Classification of in situ reflection high energy electron diffraction images by principal component analysis

Abstract

The reflection high-energy electron diffraction (RHEED) method is widely used for the in situ observation of molecular beam epitaxy (MBE). This is because the RHEED pattern dynamically changes according to the growth conditions, such as surface temperature and material supply. However, to date, the RHEED pattern has been categorized and recognized based on the experience of the researcher. In this study, we investigated the classification of RHEED pattern datasets without using labeling by the principal component analysis method that reduces the dimensionality of the data. The RHEED images were successfully classified during the MBE growth of GaAs, demonstrating that unsupervised learning can be used to recognize RHEED patterns.