3D nanostructural characterisation of grain boundaries in atom probe data utilising machine learning methods.

Area of Science:

• Materials Science
• Computational Materials Science
• Data Analysis

Background:

• Boosting algorithms are effective for object tracking in video analysis.
• Manual extraction of grain boundary information from atom probe data is time-consuming and prone to errors.

Purpose of the Study:

• To develop an automated method for extracting grain boundary location and trace from atom probe data.
• To combine boosting trackers with Hough transformation and principal component analysis for enhanced data analysis.

Main Methods:

• Implementation of a boosting tracker algorithm.
• Application of Hough transformation for feature detection.
• Utilizing principal component analysis for data dimensionality reduction.
• Analysis of atom probe data from pure aluminum bi-crystal and simulated datasets.

Main Results:

• Successful automated extraction of grain boundary location and trace.
• Demonstrated effectiveness on experimental and simulated atom probe data.
• Near-atomic resolution 3D characterization of grain boundaries.

Conclusions:

• The developed method provides an efficient and accurate alternative to manual analysis.
• Enables comprehensive 5-degree-of-freedom grain boundary characterization.
• Facilitates extraction of local atomic compositional and geometric information at interfaces.