate constant (k_et) for the initial electron transfer from Fe(C₅HMe)₂ to Co(III)OEP⁺ (k_obs = 2k_et) [32].

The same catalytic scheme can be applied for the selective two-electron reduction of O₂ by ferrocene derivatives with other monomeric metalloporphyrin complexes: CoTPP⁺, FeTPP⁺ and MnTPP⁺ (TPP²⁻ = tetraphenylporphyrin dianion) [31]. The rate constants of the rate-determining electron transfer from ferrocene derivatives to CoTPP⁺, FeTPP⁺ and MnTPP⁺ can be evaluated using the Marcus theory of outer-sphere electron transfer [33]. The Marcus relation provides that the rate constant for electron transfer from an electron donor (1) to an electron acceptor (2), k₁₂, is given by Eq. (4), where k₁₁ and k₂₂ are the rate constants for self-exchange or each component, 1 and 2, K₁₂ is the electron-transfer equilibrium constant, which is obtained from the one-electron oxidation potential of 1 and the one-electron reduction potential of 2. The parameter f in Eq. (4) is given by Eq. (5), where Z is the frequency factor (1 × 10¹¹ mol⁻¹ L s⁻¹) [33]. The k₁₁ value of ferrocene is reported to be 5.3 × 10⁶ mol⁻¹ L s⁻¹ [34], and the k₂₂ values of Co, Fe and Mn porphyrins are reported to be 20 [35], 1 × 10⁹ [36], 3.2 × 10³ mol⁻¹ L s⁻¹ [37], respectively. The results are that based on Eqs. (4) and (5) and using the k₁₁ and k₂₂ values, the rate constants for electron transfer from ferrocene derivatives to CoTPP⁺, FeTPP⁺ and MnTPP⁺ agree well with the observed rate constants [31]. Such agreement strongly indicates electron transfer from ferrocene derivatives to metalloporphyrins occurs via an outer-sphere pathway.

$$k_{12}={(k_{11}{k}_{22}{K}_{12}f)}^{1/2}$$ (4)

$$logf={(\text{log}{K}_{12})}^2/[4\text{log}({\text{k}}_{11}{\text{k}}_{22}/Z^2)]$$ (5)

The reason why only Co₂(DPX) can act as a catalyst for the selective four-electron reduction of O₂ (Fig. 2) can be understood by comparing the distances between porphyrin moieties in the cofacial porphyrins complexes, based on data from reported crystal structures of Co₂(DPB) [38], Co₂(DPA) [39], Co₂(DPX) [40,41], and Co₂(DPD)(2MeOH) [40,41] (Fig. 3). The metal-metal separations in Co₂(DPA) (0.453 nm) and Co₂(DPX) (0.458 nm) are virtually the same. However, the xanthene spacer of Co₂(DPX) is more flexible than the anthracene spacer of Co₂(DPA) and thereby more suitable for the strong binding between two cobalt nuclei and O₂. The metal-metal separation in Co₂(DPB) (0.373 nm) may be too short, whereas the separation in Co₂(DPD)(2MeOH) (0.862 nm) is too long to bind O₂ between two cobalt nuclei. Thus, the interaction of two cobalt nuclei with an active form of oxygen seems essential for the four-electron reduction of O₂. It is interesting to note that the Co-Co distance of Co₂(DPX) (0.458 nm) is nearly the same as the Co-Co distance of a dicobalt(III) μ₂-η¹:η¹-peroxo complex with the tetrapodal pentaamine ligand 2,6-bis(1′,3′-diamino-2′-methylprop-2′yl)pyridine (0.450 nm), which was structurally characterized by X-ray crystallographic analysis [42]. The Co-Co distances in dicobalt bis-μ-oxo complexes, {[Me₂NN]Co}₂(μ-O)₂ (Me₂NN = β-diketiminato) [43] and {[Tp^Me3]Co}₂(μ-O)₂ (Tp^Me3 = hydrotris(3,5-dimethyl-4-methylpyrazolyl)borate), were reported to be 0.3067 nm and 0.2724 nm, respectively [44]. The Co-Co distance of a biscobalt peroxo complex, [Co³⁺(μ,η¹:η²-O₂)(oxapyme)Co³⁺]²⁺ (oxapyme(H)₂ = 2-(bis-pyridin-2-ylmethyl-amino)-N-[2-(5-{2-[2-(methyl-pyridin-2-ylmethyl-amino)]-phenyl}-[1,3,4]-oxadiazol-2-yl)-phenyl]-acetamide), in which one of the oxygen atoms bridges the two metals and is sideways bonded to one of the metals, was reported to be 0.3339 nm [45]. Thus, the biscobalt(III) peroxo complex responsible for the catalytic four-electron reduction of O₂ may have the μ₂-η¹:η¹ bonding mode rather than bis-μ-oxo or μ,η¹:η² bonding modes.

Fig. 3.

Selected distance (nm) in Co₂(DPB) [38], Co₂(DPA) [39], Co₂(DPX) [40,41], and Co₂(DPD)(2MeOH) [40,41].

The proposed mechanism of four-electron reduction of O₂ by ferrocene derivatives is summarized as shown in Scheme 3. The initial two-electron reduction of the Co(III)₂ complex by ferrocene derivatives gives the Co(II)₂ complex, which reacts with O₂ to produce the μ-peroxo Co(III)-O₂-Co(III) complex via the intermediacy of a μ-superoxo complex (not shown). The μ-superoxo species in these cofacial dicobalt porphyrins could be separately produced by the reactions of cofacial dicobalt(II) porphyrins with O₂ in the presence of a bulky base (1-tert-butyl-5-phenylimidazole) and the subsequent one-electron oxidation of the resulting peroxo species by iodine [32]. The EPR spectra of the μ-superoxo species exhibit superhyperfine structure due to two equiv. cobalt nuclei (Fig. 4) [32]. The heterolytic O-O-bond cleavage of the Co(III)-O₂-Co(III) complex affords the high valent Co(IV)-oxo species which is reduced by ferrocene derivatives in the presence of protons to yield H₂O (Scheme 3). Alternatively the homolytic O-O bond cleavage affords two Co(III)-oxyl species which are reduced by ferrocene derivatives to H₂O in the presence of an acid. In each case, the O-O bond cleavage of the Co(III)-O₂-Co(III) complex leads to the four-electron reduction of O₂ (Scheme 3) whereas if protonation first takes place then overall two-electron reduction of O₂ to produce H₂O₂ would occur. The stronger the binding between two cobalt nuclei and oxygen in the Co(III)-O₂-Co(III) complex, the weaker is the O-O bond, and the faster is the rate of O-O bond cleavage. This seems to be the case for Co₂(DPX), which has the largest superhyperfine coupling constant of the μ-superoxo species (Fig. 4), indicating greater interaction of the superoxide moiety spin with the cobalt(III) centers, and this acts as the most efficient catalyst for the selective four-electron reduction of O₂ by ferrocene derivatives (Fig. 1).

Scheme 3.

Fig. 4.

EPR spectra of the μ-superoxo complex (~10⁻³ mol L⁻¹) produced by adding iodine (~10⁻³ mol L⁻¹) to an air-saturated PhCN solution and ESR simulation of (a) Co₂(DPB), (b) Co₂(DPA) and (c) Co₂(DPX) in the presence of 1-tert-butyl-5-phenylimidazole (5 × 10⁻³ mol L⁻¹) at 298 K [32].

