Application of the polyhedral template matching method for characterization of 2D atomic resolution electron microscopy images.

High-throughput image segmentation of atomic resolution electron microscopy data poses an ongoing challenge for materials characterization. In this paper, we investigate the application of the polyhedral template matching (PTM) method, a technique widely employed for visualizing three-dimensional (3D) atomistic simulations, to the analysis of two-dimensional (2D) atomic resolution electron microscopy images. This technique is complementary with other atomic resolution data reduction techniques, such as the centrosymmetry parameter, that use the measured atomic peak positions as the starting input. Furthermore, since the template matching process also gives a measure of the local rotation, the method can be used to segment images based on local orientation. We begin by presenting a 2D implementation of the PTM method, suitable for atomic resolution images. We then demonstrate the technique's application to atomic resolution scanning transmission electron microscopy images from close-packed metals, providing examples of the analysis of twins and other grain boundaries in FCC gold and martensite phases in 304 L austenitic stainless steel. Finally, we discuss factors, such as positional errors in the image peak locations, that can affect the accuracy and sensitivity of the structural determinations. [Display omitted] • We develop the 2D-PTM technique to analyze atomic resolution electron microscopy images. • 2D-PTM matches local atomic peak positions to predefined templates. • The technique segments images into local structural regions and maps local rotations. • We demonstrate the 2D-PTM for polycrystalline Au and martensite phases in stainless steel. • Our analysis addresses sensitivity of the 2D-PTM to positional errors in peak location. [ABSTRACT FROM AUTHOR]