Analyzing ionic liquid systems using real-time electron microscopy and a computational framework combining deep learning and classic computer vision techniques

Abstract

Electron microscopy (EM) is one of the most important methods for characterizing various systems, and it is traditionally applied to static solid structures. Remarkable recent developments have opened multiple possibilities for in situ observation of different phenomena, including liquid phase processes. In contrast to routine solid-state EM measurements with static images, electron microscopy in liquids often deals with ubiquitous dynamics, which can be recorded as video streams. Providing much information about the sample, real-time EM increases the complexity of data analysis, challenging researchers to develop new, highly efficient systems for data processing. The present work proposes a framework for data analysis in real-time electron microscopy. Multiple algorithm choices are compared, and efficient solutions are described. Using the best algorithm, combining classical computer vision methods and deep learning-based denoising, the unique anisotropic effect of the electron beam in microstructured ionic liquid-based systems was discovered. The developed method provides an efficient approach for studying the structure and transformation of soft micro-scale domains in molecular liquids. The corresponding software was made publicly available, and detailed instructions to reapply it to other problems were provided.