Detailed kinetic investigations and analyses on the rate of formation of Fe(C₅H₅)₂⁺ in the Co₂(DPX)-catalyzed electron transfer oxidation of Fe(C₅H₅)₂ by O₂ in the presence of HClO₄ revealed that the rate-determining step (Scheme 3) is a proton-coupled electron transfer from Co(III)Co(II)(DPX)⁺ to O₂ following initial electron transfer from Fe(C₅H₅)₂ to Co(III)₂(DPX)²⁺ [32]. In this case the rate of formation of Fe(C₅H₅)⁺ is proportional to the concentrations of Co(III)₂(DPX)²⁺, O₂ and HClO₄. When Fe(C₅H₅)₂ is replaced by a much stronger reductant, that is Fe(C₅Me₅)₂ (Fc*) however, the kinetics of formation of Fc*⁺ change drastically from first-order kinetics in the case of Fe(C₅H₅)₂ to zero-order kinetics, because the rate-determining step in the is changed to an O-O bond cleavage step in the Co(III)-O₂-Co(III) complex [32]. In this situation, the rate remains constant irrespective of change in concentrations of O₂ and HClO₄. The O-O bond cleavage rate has been determined as 320 s⁻¹ [32].

The electrocatalytic reduction of O₂ was examined using Co₂(DPX) and Co₂(DPD), which were adsorbed onto an electrode surface by means of a dip-coating procedure [46]. Rotating Pt ring-disk voltammograms for reduction of O₂ at pyrolytic graphite disks coated with Co₂(DPX) and Co₂(DPD) revealed the catalytic four-electron reduction of O₂ with 72% and 80% selectivity, respectively [46]. The partial formation of H₂O₂ was clearly detected by the rotating ring current [46]. Thus, the selectivity for the electrocatalytic four-electron reduction of O₂ is lower than the selectivity in the homogeneous system (Fig. 1). Such differences in selectivity for homogeneous versus heterogeneous catalysis of O₂-reducing systems will be discussed further, in the next section.

2.2. Biscobalt porphyrin-corrole complexes

The same selectivity with regard to two-electron vs. four-electron reduction of O₂ by Fe(C₅H₄Me)₂ depending on the type of linkage (Y) of biscobalt complexes (Fig. 1) was observed for the catalytic reduction of O₂ with cofacial biscobalt porphyrin-corrole complexes (Chart 1) in the presence of HClO₄ in PhCN [47]. When 1, 2, 4 or 5 is used as a catalyst, two-electron reduction of O₂ by Fe(C₅H₄Me)₂ occurred efficiently in the presence of HClO₄ in PhCN, whereas the four-electron reduction of O₂ occurred when 3 or 6 was used as a catalyst as shown in Fig. 5 [47]. Thus, in this case as well, a suitable metal-metal separation with Y = 9,9-dimethylxanthene (X) is required to produce the μ-peroxo Co(III)-O₂^{2 −}-Co(III) complex that is the key intermediate for the catalytic four-electron reduction of O₂.

Chart 1.

Structures of biscobalt porphyrin-corrole complexes.

Fig. 5.

Visible absorption spectra changes in the catalytic reduction of O₂ (1.7 × 10⁻³ mol L⁻¹) by Fe(C₅H₄Me)₂ (2.0 × 10⁻² mol L⁻¹) in the presence of HClO₄ (2.0 × 10⁻² mol L⁻¹) and 1 – 6 (2.0 × 10⁻⁵ mol L⁻¹) in PhCN at 298 K; 1 (red solid line), 2 (green solid line), 3 (blue solid line), 4 (red broken line), 5 (green broken line) and 6 (blue broken line). Black: in the absence of cobalt complex.

The catalysts 1–6 were also adsorbed onto a graphite disk by transferring aliquots of a solution in CHCl₃ directly to the electrode surface followed by evaporation of the solvent as the case of Co₂(DPX) and Co₂(DPD) (vide supra) [46]. The biscobalt porphyrin–corrole complex-modified electrodes were used to examine the electrocatalytic properties in the presence of 1 mol L⁻¹ HClO₄ in PhCN [47]. The catalytic activity was determined by cyclic voltammetry as well as by rotating disk electrode voltammetry and the results are summarized in Table 1 [47]. All six complexes catalyze the electrocatalytic reduction of O₂ at potentials close to their E_1/2 values (Table 1). The average E_1/2 value for the electroreduction of O₂ at a rotating disk electrode coated with 1–3 is 0.32 V.

Table 1.

Electroreduction of dioxygen by adsorbed biscobalt porphyrin–corrole dyads in air-saturated 1.0 mol L⁻¹ HClO₄

Bridge (Y)	(PMes2CY)Co2 (1–3)			(PCY)Co2 (4–6)
	Epa	E1/2b	nc	Epa	E1/2b	nc
O	0.25	0.32	2.4	0.34	0.41	3.547
Ox	0.27	0.33	2.5	0.35	0.40	3.1
X	0.25	0.32	2.5	0.38	0.45	3.747

^aPeak potential of the dioxygen reduction wave (V vs. SCE).

^bHalf-wave potential (V vs. SCE) for dioxygen reduction at rotating disk electrode (ω = 100 rpm).

^cElectrons consumed in the reduction of O₂ as estimated from the slope of the Koutecky–Levich plots.

The number of electrons transferred during oxygen reduction was calculated from the magnitude of the steady-state limiting current values, which were taken at a fixed potential on the catalytic wave plateaus of the different current-voltage curves, as shown in Fig. 6a [47]. If mass transport alone controls the reduction of O₂ at the 2-modified electrode, then the relationship between the limiting current and rotation rate obeys the Levich equation [Eq. (6)] [48,49]:

$$I_{\text{Lev}}=0.62{\mathit{nFAv}}^{-1/6}{D}^{2/3}[{\text{O}}_2]{\omega }^{1/2}$$ (6)

where n is the number of electrons transferred, F is the Faraday constant, A is the electrode area (cm²), v is the kinematic viscosity of the solution (cm² s⁻¹), D is the O₂ diffusion constant (cm² s⁻¹), [O₂] is bulk concentration of O₂ (mol L⁻¹), and ω is the rate of rotation (rad s⁻¹). According to Eq. (6), the plot of the limiting current density I/A vs. ω_1/2 gives a straight line intersecting the origin as shown in Fig. 6b. The deviation from the initial linearity suggests that the catalytic reaction is limited by kinetics, not by mass transport alone. In that case, the number of electrons transferred can be determined by using the Koutecky-Levich equation [Eq. (7)] [50,51]:

$${(I/A)}^{-1}={(\mathit{nFk}[{\text{O}}_2])}^{-1}+{(0.62{\mathit{nFv}}^{-1/6}{D}^{2/3}[{\text{O}}_2]{\omega }^{1/2})}^{-1}$$ (7)

where k is the second-order rate constant of electron transfer (mol⁻¹ L s⁻¹), which limits the plateau current. The number of electrons transferred in the O₂ electroreduction process involving complexes 1 – 3 ranges from 2.4 to 2.5 (Table 1 and Fig. 6c) [47]. This indicates that the electrocatalytic reduction of O₂ by the biscobalt porphyrin–corrole complexes results in formation of H₂O₂ mainly via the two-electron reduction of O₂ in the presence of HClO₄. These values are different from the biscobalt porphyrin–corrole complexes 4 – 6 where n = 3.1 to 3.7 (Table 1). In the case of 6, O₂ was mainly reduced to H₂O via a four-electron and four-proton process [47].

Fig. 6.

Electrocatalytic reduction of O₂ in 1 mol L⁻¹ HClO₄ at a rotating graphite disk electrode coated with (PMes₂CO)Co₂ 1. (a) Values of the rotation rates of the electrode (w) are indicated on each curve. The disk potential was scanned at 5 mV s⁻¹. (b) Levich plots of the plateau currents of (a) vs. (rotation rate)^1/2. The dashed line refers to the theoretical curve expected for the diffusion-convection limited reduction of O₂ by 2e⁻. (c) Koutecky-Levich plots of the reciprocal plateau currents vs. (rotation rates)^−1/2. Supporting electrolyte: 1 mol L⁻¹ HClO₄ saturated with air.

