Surface analysis insight note: Multivariate curve resolution of an X-ray photoelectron spectroscopy image

This Insight Note follows a series of three previous insight notes on X-ray photoelectron spectroscopy image analysis that focused on the importance of analyzing the raw data, the use of summary statistics, and principal component analysis (PCA). The same X-ray photoelectron spectroscopy image data set was analyzed in all three notes. We now show an analysis of this same data set using multivariate curve resolution (MCR). MCR is a widely used exploratory data analysis method. Because of MCR's nonnegativity constraints, it has the important advantage of producing factors that look like real spectra. That is, both its scores and loadings are positive, so its results are often more interpretable than those from PCA. The requirements for preprocessing data are also, in general, lower for MCR compared with PCA. To help determine the number of factors that best describe the data set, a series of MCR models with different numbers of factors was created. Based on the chemical reasonableness of its factors, a two-factor model was selected. Scores plots/images show the regions of the image that correspond to these two factors.