Inverse design of semiconductor materials with deep generative models

Abstract

To generate thermodynamically stable semiconductor materials, a generative inverse design framework that integrates VAE, GAN, and DDPM along with chemical validity testing and template-based structure prediction (TSP) was proposed.