From Magma Oceans to Solid Mantles: Experimental Insights into the Differentiation of Earth and Mars

Planetary magma oceans were ubiquitous in the early Solar System and enabled terrestrial planets to differentiate into a silicate mantle and an iron-rich core. With time, these large-scale magma oceans progressively solidified by secular cooling, and their crystallisation led to the formation of the main petrological and geochemical reservoirs in the mantle. In this thesis, both metal-silicate and magmatic differentiation mechanisms were investigated through high-pressure high-temperature experiments and applied to two planets: Mars and Earth.

The Martian core is significantly lighter than a pure Fe-Ni alloy, which suggests the presence of light elements. Because Mars is more oxidised than Earth, the core cannot contain silicon. Hydrogen was investigated but found negligible while carbon can only be present at low concentrations due to its low solubility in sulphur-rich iron alloys. Oxygen was never considered because its solubility in iron at Martian core conditions was assumed to be low, leaving sulphur as the best candidate. However sulphur alone cannot explain the large density deficit in Mars' core without violating cosmochemical and geochemical constraints, leaving us with an apparent conundrum. We addressed this question through a series of metal-silicate partitioning experiments run in large-volume presses (piston-cylinder and multi-anvil) at Martian core conditions. The recovered samples were analysed either by scanning electron microscopy or by electron microprobe and implemented in a thermodynamical model of Martian core formation. Our results show that a substantial amount of oxygen could be dissolved in a sulphur rich Martian core because of thermodynamic interactions in the metallic alloy and could explain the observed core density deficit.

Magmatic differentiation is another major differentiation event that took place during the solidification of the magma ocean and that led to the production of various petrological and geochemical reservoirs still observable at present. Yet, this process remains poorly constrained from a petrological and dynamical standpoint, in particular at lower mantle pressures. For that reason, we developed and used a new experimental protocol to study fractional crystallisation in the laser-heated diamond anvil cell. We applied this protocol to a series of experiments covering the entire pressure range of the lower mantle. The samples were analysed by transmission electron microscopy and synchrotron scanning transmission X-ray microscopy to investigate the nature, composition and iron valence state of the mineral assemblages and how they evolve during fractional crystallisation. We confirm that bridgmanite is the liquidus phase of pyrolite at all lower mantle pressures. Subsequently, it is accompanied by ferropericlase until very low degrees of residual melt are reached, when calcium perovskite crystallises. We also tracked the composition of the coexisting residual melts and established the melting phase diagrams of pyrolite at four different pressures (53, 87, 107, 131 GPa). These experimental results can then be used to constrain self-consistent thermodynamic models to predict magma ocean crystallisation sequences at all mantle conditions. We then analysed the crystal chemistry of bridgmanite during crystallisation. We found a limited solubility of calcium, even at those extreme pressures and temperatures where its solubility in bridgmanite is the highest. This implies that calcium perovskite has to be present in the solid lower mantle at all temperatures and depths. Finally, we measured the iron valence state distribution (ferrous and ferric) in bridgmanite. We found that it contains high concentrations of ferric iron, even with a starting material that contains only ferrous iron, which proves that disproportionation is at work during magma ocean solidification.