CDS&E/Collaborative Research: Machine Learning-Enabled Electronic-to-Mesoscale Modeling Framework for Analyzing Defect Dynamics in Semiconductors under Light

Iowa State University, Ames IA

Investigators

Abstract

Recent advancements in data science and machine learning have enabled the resolution of complex problems in science and engineering, for instance, to explain why and how photoplasticity occurs in semiconductors. Photoplasticity is a phenomenon where light exposure can cause materials to harden or soften. Although known since 1957, the mechanisms behind it remain unclear due to the lack of high-fidelity tools that can address the full complexity of a light-matter interaction. To meet this need, this Computational and Data-Enabled Science and Engineering (CDS&E) award supports the development of a multiscale simulation tool that fuses quantum, atomistic, and mesoscale models together through machine learning. The goal is to establish simulation tool and produce database that can advance our understanding on how semiconductors deform under light exposure. This research will promote progress in science and benefit industry technologies in flexible electronics, deformable solar panels, and photocatalysts. Additionally, this project will equip the next-generation workforce with a broad range of knowledge and skills in data science, machine learning, and computational mechanics. This project aims to perform research that attempts to establish an electronic-to-mesoscale simulation tool for predicting the deformation behavior of semiconductors under light. By choosing zinc sulfide as an example, at the electronic level, a large-scale dataset will be first created from systematic constrained density functional theory calculations. This dataset will then be consolidated into light illumination-affected interatomic force fields through machine learning. To scale up in length, such machine learning-based force fields will be informed into nanoscale molecular dynamics (MD) and mesoscale coarse-grained (CG) models to simulate the interactions between light and nanometer-/micrometer-long dislocations, which are line defects and main carriers of plastic deformation in semiconductors. To get one step closer to experimental conditions, the dislocation mobility laws and plastic flow rules will be synthesized from the MD and CG simulation data, which can be used to interpret the photoplasticity observed in laboratory tests. The resulting multiscale computational framework and knowledge database are general. They can be applied not only to photo-plasticity, but also to electro- and chemo-plasticity various materials and devices such as optical sensors, solar cells, photocatalysts, where the interaction between electrons and dislocations become important but is difficult to probe in experiments. This project is jointly funded by CMMI and the Established Program to Stimulate Competitive Research (EPSCoR). This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.