Solubility of silicon in liquid metal at high pressure: Implications for the composition of the Earth's core

Silicon solubility in liquid Fe-rich metal was measured experimentally as a function of pressure, temperature and oxygen fugacity to determine if silicon could be a major light element in the Earth's core. At the P, T, fO₂ conditions of the experiments, Si solubility in liquid metal increases with increasing pressure, increasing temperature and decreasing oxygen fugacity. Evaluating single-stage core formation scenarios, the experimental results show that if the core segregated at low pressure (∼ 2.5 GPa) and an oxygen fugacity consistent with the current FeO content (8 wt%) of the mantle, the liquid metal would contain negligible Si ( < 0.01 wt%). If, as proposed recently [1-3] however, the metallic liquid equilibrated with the mantle at the base of a deep magma ocean (∼ 700 km depth, 25 GPa) the experimental results suggest a core containing several wt% Si. For example, at the melting temperature of a peridotite mantle (∼ 2800 K), which represents the temperature at the base of a magma ocean, 2 wt% Si (conceivably as much as 6 wt%) could dissolve in liquid Fe metal at 25 GPa. In order to generate a core containing 7 wt% Si, as required by the CI chondrite Earth model, a magma ocean extending to 850 km (28-30 GPa) would be required. We conclude therefore that equilibration at the base of a deep magma ocean is consistent with Si being a major part of the light element inventory of the core. Extrapolation to the core-mantle boundary pressure shows that the solubility of Si in the metal, after increasing in the upper mantle, decreases with increasing depth in the lower mantle due to the change of Si coordination number in solid silicates from 4 to 6. This result is in agreement with predictions based on theoretical arguments [4] and on EOS measurements of FeSi alloys [5]. Therefore, a core segregated at the base of a deep magma ocean and containing several wt% Si must now be back-reacting with the mantle, expelling Si and absorbing Fe from the silicate. Progressive reaction of this kind may, in part, be responsible for a chemical boundary layer represented by the D″ layer at the core-mantle boundary.