Evaluating X-Ray Microanalysis Phase Maps Using Principal Component Analysis.

Area of Science:

• Materials Science
• Geology
• Analytical Chemistry

Background:

• Automated phase mapping is crucial for sample characterization.
• Current data quality evaluation methods (BSE image overlay, composition averages) have significant limitations.

Purpose of the Study:

• To introduce novel principal component analysis (PCA) based methods for evaluating automated phase map quality.
• To address limitations of existing evaluation techniques in materials and geological analysis.

Main Methods:

• Utilized PCA to create RGB composite images of principal components representing chemical variation.
• Developed PCA maps for individual phases to assess intra-phase chemical heterogeneity.
• Applied K-means clustering and K-nearest neighbor for phase classification.

Main Results:

• PCA-based RGB composite images provide a superior reference for phase map comparison compared to BSE images.
• PCA maps reveal chemical zoning and heterogeneity within phases, aiding in identification of unclassified phases and analytical artifacts.
• Demonstrated effectiveness on a complex geological sample.

Conclusions:

• PCA offers powerful, sensitive methods for automated phase map evaluation.
• These methods improve the reliability and interpretability of phase mapping in complex samples.
• Highlights the importance of robust data quality assessment in automated analysis.