Transfer of oxygen to Earth's core from a long-lived magma ocean

Chemical interactions between metal and silicates at the core-mantle boundary (CMB) are now thought to lead to transfer of oxygen into Earth's liquid core. Establishing the nature and extent of this transfer is important for constraining the conditions under which the core formed, the origin of a stably stratified region below the CMB and the possible precipitation of oxides within the core. Previous models of FeO transfer have considered a solid mantle; however, several lines of evidence suggest that the lowermost mantle could have remained above its solidus long after core formation was complete, which would allow much faster mass transfer. We investigate this scenario by developing a time-dependent model of FeO exchange between a diffusive stratified layer at the top of the core and a long-lived molten magma ocean. Core FeO concentration, ̄ccFeO, is evolved subject to a time-dependent mass flux at the CMB, radius rcmb, which depends on the FeO concentration at the bottom ( ̄cmFeO(rcmb)) and top ( ̄cmFeO(rbulk)) of the chemical boundary layer above the CMB. Coupled core-magma ocean evolution arises because ̄cmFeO(rcmb) and ̄ccFeO(rcmb) are linked through the partition coefficient P= ̄ccFeO(rcmb)/ ̄cmFeO(rcmb). ̄cmFeO(rbulk) is held constant in No Crystallization (NC) models and evolves in Middle-Out Crystallization (MOC) models according to the basal magma ocean model of Labrosse et al. (2007), generalised to account for FeO loss to the core. In the first 1 Gyr, FeO transfer in all models with ≥10% FeO in the magma ocean and P≥5 produces pure FeO compositions at the CMB, stably stratified layers of 60−80 km and accounts for 15−50% of the total present-day core oxygen content. In NC models the magma ocean does not completely freeze in 4 Gyr, in which time the stable layer reaches 120−150 km and FeO transfer can account for all of the present-day O in the core. However, in MOC models FeO loss to the core causes the magma ocean to completely freeze in the first 1-3 Gyrs following core formation. Our results suggest that the present-day core composition may not provide a strong constraint on models of core formation and that FeO could have precipitated at the top of the core.