Abstract
While electron microscopy offers crucial atomic-resolution insights intostructure-property relationships, radiation damage severely limits its use onbeam-sensitive materials like proteins and 2D materials. To overcome thischallenge, we push beyond the electron dose limits of conventional electronmicroscopy by adapting principles from multi-image super-resolution (MISR) thathave been widely used in remote sensing. Our method fuses multiplelow-resolution, sub-pixel-shifted views and enhances the reconstruction with aconvolutional neural network (CNN) that integrates features from synthetic,multi-angle observations. We developed a dual-path, attention-guided networkfor 4D-STEM that achieves atomic-scale super-resolution from ultra-low-dosedata. This provides robust atomic-scale visualization across amorphous,semi-crystalline, and crystalline beam-sensitive specimens. Systematicevaluations on representative materials demonstrate comparable spatialresolution to conventional ptychography under ultra-low-dose conditions. Ourwork expands the capabilities of 4D-STEM, offering a new and generalizablemethod for the structural analysis of radiation-vulnerable materials.