Machine Learning Enabled High-Throughput Screening of 2D Ultrawide Bandgap Semiconductors for Flexible Resistive Materials

Abstract

The 2D ultrawide bandgap (UWBG) semiconductors have attracted great attentions for the next generation of electronics and optoelectronics, owing to their superiority on material flexibility, device stability, and power consumption. However, few 2D UWBG semiconductors have been discovered, impeding their prosperous developments and widespread applications. Here, a high-throughput workflow is constructed to screen 2D UWBG semiconductors assisted by machine learning, and 507 potential candidates are obtained. Moreover, by learning, predicting, and screening Young's modulus and Poisson's ratio, 31 flexible 2D UWBG semiconductors are identified. Then the generation and the diffusion of anion vacancies, as well as the corresponding electronic properties are investigated by using the first-principles calculations, and 3 of them are demonstrated as the most promising candidates for the flexible resistive materials. The facile interface tunneling and the increased material conductance caused by the anion vacancies will contribute to the transition from high resistive state to low resistive state. This work provides an efficient high-throughput screening protocol to enrich the family of 2D UWBG semiconductors and is expected to foster their practical applications.