PNNL

Abstract

The inpainting of randomly sub-sampled images acquired by scanning transmission electron microscopy (STEM) is an attractive method for imaging under low-dose conditions (e.g., 1 e-/A^2) without changing either the operation of the microscope or the physics of the imaging process. We show that 1) adaptive sub-sampling increases acquisition speed, resolution, and sensitivity; and 2) random (non-adaptive) sub-sampling is equivalent, but faster than, traditional low-dose techniques. Adaptive sub-sampling opens numerous possibilities for the analysis of beam sensitive materials and in-situ dynamic processes at the resolution limit of the aberration corrected microscope and is demonstrated here for the analysis of the node distribution in metal-organic frameworks (MOFs).