Prediction of oxygen flux in Mixed Ionic-Electronic oxygen Conducting mem-branes.
Résumé
Mixed ionic-electronic oxygen conducting (MIEC) ma-terials are an important class of materials which found applications in high temperature catalysis and separation such as cathode for SOFCs and oxygen separation mem-branes. Among MIEC ceramic membranes, perovskites and related structures are conductors which exhibit highest oxygen semi-permeability. The mechanism of ionic oxygen diffusion originates from oxygen vacancies which are pre-sent in the structure (d) and relatively low energetic diffu-sion barrier for oxygen mobility. Under a concentration gradient, anionic oxygen can diffuse from the highest (PO2) to lowest compartment (PO2). Although the diffusion is well understood at molecular level, the quantitative prediction of oxygen permeability of a dense MIEC membrane is a remaining scientific challenge. This lack of understanding arises partly from grain boundaries in which the diffusion constants can be hardly determined. More importantly, the vacancy gradient across the membrane which results of the oxygen activities at both interfaces can be hardly quanti-fied; the latest activities depending on dissociative adsorp-tion and recombination desorption rates at the air and sweep sides, respectively. The lack of knowledge of diffu-sion in grain boundaries and the interlocked mechanisms of oxygen transport in the bulk and at the interfaces are scientific hurdles which have prevented the prediction of oxygen flow starting from data obtained from powders. Among attempts to establish Structure-Flux properties, Ullmann, et al. [1] observed a linear correlation of the Thermal Expansion Coefficient (TEC) and the oxygen per-meability on a series of 23 different AABBO3-d pervokistes membranes. However, this empirical trend does not hold for materials with other structures (such as LaSrGaMgO3). On a more physic-chemical basis, the oxygen permeability can be modeling by the Wagner equation which takes into account the diffusion coefficient of O2- as sole material intrinsic descriptors. This equation is not valid [2] when one of the surface steps is the rate determining step. Sunsaro, et al. [3] compared all the different models. The Xu and Thomson model accounts for surface reaction by introduc-ing k/Dv ratio in the Wagner equation, where kr and kf are the oxygen coefficient exchange and Dv the diffusion coef-ficient. The model assumes that kr/kf and Dv are constant across the membrane which is cannot be strictly the case since the re is an oxygen/vacancy gradient. In addition, k/Dv is usually measured from SIMS profile of membrane cross section which does not allowed to unravel surface ex-change rate from diffusion constant. As a result, Xu and Thomson model fail to predict oxygen permeability of MIEC membranes. The objective of this work is the development of a method which enables to predict oxygen permeability of MIEC membrane with kinetic parameters determined from powders. In contrast to the Xu & Thompson model, the developed model accounts for different surface exchanges and diffusion depending on the partial pressure of oxygen. The main originality relies on the process developed for the measurements of oxygen surface exchange coefficients and diffusion coefficient for different oxygen partial pressure by 18O/16O exchanged on powders. The materials Ba0.5Sr0.5Co0.8Fe0.2O3 (BSCFO)[4], Ba0.95La0.05FeO3-d (BLFO)[5] have been selected as first case studies.