Thermal entrainment and melting in mantle plumes

Abstract

Thermal plumes from the core-mantle boundary (CMB) are thought to cause volcanic hotspots, and starting plumes, or plume ‘heds’, are thought to cause voluminous flood basalt events at the beginning of some hotspot tracks. It has been proposed that starting plumes entrain a large volume fraction of surrounding mantle material as they rise from the CMB to the base of the lithosphere, and that this process of thermal entrainment can explain some aspects of isotope and trace element heterogeneity in hotspot and flood basalt lavas. We examine this hypothesis using numerical models of mantle plumes generated by a thermal boundary layer at the base of the mantle. Our numerical experiments are designed to simulate plumes with characteristics appropriate to explain major hotspots such as Hawaii. The flow trajectory of original plume source (boundary layer) material is mapped using a neutral buoyancy chemical tracer field. We incorporate a simple parameterized batch melting model in order to examine in detail the proportion of different mantle components (‘enriched’ boundary layer vs. ‘depleted’ ambient mantle) which undergo partial melting as starting plumes spread beneath the lithosphere. The effects of a variety of physical model parameters are explored, including temperature and depth-dependent viscosity, and phase changes in the mantle transition zone. In all cases, we find little entrainment of surrounding mantle into the region of the plume head that undergoes a significant degree of partial melting, so that the primary plume magmas represent > 90% original plume source material. These results suggest that hotspot lavas do not sample a large volume fraction of the mantle through which plumes rise, in apparent disagreement with previous interpretations of laboratory experiments on thermal entrainment in plumes. The differences arise because: (1) we examine entrainment in the melting region, rather than the entire plume head; (2) our plumes arise naturally from a boundary layer, which imposes an initial thermal gradient from plume center to periphery; (3) starting plumes in the mantle rise only about 3 plume head diameters before spreading (‘unwrapping’) and melting beneath the lithosphere; and (4) we have modeled cases in which plumes melt at sublithospheric depths, without large-scale lithospheric extension. We suggest that geochemical heterogeneity in hotspot and flood basalt lavas is mainly a reflection of either inherent plume source heterogeneity or contamination from the crust and lithosphere through which primary magmas rise.