V52A-01
The Origin of Non-chondritic HSE Ratios in the Earth’s Mantle
Abstract:
It is generally thought that Earth’s mantle abundances of highly siderophile elements (HSE) were established by the addition of a chondritic late veneer to a mantle that was stripped of HSEs by core formation. A long-standing problem with this hypothesis is that the mantle’s suprachondritic Pd/Ir and Ru/Ir ratios cannot be reconciled with any known meteorite group.To address this issue, we modelled the effect of metal-silicate segregation on abundances of the HSE and S in the Earth’s mantle by including these elements in a combined accretion/core-formation model. Because in our model only a small fraction of the mantle equilibrates with core-forming metal, the bulk mantle HSE abundances are too large by the end of accretion. Sulfur abundances also greatly exceed S-saturation levels at magma ocean crystallisation temperatures, leading to the formation of a global immiscible sulfide melt that segregated to the core, thus removing HSEs from the mantle [1].
To better constrain the role of sulfide segregation on the HSE budget of the mantle, we experimentally determined the sulfide-silicate partitioning of Pt, Pd, Ru and Ir under high P-T conditions. Results show that Pd and Ru are less chalcophile at pressures above ~20 GPa compared to Pt and Ir, as opposed to the metal-silicate system where Ru is more siderophile than Pt [2]. These results are included in our model, which now involves localized segregation of core-forming metal followed by widespread exsolution and segregation of immiscible sulfide liquids. Platinum and Ir are efficiently extracted from the mantle whereas significant concentrations of Ru and Pd remain. Late veneer addition occurs after sulfide segregation has ceased due to magma ocean solidification. This model reproduces perfectly the non-chondritic Ru/Ir and Pd/Ir ratios of the mantle, reflecting incomplete removal of Ru and Pd from the mantle with core-forming sulfide melts.
[1] O’Neill (1991) GCA 55, 1159-1172. [2] Mann et al. (2012) GCA 84, 593-613.