Advancing from phenomenological to predictive theory of ferroelectric oxide solution properties through consideration of domain walls

Prediction of properties from composition is a fundamental goal of materials science and can greatly accelerate development of functional materials. It is particularly relevant for ferroelectric perovskite solid solutions where compositional variation is a primary tool for materials design. To advance beyond the commonly used Landau-Ginzburg-Devonshire and density functional theory methods that despite their power are not predictive, we elucidate the key interactions that govern ferroelectrics using 5-atom bulk unit cells and non-ground-state defect-like ferroelectric domain walls as a simple as possible but not simpler model systems. We also develop a theory relating properties at several different length scales that provides a unified framework for the prediction of ferroelectric, antiferroelectric and ferroelectric phase stabilities and the key transition temperature, coercive field and polarization properties from composition. The elucidated physically meaningful relationships enable rapid identification of promising piezoelectric and dielectric materials.