Machine learning aided RHEED analysis of CrSb thin film growth

Reflection high energy electron diffraction (RHEED) is an in-situ technique used during molecular beam epitaxy (MBE) to understand how the film is depositing during the growth process. RHEED is a powerful, cost-effective tool for informing thin film growth if used to its full potential. RHEED can elucidate information about the surface topography, lattice constants, growth modes, and in some cases even atomic positions in the unit cell. Commonly however, RHEED is only used for surface-level quality checks of film growth, under-utilizing the technique. Previous studies suggest that machine learning (ML) could be a promising tool for detecting changes in the RHEED pattern that are imperceptible to the human eye, allowing for the opportunity to autonomously tune parameters during growth to achieve the desired results, instead of relying on ex-situ techniques to observe film quality after growth. ML-enabled RHEED could unlock the full potential of RHEED, aiding in the development of better films by providing more immediate feedback and readily accessible information. We employ some of these available RHEED ML tools to CrSb thin film growths to investigate if ML is a viable technique for improving film quality.