GSA Annual Meeting in Phoenix, Arizona, USA - 2019

Paper No. 6-9
Presentation Time: 11:00 AM


MCMILLAN, Nancy J., Geological Sciences, New Mexico State University, P.O. Box 30001, Las Cruces, NM 88003, DUTROW, Barbara L., Dept. of Geology and Geophysics, Louisiana State Univ, Baton Rouge, LA 70803-4101, HENRY, Darrell J., Dept. of Geology and Geophysics, Louisiana State University, Baton Rouge, LA 70803 and MCMANUS, Catherine E., Materialytics, LLC, P.O. Box 10988, Killeen, TX 76547

Tourmaline (tur) is an excellent provenance indicator mineral. To develop a new technique for analyzing detrital and matrix tur, with the goal of relating tur to its host rock and evaluating clastic sediment provenance, Laser-Induced Breakdown Spectroscopy (LIBS) was utilized. To optimize the chemical fingerprint obtained, novel methodologies were developed.

LIBS analysis can be done with little sample preparation. Thumbnail samples were analyzed “as-is” without cutting or polishing. Samples were cleaned in isopropyl alcohol and lint-free tissue.

Analytical settings were chosen to maximize peak-to-background. Laser power was 150 mJ, time between the laser shot and light collection was 0.5 µs, light collection time was 10 µs, and spot size was 50 µm. Initially, 8 x 8 grids (64 shots) were used for data collection with line spacing of 100 µm. However, plasma fallout of early shots contaminated late shot locations. Analysis of B peak intensities in spectra from grids placed over a tur-quartz boundary suggested that line spacing must be at least 800 µm for these settings. The number of grids required to obtain a representative analysis was investigated by acquiring 16 2 x 2 grids; spectra were averaged in groups of 4 shots up to 16 grids. In Principal Component Analysis (PCA), tur spectra converged when 10-16 grids were averaged, suggesting that 40-64 shots are necessary.

Three drift samples were analyzed first and last daily. In PCA, spectra from each sample formed morning and evening groups, defining daily drift. However, PCA of spectra from all three samples shows excellent grouping of each sample’s spectra; drift does not inhibit the ability to distinguish samples. Because drift is a non-linear function of wavelength due to spectrometer thermal expansion, it is difficult to correct mathematically for the entire spectrum in multivariate analysis.

Individual detrital tur grains were mounted on cardstock with double-sticking tape. Each grain was analyzed with three laser shots. Some grains tolerated three shots; others shattered after one or two. Spectra were inspected for a C peak from the tape; C-bearing spectra were eliminated. Spectra of each grain (1-3 shots) were averaged for modeling. With these methods, 700 thumbnail and ~1000 detrital grains have been analyzed allowing insightful provenance interpretations.