There is a lot of discussion and little agreement about the mechanisms for the oxygen reduction/oxide ion oxidation reaction ceramic electrodes in solid oxide cells, such as perovskite electrodes like the lanthanum strontium manganate – yttria stabilized zirconia composite electrode or mixed ion and electron conducting electrodes like (La_1-xSr_x)_sCoO_3-d (LSC). Many mechanisms with similarities such as electron transfer being rate determining and involving surface oxide vacancies have been proposed. Often the proposals were based on studies of oxygen isotope (¹⁶O – ¹⁸O) exchange at perovskite and fluorite structured metal oxide surfaces, which actually is not an electrically driven process. Even though oxygen exchange - on a non-polarized surface of a ceramic that is not electrically connected to an electrolyte - has electrochemical character, it does not necessarily behave the same way as an electrochemically polarized electrode. Furthermore, most proposed mechanisms assumed that the oxygen electrode reaction took place on a real perovskite surface, but during recent years, it has become generally accepted that perovskites and many other Ruddlesden-Popper phases have a thin (few nanometer) and continuous surface layer of A-site oxides, e.g. SrO and La₂O₃in the case of the popular mixed conducting LSC. This layer is observed to increasingly impede the kinetics of the reversible oxygen electrode as the AO layer grows thicker.

As the electrode reaction does not stop, but just decreases with increasing SrO thickness, we take these literature reports as an indication that the electrode reaction takes place through an AO layer even though the AO bulk phase is both ion and electron insulating. Our hypothesis is that a thin AO layer, like a 1 nm thick SrO layer, is electron and ion conducting due to space charge layers both on the oxygen gas side and on the perovskite side of it. Thus, the SrO layer will probably contain O^- and O₂^-ions, which act as electron holes, and request oxygen vacancies, which enables the oxide ion conductivity. This otherwise insulating layer then becomes a mixed conductor. We further hypothesize that the actual specific conductivity by oxygen ions as well as by electrons will be very sensitively dependent on the oxygen vacancy concentration in, and the electron Fermi level of, the perovskite layer.

Apart from the change in defect chemistry with polarization of the perovskite electrode, the SrO layer will also change stress state because of stoichiometric expansion/contraction of the perovskite due to the change in the perovskite stoichiometry with polarization. Straining of the SrO lattice may also change its conductivity.

On this background, another rate limiting process is suggested: 

                                              O₂(gas) + 2 e^-(surface) → 2 O^-(ads,SrO)                    fast                   (1)

                                         O^-(ads, SrO) + V_O(surface) → O^-(SrO,subsurface)            slow                  (2)

                                                              O^-(SrO) + e^- → O^2-(LSC)                           fast                   (3)

This mechanism is sketched in the Figure. The slow rate determining process of eq. 2 is not limited by slow electron transfer and it is not a simple ion transfer, but rather a kind of diffusion or conduction process through the thin segregated SrO surface layer.

This hypothesis seems to be able to qualitatively explain a number of otherwise puzzling observations reported in the literature. This will be explained in the paper together with more details.

The Figure shows a much simplified sketch of the proposed mechanism for the (La_0.6Sr_0.4)_0.98CoO_3-d oxygen electrode. The electrons may meet the O^- - ions inside the SrO layer. V_O means oxide vacancies and is not attempting to account for the charge imbalance.