Reactive Crystallization of the Basal Magma Ocean on Terrestrial Planets

Abstract:

Terrestrial planets evolve through several magma-ocean stages during accretion and differentiation. Any magma ocean is thought to become progressively enriched upon fractional crystallization (FC). The resulting upwards enrichment in FeO of the cumulate (=crystal) package drives gravitational over-turn(s), and ultimately stabilizes a variably FeO-enriched basal magma ocean (BMO). Upon stabilization in the deep mantle, the BMO is then also expected to undergo FC, ultimately stabilizing a strongly FeO-enriched layer at the core-mantle boundary (CMB) >50 km thick that would not be entrained by mantle convection. Such an outcome, however, is ruled out by geophysical observations, at least for Earth.

Here, we explore the consequences of BMO reactive crystallization (RC), on the initial condition of solid-state mantle convection. Due to chemical disequilibrium, the BMO reacts with mantle pyrolite to form reactive cumulates as long as cumulates are entrained by convection. For a wide range of BMO initial compositions, the reactive cumulate package consists of two discrete layers: the first is Mg-rich bridgmanite (~MgSiO₃); the second is similar to an FeO-enriched pyrolite. The first layer is readily entrained by mantle convection due to its intrinsic buoyancy, but resists efficient mixing due to its intrinsic strength, thereby potentially providing an explanation for seismic scatterers/reflectors and ancient geochemical reservoirs [Gülcher+ 2020]. The second layer is swept up into thermochemical piles due to moderate denisty anomalies, providing a candidate origin for large low shear velocity provinces. The relevant timescales of RC are several billion years for the end-member of solid-state diffusion across a sharp BMO-mantle boundary, but much shorter if partial melts are stabilized in the lowermost mantle (e.g., due to subduction of mafic materials). For such efficient RC, large rocky planets such as Earth, Venus or even Super-Earths may host a rather short-lived (<<1 Gyr) BMO. In turn, small rocky planets, such as Mars (?), may have never hosted a BMO due to low CMB pressures [Caracas+ 2019], and instead stabilize a long-lived (enriched) solid layer at the CMB. Our results have important implications for mantle stratification, and the long-term thermal evolution of terrestrial planets.