Multi-scale Modeling of Complex Interfaces Aided by Machine Learning
SPRING 2020 RESEARCH INCUBATION AWARDEES
PI: Emily M. Ryan, Assistant Professor, Mechanical Engineering, ENG
Co-PIs: Sahar Sharifzadeh, Assistant Professor, Electrical & Computer Engineering and Materials Science & Engineering, ENG. Brian Kulis, Assistant Professor, Electrical & Computer Engineering, ENG
Track: Hariri Institute for Computing
What is the Challenge?
To investigate the complex interfaces in material systems and create a multi-scale modeling framework by studying interfaces in the presence of an electric field. This includes resolving the critical chemical-physical processes that occur at complex interfaces in materials systems that would allow a better understanding of the fundamental physics at the interface and the design of improved material systems.
What is the Solution?
Emily M. Ryan investigates complex interfaces in material systems. Co-PIs Ryan and Sharifzadeh are experts in materials science and computational modeling of complex systems at the meso- and micro-scales. Co-PI Kulis is an expert in state-of-the-art machine learning techniques and stochastic modeling of rare events. Together they will create a multi-scale modeling framework to study interfaces in the presence of an electric field. Resolving the critical chemical-physical processes that occur at complex interfaces in materials systems will allow a better understanding of the fundamental physics at the interface and the design of improved material systems.
What is the Process?
Ab initio modeling would be used to understand how heterogeneities on a surface affect the surface energies and free energies, which would inform mesoscale modeling of the interfacial region to study how these heterogeneities affect the functionality of the interface when taking local mass and charge transport near the interface into account. Machine learning principles would couple the ab initio atomic scale to the meso-scale using physically informed machine learning models developed for sparse data.