Electrocatalytic O₂ Reduction by Covalently Immobilized Mononuclear Copper(I) Complexes: Evidence for a Binuclear Cu₂O₂ Intermediate

Abstract

A Cu^I complex of 3-ethynyl-phenanthroline covalently immobilized to an azide-modified glassy carbon surface is an active electrocatalyst for the 4-electron reduction of O₂ to H₂O. The rate of O₂ reduction is 2^nd order in Cu coverage at moderate overpotential, suggesting that two Cu^I species are necessary for efficient 4-electron reduction of O₂. Mechanisms for O₂ reduction are proposed that are consistent with the observations for this covalently immobilized system and previously reported results for a similar physisorbed Cu^I system.

Discrete copper complexes are potential catalysts for the 4-electron reduction of O₂ to water in ambient temperature fuel cells as evidenced by Cu-containing fungal laccase enzymes that rapidly reduce O₂ directly to water at a trinuclear Cu active site at remarkably positive potentials.^1-5 Several groups have studied molecular Cu complexes immobilized onto electrode surfaces as an entry into the study of 4-electron O₂ reduction.^6-19 In particular, physisorbed Cu^I(1,10-phenanthroline), Cu(phen^P), reduces O₂ quantitatively by 4 electrons and 4 protons to water.^8-10 Anson, et al., determined that this reaction was 1^st order in Cu coverage, suggestive of a mononuclear Cu site as the active catalyst.^8,10

In the present study, similar Cu^I complexes are covalently attached to a modified glassy-carbon electrode surface to form a species denoted Cu(phen^C), and the effect of Cu coverage on the kinetics of electrocatalytic O₂ reduction is investigated. At low overpotentials, we observe a 2^nd order dependence of the O₂-reduction rate on the coverage of Cu(phen^C), from which we infer that two physically proximal Cu(phen^C) bind O₂ to form a binuclear Cu₂O₂ species required for 4-electron reduction. We suggest that a similar binuclear species also forms in the case of Cu(phen^P)^8,10 but that rate-limiting binding of O₂ to the first Cu(phen^P) followed by rapid surface diffusion of a second Cu(phen^P) has, until now, obscured the binuclear nature of the reaction.

The covalent attachment of 3-ethynyl-1,10-phenanthroline to an azide-modified glassy carbon electrode to form Cu(phen^C) relies on the Cu^I-catalyzed cycloaddition of azide and ethynyl groups to form a triazole linker, commonly referred to as the click reaction.^20,21 The electrode is azide terminated by treating a roughly-ground, heat-treated glassy carbon surface with a solution of IN₃ in hexanes, a procedure modified from that first described by Devadoss and Chidsey.²² An XPS survey of the azide-modified surface shows two N 1s peaks at 399 eV and 403 eV in a 2:1 ratio attributable to the azide nitrogens.^22-24 Upon exposure to 3-ethynyl-1,10-phenanthroline under the click reaction conditions²⁵, the 403-eV peak disappears and the 399-eV peak broadens, consistent with the formation of the 1,2,3-triazole linker.^22,24 XPS peaks at 934 and 953 eV corresponding to the Cu 2p^3/2 and 2p^1/2 transitions²⁶ and a N-to-Cu coverage ratio of 5.3 ± 0.3 are attributable to a covalently attached Cu(3-(4-triazolyl)-1,10-phenanthroline) complex, Cu(phen^C).

A cyclic voltammogram (CV) of Cu(phen^C) on a static glassy carbon electrode shows quasi-reversible reduction and oxidation peaks under anaerobic conditions, which are assigned to the Cu^II/I redox couple (Figure 1a). The standard redox potential of the complex is taken as the average of the cathodic and anodic peak potentials: $E_{\text{Cu}}^0=275\pm 15\text{mV}$ vs. NHE.²⁷ The Cu coverage was varied by controlling the initial azide coverage of the glassy carbon electrode,²⁵ and the amount of immobilized Cu catalyst was determined by Faraday's Law from the Cu^II/I redox charge, Δq. This value was taken to be the average of the absolute values of the integrated current in the negative and positive scans of the CV corrected by the same scans after the surface was stripped of all copper.²⁸ For the CV presented in Figure 1a, Δq = 21.8 μC, which is equivalent to a coverage of 7.0 × 10¹⁴ molecules cm^-2. The relatively high surface coverage is consistent with significant roughness of the roughly-ground surface.²⁹

Figure 1.

(a) Cyclic voltammogram (CV) of Cu(phen^C) on a static glassy carbon electrode in an Ar-purged aqueous solution (scan rate = 1000 mV/s). The dashed line is the same CV repeated after Cu was removed by exposing the surface to a Cu chelating agent for 20 min while rotating the electrode at 3000 rpm. (b) Rotating-disk voltammograms for the reduction of O₂ in an O₂-saturated aqueous solution by Cu(phen^C) (scan rate = 25 mV/s). The dashed lines are the same set of rotating-disk voltammograms repeated after Cu was removed by exposing the surface to a copper-chelating agent for 20 min while rotating the electrode at 3000 rpm. The inset is a Koutecky-Levich plot of the inverse of the disk current measured at 0 mV vs. NHE as a function of square root of the inverse of the rotation rate. The dashed lines are calculated diffusion-limited currents for the reduction of O₂ by 2 and 4 electrons. The fitted line yields n = 3.6 electrons. The intercept is the inverse of the kinetically limited current, (i_K)^-1. All solutions contained 0.05M sodium acetate, 0.05M acetic acid, 1 M sodium perchlorate with measured pH = 4.8.

