STEM Diffraction Pattern Analysis with Deep Learning Networks

Accurate grain orientation mapping is essential for understanding and optimising the performance of polycrystalline materials, particularly in energy applications. Lithium nickel oxide (LiNiO2) is a promising cathode material for next-generation lithium-ion batteries, and its electrochemical behaviour is closely linked to microstructural features such as grain size and crystallographic orientation. Traditional orientation mapping methods—manual indexing, template matching (TM), or Hough transform-based techniques—are often slow and noise-sensitive, especially when processing complex or overlapping patterns, creating a bottleneck in large-scale microstructural analysis. This work presents a machine learning-based approach for predicting Euler angles directly from scanning transmission electron microscopy (STEM) diffraction patterns (DPs), enabling automated, high-resolution orientation mapping for nanoscale microstructure analysis. Three deep learning architectures—convolutional neural networks (CNNs), Dense Convolutional Networks (DenseNets), and Shifted Windows (Swin) Transformers—are evaluated using an experimentally acquired dataset labelled via a commercial TM algorithm. While the CNN model serves as a baseline, both DenseNets and Swin Transformers achieve superior performance, with the Swin Transformer yielding the highest evaluation scores and most consistent microstructural predictions. The resulting crystal maps reveal clear grain boundaries and coherent intra-grain orientation distributions, underscoring the potential of advanced machine learning models for high-throughput microstructural characterisation.