Compressive multi-beam scanning transmission electron microscopy

TL;DR

This paper introduces a multi-beam STEM imaging method that combines down-sampling with super-resolution reconstruction using compressive sensing, enabling faster imaging while maintaining high fidelity.

Contribution

It presents a novel multi-beam STEM approach with a custom aperture and a compressive sensing reconstruction framework for accelerated imaging.

Findings

High-fidelity images reconstructed from down-sampled data

Effective multi-beam probe created with custom aperture

Potential for significant acceleration of STEM imaging

Abstract

We demonstrate a multi-beam scanning transmission electron microscopy (STEM) imaging that integrates down-sampling with super-resolution image reconstruction via a compressive sensing framework. A custom condenser aperture with six randomly positioned circular holes is employed to produce a multi-beam STEM probe, with the beam shape and distribution tuned through defocus. While the raw multi-beam images exhibit overlapping patterns, reconstruction using Adam optimization and total variation normalization yields high-fidelity images that closely reproduce the original sample structures, even from substantially down-sampled data. The proposed approach offers a pathway toward significant acceleration of such techniques through multibeam sparse sampling and computational reconstruction potentially useful for the analytical scanning methods in general.