(189aa) Predicting Point Defect Concentrations in Complex, Disordered Oxides

Authors: 
Millican, S. L., University of Colorado Boulder
Deml, A. M., National Renewable Energy Laboratory
Weimer, A. W., University of Colorado Boulder
Holder, A. M., University of Colorado at Boulder
Stevanovic, V., National Renewable Energy Laboratory
Musgrave, C. B., University of Colorado Boulder
Materials-by-design in rapidly becoming a standard in materials science, driven by substantial advances in computational capability and improved integration of theory and experiment. In order to continue to advance the materials design and discover, computation must continue to drive towards experimental reality through incorporation of finite temperature effects at relevant operating and/or synthesis conditions. Chemically and structurally complex solid compounds, including those with magnetic and atomic disorder, are rapidly extending new material functionalities and device performance across applications ranging from ceramic fuel cells and thermochemical redox cycles to membrane separations and sensors. Accelerated development of these complex compounds requires accurate predictions of material properties including effective defect formation energies and equilibrium defect concentrations at finite temperatures. Traditional first principles approaches employ periodic models and systematically examine relatively ordered atomic structures to identify the lowest energy defect sites are rarely suitable for describing the disorder and combinatorial scope of complex compounds. In this work we demonstrate a new strategy that incorporates atomic and magnetic disorder, ensemble descriptions of defect sites and formation energies, and effects beyond the dilute defect limit to predict the temperature dependence of oxygen vacancy concentrations in complex perovskite oxides, including Ba0.5Sr0.5Fe0.8Zn0.2O3 (BSFZ), Ba0.5Sr0.5Co0.8Fe0.2O3 (BSCF), and BaCo1-x-y-zFexZryYzO3 (BCFZY). We compare our thermodynamic predictions against experimental data to validate our methods.