Metal-silicate partitioning of sulphur, new experimental and thermodynamic constraints on planetary accretion

- Experimental partitioning of sulphur between metal and silicate melts.
- Thermodynamic modelling of sulphur partitioning with P, T and chemical composition.
- Effects of mantle convection and fO2 on sulphur fractionation during core segregation.
- Sulphur in Earth core and mantle can reach its observed values of 2 wt% and 200 ppm.

Partitioning of sulphur between liquid Fe-rich metals and silicates (DSmet/sil) was investigated at temperatures from 1800 °C to 2400 °C, pressures from 2 to 23 GPa and oxygen fugacities from 3.5 to 1.5 log units below the iron-wüstite buffer, using multi-anvil apparatus. The results are combined with previous experimental works to refine a multi-variable thermodynamic model of DSmet/sil. Sulphur appears to become more siderophile with increasing pressure and FeO content of the silicate melt, and less siderophile with increasing temperature and with Si, C, O, Fe and Ni contents of the metal. We then modelled the behaviour of sulphur in the course of planetary accretion, using different possible scenarios of mantle dynamics and evolution with time of oxygen fugacity. We investigated three end-member models for metal-silicate segregation of the incoming impactors: (i) the planetary mantle does not mix and is kept chemically stratified, (ii) the magma ocean is continuously mixed chemically, and (iii) both the magma ocean and the solid lower mantle are well mixed.We show that if S is accreted along the accretion, whatever the oxidation path, its distribution between core and mantle can lead to the observed S concentration of the mantle (200±80ppm) and to the estimations of S content of the core (from its depletion in the mantle relative to the other elements with the same volatility). In the case of an Earth built with reduced material, to explain the present-day 200(±80)ppm S found in the mantle, it is necessary that both the magma ocean and the solid lower mantle mix at each major step of the planetary accretion. S could also be accreted in the last 10 to 20% of Earth's growth and reach its observed present terrestrial abundances if the magma ocean is chemically mixed along the accretion. Consequently, our models show that the S terrestrial abundances do not formally require an S accretion in a late veneer but can be explained by a core-mantle equilibration alone.