Evaluation of EELS spectrum imaging data by spectral components and factors from multivariate analysis

Area of Science:

• Materials Science
• Spectroscopy
• Data Analysis

Background:

• Electron energy loss spectroscopy (EELS) generates complex spectrum imaging datasets.
• Spatial variance in EELS data can often be explained by a limited number of underlying components.
• Quantitative analysis of EELS spectrum imaging data presents challenges due to its high dimensionality.

Purpose of the Study:

• To explore dimension reduction techniques for EELS spectrum imaging data.
• To facilitate quantitative analysis by focusing on spectral components rather than individual pixel spectra.
• To apply non-negative matrix factorization (NMF) for decomposing EELS datasets.

Main Methods:

• Utilized non-negative matrix factorization (NMF) for decomposing EELS spectrum imaging datasets.
• Applied the method to datasets from Fe2O3 thin films with varying Sn doping profiles on different substrates (SnO2 and Si).
• Developed Matlab codes to guide the analysis of spectral features.

Main Results:

• Demonstrated that a limited number of components capture most spatial variance in EELS datasets.
• Successfully decomposed EELS data from doped Fe2O3 thin films.
• Enabled detailed case studies analyzing spectral features like background models, signal integrals, peak positions, and widths.

Conclusions:

• Non-negative matrix factorization is an effective tool for dimension reduction in EELS spectrum imaging.
• Supervising spectral components simplifies and enhances quantitative analysis of EELS data.
• The developed methods and codes provide a practical approach for microscopists analyzing doped thin film materials.