Automatic Mineral Identification Using Raman Spectroscopy for Planetary Exploration: Implications for RLS and Supercam Instruments

The acquisition of mineralogical data of bodies beyond the Earth constitutes a technical challenge. As equipment performance increases, so do data processing techniques. Intended for that purpose, this work proposes a new signal-processing methodology in the context of space exploration using Raman Spectroscopy. This work is performed in the framework of the RLS instrument for the ExoMars mission that, after several delays, will be launched to Mars in 2028. This new method conducts an automatic detection of minerals from Raman spectra. Based on the accumulation of spectra by multiplication, this work combines several filters and algorithms to identify the mineral composition of rocks. The multiplication of several Raman spectra corresponding to the same point of a sample allows the promotion of the peak intensity associated with Raman vibrations from those that constitute the base level of noise. A detection algorithm is created by establishing a logic tree of criteria that includes filters of intensity, width, identifications per iteration and the number of iterations. This algorithm improves the number of peaks detected and reduces the false positives compared to previously existing algorithms. The method has been trained with spectra acquired with a double-track Kaiser Optical System Raman spectrometer with a wavelength of 532nm. It hasbeen verified with spectra acquired with the RLS Simulator spectrometer, with similar characteristics to the flight device of the ExoMars mission. The samples selected for the evaluation of the method have high astrobiological interest, including vermiculite, lizardite and antigorite among others. The automatic peak identification algorithm has then been coupled with an iterative Raman peak comparison process for automatic mineral identification. The proposed algorithm has drastically improved the automated classification of minerals based on their hydration level. Additionally, the final method facilitatesthe identification of the majormineral phase of a sample with a degree of success in 100% of cases of the evaluated samples, with a 50% success for at least one secondary phase in mineral mixtures. The creation ofan automaticmineral classification method could be combined with the automatic sampling that rovers, such as Rosalind Franklin will carry out on the ExoMars mission, to allow a better data classification and to maximize the efficiency and scientific return using Raman spectroscopy on planetary exploration missions.