Search papers, labs, and topics across Lattice.
This paper benchmarks several machine learning models (ResNet, VGG, custom CNN, PCA-kNN) for automating polarization direction detection in ferroelectric potassium sodium niobate using 4D-STEM diffraction patterns. The study finds a significant domain gap between synthetic training data and real experimental data, hindering direct deployment. A custom representation training regime and PCA-based methods, along with data augmentation, help bridge this gap, and the models' prediction errors correlate with crystal defects.
Despite high accuracy on synthetic data, machine learning models struggle to generalize to real-world 4D-STEM data for ferroelectric polarization mapping, highlighting a critical domain gap.
Four-dimensional scanning transmission electron microscopy (4D-STEM) provides rich, atomic-scale insights into materials structures. However, extracting specific physical properties - such as polarization directions essential for understanding functional properties of ferroelectrics - remains a significant challenge. In this study, we systematically benchmark multiple machine learning models, namely ResNet, VGG, a custom convolutional neural network, and PCA-informed k-Nearest Neighbors, to automate the detection of polarization directions from 4D-STEM diffraction patterns in ferroelectric potassium sodium niobate. While models trained on synthetic data achieve high accuracy on idealized synthetic diffraction patterns of equivalent thickness, the domain gap between simulation and experiment remains a critical barrier to real-world deployment. In this context, a custom made prototype representation training regime and PCA-based methods, combined with data augmentation and filtering, can better bridge this gap. Error analysis reveals periodic missclassification patterns, indicating that not all diffraction patterns carry enough information for a successful classification. Additionally, our qualitative analysis demonstrates that irregularities in the model's prediction patterns correlate with defects in the crystal structure, suggesting that supervised models could be used for detecting structural defects. These findings guide the development of robust, transferable machine learning tools for electron microscopy analysis.
