Machine Learning for Multi-Functional Kirigami Metamaterials

SPRING 2018 RESEARCH INCUBATION AWARDEE

PI: David Campbell, Professor, Physics, CAS

What is the Challenge?
Atomically thin two-dimensional (2D) materials such as graphene and MoS2 have been studied extensively due to their exceptional physical properties (mechanical strength, electrical and thermal conductivity, etc.). Recently, the PIs were among the first to investigate “kirigami”-based 2D metamaterials. These are made by introducing arrays of cuts into a sheet of the 2D material. However, there is currently no theory that explains the mechanisms by which such high stretchability is enabled in kirigami. It also doesn’t explain or provide guidance on the optimal cutting strategy to achieve a given result.

What is the Solution?
If successful, the usage of ML techniques will not only efficiently give us the optimal design for graphene kirigami metamaterials. However, it will also provide new mechanistic insights into the kirigami structures that govern novel metamaterial properties, including high-stretchability, and novel electromechanical coupling mechanisms.

What is the Process?
We will use the Sandia open-source molecular dynamics (MD) simulation code LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator) to generate the ground truth data for our training model, where we will use graphene as the 2D constituent material of choice for the kirigami.