Deep reinforcement learning for data-driven adaptive scanning in ptychography.

We present a method that lowers the dose required for an electron ptychographic reconstruction by adaptively scanning the specimen, thereby providing the required spatial information redundancy in the regions of highest importance. The proposed method is built upon a deep learning model that is trained by reinforcement learning, using prior knowledge of the specimen structure from training data sets. We show that using adaptive scanning for electron ptychography outperforms alternative low-dose ptychography experiments in terms of reconstruction resolution and quality.

Fast Grain Mapping with Sub-Nanometer Resolution Using 4D-STEM with Grain Classification by Principal Component Analysis and Non-Negative Matrix Factorization.

High-throughput grain mapping with sub-nanometer spatial resolution is demonstrated using scanning nanobeam electron diffraction (also known as 4D scanning transmission electron microscopy, or 4D-STEM) combined with high-speed direct-electron detection. An electron probe size down to 0.5 nm in diameter is used and the sample investigated is a gold­palladium nanoparticle catalyst. Computational analysis of the 4D-STEM data sets is performed using a disk registration algorithm to identify the diffraction peaks followed by feature learning to map the individual grains. Two unsupervised feature learning techniques are compared: principal component analysis (PCA) and non-negative matrix factorization (NNMF). The characteristics of the PCA versus NNMF output are compared and the potential of the 4D-STEM approach for statistical analysis of grain orientations at high spatial resolution is discussed.

py4DSTEM: A Software Package for Four-Dimensional Scanning Transmission Electron Microscopy Data Analysis.

Scanning transmission electron microscopy (STEM) allows for imaging, diffraction, and spectroscopy of materials on length scales ranging from microns to atoms. By using a high-speed, direct electron detector, it is now possible to record a full two-dimensional (2D) image of the diffracted electron beam at each probe position, typically a 2D grid of probe positions. These 4D-STEM datasets are rich in information, including signatures of the local structure, orientation, deformation, electromagnetic fields, and other sample-dependent properties. However, extracting this information requires complex analysis pipelines that include data wrangling, calibration, analysis, and visualization, all while maintaining robustness against imaging distortions and artifacts. In this paper, we present py4DSTEM, an analysis toolkit for measuring material properties from 4D-STEM datasets, written in the Python language and released with an open-source license. We describe the algorithmic steps for dataset calibration and various 4D-STEM property measurements in detail and present results from several experimental datasets. We also implement a simple and universal file format appropriate for electron microscopy data in py4DSTEM, which uses the open-source HDF5 standard. We hope this tool will benefit the research community and help improve the standards for data and computational methods in electron microscopy, and we invite the community to contribute to this ongoing project.

Functional Materials Under Stress: In Situ TEM Observations of Structural Evolution.

The operating conditions of functional materials usually involve varying stress fields, resulting in structural changes, whether intentional or undesirable. Complex multiscale microstructures including defects, domains, and new phases, can be induced by mechanical loading in functional materials, providing fundamental insight into the deformation process of the involved materials. On the other hand, these microstructures, if induced in a controllable fashion, can be used to tune the functional properties or to enhance certain performance. In situ nanomechanical tests conducted in scanning/transmission electron microscopes (STEM/TEM) provide a critical tool for understanding the microstructural evolution in functional materials. Here, select results on a variety of functional material systems in the field are presented, with a brief introduction into some newly developed multichannel experimental capabilities to demonstrate the impact of these techniques.

Direct measurement of nanostructural change during in situ deformation of a bulk metallic glass.

To date, there has not yet been a direct observation of the initiation and propagation of individual defects in metallic glasses during deformation at the nanoscale. Here, we show through a combination of in situ nanobeam electron diffraction and large-scale molecular dynamics simulations that we can directly observe changes to the local short to medium range atomic ordering during the formation of a shear band. We observe experimentally a spatially resolved reduction of order prior to shear banding due to increased strain. We compare this to molecular dynamics simulations, in which a similar reduction in local order is seen, and caused by shear transformation zone activation, providing direct experimental evidence for this proposed nucleation mechanism for shear bands in amorphous solids. Our observation serves as a link between the atomistic molecular dynamics simulation and the bulk mechanical properties, providing insight into how one could increase ductility in glassy materials.

Optimizing disk registration algorithms for nanobeam electron diffraction strain mapping.

Scanning nanobeam electron diffraction strain mapping is a technique by which the positions of diffracted disks sampled at the nanoscale over a crystalline sample can be used to reconstruct a strain map over a large area. However, it is important that the disk positions are measured accurately, as their positions relative to a reference are directly used to calculate strain. In this study, we compare several correlation methods using both simulated and experimental data in order to directly probe susceptibility to measurement error due to non-uniform diffracted disk illumination structure. We found that prefiltering the diffraction patterns with a Sobel filter before performing cross correlation or performing a square-root magnitude weighted phase correlation returned the best results when inner disk structure was present. We have tested these methods both on simulated datasets, and experimental data from unstrained silicon as well as a twin grain boundary in 304 stainless steel.