Amorphous materials have been used in a range of electronic and photonic applications, and the need for quantitative analytical techniques on their local structural information is growing. We present a comprehensive analysis of the atomic and electronic structures of an amorphous material, amorphous carbon (a-C), with scanning transmission electron microscopy (STEM)-derived techniques, four-dimensional STEM (4D-STEM), and STEM-electron energy loss spectroscopy (STEM-EELS). Each diffraction pattern of an a-C layer stack acquired via 4D-STEM is transformed into a reduced density function (RDF) and a radial variance profile (RVP) to retrieve the information on the atomic structures. Importantly, a machine-learning approach (preferably cluster analysis) separates distinct features in the EELS and RDF datasets; it also describes the spatial distributions of these features in the scanned regions. Consequently, we showed that the differences in the sp2/sp3 ratio and the involvement of additional elements led to changes in the bond length. Furthermore, we identified the dominant types of medium-range ordering structures (diamond-like or graphite-like nano-crystals) by correlations among the EELS, RDF, and RVP data. The information obtained via STEM-EELS and 4D-STEM can be strongly correlated, leading to the comprehensive characterization of the a-C layer stack for a nanometer-scale area. This process can be used to investigate any amorphous material, thereby yielding comprehensive information regarding the origins of notable properties.
Read full abstract