LIBS and Raman image fusion: An original approach based on the use of chemometric methodologies

Laser-induced breakdown spectroscopy (LIBS) imaging is a powerful and innovative technique with a constant increasing success and interest in many scientific fields. Using LIBS imaging, it is possible to highlight the presence of atoms in complex samples of different nature to achieve important spectral and spatial information. Simple preparation of the sample, an acquisition rate that can reach a speed of 1 kHz, a high spatial resolution (in the order of μm scale) and a sensitivity in the order of ppm are among the assets of this technique. An additional valuable aspect in the current LIBS setups is the possibility to acquire with the same LIBS platform spectroscopic responses resulting from another radiation-matter interaction, such as Raman measurements. The most common data treatment approach to LIBS imaging data is still univariate, i.e., the observation of maps at certain representative LIBS wavelength, but this prevents extracting all the useful information contained in the acquired spectra and obtaining an integral understanding of the correlation between the spatial and spectral information. Chemometrics and multivariate analysis in the framework of spectral unmixing can lead to these outcomes. Therefore, the aim of this work is to show the potential of investigating simultaneously LIBS and Raman imaging spectral data acquired on the same sample with the assistance of the unmixing method Multivariate Curve resolution – Alternating Least Squares (MCR-ALS). To illustrate the value of the thorough interpretation of fused LIBS and Raman images by unmixing analysis and the steps to take place in this kind of study, a real sample of a complex polymetallic mineral formed by several mineral phases incorporating carbonates, silicates and sulphides has been used. In this paper we will show that using a pipeline analysis already validated in another work of our group, it is possible to extract pure chemical contributions of the heterogeneous aforementioned minerals. The data analysis protocol presented is valid for the investigation of LIBS and Raman spectroscopies separately, but becomes much more valuable when the two acquired data sets for the same sample are simultaneously examined, leading to new aspects that would be hindered if not investigated at the same time with a suitable fusion approach. • An interesting data fusion strategy to manage big hyperspectral data sets • A data compression approach to keep relevant chemical information • Better spectroscopic interpretations thanks to LIBS/Raman fusion