Signal distribution impact on limit of detection in laser-induced breakdown spectroscopy

Abstract

The detailed study of the different signal statistics in laser induced breakdown spectroscopy (LIBS) measurements has been carried out. It was demonstrated that the Nd:YAG laser pulse energy distribution function is not described by the Gaussian distribution function, but can follow the normal distribution if proper lasing conditions (pumping level, Q-switch type, etc.) are chosen. The time gated LIBS signals (atomic lines intensity, plasma background emission, acoustic signal) have been shown to be biased from a Gaussian distribution function. Alternatively, distribution function of the time and spectra integrated plasma emission corresponded to the normal distribution. The normalization procedure did not change the distribution functions for LIBS signals so the generalized extreme value distribution was a better choice. In case of the generalized extreme value distribution for the LIBS signals, the limits of detection should be defined by distribution functions rather than calibration curve and “3-sigma criteria” approach.