Atomic-scale statistical analysis of interface structure and stability in Ga₂O₃-based power electronic devices

Interface stability, or lack thereof, can have enormous consequences on the performance of UWBG-based power electronic devices, which must operate reliably at high temperature and voltage. Such conditions can cause the growth or evolution of interphase boundary layers at thin film interfaces, which impact device performance. Characterization of these interphase layers via scanning transmission electron microscopy (STEM) can elucidate the atomic structure and chemistry of these defect layers, giving detailed insight into how and why they form. Traditionally, however, these analyses focus on relatively narrow fields of view. Thus, they gather only limited statistical information and may miss sparse features. Here, we use atomic-resolution STEM imaging and spectroscopy, in combination with machine learning-assisted image segmentation, to quantify interfacial defect structure distributions across hundreds of nanometers of NiO/Ga₂O₃ interfaces, which are of interest for high-power applications. We compare interphase layer evolution due to thermal annealing for multiple Ga₂O₃ interface orientations. We thus extract insights into the prevalence and relative stabilities of particular interphase layer structures and point to potential optimization strategies for NiO/Ga₂O₃ heterojunction devices.