Abstract: Quantification of annular dark field (ADF) scanning transmission electron microscopy (STEM) images in terms
of composition or thickness often relies on probe-position integrated scattering cross sections (PPISCS). In
order to compare experimental PPISCS with theoretically predicted ones, expensive simulations are needed for
a given specimen, zone axis orientation, and a variety of microscope settings. The computation time of such
simulations can be in the order of hours using a single GPU card. ADF STEM simulations can be efficiently
parallelized using multiple GPUs, as the calculation of each pixel is independent of other pixels. However, most
research groups do not have the necessary hardware, and, in the best-case scenario, the simulation time will
only be reduced proportionally to the number of GPUs used. In this manuscript, we use a learning approach and
present a densely connected neural network that is able to perform real-time ADF STEM PPISCS predictions as
a function of atomic column thickness for most common face-centered cubic (fcc) crystals (i.e., Al, Cu, Pd, Ag,
Pt, Au and Pb) along [100] and [111] zone axis orientations, root-mean-square displacements, and microscope
parameters. The proposed architecture is parameter efficient and yields accurate predictions for the PPISCS
values for a wide range of input parameters that are commonly used for aberration-corrected transmission
electron microscopes.

Abstract: Quantitative structure determination is needed in order to study and understand nanomaterials at the atomic scale. Materials characterisation resulting in precise structural information is a crucial point to understand the structure–property relation of materials. Counting the number of atoms and retrieving the 3D atomic structure of nanoparticles plays an important role here. In this paper, an overview will be given of the atom-counting methodology and its applications over the past decade. The procedure to count the number of atoms will be discussed in detail and it will be shown how the performance of the method can be further improved. Furthermore, advances toward mixed element nanostructures, 3D atomic modelling based on the atom-counting results, and quantifying the nanoparticle dynamics will be highlighted.

Abstract: HAADF-STEM tomography is a widely used experimental technique for analyzing nanometer-scale crystalline structures of a large variety of materials in three dimensions. Unfortunately, the acquisition of conventional HAADF-STEM tilt series can easily take up one hour or more, depending on the complexity of the experiment. It is therefore far from straightforward to investigate samples that do not withstand long acquisition or to acquire large amounts of tilt series during a single TEM experiment. The latter would lead to the ability to obtain statistically meaningful 3D data, or to perform in situ 3D characterizations with a much shorter time resolution. Various HAADF-STEM acquisition strategies have been proposed to accelerate the tomographic acquisition and reduce the required electron dose. These methods include tilting the holder continuously while acquiring a projection “movie” and a hybrid, incremental, methodology which combines the benefits of the conventional and continuous technique. However, until now an experimental evaluation has been lacking. In this paper, the different acquisition strategies will be experimentally compared in terms of speed, resolution and electron dose. This evaluation will be performed based on experimental tilt series acquired for various metallic nanoparticles with different shapes and sizes. We discuss the data processing involved with the fast HAADF-STEM tilt series and provide a general guideline when which acquisition strategy should be preferentially used.