Sunday, 5 October 2008
George R. Brown Convention Center, Exhibit Hall E
Determination of gemstone provenance continues to be a difficult yet important endeavor in the gem industry (pricing, insurance, appraisal, fraud) and in efforts to eliminate gems in terrorist financing. Provenance models depend on large databases; one obstacle has been finding a technique that provides a rich chemical signature of the gem while imposing negligible damage. This study uses LIBS analysis of ca. 100 beryl (Be3Al2Si6O18) samples to develop a geochemical model for beryl provenance. In LIBS analysis, a pulsed Nd-YAG laser, operating at wavelength of 1064 nm and 91 mJ, is focused on the gem surface, causing ablation and excitation of atoms in a short-lived plasma. During cooling, atoms emit photons as electrons decay to lower-energy orbitals. This light is collected with an optic fiber, diffracted by seven Echelle gratings in Czerny-Turny configurations, and recorded on CCD cameras. The spectra, which contain intensities of most elements, are analyzed by Principal Components Analysis (PCA) using the program The Unscrambler. PCA reduces the complex signal to a few chemical variables (principal components), each of which has contributions from many of the wavelengths in the spectrum. Three PCA models were tested. Using the entire broadband spectrum, all beryls grouped together; only a few regions could be cataloged by provenance. The second method used was a PCA analysis of only the most significant elements in the first model: Li, Be, Ca, and Na. The resulting model was very similar to the broadband model. The third PCA model eliminated all Li, Be, Ca, and Na peaks and only used the peaks of lower intensity. The results were a little different and displayed some provenance groupings, but not in a definitive way. This study suggests that LIBS could be used to determine gem provenance if a sufficiently large data base were available.