Patches of negative core-mantle boundary heat flux: simulations of mantle convection and implications for core dynamics

Heat flux at the Earth’s core-mantle boundary (CMB) partially controls the outer core dynamics and its associated geodynamo. On the mantle side, lateral variations in temperature above the CMB trigger lateral variations in heat flux with low temperature (typically, in and around subducted slabs) and high temperatures (at plumes roots and beneath hot thermo-chemical piles) areas being associated with high and low heat flux regions, respectively. Spatial and temporal variations in temperature are, in turn, controlled by details of mantle convection and mantle material properties. Here, we investigate the influence on CMB heat flux of two key parameters: the excess internal heating within piles of hot, dense material (also referred to as primordial material) modelling the large low shear-wave velocity provinces (LLSVPs) observed by global seismic tomography maps; and the temperature-dependence of thermal conductivity. For this, we perform a series of high-resolution numerical simulations of thermo-chemical convection in spherical annulus geometry using the code StagYY. Importantly, the total heating rate within the mantle is fixed, meaning that an excess heating within piles is balanced by a reduced heat released elsewhere. The initial condition on composition consists in a thin basal layer of chemically denser material, which subsequently evolves into piles of hot, primordial material on the top of which plumes are being generated. Our simulations show that the CMB heat flux is lower than the core adiabatic heat flux throughout the base of primordial material piles, and that it can be locally negative, i.e., heat flows from the mantle to the core. We further investigated the conditions needed for such patches to appear. As one would expect, a larger internal heating excess and a stronger temperature dependence of thermal conductivity both favor the development of negative heat flux patches. However, patches disappear if the piles excess heating gets too large. In this case, heat released in the regular mantle is strongly reduced, allowing plumes generated at the top of piles to extract more heat from these piles. Finally, our simulations predict relatively large CMB heat flux spatial heterogeneity, together with substantial temporal variations in this heterogeneity. Our findings have strong implications for core dynamics. In particular, they support the hypothesis that partial stratification at the top of the core can occur beneath LLSVPs, reconciling geomagnetic and seismic observations. In addition, and based on core dynamics studies, the CMB heat flux heterogeneity and temporal variations predicted by our simulations may play a key role in the occurrence of geomagnetic superchrons.