Abstract: We present a computational imaging mode for large scale electron microscopy data, which retrieves a complex wave from noisy/sparse intensity recordings using a deep learning approach and subsequently reconstructs an image of the specimen from the Convolutional Neural Network (CNN) predicted exit waves. We demonstrate that an appropriate forward model in combination with open data frameworks can be used to generate large synthetic datasets for training. In combination with augmenting the data with Poisson noise corresponding to varying dose-values, we effectively eliminate overfitting issues. The U-NET[1] based architecture of the CNN is adapted to the task at hand and performs well while maintaining a relatively small size and fast performance. The validity of the approach is confirmed by comparing the reconstruction to well-established methods using simulated, as well as real electron microscopy data. The proposed method is shown to be effective particularly in the low dose range, evident by strong suppression of noise, good spatial resolution, and sensitivity to different atom types, enabling the simultaneous visualisation of light and heavy elements and making different atomic species distinguishable. Since the method acts on a very local scale and is comparatively fast it bears the potential to be used for near-real-time reconstruction during data acquisition.

Abstract: In this study, we explore the possibility to use deep learning for the reconstruction of phase images from 4D scanning transmission electron microscopy (4D-STEM) data. The process can be divided into two main steps. First, the complex electron wave function is recovered for a convergent beam electron diffraction pattern (CBED) using a convolutional neural network (CNN). Subsequently, a corresponding patch of the phase object is recovered using the phase object approximation. Repeating this for each scan position in a 4D-STEM dataset and combining the patches by complex summation yields the full-phase object. Each patch is recovered from a kernel of 3x3 adjacent CBEDs only, which eliminates common, large memory requirements and enables live processing during an experiment. The machine learning pipeline, data generation, and the reconstruction algorithm are presented. We demonstrate that the CNN can retrieve phase information beyond the aperture angle, enabling super-resolution imaging. The image contrast formation is evaluated showing a dependence on the thickness and atomic column type. Columns containing light and heavy elements can be imaged simultaneously and are distinguishable. The combination of super-resolution, good noise robustness, and intuitive image contrast characteristics makes the approach unique among live imaging methods in 4D-STEM.

Abstract: Inelastic excitation as exploited in Electron Energy Loss Spectroscopy (EELS) contains a rich source of information that is revealed in the scattering process. To accurately quantify core-loss EELS, it is common practice to fit the observed spectrum with scattering cross-sections calculated using experimental parameters and a Generalized Oscillator Strength (GOS) database [1].   The GOS is computed using Fermi’s Golden Rule and orbitals of bound and excited states. Previously, the GOS was based on Hartree-Fock solutions [2], but more recently Density Functional Theory (DFT) has been used [3]. In this work, we have chosen to use the Dirac equation to incorporate relativistic effects and have performed calculations using Flexible Atomic Code (FAC) [4]. This repository contains a tabulated GOS database based on Dirac solutions for computing double differential cross-sections under experimental conditions.   We hope the Dirac-based GOS database can benefit the EELS community for both academic use and industry integration.   Database Details: – Covers all elements (Z: 1-108) and all edges – Large energy range: 0.01 – 4000 eV – Large momentum range: 0.05 -50 Å-1 – Fine log sampling: 128 points for energy and 256 points for momentum – Data format: GOSH [3]   Calculation Details: – Single atoms only; solid-state effects are not considered – Unoccupied states before continuum states of ionization are not considered; no fine structure – Plane Wave Born Approximation – Frozen Core Approximation is employed; electrostatic potential remains unchanged for orthogonal states when – core-shell electron is excited – Self-consistent Dirac–Fock–Slater iteration is used for Dirac calculations; Local Density Approximation is assumed for electron exchange interactions; continuum states are normalized against asymptotic form at large distances – Both large and small component contributions of Dirac solutions are included in GOS – Final state contributions are included until the contribution of the previous three states falls below 0.1%. A convergence log is provided for reference.   Version 1.1 release note: – Update to be consistent with GOSH data format [3], all the edges are now within a single hdf5 file. A notable change in particular, the sampling in momentum is in 1/m, instead of previously in 1/Å. Great thanks to Gulio Guzzinati for his suggestions and sending conversion script.  Version 1.2 release note: – Add “File Type / File version” information [1] Verbeeck, J., and S. Van Aert. Ultramicroscopy 101.2-4 (2004): 207-224. [2] Leapman, R. D., P. Rez, and D. F. Mayers. The Journal of Chemical Physics 72.2 (1980): 1232-1243. [3] Segger, L, Guzzinati, G, & Kohl, H. Zenodo (2023). doi:10.5281/zenodo.7645765 [4] Gu, M. F. Canadian Journal of Physics 86(5) (2008): 675-689.

Abstract: Supplementary information for the article “In-situ Plasma Studies using a Direct Current Microplasma in a Scanning Electron Microscope” containing the videos of in-situ SEM imaging (mp4 files), raw data/images, and Jupyter notebooks (ipynb files) for data treatment and plots. Link to the preprint: https://doi.org/10.48550/arXiv.2308.15123 Explanation of the data files can be found in the Information.pdf file. The Videos folder contains the in-situ SEM image series mentioned in the paper. If there are any questions/bugs, feel free to contact me at lukas.grunewaldatuantwerpen.be