Atomic-resolution imaging with scanning transmission electron microscopy is a powerful tool for characterizing the nanoscale structure of materials, in particular features such as defects, local strains, and symmetry-breaking distortions. In addition to advanced instrumentation, the effectiveness of the technique depends on computational image analysis to extract meaningful features from complex datasets recorded in experiments, which can be complicated by the presence of noise and artifacts, small or overlapping features, and the need to scale analysis over large representative areas. Here, we present image analysis approaches which synergize real and reciprocal space information to efficiently and reliably obtain meaningful structural information with picometer scale precision across hundreds of nanometers of material from atomic-resolution electron microscope images. Damping superstructure peaks in reciprocal space allows symmetry-breaking structural distortions to be disentangled from other sources of inhomogeneity and measured with high precision. Real space fitting of the wave-like signals resulting from Fourier filtering enables absolute quantification of lattice parameter variations and strain, as well as the uncertainty associated with these measurements. Implementations of these algorithms are made available as an open source Python package.
@article{arxiv.2504.01159,
title = {Quantitative approaches for multi-scale structural analysis with atomic resolution electron microscopy},
author = {Noah Schnitzer and Lopa Bhatt and Ismail El Baggari and Robert Hovden and Benjamin H. Savitzky and Michelle A. Smeaton and Berit H. Goodge},
journal= {arXiv preprint arXiv:2504.01159},
year = {2025}
}