Metal-silicate partitioning of nickel and cobalt: The influence of temperature and oxygen fugacity

Abstract

We report new metal-silicate melt partitioning experiments for Co and Ni. One atmosphere, gas-mixing (H 2CO 2 or CO-CO 2) experiments were run at 1300, 1425, and 1550°C and at oxygen fugacities between air and near or below the iron-wüstite (IW) buffer. Bulk melt composition, at the diopsideanorthite-forsterite eutectic, was constant in all runs. Metals in these experiments were platinum alloys permitting a large oxygen fugacity range to be investigated while maintaining fixed bulk melt composition. Since activity-composition relations of Pt-Ni and Pt-Co alloys are reasonably well known, we calculated equilibrium constants for the metal-silicate partitioning reactions. Variations of the equilibrium constants as a function of experimental parameters allowed us to monitor the chemical behavior of Ni and Co in the silicate melt. Nickel and Co behave as divalent species in the silicate melt over the range of conditions investigated. However, differences in solubility between low and high oxygen fugacity could not be entirely explained by a simple oxidation-reduction equilibrium constant. Differences in melt solubility were correlated with which gas-mixing system was used, H 2-CO 2 gave larger NiO solubilities at low oxygen fugacity than CO-CO 2. The difference in melt solubility was attributed to minor amounts of melt-dissolved volatile species, perhaps H 2O. Along the IW buffer, Fe metal-silicate partition coefficients for Ni and Co decrease with increasing temperature, but strengths of the temperature dependencies, even in this simple system, appear to vary with experimental conditions. Such differences highlight the danger of extrapolation to conditions far beyond the experimental regime. For example, extrapolation to magma ocean temperatures (3000–4000°C) results in uncertainties larger than the order of magnitude difference in siderophility between Ni and Co.