Effect of surfaces, domain walls and grain boundaries on ferroelectricity in lead titanate using atomic scale simulations

Ultrathin ferroelectric (FE) films have potential application in non-volatile random access memories and microelectromechanical systems. Miniaturization of such devices demands the establishment of the thickness-property relationship, a detailed understanding of the surface termination, and the effects of domain walls and grain boundaries. Density functional theory (DFT) and molecular dynamics (MD) methods are used to investigate these effects in lead titanate.;The surface effects are characterized for the (100) terminations with open circuit electrical boundary condition. The resulting surface energies indicate that the surface terminations with polarization out-of-the-surface are more stable than the cases where polarization occur into-the-surface. Analysis of atomic relaxation, surface rumpling and interlayer distance provide insight into the surface effects for each termination including polarization.;The effect of domain walls are carried out for (100) and (110) 180 degree domains. The (100) Pb-centered and (110) OO-centered domain walls are predicted to be the favorable structure with comparable energies. The examination of polarization variation across the domain wall results in the presence of in-plane polarization, which develops a small polarization rotation analogous to that observed experimentally in ferromagnetic materials. The in-plane polarization is perpendicular to the domain wall and points away from the wall. The effect of grain boundaries on polarization are calculated for sigma 5 coincident site lattice (310) tilt boundaries in strontium titanate and lead titanate. The reverse twinned structures are found to be energetically more favorable than the regular twins. The presence of the boundary resulted in a local polarization around the boundary (normal and parallel to the boundary), which vanishes away from the boundary.;This study also focuses on the challenge to simulate complex ferroelectric structures. As a first step towards achieving the goal, the existing parallel molecular dynamics code was modified to include the core-shell model, thereby allowing complex ferroelectric systems to be simulated.