Noise reduction in X-ray photoelectron spectromicroscopy by a singular value decomposition sorting procedure

Abstract

In spite of the fact that most X-ray photoelectron spectroscopy (XPS) instruments are capable of acquiring multispectral data sets, few analysts choose to do so due to the time required to obtain an adequate signal to noise ratio. A significant reduction in the acquisition time can be achieved by the use of principal component analysis to reduce noise in the data set. Additionally a reduction in computational requirements of Principal Component Analysis (PCA) can be gained by prior application of a singular value decomposition sorting procedure. Data transformation prior to processing minimises the difficulties arising due to the non-uniform distribution of Poissonian noise through the spectra. The use of the procedure is illustrated to produce quantified elemental images from wide energy spectra, and chemical state maps requiring curve fitting to high energy resolution spectra. Finally we discuss the effect sample charging during analysis of insulators has on the images.