Calculation and optimization of sample identification by laser induced breakdown spectroscopy via correlation analysis

Linear correlation analysis may be used as a technique for the identification of samples with a very similar chemical composition by laser induced breakdown spectroscopy. The spectrum of the “unknown” sample is correlated with a library of reference spectra. The probability of identification by single shot analysis can be estimated by the method described in this paper. When a right identification is not obtained by single shot correlation analysis the accuracy can be increased by averaging spectra or by averaging the correlation coefficients. The number of spectra or correlation coefficients to be averaged to obtain a 99.9% right identification is evaluated. We found out that the number of spectra to be averaged is equal to the number of correlation coefficients to be averaged. The benefit of using averaged spectra over averaging correlation coefficients is a faster calculation.