Investigating oxygen reduction pathways on pristine SOFC cathode surfaces by in situ PLD impedance spectroscopy

The oxygen exchange reaction mechanism on truly pristine surfaces of SOFC cathode materials (La0.6Sr0.4CoO3−δ = LSC, La0.6Sr0.4FeO3−δ = LSF, (La0.6Sr0.4)0.98Pt0.02FeO3−δ = Pt:LSF, SrTi0.3Fe0.7O3−δ = STF, Pr0.1Ce0.9O2−δ = PCO and La0.6Sr0.4MnO3−δ = LSM) was investigated employing in situ impedance spectroscopy during pulsed laser deposition (i-PLD) over a wide temperature and p(O2) range. Besides demonstrating the often astonishing catalytic capabilities of the materials, it is possible to discuss the oxygen exchange reaction mechanism based on experiments on clean surfaces unaltered by external degradation processes. All investigated materials with at least moderate ionic conductivity (i.e. all except LSM) exhibit polarization resistances with very similar p(O2)- and T-dependences, mostly differing only in absolute value. In combination with non-equilibrium measurements under polarization and defect chemical model calculations, these results elucidate several aspects of the oxygen exchange reaction mechanism and refine the understanding of the role oxygen vacancies and electronic charge carriers play in the oxygen exchange reaction. It was found that a major part of the effective activation energy of the surface exchange reaction, which is observed during equilibrium measurements, originates from thermally activated charge carrier concentrations. Electrode polarization was therefore used to control defect concentrations and to extract concentration amended activation energies, which prove to be drastically different for oxygen incorporation and evolution (0.26 vs. 2.05 eV for LSF).


S.I.2 Effect of Impurity Contaminations on Reaction Rates
To assess the effect of contaminations on the analysis of reaction mechanisms, one LSC thin film was analyzed at 5·10 −6 mbar and 600 • C by x-ray photoelectron spectroscopy (XPS) while performing electrochemical impedance spectroscopy. The XPS measurement showed that no contaminations were present on the surface of the thin film and the resistance value lines up perfectly with the slope measured in-situ at low oxygen partial pressures. However, samples measured ex-situ (sulphur contamination on ex-situ samples confirmed previously by XPS [1]) not only exhibit vastly inferior oxygen exchange kinetics but also a different effective p(O 2 ) dependency [2]. As adsorption kinetics are usually p(O 2 ) dependent, we suspect that many measurements of the oxygen exchange kinetics are afflicted by significant uncertainty due to contaminations by impurities in the measurement atmosphere.

S.I.3 In-Situ Degradation Phenomena
In Fig. 2, the p(O 2 ) dependence of the area specific resistance is shown for measurements from low to high oxygen partial pressures and for measurements in the reverse direction. At high oxygen partial pressures, the resistance slowly increases with time for all materials except PCO, probably due to segregation of Sr from the crystal lattice to the surface. This segregation seems to be partially reversible as the resistance decreases slowly at lower oxygen partial pressures, explaining the slight deviation between the curves.

S.I.4 Bias impedance measurements
To extract the true potential drop at the working electrode, it is essential that the semicircles corresponding to the surface exchange resistance in parallel to the chemical capacitance of the working electrode thin film and the counter electrode respectively are well separated. At 0.01 mbar, this is only the case for anodic bias voltages and with increasing bias voltage, this separation is becoming even more pronounced. An exemplary impedance curve measured during the application of 200 mV anodic bias is shown in Fig. 3. In Fig. 4, current-overpotential curves recorded during bias impedance measurements are shown. During the measurements, bias was applied until the oxygen vacancy concentration reached a previously defined level (for anodic and cathodic direction respectively), the exact procedure is detailed in the main manuscript.

S.I.5 Defect Chemistry of SOFC cathode materials
Defect model of LSF The most commonly used defect model for bulk LSF was introduced by Mizusaki et al. [3,4] with oxygen vacancies V ·· O being the main ionic defect together with electrons Fe F e and electron holes Fe · F e as electronic charge carriers. The equations governing the defect chemistry of LSF are the overall oxygen exchange reaction, the electron-hole pair formation and the according mass action laws as well as charge neutrality and the Fe balance:

Defect model of PCO
In contrast to LSF and STF, PCO crystallizes in the fluorite structure and is usually considered a solid solution of praseodymium oxide and cerium oxide. Again the oxygen nonstoichiometry is governed by the oxygen exchange reaction which includes charge transfer from Pr atoms. In literature, the electronic structure of PCO is often described by introducing an additional Pr-impurity band which allows the migration of electronic species from one to a neighbouring Pr atom [7]. The defect chemistry of PCO is well explored and the governing equations and the according mass action laws are oxygen exchange, band-band excitation, praseodymium ionization, a simplified charge neutrality equation and mass balances:

S.I.6 p(O 2 ) dependencies of different reaction mechanisms
To better understand the reaction order with regard to p(O 2 ), it is necessary to incorporate the complex defect chemistry of the investigated materials into mechanistic considerations. In Fig. 7, a normalized reaction rate is plotted over p(O 2 ) for different proportionalities to selected defect species of LSF. One sees immediately that different defects partaking in or before the rate limiting step have qualitatively very different impacts on the p(O 2 ) dependency. For example, when atomic oxygen is partaking in a charge transfer process before or during the rate limiting step, the slope should not exceed 0.5, in contrast to reactions including molecular oxygen. Moreover, due to the complex defect chemistry, defects affect the p(O 2 ) dependency differently according to the investigated p(O 2 ) regime. Nevertheless, the qualitative behaviour of the reaction rate with regard to oxygen partial pressure, when observed over a wide p(O 2 ) range, gives detailed insight into the actual oxygen exchange reaction mechanism and the partaking defects. It is important to note, that when dealing with equilibrium reaction rates, both directions of the reaction must show the same p(O 2 ) dependence as the rate limiting step must be the same. Thus, every argument given in this study formulated for the oxygen incorporation reaction translates directly to the reverse direction, i.e. oxygen release, except when discussing non-equilibrium measurements.

S.I.7 Concentration dependent contributions to the effective activation energy
The effectively measured activation energy during equilibrium experiments is naturally a convolute of an actual activation barrier and concentration dependent contributions, originating from the defect chemistry of the material. Thus, activation energies derived from equilibrium measurements are generally not suitable to discuss the actual energetic activation of the oxygen exchange reaction and the defect chemical contributions must be considered in such discussions. For illustrative pur-poses, these defect chemical contributions are shown for the two exemplary rate determining steps of the main paper. Here, it is important to distinguish between incorporation and release reaction, as different species partake in the reaction.