Aspects of using the factor analysis for XPS data interpretation

In this work, the use of factor analysis to chemical state quantification of XPS data is studied. First, the theory of the method is reviewed with a special emphasis on the issues related to XPS data analysis. In particular, we concentrate on the transformation of the abstract components into physically meaningful ones in the case where reliable reference spectra are not available. We have observed that in the commonly used iterative target transformation factor analysis (ITTFA), in which a delta peak serves as the initial guess, the shape of the obtained component depends strongly on the position of the delta peak and on the minimum allowed intensity level. We propose an approach in which these parameters are varied in order to generate different representations for each component of the data. With simulated model data we show that if the variation is done with a sufficiently small step size, the correct representation will be generated. We also show that in the case of two-component data the iteration of the components is not necessary because a position can be found where a delta peak directly transforms into the correct component without unphysical features. Besides the model data, the proposed method is applied to experimental 2p photoelectron spectra of iron and chromium oxides.