IMF2017

I will discuss a machine learning guided density functional theory (DFT) approach to accelerate the search for noncentrosymmetric (NCS) materials in the layered n = 1 Ruddlesden-Popper (RP) oxides. Our approach is built on the foundations of applied group theory, machine learning and DFT to uncover quantitative symmetry-chemistry guidelines for rapidly predicting novel compositions with NCS ground state. Group theory identifies how configurations of oxygen octahedral rotation patterns, ordered cation arrangements and their interplay break inversion symmetry, while machine learning tools allow us to learn from available data to predict candidate compositions that fulfil the group theoretical postulates. Finally, we validate the machine learning predictions using DFT phonon calculations and identify novel compositions (e.g. stannates, ruthenates etc) that show potential for NCS ground state. Our approach enables rational design of new materials with targeted crystal symmetries and functionalities.