The solubility and oxidation state of tungsten in silicate melts: Implications for the comparative chemistry of W and Mo in planetary differentiation processes

The solubility of W in 18 melt compositions in the system CaO–MgO–Al2O3–SiO2 in equilibrium with W metal was determined as a function of oxygen fugacity (fO2) at 1400 °C and atmospheric pressure, using CO–CO2 and H2–CO2 gas mixtures to control fO2. Samples were analysed by both laser-ablation ICP-MS and electron microprobe. The variation of W solubility with fO2 establishes that W dissolves predominantly as W6+, with a possible contribution from W4+ only at the very lowest fO2s accessible to the experimental method, in which regime experimental difficulties make the reliability of the results uncertain. X-ray absorption near edge structure (XANES) spectroscopy at the L3-edge of representative samples confirms the oxidation state of W as 6+, and suggests that W6+ occurs in tetrahedral coordination in silicate melts. Activity coefficients of WO3 derived from the solubility measurements correlate exactly with those of MoO3 obtained previously by similar experiments using the same melt compositions and temperature (OʹNeill and Eggins, 2002). The effect of TiO2 on W solubility is shown to be mainly one of dilution, from an investigation at one fO2 in the pseudobinary between the anorthite–diopside eutectic composition (ADeu) and TiO2. lubilities of W and also Mo may be combined with thermodynamic data from the literature for Fe–W and Fe–Mo alloys to calculate partition coefficients for W and Mo between silicate melt and Fe-rich metal. The calculated partition coefficients for W and Mo differ by ∼ 103 over the range of fO2 appropriate for equilibrium between liquid metal and silicate melt during planetary core formation at low pressures and moderate temperatures (∼ 1400 °C). Because the ratio of DWsil-melt/met/DMosil-melt/met is predicted to decrease only moderately with temperature (e.g., to ∼ 102 at 2200 °C), and is independent of fO2, melt composition and degree of partial melting, the large fractionation of Mo/W expected for equilibrium conditions could provide a useful means of discriminating between models of heterogenous and homogenous accretion and core formation, once the effect of pressure is better understood. However, comparison of our calculated metal/silicate partition coefficients with direct experimental determinations reveals an apparent lack of internal consistency among the latter, which may partly reflect a strong influence of minor components in the metal phase (e.g., carbon) on partitioning, which will also need to be understood before Mo/W systematics can be applied with confidence.