Nonfractional crystallization of a terrestrial magma ocean

Abstract

It has been suggested that evolution of a terrestrial magma ocean does not unavoidably follow a fractional crystallization scenario. Convection is able to preclude differentiation until a sharp viscosity increase occurs near some critical crystal fraction. However, this kind of crystallization and its physical and chemical consequences have not been previously studied. We consider an end‐member, called here nonfractional crystallization. We begin with a simple equilibrium thermodynamical model of partial melts which is based on an ideal three‐component phase diagram. It allows a self‐consistent calculation of physical and chemical parameters in the melting range at all interesting pressures. In particular, adiabats of the convecting magma ocean are calculated. The sharp increase in the viscosity is supposed to occur near the maximum packing crystal fraction. However, almost independently of this value, convection occurs even in the highly viscous quasi‐solid part of the magma ocean and it is strong enough to prevent differentiation in deep regions. A kind of compositional convection occurs due to the layered differentiation, although it is weaker than the thermal convection. Only a surface region undergoes an essential differentiation via melt expulsion by compaction. The thickness of this layer depends on the rheology of partial melts, critical crystal fraction, and crystal sizes but in any case the basal pressure hardly can exceed 5 – 10 GPa. Because of lower pressures in the Moon, the thickness of the differentiating layer is large and thus the entire lunar magma ocean could undergo a strong differentiation. Remelting due to the energy released by differentiation is crucial only for much deeper layers (possibly deeper than about 1000 km for the Earth). For the remaining shallow layer (p< 5 – 10 GPa) the predicted increase of the melt fraction is less than 40 % at the surface and decreases to zero at the bottom of the differentiating layer. Thus, the nonfractional crystallization is suggested to be a likely alternative to the fractional crystallization. The crucial and still poorly understood factors are suspension in convective layers, rheology of partial melts, crystal size, and surface conditions. The most pronounced chemical consequence of the nonfractional crystallization is an almost completely preserved undifferentiated lower mantle and possibly a significant undifferentiated part of the upper mantle. At all depths, in the beginning of differentiation not only the first liquidus solid phase but also subsequent phases have been partially crystallized. So, when the differentiation begins, it involves mixtures of phases. It is important for the remaining layer where differentiation is unavoidable: this layer does not have as strong differentiation of minor elements as in the case of fractional crystallization but it will still involve differentiation of major elements. Future geochemical calculations of this multiphase differentiation, considering both major and minor elements, could help to constrain the differentiation further.