On the use of laser-induced breakdown spectroscopy data for mineralogical investigations – constraints and application of a clustering method

Abstract

This study examines how laser-induced breakdown spectroscopy (LIBS) data collected using a downhole deployable LIBS prototype for geochemical analysis in a fashion that imitates downhole deployment may be used for mineralogical investigations. Two chemically and mineralogically practically identical felsic rocks, namely granite and microgranite, are used to assess the effects of rock texture on mineral classification and high-resolution SEM-TIMA (scanning electron microscope coupled with TESCAN's integrated mineral analyser) mineral maps are used to reveal mineralogical composition of each LIBS ablation crater. Additionally, in order to extend the LIBS application for fast mineralogical studies to a greenfield scenario (i.e. no previous knowledge) a clustering methodology is presented for mineralogical classification from LIBS data. Results indicate that most LIBS spot analyses sample mineral mixtures, 91.2 and 100% for granite and microgranite, respectively, which challenges mineralogical classification, particularly for fine-grained rocks. Positive identification and classification of minerals of slightly different compositions relative to the bulk rock (i.e. fluorite and biotite in granitic rocks) demonstrates how minerals or minerals groups of distinct and interesting chemical compositions (e.g. sulfides or oxides in silicate-dominated rocks) can be rapidly recognized in a mineral exploration scenario. Strategies for overcoming mineral mixture issues are presented and recommendations are given for effective workflows for mineralogical analysis using LIBS data in different mineral exploration stages. Supplementary material : Complete LIBS analyses for granite and microgranite samples (Appendix A); modal mineralogy results for each LIBS ablation crater based on TIMA analysis (Appendix B); and TIMA mineral maps for granite and microgranite samples are available at https://doi.org/10.6084/m9.figshare.c.6444482