Optimal principal component analysis of STEM XEDS spectrum images

Area of Science:

• Materials Science
• Analytical Chemistry
• Data Science

Background:

• STEM XEDS (Scanning Transmission Electron Microscopy with X-ray Energy Dispersive Spectroscopy) generates complex spectral images.
• Image noise significantly hinders accurate phase analysis in complex materials.
• Principal Component Analysis (PCA) is a powerful technique for dimensionality reduction and noise suppression.

Purpose of the Study:

• To provide a step-by-step analysis of the PCA workflow for denoising STEM XEDS data.
• To identify and solve common challenges in PCA decomposition for complex semiconductor structures.
• To introduce and validate a novel method for optimal principal component truncation in data reconstruction.

Main Methods:

• Detailed step-by-step application of PCA to STEM XEDS data from a multi-phase semiconductor.
• Identification of distortions affecting principal component decomposition.
• Development and description of a new method for optimal principal component truncation.

Main Results:

• PCA significantly reduces noise in STEM XEDS spectrum images.
• Specific challenges in PCA decomposition for multi-phase materials were identified and addressed.
• The novel truncation method demonstrated superior accuracy and robustness compared to existing techniques.

Conclusions:

• PCA is a highly effective tool for denoising STEM XEDS data.
• Addressing decomposition challenges and employing optimal truncation are crucial for successful PCA application.
• The proposed novel truncation method offers an improved approach for reconstructing denoised STEM XEDS data.