The number of electrons transferred in the O₂-electroreduction at the electrode process involving complexes 4 – 6 (Table 1 and Fig. 6) is different from that observed in the homogeneous phase in Fig. 4 except for the case of complex 6 (n = 3.7). This difference (Table 1), and similar variations observed for the biscobalt porphyrin complexes in homogeneous vs. heterogeneous solution (Fig. 1) may result from different coordination geometries of biscobalt complexes which may occur of their interactions with the surface material when they are adsorbed onto the graphite electrode. As already described, the selectivity for the four-electron reduction of O₂ is quite sensitive to the Co-Co distance of biscobalt porphyrins (Fig. 3), thus an electrode surface induced small change in the nature of the proximity of each porphyrin moiety within a synthetic dyad, i.e., the biscobalt complexes, may result in significant changes in terms of the selectivity toward differing reduction processes.

When a monomeric cobalt corrole ([10-pentafluorophenyl-5,15-dimesityl-corrole]cobalt complex, Co(F₅PhMes₂Cor), was employed as an electrocatalyst for the reduction of O₂, the slope of the Koutecky-Levich plot shows that the catalytic electroreduction of O₂ is a pure two-electron process to produce H₂O₂ [52]. The catalytic process employing Co(F₅PhMes₂Cor) was confirmed for the homogeneous phase using Fe(C₅H₄Me)₂ as a reductant in PhCN solvent [52]. Electron transfer from Fe(C₅H₄Me)₂ (E_ox = 0.26 V vs. SCE) to [Co(F₅PhMes₂Cor)]⁺ (E_red = 0.38 V) [53] occurs efficiently to produce [Fe(C₅H₄Me)₂]⁺ and Co(F₅PhMes₂Cor) [52]. The cobalt(III) corrole complex, Co(F₅PhMes₂Cor), can reduce O₂ in the presence of HClO₄. The site of electron transfer has been confirmed to be the corrole ligand, based on the finding by EPR spectroscopy of an observed g-value of 2.0032 for the singly oxidized cobalt corrole, that separately obtained by the chemical oxidation of Co(F₅PhMes₂Cor) with one equivalent of [Fe(bpy)₃]³⁺ (bpy = 2,2′-bipyridine). This signal is characteristic of an organic radical; it is quite different from the large g-value (2.037) observed for cobalt(IV) porphyrin complexes. In contrast to the case of cobalt porphyrins (Scheme 2), the cobalt corrole complex acts as an effective catalyst in the two-electron reduction of O₂ with HClO₄ via the redox couple between the cobalt(III) corrole and the cobalt(III) corrole radical cation (Scheme 4).

Scheme 4.

2.3. Cytochrome c oxidase models

In the final step in the biological respiratory chain, the four-electron and four-proton reduction of O₂ to H₂O is efficiently catalyzed by the heme/copper (heme a₃/Cu_B) heterodinuclear center in cytochrome c oxidases (CcO) (Fig. 7a) [54–56]. Biomimetic chemical modeling of the CcO active site has extensively been studied to provide not only the mechanistic insights into the four-electron and four-proton reduction of O₂ but also as a blueprint for bioinspired fuel cells [56–62]. A number of heme a₃/Cu_B synthetic analogues have been developed to mimic the coordination environment of the heme a₃/Cu_B bimetallic center in CcO [56–61]. The electrocatalytic function of heme a₃/Cu_B synthetic analogues has also been examined by using electrodes modified with the synthetic models to perform the catalytic four-electron and four-proton reduction of O₂ [59–63]. However, the structure of synthetic models may be changed when they are adsorbed on an electrode surface (vide supra). In addition, the solid supported state employed for such studies has precluded any spectroscopic monitoring or intermediates detection. Thus, the catalytic reduction of O₂ to water was examined using ferrocene derivatives as one-electron reductants and a heme/Cu functional model of CcO (⁶LFeCu, Fig. 7b) and its Cu-free version (⁶LFe, Fig. 7c) as catalysts in homogeneous solutions. [64]. The detailed kinetic analysis together with spectroscopic detection of reactive intermediates has provided new mechanistic insights into the O–O reductive cleavage process as described here [64].

Fig. 7.

(a) X-ray structures of the fully reduced bimetallic heme a₃/Cu_B center in CcO from bovine heart (Fe^II···Cu^I = 0.519 nm), (b) heme/Cu synthetic model for CcO (⁶LFeCu), (c) Cu-free version of synthetic model for CcO (⁶LFe).

The catalytic mechanism for the four-electron and four-proton reduction of O₂ by decamethylferrocene (Fc*) with ⁶LFeCu and ⁶LFe in acetone is summarized in Scheme 5a and Scheme 5b, respectively. In the presence of acid, [⁶LFe^III-O-Cu^II]⁺ is converted to [⁶LFe^IIICu^II]³⁺ by releasing water and the catalytic cycle starts via a fast reduction of the heme and then the Cu to generate the reduced complex [⁶LFe^IICu^I]⁺. Then O₂ binds to [⁶LFe^IICu^I]⁺, and this is followed by rapid protonation affording the Fe-hydroperoxo complex {⁶LFe^III-OOH Cu^II}²⁺. Reductive O-O bond cleavage is followed by further rapid reduction to produce H₂O accompanied by regeneration of [⁶LFe^IIICu^II]³⁺. The rate of formation of Fc* was zero-order and the zero-order rate constant increased proportionally with increasing the catalyst concentration, but the zero-order rate constant remained constant with variation of concentrations of TFA and O₂ at 213 K [64]. This unusual kinetics indicates that the rate-determining step is a process which does not involve reactions with Fc*, H⁺, O₂. Such a reaction is the O-O bond cleavage in {⁶LFe^III-OOH Cu^II}²⁺, followed by rapid electron transfer to complete the four-electron and four-proton reduction of O₂. This is confirmed by the steady-state observation of {⁶LFe^III-OOH Cu^II}²⁺ (λ_max = 415, 538 nm) during the catalytic reduction of O₂ by Fc* with ⁶LFeCu at 213 K. {⁶LFe^III-OOH Cu^II}²⁺ was independently generated at low temperature (193 K) by the addition of an excess of TFA to the previously well characterized peroxo complex [⁶LFe^III-(O₂²⁻)-Cu^II]⁺ [65].

Scheme 5.

In the case of the Cu-free version ⁶LFe (Scheme 5b) as well, the rate of formation of Fc* was zero-order and the zero-order rate constant increased proportionally with increasing the catalyst concentration, but the zero-order rate constant remained constant with variation of concentrations of TFA and O₂ at 213 K [64]. This again indicates that the rate-determining step in the catalytic cycle is the O-O bond cleavage of ⁶LFe^III-OOH. Surprisingly the bond cleavage rate of ⁶LFe^III-OOH is the same as that of {⁶LFe^III-OOH Cu^II}²⁺. This suggests that the Cu is not bound to the Fe^III-OOH moiety in {⁶LFe^III-OOH Cu^II}²⁺.

This rate-determining step found to occur at 213 K is changed to be the process of O₂-binding to [⁶LFe^IICu^I]⁺ at 298 K when the zero-order rate constant increases proportionally with increasing concentration of O₂ [64]. The change in the rate-determining step at 298 K was confirmed by the change in the steady-state species identified to be {⁶LFe^III-OOH Cu^II}²⁺ at 213 K instead to [⁶LFe^IICu^I]⁺ (λ_max = 422 nm) at 298 K. In contrast to the case at 213 K, however, the O₂-binding rate at 298 K to [⁶LFe^IICu^I]⁺ is significantly faster than that to ⁶LFe^II [64]. This result suggests that the role of the Cu in ⁶LFeCu, at ambient temperature, is to assist the heme and lead to faster O₂-binding during the catalytic cycle.

