Semi-Automated, Object-Based Tomography of Dislocation Structures

Area of Science:

• Materials Science
• Crystallography
• Electron Microscopy

Background:

• Characterizing 3D dislocation arrangements is crucial for materials science.
• Conventional dislocation tomography relies on intensity-based algorithms requiring extensive image data.

Purpose of the Study:

• To present a novel, semi-automated, object-based approach for 3D dislocation tomography.
• To reduce data collection requirements compared to traditional methods.

Main Methods:

• Dislocation line objects are extracted from individual frames.
• Objects are aligned and matched across a tilt series using graph theory and arc-length mapping.
• Tomographic reconstruction determines the full 3D configuration of dislocations.

Main Results:

• Demonstrated the method on a dislocation network dataset from scanning transmission electron microscopy.
• The object-based approach effectively reconstructs 3D dislocation structures.
• Uncertainty analysis provides insights for optimization.

Conclusions:

• The object-based method offers a more efficient alternative for 3D dislocation tomography.
• Potential for optimizing data collection and further automation exists.
• Innovations in graph theory enhance dislocation line representation and matching.