Perovskite Oxides for Oxygen Evolution Reaction

Photoelectrodes for efficient solar driven OER have to fulfill the following requirements:

(i) Semiconductor photoanodes with small band gaps < 2 eV are advantageous for efficient harvesting of sunlight since the solar spectrum has maximum intensity in the visible light range at ∼2.4 eV. Photoelectrodes like TiO₂, WO₃ and SrTiO₃ which are commonly used because of their high corrosion stability reveal band gaps of E_g > 3 eV. [71, 72] Thus, their performance is limited by the optical absorption which is less than 10% of the sunlight. (ii) Photo-excited carriers with a long lifetime and high transport mobilities are desired in order to achieve high transfer rates to the electrode/electrolyte interface before they are trapped or recombined. (iii) The electronic structure of the photoelectrode in its active state has to fit the respective redox potentials for enabling the transfer of the excited carriers across the interface (see Fig. 1.1). For OER involving the transfer of four excited hole carriers, this requires that the UVBE lies energetically

below the water oxidation potential. The overall water splitting reaction requires a thermodynamic potential energy of 4.92 eV which implies a minimum energy of 1.23 eV per hole transfer (in case of an ideal catalyst) and limits the band gap of the photoelectrode to E_g > 1.23 eV. However, experiments as well as potential energy calculations of the individual reaction steps of the OER cycle have shown that a certain overpotential η is required for driving the OER. [3, 24, 25]

The overpotential is defined by η=V −V₀, i.e. the difference between the experimentally observed potentialV and the thermodynamically required potentialV₀. The overpotential describes the kinetic inhibition of the OER, e.g. due to insufficient diffusion of electrolyte reactants and products, retarded adsorption and electron transfer or formation of stable intermediates. [73] Hence, the required overpotential η must be provided either by the band gap of the photoelectrode or by an applied external electric potential.

The overpotential required for driving OER at a certain electrode surface is commonly used as a measure of the electrode performance, i.e., low η are desired for efficient OER. Typical values are η ∼ 0.4 eV for a current flow of j ∼1 mA/cm² across the electrode/electrolyte interface. Another quantity, which is frequently used to characterize the OER activity, is the exchange current density j₀. It describes the charge transfer rates at the at the thermodynamic OER potential V₀. High exchange currents constitute high-performance electrodes for OER. Both quantities, η and j₀, can be measured by conventional cyclovoltammetry. A detailed description of this method can be found in [73].

For perovskite oxide electrodes in alkaline electrolyte (pH > 7) the B-site is usually con-sidered as the active surface site for OER, where the adsorption of OH-molecules leads to a reduction of symmetry breaking at the surface (see Fig 1.3a). On the basis of the redox-active B-site, the 4-step OER cycle has been elucidated in several experimental and theo-retical studies of different doped and undoped perovskite oxide electrodes.⁶ For instance, early investigations by Bockris et al. in 1984 combine experimental cyclovoltammetry with theoretical calculations of the potential energy of the individual reaction intermedi-ates. [3] Recently, density functional theory (DFT) calculations have been performed in order to analyze OER at well established oxide catalysts (RuO₂, IrO₂ and TiO₂) by Ross-meisl et al. [24] and perovskite oxide catalysts by Man et al. [25] The experimental work by Suntivich et al. covers numerous doped and undoped perovskite compounds including manganites. [2] The key objective of these studies is to identify a universal descriptor allowing for prediction of the catalytic activity of a certain perovskite oxide electrode.

The individual reaction steps of the OER cycle involve the formation and subsequent reactions of {O,H}-intermediates, which are bound to the B-site with different bonding strengths. Accordingly, the energy barriers governing the individual reaction steps differ from the average thermodynamic potential of 1.23 eV per hole transfer. The step with the highest energy barrier (the rate-determining step) determines the required reaction overpotential and the minimum band gap of the photoelectrode for OER.

6These works consider electrochemical OER where ground state hole carriers are transferred from the conduction band (in solar driven OER, optically excited holes are transferred from the valence band)

1.2 Scientific Background

FIG. 1.3 a) Scheme of an OH adsorbate on a perovskite surface assuming the B-site as the redox-active site. According to molecular orbital theory the bonding of an OH molecule to a surface B-site is energetically favored because of the symmetry enhancement due to the recovery of the (bulk-like) BO6

octahedral configuration instead of a BO5 pyramidal surface configuration. b) Volcano-shaped trend of the overpotential as an indicator for oxygen evolution activity taken from Ref. [2].

In the studies mentioned above a strong correlation between the strength of B-{O,H}

bond and the OER activity as well as the electrode stability has been observed. [2,3,24,25]

The OER activity as a function of the bonding strength between {O,H}-species and the B-site exhibits a volcano-shaped curve: In case of weak B-O bonding the {O,H}-adsorption and bonding to the surface is the rate-determining step. Concomitantly, the thermodynamic stability of the perovskite lattice decreases with decreasing B-O-bond strength and the electrode becomes susceptible to corrosion. [3] On the other hand, too strong B-O bonding forms a barrier for subsequent reaction steps and hinders the desorption of the product. On the basis of molecular orbital theory, Bockris et al. relate the B-O bond strength to the occupation of anti-bonding d states, i.e. the higher the occupation of anti-bonding states, the lower the bonding strength. Advancing the work by Bockris et al., Suntivich et al. consider the enhanced influence of the 3d−e_g states⁷ due to their σ-bonding to the oxygen anions. They found that the occupation of the e_g orbitals represents a universal descriptor for OER activity with highest activity at an e_g occupation close to unity (Fig. 1.3b).

In the OER reaction cycle proposed by Rossmeisl et al. and Man et al.

and adopted by Suntivich et al. the sequential formation of diverse B-{O,H} intermediates involves changes of the oxidation state of the B-site, e.g.

B^(m+1)+ −O²⁻ +OH⁻ → B^(m)+ −OOH⁻+e⁻. [2, 24, 25] This requires the respective flexibility of the electronic surface structure of the electrode without leading to electrode degradation.

7Crystal field splitting in an octahedral metal-ligand configuration leads to a lift of degeneracy of thed states and an electronic level splitting intoπ-typet2g andσ-typeeg electronic orbitals.

Although Bockris and Suntivich consider the surface B-ions as the active sites, they recognize a correlation between the OER activity and the degree of covalency in the B-O bond, i.e. a higher contribution of O 2p states to the UVBE enhances OER activity.

Recent results by Mueller et al. even reveal that oxygen anions are the active surface sites for OER in many transition metal oxides. The authors suggest that oxygen vacancy formation and incorporation play a key role for OER activity and electrode stability. [29]

1.2.3 In Situ Environmental Transmission Electron Microscopy