Defect Modeling and Device Optimization in Chalcogenide Photovoltaics from First Principles to AI-assisted Design

Abstract

This dissertation presents a comprehensive framework for defect modeling and device optimization in chalcogenide photovoltaics, focusing on Cu(In,Ga)Se2 (CIGS) and CdSeTe materials. By integrating first principles density functional theory (DFT), continuum modeling, and AI-driven methods, a robust predictive model for electronic performance and optimization of thin-film solar cells has been developed. Detailed analyses of defect formation and diffusion mechanisms were conducted using DFT calculations, enabling precise control over defect profiles to optimize material properties and enhance device performance. Continuum modeling and device modeling were employed to bridge atomic-scale defect properties with macroscopic device behavior, successfully predicting the impact of defect distributions on carrier lifetimes, providing insights into manufacturing process impacts and solar cell performance optimization. The developed predictive Technology Computer Aided Design (TCAD) model offers a detailed depiction of defect behaviors and their impact on device performance, serving as a powerful tool for guiding experimental efforts and advancing the development of high-efficiency thin-film solar cells. The integration of AI-driven methods positively influences TCAD simulations by employing machine learning (ML) for defect property prediction, thereby accelerating technology advancement. A framework powered by ML and DFT has been developed for predicting and screening functional impurities in semiconductors. This framework can be further implemented into the TCAD model to explore new dopants in solar cells and other electronic devices.