An automatic method for atom identification in scanning tunnelling microscopy images of Fe‐chalcogenide superconductors

Summary We describe a computational approach for the automatic recognition and classification of atomic species in scanning tunnelling microscopy images. The approach is based on a pipeline of image processing methods in which the classification step is performed by means of a Fuzzy Clustering algorithm. As a representative example, we use the computational tool to characterize the nanoscale phase separation in thin films of the Fe‐chalcogenide superconductor FeSexTe1‐x, starting from synthetic data sets and experimental topographies. We quantify the stoichiometry fluctuations on length scales from tens to a few nanometres. Lay description This paper describes an automatic computational method for the localization and recognition of atoms in high‐resolution Scanning Tunnelling Microscopy images of crystal lattice surfaces, showing the coexistence of different atomic layers and species. The method is based on image processing, machine learning and pattern recognition techniques to identify the layer of interest, to characterize the basic crystal structure, to perform a tessellation of the image and, finally, to classify atomic species by means of clustering techniques. We use this computational tool to characterize the nanoscale phase separation in thin films of the Fe‐chalcogenide superconductors, starting from synthetic data sets and experimental topographies. As a result, we are able to quantify the stoichiometry fluctuations on length scales from tens to a few nanometres.