O₂ reduction by Cu(phen^C) was measured at a rotating-disk electrode (RDE) at several rotation rates (Figure 1b).³⁰ The currents show an onset of O₂ reduction negative of $E_{\text{Cu}}^0$. The currents are independent of the scan rate at or below 25 mV/s (not shown), but increase with increasing rotation rate as shown. In the limit of high rotation rate, the current is expected to asymptotically approach a potential-dependent kinetic current, i_K, independent of the rate of mass transfer to the electrode surface and limited only by O₂-binding and reduction kinetics. A Koutecky-Levich plot of the inverse of the current as a function of the inverse square root of the rotation rate yields the inverse of the kinetic current, i_K^-1, as the intercept (Figure 1b, inset).³¹ The slope of the Koutecky-Levich plot is inversely proportional to the number of electrons, n, by which O₂ is reduced.³¹ The measured slope in Figure 1b yields n = 3.6,³² suggesting preferential catalysis of 4-electron, rather than 2-electron, reduction of O₂.

The dependence of the rate of O₂ reduction on catalyst coverage was determined from the difference in the measured kinetic currents with and without Cu on the electrode surface, Δi_K. This procedure was chosen to compensate for variable redox charge and kinetic current baselines. A plot of Δi_K as a function of Δq measured at 0 mV vs. NHE shows a 2^nd order dependence (Figure 2), confirmed by a log/log plot slope of 1.98 (Figure 2, inset).³³

Figure 2.

Difference in kinetic currents for O₂ reduction with and without Cu, Δi_K, at 0 mV vs. NHE, for different Cu(phen^C) coverages as measured by the redox charge Δq. The solid line is the best fit of the form y = ax². The dashed line is the best fit of the form y = ax. The inset is a log-log plot of the same data. The solid line in the inset is a linear fit with slope = 1.98. The dashed line has slope = 1 for comparison.

This 2^nd order dependence on coverage along with the onset of the O₂-reduction current negative of the Cu^II/I redox potential suggests a potential-dependent rate of reduction of an O₂ species ligated by 2 proximal Cu^I(phen^C) sites as presented in Equations 1-3.³⁵ The proximal sites, {2 Cu^I}, reversibly bind O₂ to form {Cu₂O₂} (Equation 1, Figure 3), followed by a potential-dependent reduction step (Equation 2). Rapid protonation and further reduction regenerate the proximal Cu^I sites {2 Cu^I} and release water (Equation 3).³⁶

Figure 3.

The proposed binding event between 2 proximal Cu centers and a single O₂ molecule to form a Cu₂O₂ intermediate species.³⁴

$$\{2{\text{Cu}}^{\text{I}}\}+{\text{O}}_2\underset{{\text{k}}_{-1}}{\overset{{\text{k}}_1}{\rightleftharpoons }}\{{\text{Cu}}_2{\text{O}}_2\}$$ (1)

$$\{{\text{Cu}}_2{\text{O}}_2\}+{\text{e}}^{-}\stackrel{{\text{k}}_2(E)}{\to }\{{\text{Cu}}_2{{\text{O}}_2}^{-}\}$$ (2)

$$\{{\text{Cu}}_2{{\text{O}}_2}^{-}\}+4{\text{H}}^{+}+3{\text{e}}^{-}\stackrel{\text{fast}}{\to }\{2{\text{Cu}}^{\text{I}}\}+2{\text{H}}_2\text{O}$$ (3)

The potential-dependent rate constant for the reduction of {Cu₂O₂}, k₂(E), can be expanded using the Butler-Volmer parameterization (Equation 4):

$$k_2(E)=k_{\{{\text{Cu}}_2{\text{O}}_2\}}^{0}exp\left(\frac{-\alpha F}{RT}\left(E-E_{\{{\text{Cu}}_2{\text{O}}_2\}}^0\right)\right)$$ (4)

where $k_{\{{\text{Cu}}_2{\text{O}}_2\}}^0$ is the standard rate constant of the electron-transfer reaction, F is Faraday's constant, α is the transfer coefficient typically taken to be 0.5, R is the ideal gas constant, T is the absolute temperature, and $E_{\{{\text{Cu}}_2{\text{O}}_2\}}^0$ is the standard reduction potential of {Cu₂O₂}.

At high Cu coverage, the dependence of Δi_K on potential (Figure 4a, solid) approximates the sigmoidal curve expected for an electrocatalytic process that is limited at high overpotential by a non-electrochemical step such as that in Equation 1.³¹ The deviation from sigmoidal behavior at more negative potentials suggests an additional O₂-reduction pathway at greater overpotential. More striking evidence of this pathway is observed at low Cu coverage (Figure 4b, solid) with an approximately exponential rise in O₂ reduction below 0 mV vs. NHE.

