Non-Gaussian Signal Statistics’ Impact on LIBS Analysis

Abstract

A detailed study has been carried out to reveal signal statistics’ impact on analysis sensitivity in laser-induced breakdown spectroscopy (LIBS) measurements. For several signals measured simultaneously, it was demonstrated that space-, spectra- and time-integrated plasma emission followed a normal distribution while the spectra- and time-resolved LIBS signal (atomic line intensity, plasma background emissions) distribution functions were biased compared to a Gaussian distribution function. For the first time in LIBS, the impact of a non-Gaussian distribution function on the limit of detection (LOD)’s determination has been studied in detail for single-shot spectra as well as for averaged spectra. Here, we demonstrated that the non-symmetrical distribution of the LIBS signals influenced the estimated LODs, so knowledge of a LIBS signal’s distribution function provides more reliable results, and the analysis sensitivity can be wrongly estimated if Gaussian distribution is presumed.

Keywords

• laser-induced breakdown spectroscopy
• single-shot measurements
• signal distribution function
• limits of detection
• non-Gaussian distribution
• generalized extreme value distribution