Use of Principal Component Analysis for the Identification and Mapping of Phases from Energy-Dispersive X-Ray Spectra
NAVAL POSTGRADUATE SCHOOL MONTEREY CA
Pagination or Media Count:
Multivariate statistical analysis methods such as Principal Component Analysis PCA are now finding applications in electron microscopy including the analysis of energy dispersive x-ray spectra EDS. The aim in this thesis is to extend EDS beyond its conventional use in the measurement of elemental distributions to allow both the identification of chemical phases and the mapping of their distribution. In the present work, PCA was applied to the analysis of modeled spectra representing interfaces where diffusion andor an interface reaction had occurred. A search routine was developed to find physically possible interface phases using the principal components found by PCA. From the modeled data, it was shown that an interface phase could, in principle, be found using PCA but that it is embedded in a cluster of physically possible spectra. The technique was then applied to experimental data obtained from an interface between chemically vapor deposited diamond CVDD and Cr2O3. The results followed the same pattern as was seen with the modeled data. Criteria for experimental EDS spectra other than those used to define a physically meaningful spectrum are discussed. These should help further limit the cluster of possible answers found allowing a correct determination of the real interface phase.
- Statistics and Probability
- Atomic and Molecular Physics and Spectroscopy