2015 GSA Annual Meeting in Baltimore, Maryland, USA (1-4 November 2015)

Paper No. 323-11
Presentation Time: 4:20 PM

­CLASSIFICATION AND GEOGRAPHIC ORIGIN OF GARNETS USING LASER-INDUCED BREAKDOWN SPECTROSCOPY (LIBS)


HARK, Richard R., Dept. of Chemistry, Juniata College, Huntingdon, PA 16652, WISE, Michael A., Dept. of Mineral Sciences, Smithsonian Institution, P.O. Box 37012, Washington, DC 20013-7012, DEFNET, Peter A., Department of Chemistry, Juniata College, Huntingdon, PA 16652 and HARMON, Russell S., Dept. of Marine, Earth, & Atmospheric Sciences, North Carolina State University, Raleigh, NC 27695, wisem@si.edu

The garnet supergroup is distinguished by silicate minerals that are isostructural with garnet and conforms to the general compositional formula {X3}[Y2](Z312, where the ‘X’ site is occupied by rather large divalent cations (e.g., Ca2+, Mg2+, Fe2+, or Mn2+), the ‘Y’ site hosts smaller trivalent cations (e.g., Al3+, Fe3+, or Cr3+), the ‘Z’ site is filled mainly by Si4+ and the anionic site “φ” contains O, OH or F (Grew et al. 2013). The supergroup also includes some oxides, hydroxides, vanadates, arsenates and one halide, but the most widespread members are the six common silicate species that belong to the garnet group (i.e., almandine, spessartine, pyrope, grossular, andradite, and uvarovite). Previously, Alvey et al. (2010, Appl. Opt. 49, C168) analyzed 157 samples described as garnets from 92 locations worldwide using laser-induced breakdown spectroscopy (LIBS) and demonstrated that LIBS has the potential to discern garnet geographic origin using the concept of ‘geochemical fingerprinting’. In this study, an expanded suite of over 203 garnets was examined using electron microprobe microanalysis (EPMA) to correctly classify the samples on the basis of their major element composition and then ascertain the efficacy of LIBS for classification and provenance determination. The LIBS spectral data were processed using multivariate statistical pattern recognition methods (e.g., PCA, PLSDA) and the resulting classification models then used to successfully classify unknown garnet samples of a specific compositional type according to their geographic origin. Various preprocessing techniques (normalization, spectral outlier removal) were implemented to optimize the LIBS classification results. This study demonstrates that LIBS offers a good alternative for identification of unknown garnet samples based on their geographic provenance.