Robust Mineralogy Analysis Utilizing Laser-Induced Breakdown Spectroscopy and Dispersive X-Ray Spectroscopy

This research delves into the utilization of Laser-Induced Breakdown Spectroscopy (LIBS) for automating mineralogy, particularly within the realm of in-situ resource exploitation This includes a combination of elemental detection, mineral recognition as well as specific mineralogy processes present in the mineral forming phase. With an increasing interest in rare-element deposits, particularly those housing lithium (Li), there's a pressing demand for dependable detection and spatial 2D elemental mapping of light elements like Li, beryllium (Be), boron (B), or fluorine (F). Conventional analytical methods face challenges in this specialized domain, where the emphasis lies on swift, straightforward, and accurate detection with potential for in-situ deployment. However, Laser-Induced Breakdown Spectroscopy (LIBS) coupled with sophisticated data processing emerges in combination with a precise reference analytical method, Energy-Dispersive X-ray Spectroscopy (EDX) as a promising solution to address these challenges. In this work, we focused on the analysis of large-scale geological samples combining LIBS and EDX. Hence advantages of both methods are being fully exploited, precise mineral identification based on extensive mineral libraries by the EDX in combination with multi-elemental analysis with relatively high sensitivity for selected light elements by the LIBS. By combining these methods, we were able to not only determine the accurate mineral composition of the selected samples by both methods but also observe specific mineral and phase processes, bringing more insight into the sample origin and formation. This information and methodology can be further exploited in inter-celestial analysis. As for the EDX, the mineralogy is fully automated in the data processing phase. As the goal is to focus on the possibility of generating a similar process for the LIBS, the main focus of this research is on the analytical method. The research showcases the strong suitability of LIBS in identifying and spatial mapping the major elements (Si, Al, Fe, Mg, Mn, Ca, Na), light elements (Li, Be, B, F), and trace elements (Ge, Ga, Cu, Zn, Ti, Sr, Ba) within pegmatites. It elucidates their distinct geochemical fingerprints across various stages of rock evolution, including magmatic, metasomatic, and hydrothermal processes. To use the LIBS for mineral identification, the k-means clustering method was utilized on the hyperspectral cube generated for each analyzed sample. As this method is in the category of unsupervised machine learning, we used the reference information from the EDX to specify the minerals within the sample.