Improving the understanding and optimization of LIBS analysis using chemometrics
Abstract
Laser-Induced Breakdown Spectroscopy is relatively simple to implement experimentally, but its underlying physics is very complex. This is especially due to two main factors. First, sampling and excitation occur in the same process, driven by the laser-matter interaction, and therefore are strongly coupled. Secondly, laser ablation, plasma formation and emission are transient by nature, and plasma features vary very rapidly with time.As a consequence, this observation has stimulated the use of more empirical approaches based on chemometrics, which have been more and more widespread in LIBS for a dozen years. The power of multivariate methods for LIBS data treatment was indeed demonstrated in many papers. However, in most cases, those approaches aim at retrieving some chemical information on the sample, either qualitative (presence of some elements, sample nature) or quantitative (concentration measurements).But LIBS spectra contain more information than the chemical one that could also be exploited by chemometric methods. Thus, multivariate models can be used or interpreted in order to diagnose possible instrumental issues, to optimize a LIBS system or analysis for a given application, or to improve our physical understanding of LIBS data. The talk will illustrate the use of chemometric methods to address these objectives.
Origin : Files produced by the author(s)