Reactive Crystallization of the Basal Magma Ocean: Consequences for present-day mantle structure

Terrestrial planets evolve through multiple magma-ocean stages during accretion and differentiation. Magma oceans become progressively enriched upon fractional crystallization (FC), which should be dominant at least in the upper mantle. The resulting upwards enrichment of the cumulate package drives gravitational overturn(s), and ultimately stabilizes a FeO- and SiO₂-enriched basal magma ocean (BMO) [1]. Alternatively, a ~pyrolitic BMO may be formed due to a liquid-solid density crossover at high pressures [2,3]. In any case, the slowly cooling BMO is very likely to freeze by FC. However, we find that the consequences of FC of the BMO are inconsistent with geophysical constraints for Earth (Ismail+, this meeting). For FC, the final-stage cumulates are expected to be strongly FeO-enriched (~eutectic), stabilizing a layer at the base of the mantle with density anomalies >2,000 kg/m³. Such a layer should be extremely long-lived, but is not detected by seismic imaging.

Using a thermodynamic model [4], we here investigate the chemical consequences of an alternative scenario, in which the BMO interacts with (partially) molten basaltic material in the lower mantle. We refer to such a scenario as reactive crystallization (RC). Even in the present-day, the core-mantle boundary may be hotter than the solidus of subducted basalt [5]. Accordingly, any recycled Hadean/Archean is likely to have undergone (partial) melting in the lowermost mantle, and mixed with the BMO. This scenario is attractive, because large volumes of crust may be readily delivered to the lowermost mantle, and will produce dense magmas there, which sink into the BMO to promote efficient reaction.

We find that the first BMO cumulates due to RC are Mg-rich bridgmanite (~MgSiO₃). With progressive addition of basaltic material, Al₂O₃ becomes enhanced in the BMO to promote FeO-disproportionation, leading to loss of elemental Fe to the core and crystallization of FeAlO₃. With ongoing cooling, the BMO starts effectively shrinking, and final BMO cumulates are similar in composition than, and slightly enriched compared to, basalt. The associated intrinsic density anomalies are 300~350 kg/m³, i.e., much more moderate than for FC of the BMO. These predicted densities and cumulate compositions (bridgmanitic with high FeAlO₃) are in very good agreement with the geophysical signatures of large low-velocity provinces [6]. In turn, the predicted final composition of the BMO itself may correspond to that of seismically-detected ultra-low velocity zones.

Our results imply that large rocky planets such as Earth, Venus or even Super-Earths may host only a rather short-lived BMO due to efficient crustal recycling. In turn, small stagnant-lid planets with limited crustal recycling, such as e.g. Mars, may host longer-lived BMOs (Cheng+, this meeting). These predictions have important implications for the long-term thermal and chemical evolution of terrestrial planets.

 

[1] Ballmer+, G-cubed, 2017; [2] Labrosse+, nature, 2007; [3] Caracas+, EPSL, 2019; [4] Boukare+, JGR Solid Earth, 2015; [5] Adrault+, science 2014; [6] Vilella+, EPSL, 2021