A neutron powder diffraction study of cation ordering in high-temperature synthetic amphiboles
Welch, Mark D.; Knight, Kevin S.
European Journal of Mineralogy Volume 11 Number 2 (1999), p. 321 - 332
published: Apr 19, 1999
manuscript accepted: Nov 23, 1998
manuscript received: Dec 8, 1997
ArtNo. ESP147051102007, Price: 29.00 €
Abstract The structures of synthetic fluoro-edenite and pargasite end-members have been refined by Rietveld analysis of neutron powder diffraction data to Rwp values of 1.4 % and 1.9 %, respectively. The quality of the Rietveld refinements is much higher than has been obtained for synthetic amphiboles by X-ray powder diffraction. The distribution of Al and Si over T(1) and T(2) has been determined from mean T(1)-O bond lengths. Pargasite crystallized at 1 kbar, 932 °C has 1.71 ± 0.11 Al apfu at T(1) and 0.28 + 0.11 Al apfu at T(2), implying at least 15 % long-range Al-Si disorder. Within experimental error fluoro-edenite crystallized at 2 kbar, 1006°C is long-range-ordered, with all Al at T(1). The size of the error on the long-range Al-Si order parameter (Ψ) scales approximately with 2/Al. This means that an amphibole with one Al apfu will have an error on Ψ that is approximately twice that for an amphibole with two Al apfu. Consequently, Al-Si disorder can be quantified more accurately for pargasites than hornblendes. The Al-Si distribution of synthetic pargasite is similar to that of natural high-temperature pargasites. The results of this study indicate that long-range Al-Si order-disorder in pargasites rather than hornblendes is worth investigating as a geothermometer because of the inherently larger errors on Ψ for hornblendes. Mg-Al ordering on octahedral sites in pargasite has been determined from mean octahedral bond lengths. Al is at M(2) and M(3) sites, with little or none at M(l): M(2) = 1.5Mg + 0.5Al and M(3) = 0.5Mg + 0.5Al. These occupancies accord very well with IR and 1H MAS NMR spectra of synthetic end-member pargasite and are different from those of high-temperature natural pargasites which have M3+ distributed statistically over M(2) and M(3), with none at M(1): M(2) = 1.33M2+ + 0.67M3+ and M(3) = 0.67M2+ + 0.33M3+. The difference in M(2) and M(3) site occupancies for natural and synthetic pargasites may reflect fundamentally different growth mechanisms. However, both natural and synthetic pargasites show qualitatively similar ordering behaviour for octahedral cations, pointing to a shared crystal chemistry.