Topological data analysis of TEM-based structural features affecting the thermal conductivity of amorphous Ge

Abstract

Compared to crystalline materials, amorphous materials lack periodicity and exhibit distinct thermal and lattice vibration properties. Hence, analyzing atomic networks in the transmission electron microscopy (TEM) images of amorphous materials is challenging. In this study, we employ topological data analysis (TDA) and principal component analysis (PCA) to extract structural features from the TEM images of amorphous germanium (a-Ge) and analyze its atomic networks. Our findings demonstrate that the thermal conductivity of a-Ge is influenced by larger atomic rings with more vertices, which facilitates the heat transfer through longer atomic chains and results in a higher thermal conductivity. A comparison of the experimental and simulation data confirms the non-random nature of the atomic arrangements in a-Ge. We propose herein an approach that employs the TDA to identify and analyze the atomic networks in amorphous materials, establishing connections to their thermal properties. This study enhances our understanding of amorphous materials and paves a way for tailored material design and engineering to achieve the desired thermal properties.