High-throughput quantitative atomic resolution electron microscopy using real-time image simulations.

The goal of my proposal is to develop a powerful method in order to evolve toward four-dimensional (4D=3D+time) quantification of nanostructures of arbitrary shape, size and atom type at the atomic scale. Therefore, novel quantitative measurement tools will be combined with aberration-corrected scanning transmission electron microscopy (STEM). Quantitative 3D characterisation of nanostructures can nowadays be achieved with high reliability for model-like systems with 1 type of chemical element present. Also for some heteronanostructures, a 3D visualisation at the atomic scale is possible using state-of-the-art STEM. However, high-precision quantification often involves a meticulous analysis using advanced methods. This impedes high-throughput analyses which are increasingly important for the study of dynamical processes induced by heating, under de flow of a selected gas, or by the electron beam. In this project, the initiation of real-time image simulations will be a giant leap forward for the 4D characterisation of nanomaterials. This highly challenging and innovative objective will be reached by introducing deep learning architectures into quantitative STEM. This unique approach will allow simulating images in real time using a fully physics-based description of the experimental intensities. The outcome of this project will deliver all necessary input for understanding and predicting the properties in complex nanostructures and their dynamical processes.

Keywords:ELECTRON MICROSCOPY (TRANSMISSION)

Disciplines:Statistics, Classical physics not elsewhere classified, Metrology, Quantum physics not elsewhere classified, Modelling and simulation