The electrocatalytic reduction of O₂ was examined using an edge plane pyrolytic graphite (EPG) disk shape electrode, which was modified with ⁶LFeCu [66]. The modified EPG electrode was prepared by transferring an MeCN solution of [⁶LFe^IICu^I]⁺ on the EPG surface and allowing the solvent to evaporate in an argon atmosphere giving a dried surface [66]. Levich and Koutecky-Levich plots [Eqs. (6) and (7)] are shown in Fig. 8a and 8b, respectively [66]. The dashed lines were obtained from the calculated diffusion-convection controlled currents based on the Levich equation assuming the number of electrons for O₂ reduction as two or four. The open circles were from the measured plateau currents and they were higher than those with the two-electron reduction. The deviation from linearity of the four-electron transfer plot suggests that the catalytic reaction is limited by kinetics in addition to the mass-transfer process. In the Koutecky-Levich plot in Fig. 8b obtained from the data in Fig. 8a, the measured values (open circles) were parallel to the line of four-electron reduction. This indicates that the adsorbed complex [⁶LFe^IICu^I]⁺ catalyzed the four-electron reduction of O₂ to H₂O as is the case in the homogeneous solution (vide supra). The four-electron reduction of O₂ to H₂O was confirmed by using rotating EPG disk-platinum ring electrodes to measure the fraction of O₂, which was reduced to H₂O, rather than to H₂O₂. With adsorbed [6LFe^IICu^I]⁺, a small anodic ring plateau curve was observed during the reduction of O₂ at the EPG disk electrode and it was found that most O₂ (85 %) was reduced to H₂O [66].

Fig. 8.

(a) Levich plot from the plateau currents in plot of current vs. (rotation rate)^1/2. (b) Koutecky-Levich plot from the plateau currents in Fig. 2 [(current)⁻¹ vs. (rotation rate)^1/2]. The dashed lines in (a) and (b) were obtained from the calculated diffusion-convection controlled currents for the reduction of O₂ assuming the number of electrons as two and four.

2.4. Cu complexes

There has been considerable interest in the use of copper in O₂-reduction chemistry. One reason is the existence of copper containing enzymes which efficiently effect the four-electron four-proton reduction to water as part of their function. These include so-called multi-copper oxidases (MCO’s) [67–70] wherein an array of copper ions arranged in a tricopper cluster plus a separate but electronically linked copper ion, the latter which effects one-electron oxidations (at a time) of biological substrates, e.g., ascorbate, phenols, diamines, Fe(II) and Cu(I). Such studies have lead to the use of such enzymes, supported on electrodes, to carry out electrocatalytic dioxygen reductions in fact having very high activity [71–73].

However, the existence of MCO’s and particularly inspired by the existence of dicopper enzymes which are blood oxygen carriers in mollusks and arthropods (i.e., hemocyanins) [71, 73], or biologically ubiquitous tyrosinases which ortho-hydroxylate phenols giving catechols and/or quinone products which are converted to melanin pigments [70, 72], has led synthetic bioinorganic chemists to vigorously pursue the study of dioxygen binding and reactivity with discrete copper ion complexes [74–78]. Such investigations have revealed many new fundamental aspects in the field, including the uncovering of a number of now very well defined copper(I)-O₂ derived complexes, existing in different structural forms, although all possessing the Cu₂-O₂ core. Four structural types are known [75, 77, 78], and the generation of one or another depends primarily on the nature of the nitrogenous ligand employed, and the compounds formed can arise from differing mononuclear copper-ligand complexes or those derived from the use of binucleating ligands. Some of these are shown in Fig. 9, along with diagrams of the core peroxo-dicopper(II) or bis-μ-oxo-dicopper(III) structures.

Fig. 9.

Complexes used as catalysts for the solution four-electron four-proton reduction of O₂ by ferrocene derivatives and the copper(I)-dioxygen derived complex intermediates known to form during the course of reaction. See text for discussions.

While the activity of supporting discrete ligand-metal complexes on electrode surfaces is an active area of research, including for copper (see below), homogeneous catalytic O₂-reduction using the complexes shown in Fig. 9 in water and/or organic solvents comes about when employing ferrocene derivatives as one-electron outer-sphere reductants and acids as proton sources [79, 80]. Electrocatalysis certainly has advantages, including the greater likelihood for practical application, however solution investigations enable kinetic and spectroscopic (e.g., UV-vis, EPR) monitoring of key steps during catalysis. This can and does provide for insights into mechanism, via the determination of the identity of intermediates and the order of reaction of a particular step with respect to electron and/or proton concentrations. Such information can provide for elaboration or altering of the catalyst complex ligand, so as to improve future performance.

In fact, all of the Fig. 9 complexes can catalyze O₂-reduction chemistry [79, 80]. First using Fc* and perchloric acid in acetone solutions, the catalytic 4e⁻/4H⁺ reduction of O₂ to water with [(tmpa)Cu^II(H₂O)]²⁺ was examined by following detailed spectroscopic and kinetic monitoring [79]. The mechanism derived is shown in Scheme 6 [79]. The addition of a catalytic amount of [(tmpa)Cu^II(H₂O)]²⁺ to an O₂-saturated acetone solution of Fc* and HClO₄ resulted in the efficient oxidation of Fc* by O₂ to afford ferrocenium cation (Fc*⁺). Fig. 10 shows the spectral changes observed for stepwise addition perchloric acid. For each time period of acid addition (see Fig. 10 Inset), the concentration of Fc*⁺ which was formed immediately (absorption increases at λ_max = 380 and 780 nm) was the same as the mole-amount of added HClO₄.

Scheme 6.

Fig. 10.

UV-vis spectral changes in four-electron reduction of O₂ by Fc* (1.5 × 10⁻³ mol L⁻¹) with [(tmpa)Cu^II(H₂O)]²⁺ (9.0 × 10⁻⁵ mol L⁻¹) in the presence of HClO₄ in acetone at 298 K. The inset shows the changes in absorbance at 380 and 780 nm due to Fc*⁺ produced by stepwise addition of HClO₄ (0.18 – 1.44 × 10⁻³ mol L⁻¹) to an O₂-saturated acetone solution ([O₂] = 11 × 10⁻³ mol L⁻¹) of Fc* and [(tmpa)Cu^II(H₂O)]²⁺.

The production of water was confirmed by mass spectrometry when running reactions using ¹⁸O water. Two-electron O₂-reduction chemistry could be ruled out since iodometric titration experiments yielded no detectable hydrogen peroxide. In the presence of limiting [O₂], and excess Fc*, still only 4 equiv. Fc*⁺ was formed in the presence of 4 equiv. of HClO₄. Thus, these experiments (and others) demonstrated the overall reaction stoichiometry, as catalyzed by [(tmpa)Cu^II(H₂O)]²⁺, to be:

$$4{\text{Fc}}^{\ast }+{\text{O}}_2+4{\text{H}}^{+}\to 4{\text{Fc}}^{\ast +}+2{\text{H}}_2\text{O}$$

It has been demonstrated that the rate limiting step in the reaction cycle was reduction of [(tmpa)Cu^II(H₂O)]²⁺ by Fc* [79]. Then, under conditions where Fc* and acid were depleted, stopped-flow kinetic measurements revealed the very rapid generation of the well known dioxygen adduct, peroxo complex [(tmpa)Cu^II(O₂)Cu^II(tmpa)]²⁺ [75, 77–79] (Scheme 6). Separately, this complex could be cleanly generated at 193 K in solution and subjected to Fc* and/or acid; such experiments led to the finding that when both are present, electron transfer reductive cleavage of the peroxidic O–O bond in [(tmpa)Cu^II(O₂)Cu^II(tmpa)]²⁺ occurs, but only if acid is present. Thus, this key peroxo-dicopper(II) complex intermediate is reduced faster than it is simply protonated (which would yield H₂O₂), leading to overall 4e⁻/4H⁺ O₂ reduction to water [79].

