Modeling solidification, phase separation and convection for magma oceansle - Charles-Edouard Boukaré

IMPMC - UPMC, 4, Place Jussieu - 75005 Paris. Tour 23 - Barre 22-23 - 4e étage, salle 401

Mardi 30 mai 2017 à 13 h 30

Charles-Edouard Boukaré - Brown University, Providence, Rhode Island, USA

Abstract

Since the earliest study of the Apollo lunar samples, the magma ocean hypothesis has received increasing consideration for explaining the early evolution of terrestrial planets. Giant impacts seem to be able to melt significantly large planets at the end of their accretion. The evolution of the resulting magma ocean would set the initial conditions (thermal and compositional structure) for subsequent long-term solid-state planet dynamics. However, magma ocean dynamics remains poorly understood.

The major challenge relies on understanding interactions between the physical properties of materials (e.g., viscosity (liquid or solid state), chemical buoyancy...) and the complex dynamics of a multiphase convecting system. Such complexities might be neglected in cases where liquidus/adiabat interactions and density stratification leads to stable situations. However, interesting possibilities arise when exploring magma ocean dynamics in other regime. In the case of the Earth, recent studies have shown that the liquidus might intersect the adiabat at mid-mantle depth and/or that solids might be buoyant at deep mantle conditions. These results require the consideration of more sophisticated scenarios. For instance, how does bottom-up crystallization look with buoyant crystals?

My talk will be divided into two parts. I will first focus on the petrology of a crystallizing system at deep mantle conditions. Then, I will present a numerical model of multiphase convection which can be used to investigate the impact of the petrology on the multiphase dynamic.

We build a solid-liquid thermodynamic database for silicates in the MgO-FeO-SiO2 system from 20 GPa to 140 GPa [Boukaré et al., 2015, JGR]. We compute the ternary phase diagram in the MgO-FeO-SiO2$_2$ system as a function of temperature and pressure. This self-consistent approach allows us to predict crystallization sequences at deep mantle conditions. We confirm that the melt is lighter than the solid of same composition for all mantle conditions but at thermodynamic equilibrium, the iron-rich liquid is denser than the solid in the deep mantle.

To understand this complex dynamics due to potential density cross-over between melt and solids, we develop a multiphase numerical code that can handle simultaneously phase change, convection in each phase and in the slurry, as well as the compaction or decompaction of the two phases [Boukaré et al., 2017, G3, in review]. Although our code can only run in a limited parameter range (Rayleigh number, viscosity contrast between phases, Prandlt number), it provides a rich dynamics that illustrates what could have happened. For a given liquidus/adiabat configuration, we explore magma ocean scenarios by varying the two main aspects controlling phase separation efficiency: the grain size and the iron partitioning between melt and solids. For what concerns the issue of a basal magma ocean, our study suggests that the location of a density contrast between solid and magma must be considered of equal importance with that of the intersection between liquidus and isentrope.