Machine learning based determination of crystal orientation in field ion micrograph

Abstract

Field ion microscope (FIM) has the capability to observe the atomic arrangement of the surface of a metal tip. Using FIM images continuously taken during field evaporation, the internal structure of the specimen can be observed by three-dimensional reconstruction of the specimen. However, the compression factor is needed for three-dimensional reconstruction. In order to calculate the compression factor, we automatically identified the crystal orientation of tungsten FIM images using machine learning. As a result, the correct crystal orientation was output for 88% of the total data, indicating the effectiveness of this method.