Extraction of structural and chemical information from high angle annular dark-field image by an improved peaks finding method.

Abstract

With the development of spherical aberration (Cs) corrected scanning transmission electron microscopy (STEM), high angle annular dark filed (HAADF) imaging technique has been widely applied in the microstructure characterization of various advanced materials with atomic resolution. However, current qualitative interpretation of the HAADF image is not enough to extract all the useful information. Here a modified peaks finding method was proposed to quantify the HAADF-STEM image to extract structural and chemical information. Firstly, an automatic segmentation technique including numerical filters and watershed algorithm was used to define the sub-areas for each atomic column. Then a 2D Gaussian fitting was carried out to determine the atomic column positions precisely, which provides the geometric information at the unit-cell scale. Furthermore, a self-adaptive integration based on the column position and the covariance of statistical Gaussian distribution were performed. The integrated intensities show very high sensitivity on the mean atomic number with improved signal-to-noise (S/N) ratio. Consequently, the polarization map and strain distributions were rebuilt from a HAADF-STEM image of the rhombohedral and tetragonal BiFeO3 interface and a MnO2 monolayer in LaAlO3 /SrMnO3 /SrTiO3 heterostructure was discerned from its neighbor TiO2 layers. Microsc. Res. Tech. 79:820-826, 2016. © 2016 Wiley Periodicals, Inc.