Big data in reciprocal space: Sliding fast Fourier transforms for determining periodicity

Abstract

Significant advances in atomically resolved imaging of crystals and surfaces have occurred in the last decade allowing unprecedented insight into local crystal structures and periodicity. Yet, the analysis of the long-range periodicity from the local imaging data, critical to correlation of functional properties and chemistry to the local crystallography, remains a challenge. Here, we introduce a Sliding Fast Fourier Transform (FFT) filter to analyze atomically resolved images of in-situ grown La5/8Ca3/8MnO3 (LCMO) films. We demonstrate the ability of sliding FFT algorithm to differentiate two sub-lattices, resulting from a mixed-terminated surface. Principal Component Analysis and Independent Component Analysis of the Sliding FFT dataset reveal the distinct changes in crystallography, step edges, and boundaries between the multiple sub-lattices. The implications for the LCMO system are discussed. The method is universal for images with any periodicity, and is especially amenable to atomically resolved probe and electron-microscopy data for rapid identification of the sub-lattices present.

Mapping mesoscopic phase evolution during E-beam induced transformations via deep learning of atomically resolved images

“…6,7,12,21,22,25,41 We therefore investigate an automated routine for identifying phase evolution at the atomic scale. For each frame, we computed the local FFT spectra via the sliding window method, 30 with a window size of 128px and step size of 16px (the outline of the window is shown in Fig. 6(a) in white).…”

Direct atomic fabrication and dopant positioning in Si using electron beams with active real-time image-based feedback

“…As a model sample, we have chosen amorphous silicon grown on a crystalline Si substrate. The STEM image prior to e-beam crystallization is shown in Figure 1 To obtain further insight into the structure of the newly formed crystalline Si, we perform comparative crystallographic image analysis 27,28 . In this method, a sliding window is scanned across the image, generating a stack of sub-images.…”

Merging Deep Learning, Chemistry, and Diffraction for High-Throughput Material Structure Prediction

“…Combining diffraction and EELS/EDS data sets that are collected simultaneously or separately, it is possible to improve the accuracy of these models trained solely on crystallographic data. [2,3] The additional information provided by chemistry data augments the model's understanding of higher-level structural classifications by drawing on the Open Crystallography Database, Materials Project Database, and experimental data, assembled in a robust training set that combines diffraction and chemistry.…”