Multiscale Neural-Network Quantum Molecular Dynamics and Molecular Mechanics for Polar Topological Structures

Molecular dynamics (MD) simulation based on neural network force field has been used in a wide variety of materials applications with great successes to date. However, it is known that even a well-trained machine learning model suffers from fidelity problems because of the lack of built-in physics, such as irreproducibility of simple constitutive relationship and unphysical sensitivities to external stimuli. This poses serious challenges to investigate recently discovered polar topological structures (e.g., skyrmions and merons) in ferroelectric/paraelectric heterostructures, in which a robust theoretical framework to describe complex interplay between photoexcitation, electric field and mechanical strain is crucial. We address this problem using a multiscale neural-network quantum MD and molecular mechanics (NNQMD/MM) approach, where NNQMD describes photoinduced topological dynamics in ferroelectric materials embedded in MM to accurately describe mechanical properties and stress/strain fields. Our NNQMD/MM scheme allows to study light, electric-field and strain control of topological structures in ferroelectric heterostructures. In this talk, we will discuss the NNQMD/MM scheme and its applicability for opto-electro-mechanical control of ultrafast, ultralow-power polar topotronic device simulations.<br/><br/>This work was supported as part of the Computational Materials Sciences Program funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, under Award Number DE-SC001460. Simulations were performed at the Argonne Leadership Computing Facility under the DOE INCITE and Aurora Early Science programs and at the Center for Advanced Research Computing of the University of Southern California.