AUTHOR=Han Shujun , Lӧhr Stefan C. , Abbott April N. , Baldermann Andre , Farkaš Juraj , McMahon William , Milliken Kitty L. , Rafiei Mehrnoush , Wheeler Cassandra , Owen Michael TITLE=Earth system science applications of next-generation SEM-EDS automated mineral mapping JOURNAL=Frontiers in Earth Science VOLUME=10 YEAR=2022 URL=https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2022.956912 DOI=10.3389/feart.2022.956912 ISSN=2296-6463 ABSTRACT=

Sedimentary rocks contain a unique record of the evolution of the Earth system. Deciphering this record requires a robust understanding of the identity, origin, composition, and post-depositional history of individual constituents. Petrographic analysis informed by Scanning Electron Microscope - Energy Dispersive Spectroscopy (SEM-EDS) mineral mapping can reveal the mineral identity, morphology and petrological context of each imaged grain, making it a valuable tool in the Earth Scientist’s analytical arsenal. Recent technological developments, including quantitative deconvolution of mixed-phase spectra (producing “mixels”), now allow rapid quantitative SEM-EDS-based analysis of a broad range of sedimentary rocks, including the previously troublesome fine-grained lithologies that comprise most of the sedimentary record. Here, we test the reliability and preferred mineral mapping work flow of a modern Field-Emission scanning electron microscope equipped with the Thermofisher Scientific Maps Mineralogy mineral mapping system, focusing on mud/siltstones and calcareous shales. We demonstrate that SEM-EDS mineral mapping that implements 1) a strict error minimization spectral matching approach and 2) spectral deconvolution to produce ‘mixels’ for mixed-phase X-ray volumes can robustly identify individual grains and produce quantitative mineralogical data sets comparable to conventional X-ray diffraction (XRD) analysis (R2 > 0.95). The correlation between SEM-EDS and XRD-derived mineralogy is influenced by mineral abundance, processing modes and mapped area characteristics. Minerals with higher abundance (>10 wt%) show better correlation, likely the result of increased uncertainty for XRD quantification of low-abundance phases. Automated spectral deconvolution to produce ‘mixels’ greatly reduces the proportion of unclassified pixels, especially in the fine-grained fraction, ultimately improving mineral identification and quantification. Mapping of larger areas benefits bulk mineralogy analysis, while customized area size and shape allows high-resolution in situ mineralogical analysis. Finally, we review SEM-EDS-based mineral mapping applications in the Earth Sciences, via case studies illustrating 1) approaches for the quantitative differentiation of various mineral components including detrital (allogenic), syndepositional (authigenic) and burial diagenetic phases, 2) the origin and significance of lamination, 3) the effectiveness and appropriateness of sequential leaching in geochemical studies, and 4) the utility of mineral maps to identify target grains within specific petrological contexts for in situ geochemical or geochronological analysis.