First-principles theory, coarse-grained models, and simulations of ferroelectrics.

CONSPECTUS

A ferroelectric crystal exhibits macroscopic electric dipole or polarization arising from spontaneous ordering of its atomic-scale dipoles that breaks inversion symmetry. Changes in applied pressure or electric field generate changes in electric polarization in a ferroelectric, defining its piezoelectric and dielectric properties, respectively, which make it useful as an electromechanical sensor and actuator in a number of applications. In addition, a characteristic of a ferroelectric is the presence of domains or states with different symmetry equivalent orientations of spontaneous polarization that are switchable with large enough applied electric field, a nonlinear property that makes it useful for applications in nonvolatile memory devices. Central to these properties of a ferroelectric are the phase transitions it undergoes as a function of temperature that involve lowering of the symmetry of its high temperature centrosymmetric paraelectric phase. Ferroelectricity arises from a delicate balance between short and long-range interatomic interactions, and hence the resulting properties are quite sensitive to chemistry, strains, and electric charges associated with its interface with substrate and electrodes. First-principles density functional theoretical (DFT) calculations have been very effective in capturing this and predicting material and environment specific properties of ferroelectrics, leading to fundamental insights into origins of ferroelectricity in oxides and chalcogenides uncovering a precise picture of electronic hybridization, topology, and mechanisms. However, use of DFT in molecular dynamics for detailed prediction of ferroelectric phase transitions and associated temperature dependent properties has been limited due to large length and time scales of the processes involved. To this end, it is quite appealing to start with input from DFT calculations and construct material-specific models that are realistic yet simple for use in large-scale simulations while capturing the relevant microscopic interactions quantitatively. In this Account, we first summarize the insights obtained into chemical mechanisms of ferroelectricity using first-principles DFT calculations. We then discuss the principles of construction of first-principles model Hamiltonians for ferroelectric phase transitions in perovskite oxides, which involve coarse-graining in time domain by integrating out high frequency phonons. Molecular dynamics simulations of the resulting model are shown to give quantitative predictions of material-specific ferroelectric transition behavior in bulk as well as nanoscale ferroelectric structures. A free energy landscape obtained through coarse-graining in real-space provides deeper understanding of ferroelectric transitions, domains, and states with inhomogeneous order and points out the key role of microscopic coupling between phonons and strain. We conclude with a discussion of the multiscale modeling strategy elucidated here and its application to other materials such as shape memory alloys.