Hydrogen partitioning between silicate melt and liquid sulfide: an early oxidation mechanism for the mantle?

During core formation, the presence of metallic iron in the mantle would have imposed an oxygen fugacity below the iron-wüstite buffer. Today, however, the Earth's upper mantle is uniformly about 5 orders of magnitude more oxidised, indicating that one or more processes increased the fO2 of the mantle after the end of core formation. The nature and timing of this oxidation mechanism has implications for the proportion and distribution of volatile elements accreted and retained within the Earth, as well as the composition of the earliest atmosphere.

We explore the possibility that the accretion of H₂O-rich material led to the gradual oxidation of the mantle whilst core formation was ongoing. This process would only be plausible if it could have continued once the fO₂ of the mantle had risen above the level where metallic Fe would be stable. This implies the core-forming material would have been an FeS melt, and the oxidation reaction would then have been essentially FeS + H₂O = H₂S + FeO. The resulting H₂S would have to have been removed from the mantle (i.e., sequestered in the core) in a way that would prevent it from subsequently reducing other regions of the mantle and thus reversing the above reaction.

In order to constrain the importance of this process, we have determined the partitioning behaviour of H between coexisting silicate and sulphide melts. FeS and a MORB-composition silicate with variable Fe and FeO contents, respectively, were equilibrated at 3 GPa and superliquidus temperatures in a multianvil press. Major elements in the silicate glass and quenched sulphide were determined with EPMA. H contents of both phases were determined by elastic recoil detection analyses at the nuclear microprobe at CEA Saclay, and Fe³⁺ content of the silicate was determined by transmission Moessbauer spectroscopy in order to constrain the fO₂ of the experiments. D­_H(sulphide/silicate) was found to be 0.23 ± 0.11, with little variability as a function of fO₂, implying that some H would have dissolved into a core-forming sulphide melt. However, an even greater amount of oxygen was found to have partitioned into the sulphide, and so any oxidising effect of removing H would have been compensated for by the simultaneous removal of oxygen. It is therefore unlikely that interaction between H₂O and sulphide could have led to mantle oxidation.