The dynamical and thermal structure of deep mantle plumes

If the interpretation of the D″ layer at the base of the mantle as a thermal boundary layer, with a temperature increment in the order of 800 K, is correct, then the formation of deep-mantle plumes to vent material from it appears inevitable. We demonstrate quantitatively that the strong temperature dependence of viscosity guides the upward flow into long-lived chimneys that are ∼ 20 km in diameter near the base of the mantle and decrease in width with progressive upward softening and partial melting of plume material. The speed of flow up the axis of the plume is correspondingly fast; 1.6 m y at the base and 4.8 m y at 670 km depth. Thermal diffusive spreading of a heated plume is compensated by a slow horizontal convergence of mantle material toward the chimney in response to the lower pressure there. This convergence, which contributes only a small increment to the flux of material up the plume, also serves to throttle the flow in the chimney. The global plume mass flux necessary to transport 1.6 × 10 W of core heat upward through the mantle is 1.8 × 10 kg s. At its base, plume material is probably still significantly below its solidus or eutectic temperature, but substantial partial melting is very likely as it rises. We speculate that a small fraction of this fluid component eventually emerges at the surface in "hot spots", with the fate of the remainder being unknown. The behaviour and properties of D″ and of plumes are closely coupled. Not only are plumes a necessary consequence of a thermal boundary layer, but their existence is impossible without that layer