Oxygen Solubility in Liquid Iron and Consequences for the Early Differentiation of Earth and Mars

Introduction: It has been considered that the Earth accreted from carbonaceous chondrite-like material (e.g., [1], [2]). This hypothesis is based on the observation that the refractory element ratios of Earth’s mantle are close to carbonaceous chondrite values, although the ratio of metallic Fe to total Fe is much closer to that of enstatite chondrite (e.g., [3]). Recently, Halliday [4] suggested that the proto-Earth might have been volatile-rich and oxidized, like the current Martian mantle. In this case, it is important to clarify when and how FeO reduction occurred in the Earth. One possibility is during a magma ocean stage, when the segregation of core and mantle material occurred. Rubie et al. [5] suggested that FeO could be partitioned into Fe metal in a terrestrial magma ocean because of very high temperatures. In addition, their experimental data suggested that oxygen solubility in liquid Fe alloy decreases with increasing pressure and that the equilibrium solubility of oxygen in the Earth’s core may be close to zero. However, the effect of pressure on oxygen solubility in liquid Fe has been a controversial topic for the last 3 decades. Based on phase relations in the Fe-FeO system, it has been proposed that oxygen solubility in liquid iron increases with increasing pressure (e. g., [6], [7]). On the otherhand, studies of the partitioning of oxygen between magnesiowüstite and liquid Fe suggest that oxygen solubility in liquid iron decreases with increasing pressure. Here we report new data on oxygen partitioning between magnesiowüstite and liquid iron in the Fe-FeO-MgO system, over a wide range of pressures and temperatures (3-25 GPa and 2273-3173 K), which resolve this apparent inconsistency. Using the present results, we discuss the possibility of FeO reduction in early terrestrial and Martian magma oceans and the oxygen content of metallic cores. Experimental procedure: High-pressure and high-temperature experiments were performed using a Kawai-type multi-anvil apparatus. The starting materials were mixtures of Fe metal and Fe2O3 oxide powders with a range of bulk oxygen contents. We used MgO single crystals as sample containers. Chemical analysis and imaging analysis of the recovered samples were conducted with an electron microprobe and a scanning electron microscope. Results: We consider the reaction of oxygen between magnesiowüstite and Fe liquid: Fe metal + O metal = FeO magnesiowüstite (1). The distribution coefficient, Kd, of this reaction is expressed as Kd = XOXFe/XFeO, where XO, X Fe, XFeO are the mol fractions of oxygen in metal, Fe in metal and FeO in magnesiowüstite, respectively. LnKd increases with increasing temperature over the range of experimental conditions, and shows a linear relationship with 1/T at 15-24.5 GPa, and 2273-3173 K. We observe that lnKd initially decreases with increasing pressure, reaches a minimum at 10-15 GPa and then increases at higher pressures. This means that the volume change (ΔV) of the reaction changes from being negative at pressures below 10-15 GPa to positive at higher pressures. This can be explained by a difference in the compressibilities of the FeO component in magnesiowüstite and liquid iron respectively.