THE DETERMINATION OF UNCERTAINTIES IN QUANTITATIVE XPS AES AND ITS IMPACT ON DATA ACQUISITION STRATEGY

The accuracy of quantified XPS/AES compositions for a given sample depends on both systematic and random errors. Parameters such as escape depths, transmission functions, sensitivity factors and whether the model of the surface physical structure used in the quantification procedure is appropriate contribute systematic errors to the quantification. Random errors arise from the Poisson statistics of electron detection. However, in many applications of practical interest, where comparison between samples is the preferred approach to problem solving, precision is more important than accuracy. In such cases, the systematical errors are very similar and the random errors arising from the Poisson statistics of electron detection dominate the-problem-solving exercise. The raw spectra should be capable of providing sufficient precision in the quantified elemental compositions so that the comparison can be made with statistical confidence. This is a particular problem where the elements of interest are present at low levels and there is concern over sample degradation. This paper derives routines to determine the precision (arising from the Poisson counting statistics) in single intensity, peak height, peak area and quantified composition measurements for any acquired spectra, revealing some counter-intuitive results. The beneficial effects of smoothing and background averaging are also considered. The knowledge is used to suggest novel spectral acquisition strategies to be implemented on data systems, and how they may be optimized when a result to a desired precision is required.