Density-based clustering of crystal (mis)orientations and the orix Python library.

Crystal orientation mapping is a widely used technique to study microstructure in materials. It captures orientation data within grains and across grain boundaries. These data often cluster in orientation space due to crystal symmetry and preferred orientation relationships. Prior research has demonstrated that such clustering reflects physical phenomena like deformation or twinning. However, identifying and interpreting these clusters remains a challenge. Existing methods lack robust computational tools for clustering (mis)orientations. This gap motivated the need for a systematic approach. No prior work had resolved how to efficiently detect and visualize clusters in orientation data. This paper introduces a solution using density-based clustering algorithms.

The study aimed to develop a reliable method for clustering crystal (mis)orientation data. It sought to incorporate crystal symmetry into distance metrics for accurate clustering. The goal was to identify frequently measured (mis)orientations linked to specific microstructural features. The researchers aimed to visualize clusters in both spatial and orientation space. They also aimed to provide a practical tool for this analysis. The motivation was to improve the interpretation of orientation data in materials science. This approach could help identify deformation mechanisms like twinning. The study focused on applying this method to titanium deformation twins.

The study used distance metrics that account for crystal symmetry in orientation space. It applied the DBSCAN algorithm for density-based clustering of (mis)orientations. The method identified clusters of similar orientations within grains and similar misorientations across boundaries. The approach visualized clusters in three-dimensional orientation space. It also mapped these clusters spatially within the material. The researchers used titanium deformation twinning as a test case. They implemented the method in a new Python library called orix. The library provides tools for orientation clustering and visualization.

The clustering method successfully identified frequently measured (mis)orientations in titanium. It detected clusters corresponding to deformation twins with specific orientation relationships. The clusters were visualized in both spatial and orientation space. The DBSCAN algorithm effectively grouped similar orientations and misorientations. The results showed distinct clusters in orientation space for twinned regions. The method revealed orientation relationships consistent with known twinning modes. The orix library enabled these computations and visualizations. The approach demonstrated utility in identifying microstructural features from orientation data.

The study demonstrated that density-based clustering of (mis)orientations can reveal microstructural features. The method successfully identified deformation twinning in titanium through clustering. The DBSCAN algorithm proved effective for grouping similar orientations and misorientations. The approach incorporates crystal symmetry into distance metrics for accurate clustering. The orix library provides a practical tool for orientation clustering and visualization. The results suggest that this method can improve the interpretation of orientation data. The study highlights the importance of orientation relationships in microstructure analysis. The findings support the use of density-based clustering in materials science research.