Studies of 2D Materials Beyond Graphene: From First-Principles to Machine Learning Approaches

Speaker: Paul Hanakata, Boston University, Physics Department

When: March 14, 2019 (Thu), 01:30PM to 02:30PM (add to my calendar)
Location: PRB 595

This event is part of the Departmental Seminars.

Dissertation Committee: David Campbell,Harold Park, Pankaj Mehta, Anna Swan, So-Young Pi

Abstract: Monolayers and heterostructures of 2D electronic materials offer promise for observing many novel physical effects. Most theoretical studies of these 2D materials are based on quantum mechanical simulations described by the density functional theory (DFT). This approach however is limited to small simulation sizes (several nanometers), and thus inhomogeneous strain and boundary effects that often dominate the behaviors observed experimentally cannot be simulated within a reasonable time. In the first part of the talk, I will present a tight-binding model using continuum mechanics, validated by DFT calculations, to predict strain-dependent changes in the spin and electronic properties of ferroelectric Rashba materials. In the second part of the talk I will introduce a machine learning approach based on convolutional neural network to effectively search optimal designs in an exponentially large design space. With this machine learning approach, one is able to search for the most stretchable graphene kirigami design with 1000 training data in a design space of size 4,000,000. The continuum and machine learning methods I have developed are important for studies and designs of graphene and other 2D materials.