Laser-induced breakdown spectroscopy assisted by machine learning for olive oils classification: The effect of the experimental parameters

Abstract

Laser Induced Breakdown Spectroscopy (LIBS) was combined with some machine learning techniques, such as Principal Components Analysis (PCA) and Linear Discriminant Analysis (LDA) in order to obtain useful information concerning the classification of olive oil samples (authentication of their geographic origin) and the detection of adulteration. Since the plasma characteristics can depend substantially on the state of a sample (solid, gas or liquid) three different configurations of handling the sample were examined, spray of olive oil, a thin laminar flow and the free surface of few gr of olive oil sample. Then, the effects of experimental parameters, such as the laser energy, the temporal gating conditions (i.e., delay time and integration time) of the CCD detector of the spectrometer, on the plasma characteristics and subsequently on the classification results, were thoroughly investigated and analyzed. The combination of LIBS with the machine learning techniques used, resulted in excellent classification results of the olive oils studied, achieving classification accuracies of 100%.