The effect of post-acquisition data misalignments on the performance of STEM tomography

Abstract

Scanning transmission electron microscopy (STEM) tomography is widely used to reveal three-dimensional (3D) structure of the specimens from nanometers to atomic resolution. Advances in transmission electron microscope enable high resolution and low dose data acquisition for 3D reconstruction to resolve 3D coordinates of single atoms. Meanwhile, the developments of reconstruction algorithms increase the reconstruction quality. However, the as-acquired raw datasets often require alignments before they can be used as inputs to feed a reconstruction algorithm. The effect of post-acquisition data misalignments on the final reconstruction quality of STEM tomography is less discussed. Here, using phantom datasets with a focus on atomic features, we simulate a range of misalignments on each alignment step and correlate the level of misalignments to the final reconstruction quality. The study pinpoints that accurate alignments of tilt angles and stack positions and minimizing scan distortion are important to the reconstruction quality, while the sequence of alignment steps do not affect the reconstruction quality. The study not only emphasizes the most important alignment steps that should be pay attention to for a high-quality 3D reconstruction, but also summarizes post-acquisition data alignment procedures as a primer for new practitioners to use STEM tomography.