Other complexes shown in Fig. 9 also have been shown to catalyze the O₂-reduction to water, even though they form very different copper-dioxygen adduct structures. [Cu^II₂(N3)(H₂O)₂](ClO₄)₄, following reduction, reacts rapidly with molecular oxygen to form a side-on bound μ-η²:η²-peroxo dicopper(II) complex [Cu^II₂(N3)(O₂)]²⁺ (Fig. 9, Fig. 11a), as previously demonstrated [81–83]. When a catalytic amount of [Cu^II₂(N3)(H₂O)₂]²⁺ is added to and O₂-saturated acetone solution with Fc* and trifluoroacetic acid, O₂ is reduced to water and four equiv. Fc*⁺ are formed, even if excess Fc* relative to O₂ (i.e., limiting [O₂]) and acid are present (also see the inset for Fig. 11b) [80]. Along with the finding by iodometric titration that no hydrogen peroxide is produced, it was concluded that [Cu^II₂(N3)(H₂O)₂]²⁺ is an efficient catalyst for the four-electron reduction of O₂ by Fc*. The same results were obtained using the weaker reductant, octamethylferrocene (Me₈Fc), however no reaction occurred with 1,1′-dimethylferrocene [80].

Fig. 11.

Depiction of the structure of the μ-η²:η²-peroxo dicopper(II) complex [Cu^II₂(N3)(O₂)]²⁺ (b) Formation of the η²:η²-peroxo complex (λ_max = 490 nm) in the reaction of [Cu^I₂(N3)]²⁺ (1.0 × 10⁻⁴ mol L⁻¹) with O₂ in the presence of Fc* (8.0 × 10⁻² mol L⁻¹) in acetone at 193 K. The Inset shows the time profiles of the absorbance at 490 nm (black line) and 780 nm (red line) due to [Cu^II₂(N3)(O₂)]²⁺ and Fc*⁺, respectively. (c) UV/Vis spectral changes observed in the four-electron reduction of O₂ (0.22 × 10⁻³ mol L⁻¹) by Fc* (3.0 × 10⁻³ mol L⁻¹) at 298 K and with TFA (1.0 × 10⁻² mol L⁻¹) catalyzed by [Cu^II₂(N3)(H₂O)₂]²⁺ (1.0 × 10⁻⁴ mol L⁻¹).

To further break down the individual reaction steps, experiments were carried out on [Cu^II₂(N3)(O₂)]²⁺ (λ_max = 365 and 490 nm, Fig. 11c) which was separately generated at 193 K [80]. Electron transfer reduction of this complex by Fc* occurs with concomitant production of two equiv. Fc*⁺ in acetone, see the decay of absorbances due to [Cu^II₂(N3)(O₂)]²⁺ which are accompanied by the rise of absorbance at 780 nm due to Fc*⁺ (Fig. 11c Inset). Kinetic interrogation on the rate of electron-transfer led to determination that Fc* reduction of [Cu^II₂(N3)(O₂)]²⁺ occurred with k_et = 18 mol⁻¹ L s⁻¹ (193 K), a value that was not altered in the presence of one equiv. acid. Me₂Fc was not able to effect this reduction. The latter finding indicates that the electron transfer is not coupled with the protonation of [Cu^II₂(N3)(O₂)]²⁺ [80].

Overall, the data supported a reaction mechanism where the dicopper(II) catalyst was reduced by Fc* to a dicopper(I) complex, previously established to be [Cu^I₂(N3)]²⁺, which reacts very rapidly with O₂, giving [Cu^II₂(N3)(O₂)]²⁺. In a rate-limiting step, this is reduced likely to a transient bis-μ-oxo-dicopper(II) species which is very rapidly protonated to give back the catalyst and two mol-equiv H₂O, thus completing the 4e⁻/4H⁺ reduction of dioxygen, Scheme 7 [80]. Consistent with the proposed mechanism and a piece of valuable new information derived from the studies was a very good estimate of the potential for reduction of [Cu^II₂(N3)(O₂)]²⁺. Based on the known one-electron oxidation potentials for ferrocene derivatives and that of N,N,N′,N′-tetramethylphenylenediamine (TMPD) (E_ox = 0.12 V vs. SCE) [80], which also effected the reduction of [Cu^II₂(N3)(O₂)]²⁺ accompanied by the formation of TMPD^•+, the one-electron reduction potential of [Cu^II₂(N3)(O₂)]²⁺ could be estimated to be E_red = 0.19 (± 0.07) V vs. SCE [80]. That value is significantly lower than the E_red value of (0.37 V vs. SCE) for the catalyst [Cu^II₂(N3)(H₂O)₂]²⁺, these results being consistent with rapid reduction of the latter followed by rate limiting reduction of the μ-η²:η²-peroxo dicopper(II) complex [Cu^II₂(N3)(O₂)]²⁺ formed during the catalytic cycle [80]. There now exists a huge literature describing how μ-η²:η²-peroxo dicopper(II) complexes [Cu…Cu ~ 0.36 nm; O–O ~ 0.14 – 0.15 nm; ν_O–O < 760 cm⁻¹, resonance Raman spectroscopy (rR)] not infrequently exist in equilibrium with a bis-μ-oxo-dicopper(III) isomer Cu···Cu ~ 0.28 nm, O···O ~ 0.23 nm, ν_Cu–O ~ 600 cm⁻¹ (intense) [75, 77, 83–85]. While rR measurements on [Cu^II₂(N3)(O₂)]²⁺ do not show any hint of the presence of a species [Cu^III₂(N3)(O)₂)²⁺ [83], it still had to be considered that it might be formed here in this catalytic process, and that the bis-μ-oxo-dicopper(III) species was actually the species which was being reduced by Fc*, thus following and not prior to O–O cleavage. However, this could be ruled out, as follows. The temperature dependence of the various electron-transfer steps were evaluated, leading to the finding that electron transfer from Fc* or Me₈Fc to both [Cu^II₂(N3)(H₂O)₂]²⁺ and [Cu^II₂(N3)(O₂)] ²⁺ virtually the same values for ΔS^≠ (~ 0, which is expected and typical for outer-sphere electron transfer chemistry) indicating a direct process, not what one would find if a conversion of [Cu^II₂(N3)(O₂)]²⁺ to bis-μ-oxo-dicopper(III) species [Cu^II₂(N3)(O)₂]²⁺⁺ occurred [80].

Scheme 7.

With the value E_red = 0.19 (± 0.07) V vs. SCE determined for peroxo complex [Cu^II₂(N3)(O₂)]²⁺, ΔG_et values for Fc* and Me₈Fc could be determined, (−0.27 ± 0.07) and (−0.23 ± 0.07) eV, respectively. The λ value for electron transfer from Fc* and Me₈Fc to [Cu^II₂(N3)(O₂)]²⁺ could then be estimated [80] and found to be 2.2 (± 0.1) eV, which was significantly larger than the corresponding value for electron transfer from ferrocene derivatives to the dicopper(II) catalyst [Cu^II₂(N3)(H₂O)₂]²⁺. This finding is also consistent with direct electron transfer from ferrocene derivatives to the peroxo-dicopper(II) complex intermediate, as large structural changes would of course be a part of an O–O cleavage process, but not so for the case where a bis-μ-oxo-dicopper(III) complex with already cleaved O–O bond were undergoing simple electron-transfer reduction to give the corresponding bis-μ-oxo-dicopper(II) complex [Cu^II₂(N3)(O)₂]²⁺ (that in Scheme 8) [80].

Scheme 8.

Investigations on O₂-reduction catalys is with [Cu^II(BzPY1)]²⁺ (Fig. 9) were carried out in a similar manner to those discussed for [Cu^II₂(N3)(H₂O)₂]²⁺ (vide supra) [80]. The findings can be summarized as follows (Scheme 9) [80]: [Cu^II(BzPY1)]²⁺ is reduced by Fc* giving copper(I) species [Cu^I(BzPY1)]²⁺ whose reaction with O₂ is known to afford the bis-μ-oxo dicopper(III) complex [{Cu^III(BzPY1)}₂(O)₂]²⁺ (λ_max = 390 nm). Electron transfer reduction by Fc* occurs rapidly upon mixing in acetone even at 193 K to produce two equiv. of Fc*⁺ and the reaction rate was not affected by TFA. Fast protonation gives two mole-equiv. H₂O and the catalyst. Reaction of [{Cu^II(BzPY1)}₂(O)₂]²⁺ also occurs with weaker electron donors, Me₂Fc and even Fc itself. In the latter case however, no catalysis occurs because these latter reductants can’t even reduce [Cu^II(BzPY1)]²⁺ to [Cu^I(BzPY1)]⁺ as ΔG_et > 0, and therefore the catalytic cycle cannot begin [80].

