Solubility of tungsten in a haplobasaltic melt as a function of temperature and oxygen fugacity

The solubility of tungsten (W) in a haplobasaltic melt has been determined as a function of oxygen fugacity in the temperature range 1300-1500°C using the mechanically assisted equilibrium technique of Dingwell et al. (1994), and at 1600-1700°C by the wire loop method. Quenched samples were analysed for W by using ICP-AES as well as INAA, and sample major element compositions were checked by electron microprobe. W concentrations ranged from 20 ppm to 17 wt%, and the solution of WO₂ in the melt may be described by Henry's Law up to remarkably high concentrations (e.g., 14 wt% at 1500°C). W dissolves in the melt with a quadrivalent (4+) formal oxidation state over the entire range of oxygen fugacity and temperature investigated. The solubility of W decreases strongly with increasing temperature at constant oxygen fugacity. The solubility data have been used to calculate trace distribution coefficients for W between Fe-rich metal and silicate melt, using literature values for the activity coefficient of W in liquid Fe. Comparison of our data with the distribution coefficients for Mo calculated from the analogous Mo solubility data of Holzheid et al. (1994) shows that the ratio of the metal-silicate distribution coefficient D^met/sil_Mo/D^met/sil_W remains very high (∼10³) at all T-fO₂ conditions. However, in the Earth's mantle, Mo is relatively more depleted than W only by a factor of three. The relative abundances of W and Mo in the Earth's mantle cannot, therefore, be explained by core formation from a homogeneously accreted Earth whatever the temperature at which metal/silicate equilibrium may have ocurred might have been. Their abundances may be quantitatively accounted for by a heterogeneous accretion model such as that of O'Neill (1991).