SODIUM PARTITIONING BETWEEN CLINOPYROXENE AND SILICATE MELTS

We present a quantitative model for Na partitioning between clinopyroxene and silicate melt, applicable in the range 0.1 MPa to 6.0 GPa, 1000-2000 degrees C over a wide span of composition. Our model is derived from the jadeite melting equilibrium, via the relationship Delta G(f(P,T)) = RT1nK(f) = RT1nD(Na)*, where Delta G(f(P,T)) is the Gibbs free energy of fusion of jadeite at the pressure (P) and temperature (T) of interest, T is in kelvins, R is the gas constant, K-f is the equilibrium constant for the the melting reaction and D-Na*, is a molar partition coefficient, defined here as the molar ratio of Na in crystal to Na in melt on a six-oxygen basis. An expression for Delta G(f(P,T)) is obtained from published experimental data on the jadeite melting curve from 2.8 to 16.5 GPa, combined with available (or estimated) thermochemical data for jadeite crystal and jadeite melt. The model is tested against: (1) new experimental data in the system diopside-albite from 0.1 MPa to 6.0 GPa; (2) new analyses of clinopyroxene and glass from the mid-ocean ridge basalt-pyrolite sandwich experiments of Falloon and Green (1988); and (3) published experimental data from natural and synthetic systems. The model is in good agreement with the experimental data over the entire P-T range investigated. The only systematic deviations occur in oxidizing systems (f(O2)FMQ+1), due to stabilization of the NaFe3+Si2O6 (acmite) component in clinopyroxene, and on the diopside-albite join at 0.1 MPa. The latter can be quantitatively ascribed to nonideality of melts on the diopside-silica join in the system CaMgSi2O6-NaAlSi2O6-Si3O6. The activity-composition relationships adopted for crystal and melt do not require prior knowledge of the aluminum content of either phase. As D-Na* is typically within 3% relative of the weight partition coefficient (D-Na) our thermodynamic model provides a quantitative description of Na partitioning between clinopyroxene and melt in the upper mantle.