Scheme 9.

To summarize (Scheme 10), different copper(II) complexes giving rise to completely very different Cu₂O₂ structures following reduction and reaction with molecular oxygen, all are readily reduced by ferrocene one-electron reductants, leading to the copper ion catalyzed overall 4e⁻/4H⁺ reduction of dioxygen to water [80]. The key step leading to this reaction stoichiometry is ferrocenyl reductive O–O bond cleavage (by two-electron equiv.) at the peroxo-dicopper(II) level, at least for two of the three cases. In the other, with catalyst [Cu^II(BzPY1)]²⁺, the copper(I)/O₂ chemistry already gives rise to O–O cleavage and a bis-μ-oxo-dicopper(III) intermediate which is then reduced (Scheme 10). To finish (or start) the catalytic cycle, the copper(II) complex(es) then formed are reduced to copper(I) for further O₂-reaction. With the great variations in ligand design for copper complexes which are possible, and can lead to varying patterns of Cu(II)/Cu(I) or Cu(III)/Cu(I) redox cycles, the possibility of developing systems which may effect the catalytic 2e⁻/2H⁺ O₂-reduction to hydrogen peroxide seems real.

Scheme 10.

The electrocatalytic four-electron reduction of O₂ has also been achieved using electrodes on which Cu complexes are immobilized [86–96]. Whether a mononuclear Cu complex or a dinuclear Cu complex is responsible for the four-electron reduction of O₂ was examined by determining the dependence of the rate on the Cu coverage. Anson and coworkers determined that the electrocatalytic rate of the O₂-reduction was first-order in Cu coverage, suggestive of a mononuclear Cu site as the active catalyst [86,87]. On the other hand, the rate of O₂ reduction with a Cu^I complex of 3-ethynyl-phenanthroline covalently immobilized onto an azide-modified glassy carbon surface is second-order in the Cu coverage at moderate overpotential, suggesting that two Cu^I species are necessary for the efficient four-electron four-proton reduction of O₂ [95,96]. In this case, a dinuclear peroxo-Cu complex is proposed to be a key intermediate for the four-electron reduction of O₂, in a reaction cycle which is similar to that shown in Scheme 6. In the context of the present review, it should be noted that for recent cases of electrocatalytic systems, and depending on conditions of applied potential or pH, increases in the relative amount of two-electron O₂-reduction to H₂O₂ may be observed [96,97].

In the context of the chemical reduction of molecular oxygen and production of hydrogen peroxide which may come about with copper complexes, there is in fact a pertinent literature concerning the (di)copper complex mediated catalytic oxidation of catechols to quinones [98–100]. This research is bioinspired as enzyme catechol oxidases are known to effect this reaction at an active site comprising two adjacent (0.3 – 0.4 nm) copper ions. The stoichiometry of reaction for the enzyme is thought to be:

$$2\text{Ar}-{(\text{OH})}_2+{\text{O}}_2\to 2\text{Ar}-{(=\text{O})}_2+2{\text{H}}_2\text{O}.$$

However, for a number of chemical model system studies aimed at exploring mechanistic details and typically employing 3,5-di-tert-butylcatechol (DTBC) as a convenient substrate; 3,5-di-tert-butyl-1,2-benzoquinone (DTBQ) along with hydrogen peroxide are produced stoichiometrically, according to

$$\text{DTBC}+{\text{O}}_2\to \text{DTBQ}+{\text{H}}_2{\text{O}}_2.$$

Mechanisms proposed by several groups [101–106] involve the interaction of DTBC at a one ligand-copper(II) site, whereupon valence tautomerism to give a Cu(I)-semiquinone species leads to O₂-reactivity giving a copper(II)-superoxo species which in a rate limiting step abstracts a substrate hydrogen atom, eliminating DTBQ and hydrogen peroxide while regenerating the catalytic copper(II) containing complex. Another possible mechanism would proceed via acid-base chemistry where a Cu(II)-peroxo-Cu(II) intermediate reacts with a catechol, releasing H₂O₂ and leaving a bound catecholate dicopper(II) center; the latter would undergo internal redox to give quinone product and a dicopper(I) moiety which reacts with dioxygen to repeat the cycle [99, 104,107]. With such known chemistries, it may well be that investigation of copper complex electrocatalytic or chemically selective reduction to H₂O₂ should be pursued.

In general, copper complexes, as mentioned here, have seen increasing use and success in O₂-reduction electrocatalytic applications [73,94–97,108]. But, also as additives or other enhancements of various types, copper in various forms has been shown to boost O₂-reduction electrocatalytic behavior. Some recent examples include (i) a system in which a multicopper oxidase is linked covalently to a multiwall carbon nanotube [109], (ii) a fuel cell which utilizes a Cu-Cu₂O redox cycle [110], (iii) ACu(II) grafted TiO₂ catalyst, wherein reduced copper(I) formed from a photolytic process facilitates O₂-reduction [111], (iv) copper nanoclusters deposited onto a glassy-carbon electrode showed very favorable O₂ reductive electrocatalytic activity [112] and (v) a modified platinum electrode, where sub-monolayer quantities of copper were incorporated into Pt(111) resulted in an eight-fold enhancement in O₂-reduction electrocatalytic activity [113].

3. H₂O₂ formation by solar cell

In order to utilize H₂O₂ as an energy carrier, hydrogen peroxide should be produced by a convenient method without any special equipment. One such method is electroreduction of O₂ with a solar cell in an acidic aqueous solution under air [24]. The solar cell is a convenient power source usable anywhere during the sun-shiny days. H₂O₂ can be produced anywhere by plugging electrodes with an O₂ reduction catalyst into solar cells in an acidic solution. H₂O₂ production by the electrocatalytic reduction of O₂ under air with electrical power generated by a photovoltaic solar cell has been performed using Co-porphyrin compounds. This is depicted in Fig. 12 for catalysts in an acidic solution at room temperature [24], because cobalt porphyrins act as efficient selective two-electron reduction of O₂ as described in a previous section [31,32].

Fig. 12.

Cobalt porphyrins employed for electrocatalytic reduction of O₂ to H₂O₂ [24].

Fig. 13 shows the O₂ reduction current recorded against voltage vs. saturated calomel electrode (SCE) using a glassy carbon electrode modified by the cobalt porphyrins shown in Fig. 12 in an aqueous solution containing hydrosulphuric acid (1 mol L⁻¹) [24]. Dioxygen reduction currents were observed when using [Co(TCPP)], [Co(TPP)] and [Co(OEP)], the details depending on the particular porphyrin structure. The highest on-set potential of 0.3 V was observed with [Co(TCPP)] [24]. The thermodynamic potential for the two-electron reduction of O₂ at a given pH is 0.65 V, thus, the overpotential for O₂ reduction was 0.35 V for [Co(TCPP)]. The onset potential observed on [Co(TPP)] was around 0.25 V comparable to that of [Co(TCPP)]. The onset potential of [Co(OEP)] was 0.1 V, thus indicating a larger overpotential is required for O₂ reduction.

Fig. 13.

(a) Electrocatalytic current at oxygen reduction. ([Co(DPP)] – –, [Co(OEP)] – · –, [Co(TPP)] - - -, [Co(TCPP)] —). (b) CV of [Co(TCPP)] at scan rate = 20 mV s⁻¹ (with 3 × 10⁻³ mol L⁻¹ H₂O₂ – –, with oxygen bubbling: ——). The measurements were performed in 1.0 × 10⁻¹ mol L⁻¹ hydrosulphuric acid at 298 K. To prevent contamination of the coordinating ion or ligands, ultrapure water was used in the experiments.

