COPPER-BEARING TOURMALINE SOURCES: EVIDENCE FROM LASER-INDUCED BREAKDOWN SPECTROSCOPY (LIBS) AND ELECTRON MICROPROBE ANALYSES (EMP)

Copper-bearing tourmalines (turs) exhibit some of the most exquisite colors in the mineral kingdom. Their saturated, vivid ‘neon’ blues and greenish blues are highly desired gemstones, yet other colors, from purples to greens to yellows, also exist. Most gem species are elbaites and fluor-elbaites (Na-Li-Al tur) or liddicoatites (Ca-tur) that contain significant Cu (0.5-3 wt% CuO) and Mn (0.2-2 wt% MnO), as shown by EMP. The spectacularly colored Cu-bearing turs, originally discovered at São José da Batalha (SJdB), Paraíba, Brazil, and later from other localities in Brazil as well as from Mozambique and Nigeria, are commonly referred to as “Paraíba” tur. However, the provenance or country may have a substantial impact on the gemstone price with the Brazilian Paraíba tur commanding the highest price. Most studies determine the likely source area by evaluating combinations of tur trace elements obtained by LA-ICP-MS.

To explore an additional method for determining provenance, 54 samples of Cu-bearing turs from known source locations were analyzed using Laser-Induced Breakdown Spectroscopy (LIBS) and results evaluated with multivariate statistics (PCA, PLSR). Individual grains ranged in size from ~1 mm to > 1 cm. While the sample set is limited: with 20 grains from Rio Grande del Norte (RGdN) and SJdB, Brazil; 24 from Mozambique; and 10 from Nigeria; these data provide initial indications of the efficacy of this method. Up to 64 laser shots, depending on grain size, were obtained per sample. Spectra consisted of relative intensities of > 26,000 wavelengths recorded from each shot and were averaged to produce a single spectrum for each grain. PCA modeling of the spectra reveal a tight clustering of the SJdB samples with no overlap with other groups. Mozambique grains display the largest variation and Nigerian samples fall into two distinct clusters that do not correlate with color or number of analyses. PLSR modeling of the spectra clearly distinguished the various localities. SJdB was modeled first resulting in a tight cluster, these samples were then removed from the matrix; RGdN followed, which also clustered tightly with no overlap, followed by the Nigerian samples. The final grouping resulted in the separation of Mozambique samples from others. These techniques may provide an additional tool for deciphering Cu-tur sources.