Ultra-fast flash light revealing mystery of atoms, into the world of attoseconds (10-18 second) Beyond the femtosecond (10-15 second) barrier

Ferroelectric RAM (FeRAM) is a type of non-volatile computer memory, similar to dynamic random access memory (DRAM) which can be found in majority of our computers. FeRAM uses a ferroelectric layer to achieve non-volatility and offers a number of advantages from lower power usage to faster writing speeds.
Ferroelectrics such as PbTiO3 have usually two stable oppositely-polarized states without external electric fields, and the negatively-polarized state can be reversed to the positively polarized state by applying positive external electric field, and vice versa. Since the polarized states remain stable in zero electric fields, this material has been known as one of the candidates for the application to the non-volatile random access memory (NVRAM). Storing a data bit in this material means increasing the size of one polar region at the expense of another, hence the movement of a domain wall separating the two oppositely-polarized states. This polarization reversal process on the ferroelectric domain wall was modeled by Miller and Weinreich based on the nucleation and growth process in 1960. However, their model couldn’t explain the low activation energy for the domain-wall motion which was obtained from recent experiments.

Professor Young-Han Shin, together with a research team at University of Pennsylvania began investigating ferroelectric domain-wall motion using theoretical tools (density-functional theory, molecular dynamics, kinetic Monte Carlo, and Landau-Ginzburg-Devonshire theory) in 2005. The team succeeded in developing a classical potential model for perovskite ferroelectrics, and this model potential made it possible to simulate the ferroelectric dynamics in a longer-time and larger-size scale.

Since the research on the ferroelectric domain-wall motion requires large-scale calculations, such a multi-scale approach is highly demanded. The team illustrated that the ferroelectric domainwall motion could be successfully explained using the nucleation and growth model. Molecular dynamics simulations based on the first principles calculations were performed for this purpose. They found that a critical nucleus is formed on the domain wall, and its size and the activation energy are smaller than the estimates by Miller and einrech. Nucleation and growth rates were obtained from the molecular dynamics simulations. They suggested a model to describe the critical nucleus on the ferroelectric domain wall using the Landau-Ginzburg-Devonshire theory. The main difference between the group’s model and Miller-Weinreich model is the diffused boundary around the critical nucleus on the ferroelectric domain wall.

They found that the size and the activation energy from their model are closer to molecular dynamics results.
The team performed the kinetic Monte Carlo simulations using nucleation and growth rates obtained from the molecular dynamics simulations. Overall domain-wall speeds at various temperatures and external fields were compared to the experimental values. The calculated speed is an intrinsic domain property without any defect with the upper bound of experiments where the defect is hardly avoided.

The research was published in October issue of Nature entitled, “Nucleation and growth mechanism of ferroelectric domain-wall motion”.

Professor Young-Han Shin
Department of Materials Science and Engineering
Center of Futuristic Material-Systems
Tel: +82-54-279-2795
Fax: +82-54-279-2399
Email: yhshin@postech.ac.kr