A cyclic voltammogram of [Co(TCPP)] in the presence of oxygen gas and H₂O₂ (3 × 10⁻³ mol L⁻¹) (Fig. 13b) exhibits a reduction peak which appears around 0 V in the presence of O₂, corresponding to two-electron reduction of O₂ to produce H₂O₂ [24]. The selective production of H₂O₂ can be achieved with the [Co(TCPP)] catalyst when the applied potential is controlled to around 0 V. The current efficiency for H₂O₂ production was nearly 100% in 1.0 × 10⁻¹ mol L⁻¹ hydrosulphuric acid [24].

H₂O₂ production by the reduction of O₂ in an acidic aqueous solution was also performed using a conventional Si photovoltaic solar cell, which behaves as an electric power source, with an output of 0.5 V at 2.5 mA [24]. A glassy carbon electrode (~1 cm²) mounted with a Co-porphyrin and Pt wire electrode were connected to the negative and positive electrodes. After 11 hours running in a 1.0 × 10⁻¹ mol L⁻¹ hydrosulphuric acid (~7 mL), solution, the amount of H₂O₂ produced reached 1.46 × 10⁻⁵ mol with [Co(TCPP)] [24]. Thus, H₂O₂ was conveniently produced by the electrocatalytic reduction of O₂ with a conventional photovoltaic solar cell although the catalytic behavior as well as cell and electrode structures will require improvement for practical application.

4. Direct H₂O₂ fuel cells

Fuel cells require oxygen as the oxidant and therefore the use in air-free environments such as outer space and underwater require a compressed oxygen tank. Hydrogen peroxide (H₂O₂) has been utilized as an alternative liquid oxidant in place of gaseous O₂. A number of fuel cells with H₂O₂ as the oxidant have been developed using borohydride [114–116], metals [117–119], methanol [120,121], hydrazine [122] and biofuels [123] as reductants. H₂O₂ can also be used as a reductant and direct H₂O₂ cells have recently been developed [124–127]. Direct H₂O₂ fuel cells have number of merits as compared with other fuel cells: (1) H₂O₂ is liquid and soluble in water and thereby easy to store and carry: (2) H₂O₂ has higher standard reduction potential than O₂ (e.g., 1.776 V vs. SHE for H₂O₂ and 1.229 V vs. SHE for O₂ in acidic medium [124]; (3) there is no need to use membranes, because H₂O₂ can act both oxidant and reductant. Thus, fuel cells with H₂O₂ used as both oxidant and reductant can have much simpler cell structure (i.e., one compartment cell without membrane), theoretically providing higher power output than those with oxygen used as an oxidant [124]. The reactions occurring at anode and cathode of the H₂O₂ fuel cell are given by Eqs. (8) – (10) [24,125–127].

$$\text{Anode}:{\text{H}}_2{\text{O}}_2\to {\text{O}}_2+2{\text{H}}^{+}+2{\text{e}}^{-}0.68\text{V}$$ (8)

$$\text{Cathode}:{\text{H}}_2{\text{O}}_2+2{\text{H}}^{+}+2{\text{e}}^{-}\to 2{\text{H}}_2\text{O}1.77\text{V}$$ (9)

$$\text{Total}:2{\text{H}}_2{\text{O}}_2\to {\text{O}}_2+2{\text{H}}_2\text{O}$$ (10)

Thus, H₂O₂ fuel cells emit only oxygen after electrical power generation. The theoretical maximum of the output potential of the H₂O₂ fuel cell is 1.09 V [125–127], which is comparable to those of a hydrogen fuel cells (1.23 V) and a direct methanol fuel cell (1.21 V) [127]. A one compartment H₂O₂ fuel cell has been constructed by using an Au plate as an anode and an Ag plate as a cathode, because these metal plates act as selective oxidation and reduction catalyst toward H₂O₂, respectively [126]. Such a one-compartment structure without membrane is certainly more promising for development of low-cost fuel cells as compared with two compartment fuel cells with membranes. However, the voltage achieved from the fuel cell described above was 0.12 V, which is lower than the theoretically achievable voltage of 1.09 V; this was due to a large over-potential at the Ag cathode [125].

Increasing the specific surface area is one of the easiest means to achieve high catalytic activity per weight. In order to increase the surface area of Ag plate, nanoparticles of Ag with high specific surface area were examined as a cathode of the one-compartment H₂O₂ fuel cell [24]. Also addition of foreign elements of Pb to modify the electronic structure of Ag nanoparticles was examined [24]. Fig. 14 shows the TEM images of Ag and Ag-Pb alloy nanoparticles used for the construction of one-compartment H₂O₂ fuel cells. The average Ag nanoparticles diameter was 41 nm.

Fig. 14.

TEM images of Ag or Ag-Pb alloy nanoparticles. (a) Ag nanoparticles, (b) Ag-Pb alloy (Ag:Pb = 9:1), (c) Ag-Pb alloy (7:3) and (d) Ag-Pb alloy (Ag:Pb = 6:4).

The Ag based nanoparticles were mounted on a glassy carbon electrode by a drop-casting method [24] and the one-compartment H₂O₂ fuel cells constructed were operated under basic conditions using an aqueous solution containing NaOH (1.0 mol L⁻¹) and H₂O₂ (3.0 × 10⁻¹ mol L⁻¹). Fig. 15 shows I-V and I-P curves for such H₂O₂ fuel cells where the current density was normalized by the geometric surface area of the glassy carbon electrode [24]. The results obtained with Ag nanoparticles without Pb addition are plotted in black color in Fig. 15. The performance of the H₂O₂ fuel cell using Ag nanoparticles was comparable to that using Ag plate. On the other hand, when Ag-Pb alloys were used as cathodes [Ag:Pb = 9:1 (blue), 7:3 (red), 6:4 (green) in Fig. 15], cell performance was improved, i.e., higher power densities, open circuit voltages and short-circuit currents were obtained compared to those using the Ag nanoparticles as the cathode. Large values were achieved on Ag-Pb alloys with the ratios of 9:1 and 7:3 although the open-circuit potential of ca. 150 mV was still far from the theoretical potential (1.09 V) [24].

Fig. 15.

I-V and I-P curves of a one-compartment H₂O₂ fuel cell with Ag or Ag-Pb alloy cathode. (Au anode. 1.0 mol L⁻¹ NaOH, 3.0 × 10⁻¹ mol L⁻¹ H₂O₂. black: Ag, green: Ag:Pb = 6:4, red: Ag:Pb = 7:3 and blue: Ag:Pb = 9:1).

In order to achieve a more efficient system, the medium conditions should be the same or similar for H₂O₂ production and power generation. Hydrogen peroxide fuel cells using an Ag-based cathode can generate power only in basic media, however, H₂O₂ production by O₂ reduction was performed under acidic conditions [24]. H₂O₂ is not stable under basic conditions, thus, there is a tremendous need to develop an H₂O₂ fuel cell which can be operated under acidic conditions for the ideal combination with the production of H₂O₂ via the two-electron reduction of O₂ with a solar cell.

A one-compartment H₂O₂ fuel cell using an iron phthalocyaninato complex can be operated under acidic conditions [126]. The choice of iron porphyrin and analogous compounds for H₂O₂ reduction is quite reasonable because in natural systems, the reduction of hydrogen peroxide is owing to hydroperoxidases, which contain iron(III)-porphyrins in their active sites [128–130]. The activity of iron complexes depicted in Fig. 16 was examined for H₂O₂ reduction [126]. Fig. 17 shows the cyclic voltammograms (CV) of H₂O₂ with an iron complex mounted glassy carbon as a working electrode in an aqueous solution of acetate buffer (pH 4) containing 3.0 × 10⁻¹ mol L⁻¹ H₂O₂ [126]. The black curves in Fig. 17 are CVs of each iron complex in an acetate buffer solution without H₂O₂. A catalytic current of the H₂O₂ reduction was observed in a cathodic sweep with all the Fe complexes, indicating they act as catalysts for the H₂O₂ reduction in acidic media. The onset potentials for H₂O₂ reduction on electrodes with [Fe^III(OEP)Cl], [Fe^III(TPP)Cl] and [Fe^III(Pc)Cl] were ca. 0.2 V, 0.2 V and 0.5 V (vs. SCE), respectively. Thus, [Fe^III(Pc)Cl] showed the smallest overpotential in this acidic solution, in comparison to the Fe^III porphyrin complexes.

