Fluid dynamics of core formation

Abstract

Past discussions of core formation are incorrect or incomplete because they assume that metallic iron-rich liquid is able to migrate through a mostly solid silicate matrix by percolation, prior to macrosegregation and diapiric descent. Experimental and theoretical considerations suggest that percolation is largely prevented because of the high surface tension of iron. Two alternative views of core formation are offered. One assumes that percolation is possible in the deep mantle (where perovskite is the major phase). Iron is then supplied to the deep mantle by Rayleigh-Taylor instabilities of a silicate-iron suspension in the shallow mantle, and drains efficiently from the deepest mantle into the core by Darcy flow. The other model assumes complete or nearly complete melting of all or part of the mantle. Despite vigorous convection, iron droplets approximately one centimeter in radius are predicted and settle rapidly by Stokes flow, either to the core or into a layer or ponds that provide iron-rich diapirs that can descend to the core. These stories generally suggest very efficient core formation in the sense that the typical residence time of metallic iron in the mantle is orders of magnitude shorter than the formation time of Earth (~10^8 years). Good chemical equilibrium between mantle and core phases is also predicted in many cases. Geochemical constraints and implications relevant to these scenarios are discussed but are largely inconclusive. The tentative inference of rapid core formation on Mars suggests a magma ocean and iron rainout.

Additional details