Principal component analysis: a versatile method for processing and investigation of XPS spectra.

摘要: Given the relevance of principal component analysis (PCA) to treatment spectrometric data, we have evaluated potentialities and limitations such useful statistical approach for harvesting information in large sets X-ray photoelectron spectroscopy (XPS) spectra. Examples allowed highlighting contribution PCA data by comparing results this with those obtained usual XPS quantification methods. was shown improve identification chemical shifts interest reveal correlations between peak components. First attempts use method led poor results, which showed mainly distance series samples analyzed at different moments. To weaken effect variations minor interest, a normalization strategy developed tested. A second issue encountered spectra suffering an even slightly inaccurate binding energy scale correction. Indeed, channels lead being performed on incorrect variables consequently misleading information. In order correction speed up step pretreatment, processing based used. Finally, overlap sources variation studied. Since intensity given channel consists electrons from several origins, having suffered inelastic collisions (background) or not (peaks), cannot compare them separately, may confusion loss By extracting peaks background considering as new variables, elemental composition could be taken into account case very backgrounds. conclusion, is diagnostic tool interpretation spectra, but it requires careful appropriate pretreatment.