A framework for analyzing real-time electron microscopy data was developed in the ZIOC

Electron microscopy is one of the most widely used methods for studying the structure of materials, biological objects, and complex chemical systems of various scales, even reaching the monomolecular level. The development of modern nanotechnologies, catalytic systems, as well as the production of semiconductors and fuel cells cannot be imagined without electron microscopy. One of the most interesting and promising areas of research in the field of electron microscopy is the analysis of video fragments obtained using an electron microscope, which reflect the evolution of the morphology of the studied samples. This approach makes it possible to study the dynamics of various systems and draw conclusions about the structure-property relationship. However, the large volumes of data generated during these experiments are extremely difficult to process manually. A promising solution to this problem is the use of machine learning methods.

In recent years, scientists from the Laboratory of Metal Complex and Nanoscale Catalysts of the Zelinsky Institute have been actively working on the implementation of machine learning methods for analyzing large amounts of experimental data. In one of their latest studies, they succeeded in developing a framework for analyzing real-time electron microscopy data. Efficient data processing is provided by a whole range of denoising, binarization, segmentation, and tracking modules. The developed computing base is of particular importance and applicability in liquid-phase systems. Experimental verification of the proposed approach led to the discovery of the anisotropic effect of the electron beam in microstructured liquid systems. As a tentative explanation for this phenomenon, it can be assumed that the non-uniform interaction of the focused electron beam with the sample, which occurred due to the specific shape of the scanning pattern typically used in scanning electron microscopy, led to the formation of non-equilibrium liquid zones located along the direction of beam movement. The discovery of such an effect opened up new possibilities for direct control of the state of the microphase system with a change in the raster scanning pattern.
 

Source:

Daniil A. Boiko, Alexey S. Kashin, Vyacheslav R. Sorokin, Yury V. Agaev, Roman G. Zaytsev, Valentine P. Ananikov Analyzing ionic liquid systems using real-time electron microscopy and a computational framework combining deep learning and classic computer vision techniques // J. Mol. Liq., 2023, 121407. DOI: 10.1016/j.molliq.2023.121407.