Deep Learning Method to Accelerate Discovery of Effective Doping in Sr_1-xR_xFe_1-yM_yO_3-δ Oxygen Carriers for Chemical Looping Air Separation

Design and discovery of high-performance oxygen carrier materials play a crucial role in the energy applications (i.e. oxide fuel cells, cleaner fossil fuel combustion, etc.). Perovskite-type ABO_3-δ oxides as oxygen carriers are receiving much attention recently due to their high thermal stability, good mechanical properties, and ability to reversibly and rapidly uptake and release oxygen. Furthermore, the flexibility in choosing the elemental composition of the A and B sites allows for the synthesis of many different perovskite-structured materials with inherently distinct oxygen storage properties. SrFeO₃ is a promising oxygen carrying material due to its effectiveness and the low cost of iron. In the current work, the dopant chemical space in Sr_1-xR_xFe_1-yM_yO_3-δ (R= La, K, Rb, Cs, Ca, Ba, Pd, Cu, Ag, Au, Cd, Hg, Tl, Pb, M= Co, Ni, Mn, Mo, Ti, Cu, Zn, x, y = 0, 0.125, 0.25, 0.375, 0.5, and δ = 0, 0.0625) is systematically explored using density functional theory (DFT) computations in combination with machine learning (ML) methods. We study a range of cationic dopants including alkali, alkaline earth metals, 3d, 4d, and 5d transition metal elements with and without an adjacent O vacancy. The effect of A and B site doping, both individually and in combination, on the oxygen ion diffusion considering oxygen vacancy formation energy is investigated utilizing DFT calculations. A linear programming approach is used to determine the energetically most favorable decomposition pathway (or products) and the corresponding decomposition energy. The dopants are then assessed based on the resistance of the doped oxide to decomposition, the tendency of O vacancy formation, and the site preference based on the decomposition energy. At the end, a predictive machine learning (ML) model is developed based on the data from DFT calculations and experiments for rational materials design and discovery by establishing a relationship between dopant features and the oxygen formation energy.