Phase relations in iron-rich systems and implications for the Earth's core

Abstract

Recent experimental data concerning the properties of iron, iron sulfide, and iron oxide at high pressures are combined with theoretical arguments to constrain the probable behavior of the Fe-rich portions of the Fe-O and Fe-S phase diagrams. We infer that a solid solution exists between ϵ-Fe and S at high pressures. This is based on the similarity in the atomic radii of Fe and S (rFe/rS ≈0.97), and on the observation that S apparently does not form a solid solution with the γ-phase of Fe. We suggest that the ϵ-Fe S system may, to first order, be modeled as an ideal solid solution and, therefore, will not have a eutectic. The solid solution probably is not ideal, but there are insufficient data to constrain the non-ideal behavior. The ϵ-Fe-O system, on the other hand, probably has very little solid solution (rO/rFe ≈0.84), although there is recent evidence that a highly non-ideal solid solution does exist between γ-Fe and O. Experimental data extending to > 100 GPa suggest that solid FeO may remain a stable compound at high pressures, with a possible miscibility gap between FeO and Fe. This is based on the observation that FeO melts at higher temperatures than either Fe or O at high pressures. Theoretical extrapolation of the melting curve of FeO indicates that this behavior should continue throughout the pressure range relevant to the Earth's core. Thus, the Fe-rich portion of the Fe-O phase diagram is predicted to display a eutectic between ϵ-Fe and FeO at core pressures. Comparison of the predicted composition of the Fe-FeO eutectic at inner core pressures with the amount of oxygen required to give the outer core its observed density (28 at. % O) indicates that, if the only light element in the core were oxygen, the core composition would lie significantly toward the FeO side of the eutectic and FeO would be the solid phase of the inner core. This is inconsistent with the properties of the inner core. This argument indicates that, although oxygen is probably present in non-trivial quantities, it cannot be the only light element in the core, and may not even be the most abundant light element in the core. The results of our analysis are compatible with sulfur as another major light element in the core, as the Fe-S solidus is always more Fe-rich than the coexisting liquidus.

Additional details