Previous page     Contents     Next page

Melt-solid fractionation processes are responsible for the formation of chemical heterogeneities in planetary interiors and play an important role in the evolution of the Earth's mantle. The earliest differentiation of the mantle, possibly occurring during the crystallization of a magma ocean, and later melt extraction to form oceanic and continental crusts were governed by mineral-melt fractionation. Melt extraction in the present mantle occurs mostly at pressures below 5 GPa. Hotter plumes rising through the Archean mantle, however, could have experienced initial melting in the upper part of the lower mantle. The lack of precise and reliable trace element partitioning coefficients between melt and residual minerals stable in the lower mantle and the transition zone have caused uncertainties and debate over the possible consequences of fractionation processes related to the crystallization of a magma ocean and to melt separation from hot plumes.

The mineral-melt partitioning of trace elements is fairly well documented at pressures up to about 10 GPa. At higher pressures, however, melting experiments and partitioning studies conducted in the multianvil apparatus have been plagued by large thermal gradients of more than 200^oC/mm across the samples. Such thermal gradients are essentially unavoidable when the cylindrical heaters in the pressure cells have compressed lengths of only 6-4 mm and mid-point temperatures of 1800-2400°C. Melt migration toward the hot spot by solution and precipitation processes characterises such experiments and true equilibrium within the experimental charge is prevented, even if local equilibrium is achieved. In this study, melting relations and trace element partitioning in the 20-26 GPa range are being determined using larger volume 18/8 and 10/4 type assemblies with the use of 1000 and 5000 tonne multianvil presses, respectively. As part of this study a new 18/8 configuration was developed and calibrated. High-temperature calibrations are performed at 2000-2200°C and 20-26 GPa with compositions on the enstatite-pyrope join.

Two different peridotite compositions (KLB-1 and model pyrolite) under both anhydrous and slightly hydrous (1 wt% H₂O) conditions were used, and the effect of oxygen fugacity variations on phase compositions was evaluated by using Re-capsules with some duplicate experiments using separate additions of Fe metal and ReO₂. Oxide starting mixes were doped at the 50-500 ppm level with a series of trace elements (REE, Rb, Sr, Ba, Sc, Y, Zr, Nb, Hf, Pb, Th, U), facilitating the analyses of resulting phases by ion-probe. The phase relations were verified by a selection of reversal experiments based on separately synthesised phases.

Preliminary results indicate that at 22 GPa the sequence of crystallisation observed is garnet followed by magnesiowüstite with a solidus temperature of approximately 2170°C. At 24 GPa magnesiowüstite is on the liquidus with garnet then crystallising before Mg-silicate perovskite. The solidus temperature at 24 GPa is approximately 2200°C. Between 22 and 24 GPa Ca-silicate perovskite forms just below the solidus but would appear to scavenge the majority of added trace elements in the charge.