Thermometry of the magma ocean: Controls on the metal–silicate partitioning of gold

We have performed experiments to investigate the solubility and metal–silicate partitioning of gold as a function of metal sulphur content (XS), silicate melt polymerization (NBO/T) and pressure (P  ). These experiments show that Au becomes less siderophile both with increasing pressure and as the metal phase becomes more sulphur-rich. For the studied range of compositions, melt polymerization has no effect on the solubility of Au. The reduction in the siderophile tendency of gold with increasing metal sulphur content is greater than expected on the basis of activity–composition relationships in the metal phase. This suggests a significant role for complexing between Au and S in the silicate melt. Our new experimental results are combined with literature data to yield a parameterisation for the exchange coefficient of Au (KdAuMet/Sil) as a function of P, T and XS:logKdAuMet/Sil=-1.10(0.04)+1.11(0.01)104T(K)-67.8(10.9)P(GPa)T(K)+5.81(0.24)log(1-XS)Using this relationship, alongside similar parameterisations for Ni and Co selected from the literature, we performed continuous accretion models to delineate regions of parameter space in which Au and moderately siderophile element (MSE) abundances in the primitive upper mantle (PUM) could be reproduced. These models suggest that for metal–silicate equilibrium at very high pressures, Au will be overabundant in the PUM if equilibrium also occurs at extreme temperatures. Instead, most successful models are found when equilibrium occurs at high pressure but sub-liquidus temperatures. This result is in keeping with the physical conditions expected for a scenario where core-forming metal ponds and equilibrates at the rheological base of a magma ocean (e.g. Wade and Wood, 2005).