Micro-analytical study of the optical properties of rainbow and sheen obsidians

Samples of Mexican obsidian that exhibit either "sheen" or ''rainbow" optic al properties were examined with a combination of EMP, SEM, TEM, as well as visible and IR spectroscopy. Electron-microprobe analyses of the sheen mat rix give (in wt.%): 76.2 SiO2, 0.2 TiO2, 11.6 Al2O3, 2.2 FeOTOT, 0.07 MgO, 0.1 CaO, 4.8 Na2O and 4.3 K2O. The sheen is attributed to the presence of a ligned flow-stretched hollow vesicles in the gemological literature, SEM im ages show that many of the flow-aligned lenticular areas are a second rhyol ite glass with, on average, 74.6 SiO2 0.2 TiO2, 12.7 Al2O3, 2.1 FeOTOT, 0.1 MgO, 0.9 CaO, 5.6 Na2O and 4.6 K2O. These two compositions do not overlap at the 2 sigma level. Their inferred indices of refraction differ by as muc h as 0.04, leading to optical interference along the elliptical interfaces of the two glasses. Thus we postulate that the sheen reflects differ differ ences in indices of refraction (eta) between the matrix obsidian and the lo wer eta of either gas-filled or glass-filled vesicles. In our sample, the p resence of the second glass probably correlates with incorporation (and rem elting?) of fragments of an earlier rhyolitic ash or tuff. Two different ty pes of Mexican rainbow obsidian were studied. The first has layers of numer ous trachytically oriented rods (0.2-2 by 10-20 mum) of hedenbergire (Ca0.8 8Mg0.07Fe0.98Mn0.06Si2.01O6). The composition of the matrix is: 76.3 SiO2, 12.5 Al2O3, 1.7 FeOTOT, 0.01 MgO, 0.16 CaO, 4.4 Na2O and 4.6 K2O. The secon d type has trachytically aligned plagioclase (similar to An(20)), also rod- shaped (as small as 0.5 x 2.0 mum). The composition of the matrix is: 76.1 SiO2, 13.5 Al2O3, 0.7 FeOTOT, 0.09 MgO, 0.7 CaO, 3.75 Na2O and 4.85 K2O. Mu ltiple hypotheses are considered for the possible cause ol the rainbow effe cts: gas or fluid inclusions, a small component of scattering centers, diff erential indices of refraction, Bragg diffraction of visible light, and thi n-film interference. Our data support the last hypothesis.