Rapid geochemical imaging of rocks and minerals with handheld laser induced breakdown spectroscopy (LIBS)

• Published in: Journal of geochemical exploration 2021-03, Vol.222, p.106694, Article 106694
• Main Authors: Lawley, Christopher J.M., Somers, Andrew M., Kjarsgaard, Bruce A.
• Format: Article
• Language: English
• Subjects: Element mapping; Geochemical imaging; LIBS; Machine learning; Mantle xenolith; Spectroscopy
• ISSN: 0375-6742; 1879-1689
• DOI: 10.1016/j.gexplo.2020.106694

Geochemical imaging is a powerful tool for unravelling the complex geological histories of rocks and minerals. However, its applications have until recently been restricted to geological research in a lab environment due to the cost and size of conventional instrumentation, long analysis times, and extensive sample preparation for some methods. Herein we present a rapid, qualitative geochemical imaging method for rocks and minerals using handheld LIBS. Analyses were completed directly on sawed drill core surfaces for a suite of kimberlite-hosted mantle xenoliths (Jericho and Muskox kimberlites, Nunavut, Canada). Semi-automated LIBS spectral processing following a new open-source workflow allows stitching of multiple small-area maps (each approximately 3 × 3 mm that take 2–3 min to complete) to produce cm-scale geochemical images of variably altered mantle xenolith samples (total data acquisition in 1–2 h). Replicate analyses of a ZnAl alloy reference material (NZA-1; CANMET) that were undertaken during standard-sample bracketing suggests that the relative standard deviation (RSD) is typically 15–20% for sum-normalized emission intensities above the estimated background. We demonstrate with open-source machine learning tools how qualitative LIBS spectral data can be converted to Feature-Of-Interest (FOI) maps to distinguish a variety of metasomatic and alteration features (e.g., Cr-diopside, kelyphite rims on pyrope garnet, and calcite veinlets) from the primary mantle mineralogy (e.g., olivine and orthopyroxene). Our results further demonstrate that the resolution of handheld LIBS-based geochemical imaging is sufficient to map veinlets and grain boundaries lined with metasomatic minerals. The LIBS approach is particularly sensitive for mapping the microscale distribution of elements with low atomic number (e.g., Li and Na). These light elements are difficult to detect at low concentrations with other handheld and field-portable technologies, but represent important geochemical tracers of hydrothermal and magmatic processes. Rapid LIBS mapping thus represents an emerging geochemical imaging tool for unravelling the complex geological history of rocks and minerals in the field with minimal to no sample preparation. [Display omitted] •New open-source workflow for geochemical imaging with handheld LIBS•Analyses completed on sawed drill core with minimal sample preparation•Qualitative LIBS spectra converted to mineralogy with machine learning tools•Results used