Direct motif extraction from high resolution crystalline STEM images.

Area of Science:

• Materials Science
• Crystallography
• Data Analysis

Background:

• Recent advancements in automatic data analysis for crystal structures include unit cell extraction and defect detection.
• However, automatic and unsupervised motif extraction methods are still lacking.

Purpose of the Study:

• To introduce a novel, unsupervised method for automatic motif extraction in real space from crystalline images.
• To address the gap in automated analysis tools for crystal structure motifs.

Main Methods:

• A variational approach utilizing the unit cell projection operator for motif extraction.
• A multi-stage algorithm involving primitive unit cell determination, motif image estimation, and atomic position identification.
• Testing on synthetic and experimental High-Angle Annular Dark-Field Scanning Transmission Electron Microscopy (HAADF STEM) images.

Main Results:

• Successful automatic extraction of motifs, represented as images and atomic positions.
• Accurate determination of primitive unit cell vectors.
• Generation of denoised and modeled reconstructions of input images.
• Application to complex μ-phase structures (Nb6.4Co6.6 and Nb7Co6) revealing subtle interplanar spacing differences.

Conclusions:

• The proposed method provides a robust tool for automatic motif extraction in crystalline materials.
• It enables detailed analysis of complex structures and identification of subtle structural variations.
• This technique advances automated crystallographic analysis.