The partitioning of water between olivine, orthopyroxene and melt synthesised in the system albite–forsterite–H 2O

Abstract

Coexisting forsterite, enstatite and silicate glass were synthesised from a water-bearing basalt analogue in programmed cooling experiments at 1.0, 1.5, 2.0 and 2.5 GPa. Hydrous species in each phase were characterised and quantified using infrared absorbance spectroscopy. Al 3+-related hydroxyl defects, with vibrational energies between 3400–3200 cm − 1 , dominate the hydroxyl budget of synthetic forsterite. The vibrational characteristics of these defects indicate that similar absorptions in Fe-bearing systems result from Fe 3+-related OH species. This interpretation resolves a previously recognised, but unexplained, relationship between oxygen fugacity and hydroxyl speciation in experimentally annealed olivines, and suggests that the intensities of bands at 3400–3300 cm − 1 could form the basis of an oxybarometer. OH defect speciation in enstatite is similar to that in aluminous crystals synthesised at high water fugacity and many natural orthopyroxenes. Hydroxyl speciation in orthopyroxene is not, therefore, controlled by water activity. Total water contents of synthetic enstatites from any single experiment vary enormously. This heterogeneity is attributed to variable contributions from a non-intrinsic broad-band hydrous component. Intrinsic hydroxyl concentrations dissolved in enstatite are identical in all crystals from an experiment. The partitioning of water between forsterite, enstatite and coexisting silicate melt is independent of pressure, temperature and total water content over the range of conditions studied, as long as the influence of these parameters on phase compositions is taken into account. D H 2O Fo/melt = 0.0001–0.0003 ( n = 4) and is insensitive to pressure or olivine aluminium contents. D H 2O En/melt is strongly controlled by Al 2O 3 concentration and ranges from 0.003–0.016 ( n = 6). Two samples contain both forsterite and enstatite and yield D H 2O En/Fo = 25 ± 1.