Speaker
Description
This talk presents our progress in developing autonomous workflows that generate DFT data and fit both reactive and machine-learned force fields to guide the molecular beam epitaxial synthesis and characterization of the van der Waals–layered topological insulator Bi₂Se₃. I will introduce pyRMG, our Python package for high-throughput real-space multigrid (RMG) DFT calculations on Frontier, and show how it can be used to explore the potential-energy surface and electronic structures of Bi₂Se₃ films with variable twist angles on NbSe₂ superconductor substrates. Next, I’ll describe our fitting approaches for ReaxFF and ML (MACE) force fields for the Bi-Se system, and how we've applied Bayesian optimization to navigate the complex MD parameter space for targeted synthesis (e.g., recrystallization pathways). Finally, I’ll outline future directions for extending these highly-integrated computational methods to other low-dimensional topological materials and compositional systems.
| Topical Area | Hard matter: quantum, electronic, semiconducting materials |
|---|
