New insights into magma ocean solidification dynamics

At the end of planetary accretion, magma ocean (MO) evolution is thought to set the initial conditions for the evolution of terrestrial planets. For subsequent long-term mantle dynamics and its associated magmatism, key aspects of magma ocean evolution are the degree of silicate differentiation (that forms distinct geochemical reservoirs) and the degree of mantle mixing at the end of MO solidification. Most aspects of MO dynamics are derived from lunar MO models based on data of the Apollo mission obtained more than forty years ago. However, the actual geodynamic evolution of terrestrial mantle from a partially (or fully) MO to a solid mantle remains poorly understood and the lunar MO model might not be the rule for all terrestrial mantle. The issue of primordial mantle mixing is closely related to the onset (and vigor) of solid-state mantle dynamics. In the canonical lunar MO model, the solid-state mantle dynamics starts with a mantle overturn near or after the end of MO solidification. Such late overturn is expected to reverse but preserve most of the early formed lunar mantle layering. Recent studies show that early solid-state convection in the cumulate can significantly mix the young mantle before the end of MO solidification. Using numerical models of mantle convection, we show that the timing of mantle overturn can be quantitatively estimated. For the Moon, quantitative estimates of overturn timing offer new dynamic constraints on the relationships between the age of the anorthosites, the age of the Mg-Suite and the mechanism of mare basalt generation. In the canonical lunar MO model, solidification occurs from the bottom-up. Recent studies suggest that top-down MO solidification might occur in larger terrestrial body (e.g., Earth or Super-Earths) because liquidus could first intersect adiabat at mid-mantle depth and/or crystal could become buoyant at high pressure. A basal magma ocean (BMO) solidifies very slowly and and progressively forms deep mantle reservoir(s) enriched in incompatible elements. Whereas there are several mineralogical and thermodynamic arguments that favor the BMO scenario, there was yet no fluid dynamics investigation of BMO formation. We develop a multiphase flow model that allows to track the dynamics of a solidifying system where crystal can precipitate, sink (or float) and remelt according to prescribed melting relations. Our models show interesting transcient behavior when thermodynamic equilibrium competes with gravitational equilibrium, i.e., phase buoyancy. The ease of phase segregation selects the winner. If crystal settling is not a limiting factor, our model suggests that density cross-over between melt and solid – that is easily achieved by iron enrichment of the melt - is able to generate a BMO regardless of the interplay between liquidus and adiabat.