APPLICATION OF ELECTRON SPECTROSCOPIES AIDED BY THE PATTERN-RECOGNITION METHOD FOR QUANTITATIVE-ANALYSIS OF SOLID-SURFACES

Quantitative analysis of solid surfaces by AES or XPS involves determi nation of intensity of selected spectral features, usually the most pr onounced, recorded for the constituents of a given sample. Such experi ments are frequently accompanied by intensity measurements of the same features on standard samples. Differences between the spectra shapes corresponding to samples with different concentrations are neglected w ithin this procedure. Much less frequently, the quantitative analysis is based on variation of the shape of the selected feature in the ener gy spectra. An attempt is made in this article to perform the quantita tive analysis by relating the spectra shapes to composition using rout ines of the pattern recognition method. The proposed approach is also based on the set of standards, i.e. a set of samples with known compos ition. The photoelectron or the Auger electron peaks recorded for all standards, with height normalized to a constant value, create the 'exp erience' of a computer for the identification of samples with unknown composition. It has been found that one of the algorithms of the patte rn recognition method, the so called k-nearest-neighbour (kNN) rule, c an be effectively used in the calculations. An extensive user-friendly expert system has been developed to facilitate the calculations. Perf ormance of the proposed method is demonstrated on a set of AuPd alloys .