Spectroelectrochemical analysis of the mechanism of (photo)electrochemical hydrogen evolution at a catalytic interface

Abstract

Multi-electron heterogeneous catalysis is a pivotal element in the (photo)electrochemical generation of solar fuels. However, mechanistic studies of these systems are difficult to elucidate by means of electrochemical methods alone. Here we report a spectroelectrochemical analysis of hydrogen evolution on ruthenium oxide employed as an electrocatalyst and as part of a cuprous oxide-based photocathode. We use optical absorbance spectroscopy to quantify the densities of reduced ruthenium oxide species, and correlate these with current densities resulting from proton reduction. This enables us to compare directly the catalytic function of dark and light electrodes. We find that hydrogen evolution is second order in the density of active, doubly reduced species independent of whether these are generated by applied potential or light irradiation. Our observation of a second order rate law allows us to distinguish between the most common reaction paths and propose a mechanism involving the homolytic reductive elimination of hydrogen.

Figure 1. Schematic representation of the dark and light electrocatalytic systems.

(a) RuO_x/FTO cathode and (b) FTO/Cu₂O/AZO/TiO₂/RuO_x photocathode. The diagram is not scaled.

Figure 2. J-E behaviour of the dark electrocatalyst and the illuminated photocathode.

(a) J-E characteristics normalized for clarity at an applied potential where HER occurs in both systems and (b) J-E characteristics represented as log [−J or −J^ph] versus E. The samples were measured in a pH=5 phosphate-sulfate electrolyte.

Figure 3. Spectroelectrochemical spectra at different applied potentials.

Spectra showing the changes in absorbance (ΔA × 10⁻³) when RuO_x (a) acts as a dark electrocatalyst and (b) is part of a [Cu₂O] photocathode under irradiation. For the lower box in a, ΔA is plotted relative to the absorbance spectrum at open circuit potential (OCP=0.78 V); for the upper box, showing ΔA spectra at potentials negative of the hydrogen reduction current onset potential (E_onset=0 V versus RHE), the spectra are plotted relative to that at this onset potential. For the [Cu₂O]/RuO_x photocathode (b), at all potentials ΔA is plotted as the difference between light on and light off. For reference, the ΔA with respect to the absorbance at OCP and the simultaneously measured currents for all applied potentials are shown in Supplementary Figs 3 and 4. For both device types, the negative ΔA signals observed for E>E_onset are assigned to RuO_x reduction to RuO_x(−1), and the positive ΔA signals observed for E<E_onset are assigned to the further reduction to RuO_x (−2) species. For (b) the irradiation conditions were ∼5-6 s (25 s for back irradiation), 365 nm illumination (∼1.5 mW cm⁻²). See Methods for experimental details.

Figure 4. Correlation between the changes in absorbance and current.

The ΔA was calculated as an average of the optical signal between 850 and 900 nm and represented in a normalized scale for clarity. The steady state current density was calculated as the average current measured within 300–350 s (see Supplementary Fig. 4).

Figure 5. Absorbance and photocurrent changes under conditions of HER.

(a) Time-dependent photoinduced absorbance changes of the photocathode upon 365 nm light illumination at different photon fluxes (0.5–1.5 mW cm⁻²) at a fixed applied potential of 0.1 V_RHE and (b) the photocurrent measured simultaneously.

Figure 6. Relationship between (photo)current and absorbance changes.

(a) Current-absorbance characteristics of the RuO_x electrocatalyst represented as the log [−J] versus the absorbance of the catalytic RuO_x species at different applied potentials (−0.05 to −0.2 V versus RHE). (b) Photocurrent-absorbance characteristics of the photocathode represented as the log [−J^ph] versus the photoinduced absorbance of the catalytic RuO_x species generated at 0.1 V versus RHE upon 365 nm illumination at different photon fluxes (∼0.5 to 1.5 mW cm⁻²). The regression of log [−J] versus log [ΔA] yielded slopes of 2.1±0.1 (correlation coefficient, R=0.997) for the electrocatalyst. For the photocatalyst two sets of data are shown, collected with (empty circles, see Supplementary Fig. 8) and without (solid circles, see Fig. 5) the addition of 0.01 M surfactant (Tritron X-100) used to promote facile bubble release. The independent data sets have been normalized to be plotted in the same graph. The regression of log [−J^ph] versus log [ΔA] of each dataset yielded identical slopes of 1.9±0.1 (R=0.992).