Implementing dynamic high-performance computing supported workflows on Scanning Transmission Electron Microscope

TL;DR

This paper demonstrates the integration of STEM-EELS with machine learning and high-performance computing to improve material characterization workflows, enabling real-time analysis and active learning in electron microscopy.

Contribution

It introduces novel workflows that combine STEM, ML, and HPC, including object finding and Deep Kernel Learning, to enhance efficiency and capabilities in material analysis.

Findings

Enhanced efficiency of STEM workflows with ML and HPC

Successful implementation of object finding and DKL techniques

Improved real-time data analysis in electron microscopy

Abstract

Scanning Transmission Electron Microscopy (STEM) coupled with Electron Energy Loss Spectroscopy (EELS) presents a powerful platform for detailed material characterization via rich imaging and spectroscopic data. Modern electron microscopes can access multiple length scales and sampling rates far beyond human perception and reaction time. Recent advancements in machine learning (ML) offer a promising avenue to enhance these capabilities by integrating ML algorithms into the STEM-EELS framework, fostering an environment of active learning. This work enables the seamless integration of STEM with High-Performance Computing (HPC) systems. We present several implemented workflows that exemplify this integration. These workflows include sophisticated techniques such as object finding and Deep Kernel Learning (DKL). Through these developments, we demonstrate how the fusion of STEM-EELS with ML and HPC enhances the efficiency and scope of material characterization for 70% STEM available globally. The codes are available at GitHub link.