Multivariate Statistics Applications in Scanning Transmission Electron Microscopy X-Ray Spectrum Imaging

• Published in: International review of neurobiology Vol. 168; pp. 249 - 295
• Main Author: Parish, Chad M
• Format: Journal Article
• Language: English
• Published: 01.01.2011
• ISSN: 0074-7742
• DOI: 10.1016/B978-0-12-385983-9.00005-3

A modern scanning transmission electron microscope (STEM) fitted with an energy-dispersive X-ray spectroscopy (EDS) system can quickly and easily produce spectrum image (SI) datasets that contain so much information (hundreds to thousands of megabytes) that they cannot be comprehensively interrogated by a human analyst. Therefore, advanced mathematical techniques are needed to glean materials science and engineering insight into the processing-structure-property relationships of the examined material from the SI data. This review discusses recent advances in the application of multivariate statistical analysis (MVSA) methods to STEM-EDS SI experiments. In particular, the fundamental mathematics of principal component analysis (PCA) and related methods are reviewed, and advanced methods such as multivariate curve resolution (MCR) are discussed. The applications of PCA- and MCR-based techniques to solve difficult materials science problems, such as the analysis of a particle fully embedded in a matrix phase, are discussed, as well as the effects that can confuse the results of MVSA computations. Possible future advances and areas in need of further study are also discussed.