Effects of thermo-chemical mantle convection on the thermal evolution of the Earth’s core

A coupled core-mantle evolution model that combines a global heat balance in the core with a fully dynamic thermo-chemical mantle convection model is developed to investigate the thermal evolution of the core over the 4.5 Gyr of Earth history. The heat balance in the core includes gravitational energy release, latent heat release and compositional convection associated with inner core growth. In the mantle convection model, compositional variations, plate-like behavior, phase changes and melting-induced differentiation are included. For mantle compositional variations, three idealized situations are considered: no variations (isochemical), variations resulting from a layered initial condition, and variations resulting from melting-induced differentiation from a homogeneous start. Only models whose thermal evolution satisfies three criteria are judged to be ‘successful’, with the criteria based on: (1) the radius of the inner core, (2) the heat flux through the core-mantle boundary (CMB), and (3) the heat flux through the surface. The radius of the inner core is the strictest criterion of these three. Models with an isochemical mantle fail because the inner core becomes much larger than the current size of the inner core. The final inner core radius is quite sensitive to mantle chemical buoyancy ratio. Models that fully satisfy all three criteria have a 1.5–2% compositional density difference, and either initial layering or compositional layering generated from melt-induced differentiation. These results imply that the heat flux buffering effect of a compositionally-dense layer in the deep mantle may be required to explain the thermal evolution of the core, when the heat flux through the CMB is calculated using a fully dynamical mantle convection model. Considering geochemical constraints, the compositional layering could be generated a combination of melt-induced differentiation and primordial layering. However, while the observed trends are robust, the models include various approximations and uncertainties, with the core model not including the effects of heat generated by radioactive element, so further investigations are warranted.