Fig. 16.

Chemical structures of porphyrin and phthalocyanine iron(III) complexes similar to active site structures of hydroperoxidases as candidates of cathodes for an H₂O₂ fuel cell. (a) [Fe^III(Pc)Cl], (b) [Fe^III(OEP)Cl] and (c) [Fe^III(TPP)Cl].

Fig. 17.

Cyclic voltammograms of H₂O₂ on glassy carbon electrodes modified with Fe^III complexes. (a) [Fe^III(Pc)Cl], (b) [Fe^III(OEP)Cl] and (c) [Fe^III(TPP)Cl]. The measurements were performed in an acetate buffer solution (pH 4) containing 3.0 × 10⁻³ mol L⁻¹.

A one-compartment H₂O₂ fuel cell working in an acidic media was constructed with a glassy carbon electrode mounting [Fe^III(Pc)Cl] as a cathode and Ni metal as an anode [126]. The cell performance was evaluated by dipping the anode and cathode in a buffer solution containing 3.0 × 10⁻¹ mol L⁻¹ H₂O₂. Fig. 18 shows the cell performances of the H₂O₂ fuel cells in acidic solutions containing 3.0 × 10⁻¹ mol L⁻¹ H₂O₂ at 293 K [126]. Fig. 18a displays the I–P and I–V curves obtained by the operation of the fuel cell under the conditions of pH 4 (black) and pH 5 (blue). In both cases, the open circuit potentials were ca. 0.24 V, which is significantly higher than that (0.15 V) of the H₂O₂ fuel cell using a Ag cathode working under basic conditions [24,125]. The maximum power density was improved from 0.39 μWcm⁻² at pH 5 to 0.44 μWcm⁻² at pH 4 by decreasing the pH. The I–V and I–P curves obtained for the H₂O₂ fuel cell operated at pH 3 were much higher in both open circuit potential and power density as displayed in Fig. 18b. The power density reached to 10 μWcm⁻², which is more than 20 times higher than that obtained under pH 4 and pH 5 conditions. The open-circuit potential of 0.5 V is more than three times higher than that (0.15 V) of the one-compartment H₂O₂ fuel cell operated under basic conditions [24,125]. A further decrease in pH resulted in the decomposition of H₂O₂ on the surface of the Ni anode. Thus, the conditions of pH 3 are the optimum for operating the H₂O₂ fuel cell using [Fe^III(Pc)Cl] as the cathode.

Fig. 18.

I–V and I–P curves of a one-compartment H₂O₂ fuel cell with Ni anode and [Fe^III(Pc)Cl] cathode. Performance tests were conducted in an acetate buffer containing 3.0 × 10⁻¹ mol L⁻¹ H₂O₂. The pH of the solutions was fixed to 5 (a, blue), 4 (a, black) or 3 (b, red). Currents and powers were normalized by a geometric surface area of electrode.

The chemical structure of [Fe^III(Pc)Cl] under acidic condition was investigated by an acid titration of [Fe^III(Pc)Cl] with trifluoroacetic acid (TFA) using UV-vis spectroscopy as shown in Fig. 19 [126]. To the benzonitrile solution of [Fe^III(Pc)Cl] (4.0 × 10⁻⁵ mol L⁻¹), a known amount of TFA (2 – 16 × 10⁻³ mol L⁻¹) was added and UV-vis spectral changes monitored. By the acid addition TFA, an intense absorption band around 656 nm assigned to [Fe^III(Pc)Cl] became weaker and new absorption bands appeared around 520 and 720 nm (Fig. 19a), which were assigned to the protonated iron phthalocyanate complex [131]. The slope of Hill plot by using absorption change at 520 nm shown in Fig. 19b was 1.2, indicating that one proton is associated with in the change of UV-vis absorbance. The change was owing to the protonation of one of imine group of the phthalocyanine ligand. The protonation constant K was determined from the titration in Fig. 19 to be 160 mol⁻¹ L [126]. The protonation to the phthalocyanine ligand provided a positive shift of a redox potential of Fe(II)/Fe(III) and stabilization of the reduced state of Fe(II) then suitable for H₂O₂ reduction.

Fig. 19.

(a) UV-vis absorption change of a benzonitrile solution of [Fe^III(Pc)Cl] (4.0 × 10⁻⁵ mol L⁻¹) by adding trifluoroacetic acid (2 – 16 × 10⁻³ mol L⁻¹). (b) The Hill plot obtained from monitoring of the absorption changes at 520 nm.

The protonation of [Fe^III(Pc)Cl] improves the catalytic activity of [Fe^III(Pc)Cl] for H₂O₂ reduction, however, the cationic nature of [Fe^III(PcH)Cl]⁺ decreases the stability because its solubility in water increases [126]. As a result, a serious deterioration in the cell performance was observed during the course of repetitive tests. Fig. 20 (black) shows the resulting potential decrease of the power density of 20 μA cm⁻² (not open circuit potential) by the repetitive tests [126]. A high potential of 0.28 V observed in the first test decreased to ca. 0.1 V after repetition. The robustness of [Fe^III(PcH)Cl]⁺ was improved by coating of the complex with a protic membrane, Nafion^®, which is a highly acidic polymer containing sulfonate groups at the end of side chains [126]. A glassy carbon electrode was coated with Nafion^® after mounting [Fe^III(Pc)Cl], and an H₂O₂ fuel cell was constructed with the electrode. Subsequent repetitive tests on this H₂O₂ fuel cell were performed at pH 3. As indicated in Fig. 20 (red), the potential of more than 0.25 V observed in the first cycle was maintained in the repetitive tests [126]. Thus, the coating with Nafion^® effectively improved the stability of the cell performance.

Fig. 20.

Potential changes by repetitive measurements of H₂O₂ fuel cells with the Nafion^® coated [Fe^III(Pc)Cl] cathode (red) and without Nafion^® coating (black). Potentials required to achieve the power density of 20 μA cm⁻² were recorded.

5. Conclusions

In this review, it has been shown that a variety of metal complexes can catalyze reduction of O₂ by one-electron reductants in the presence of an acid in homogeneous solution. Electrocatalytic reduction of O₂ can also be achieved using electrodes modified with metal complexes. The number of electrons transferred during O₂ reduction is two or four depending of type of metals and ligands. In the case of biscobalt porphyrins and biscobalt porphyrin-corrole complexes, the Co-Co distance is crucial to attain the four-electron reduction of O₂, because the formation of the dinuclear Co(III) μ₂-η¹:η¹-peroxo complex with suitable Co-Co distance and the cleavage of the O-O bond are required for the four-electron reduction of O₂. In the case of cytochrome c oxidase model compounds, however, a Fe porphyrin without a Cu unit can catalyze the four-electron reduction of O₂. In this case, an Fe(III)-hydroperoxo complex can be further reduced by one-electron reductants in the presence of an acid to produce water. Electrocatalytic four-electron reduction of O₂ can also be attained using electrodes modified with cytochrome c oxidase model complexes. Cu complexes alone can also catalyze the four-electron reduction of O₂ by one-electron reductants in the presence of an acid in homogeneous and heterogeneous systems. Hydrogen peroxide produced by electrocatalytic reduction of O₂ using electrodes modified with metal complexes acting as catalysts for selective two-electron reduction of O₂ can be used as a fuel in a hydrogen peroxide fuel cell. A hydrogen peroxide fuel cell has great advantages in comparison with other fuel cells which require O₂ as the oxidant, because it can be operated using a one-compartment cell without a membrane and in air-free environments such as in outer space and underwater. Future scrutiny is desired to improve the catalytic activity for the selective two-electron reduction of O₂ to H₂O₂, the latter being promising candidates as renewable and clean energy sources.