Three-dimensional deconvolution for large-angle illumination annular dark-field scanning transmission electron microscopy depth sectioning

To fully understand materials’ properties, it is essential to identify three-dimensional (3D) distributions of point defects at the atomic scale. With recent progress in the aberration correction electron optics, the 3D spatial resolution in scanning transmission electron microscopy (STEM) has been substantially improved, making 3D atomic-resolution observation increasingly feasible. Annular dark-field (ADF)-STEM depth sectioning is one of the promising 3D electron microscopy imaging, which optically sections a sample along the depth direction with a focal series of ADF-STEM images. Using ADF-STEM depth sectioning, it has already become possible to determine the 3D locations of single dopants embedded in a crystal and visualize the atomistic defects formed on oxides surfaces. However, resolving 3D atomic locations by ADF-STEM depth sectioning remains a challenge due to the limited depth resolution. Here, we develop a 3D deconvolution algorithm, and we demonstrate through simulations that it substantially improves the depth resolution of depth sectioning images, enabling the precise identification of closely distributed dopants in 3D. Additionally, we demonstrate that the sub-nanometer depth resolution is attainable at surfaces with the illumination semi-angle of 60 mrad and the acceleration voltage of 300 kV, and the deconvolved image allows for the precise reconstruction of the surface atomic structures. Combining with the proposed 3D deconvolution algorithm, ADF-STEM depth sectioning imaging will surpass the conventional resolution limit and potentially enable 3D atomic-resolution imaging in solids.