Figure 4.

Difference in kinetic currents for O₂ reduction with and without Cu, Δi_K, (a) high coverage (Δq = 41.5 μC) and (b) low coverage (Δq = 4.0 μC). The solid black line is the measured Δi_K. The dash-dotted green line is a proposed fit to the data comprised of the sum of the potential-dependent currents from two competing pathways (Equation 8): a binuclear 4-electron O₂-reduction pathway (red dashed) and a mononuclear 2-electron O₂-reduction pathway (blue dotted).

We propose that the second O₂-reduction pathway is due to site-isolated mononuclear Cu complexes, {Cu^I}, which reduce O₂ by 2 electrons and 2 protons to H₂O₂ at more negative potentials (Equations 5-7). This model is supported by the increased H₂O₂ production detected in rotating ring-disk voltammetry experiments at high overpotentials.³⁷

$$\{{\text{Cu}}^{\text{I}}\}+{\text{O}}_2\underset{{{\text{k}}^{\prime }}_{-1}}{\overset{{{\text{k}}^{\prime }}_1}{\rightleftharpoons }}\{{\text{CuO}}_2\}$$ (5)

$$\{{\text{CuO}}_2\}+{\text{e}}^{-}\stackrel{{\text{k}}_2^{\prime }(E)}{\to }\{\text{Cu}{{\text{O}}_2}^{-}\}$$ (6)

$$\{\text{Cu}{{\text{O}}_2}^{-}\}+2{\text{H}}^{+}+{\text{e}}^{-}\stackrel{\text{fast}}{\to }\{{\text{Cu}}^{\text{I}}\}+{\text{H}}_2{\text{O}}_2$$ (7)

k₂′(E) can be expanded using the Butler-Volmer model in a similar manner as shown in Equation 4. The total kinetic current for 4- and 2-electron reduction of O₂ expected by the two pathways is given in Equation 8.²⁵

$$i_{\text{K}}=\frac{4FA{k}_1{\mathrm{\Gamma }}_{\{2{\text{Cu}}^{\text{I}}\}}[{\text{O}}_2]k_{\{{\text{Cu}}_2{\text{O}}_2\}}^{0}exp\left(\frac{-\alpha F}{RT}\left(E-E_{\{{\text{Cu}}_2{\text{O}}_2\}}^0\right)\right)}{k_{-1}+k_{\{{\text{Cu}}_2{\text{O}}_2\}}^{0}exp\left(\frac{-\alpha F}{RT}\left(E-E_{\{{\text{Cu}}_2{\text{O}}_2\}}^0\right)\right)}+\frac{2FA{k}_1^{\prime }{\mathrm{\Gamma }}_{\{{\text{Cu}}^{\text{I}}\}}[{\text{O}}_2]k_{\{\text{Cu}{\text{O}}_2\}}^{0}exp\left(\frac{-\alpha F}{RT}\left(E-E_{\{\text{Cu}{\text{O}}_2\}}^0\right)\right)}{k_{-1}^{\prime }+k_{\{\text{Cu}{\text{O}}_2\}}^{0}exp\left(\frac{-\alpha F}{RT}\left(E-E_{\{\text{Cu}{\text{O}}_2\}}^0\right)\right)}$$ (8)

This model provides a good fit to the observed kinetic currents for O₂ reduction by Cu(phen^C) at both high and low coverage (Figure 4a-b).

The 2^nd order dependence of O₂ reduction on the coverage of Cu(phen^C), best observed at positive potentials, is distinct from the 1^st order dependence on Cu coverage determined by Anson et al. for Cu(phen^P) on edge-plane graphite.¹⁰ We have confirmed the Anson result at 0 mV vs. NHE,²⁵ and have previously reported that there is no rate-limiting step for electrocatalytic O₂ reduction by Cu(phen^P) subsequent to O₂ binding. ^{19, 38} In the physisorbed case, the expected high surface lateral mobility of Cu(phen^P)³⁹ will ensure that once one Cu(phen^P) coordinates O₂, another Cu(phen^P) will be able to combine rapidly in an unconstrained manner to form a Cu₂O₂(phen^P)₂ complex. This would lead to the observed 1^st order dependence on the Cu coverage if all reduction and protonation steps beyond the peroxide level intermediate are fast.

In conclusion, controlling the coverage of a covalently attached electrocatalyst allows for detailed mechanistic study of electrocatalytic processes. Here, this method yields evidence that the electrocatalytic 4-electron reduction of O₂ to water by adsorbed discrete Cu catalysts requires two proximal Cu^I sites to coordinate and efficiently reduce O₂. A general mechanism is proposed for electrocatalytic O₂ reduction by adsorbed Cu catalysts that is consistent with both the observed 2^nd order rate dependence on Cu coverage for covalently-immobilized Cu(phen) and the observed 1^st order rate dependence on Cu coverage for physisorbed Cu(phen). This study suggests that having a polynuclear assembly is the key to attain 4-electron reduction of O₂ at low overpotentials.