Differential evolution optimization of Rutherford backscattering spectra

Abstract

We investigate differential evolution optimization to fit Rutherford backscattering data. The algorithm helps to find, with very high precision, the sample composition profile that best fits the experimental spectra. The capabilities of the algorithm are first demonstrated with the analysis of synthetic Rutherford backscattering spectra. The use of synthetic spectra highlights the achievable precision, through which it becomes possible to differentiate between the counting statistical uncertainty of the spectra and the fitting error. Finally, the capability of the algorithm to analyze large sets of experimental spectra is demonstrated with the analysis of the position-dependent composition of a Sr[Formula: see text]Ti[Formula: see text]O[Formula: see text] layer on a 200 mm silicon wafer. It is shown that the counting statistical uncertainty as well as the fitting error can be determined, and the reported total analysis uncertainty must cover both.