Diamonds sampled by plumes from the core-mantle boundary

Diamonds are formed under high pressure more than 150 kilometres deep in the Earth's mantle and are brought to the surface mainly by volcanic rocks called kimberlites. Several thousand kimberlites have been mapped on various scales, but it is the distribution of kimberlites in the very old cratons (stable areas of the continental lithosphere that are more than 2.5 billion years old and 300 kilometres thick or more) that have generated the most interest, because kimberlites from those areas are the major carriers of economically viable diamond resources. Kimberlites, which are themselves derived from depths of more than 150 kilometres, provide invaluable information on the composition of the deep subcontinental mantle lithosphere, and on melting and metasomatic processes at or near the interface with the underlying flowing mantle. Here we use plate reconstructions and tomographic images to show that the edges of the largest heterogeneities in the deepest mantle, stable for at least 200 million years and possibly for 540 million years, seem to have controlled the eruption of most Phanerozoic kimberlites. We infer that future exploration for kimberlites and their included diamonds should therefore be concentrated in continents with old cratons that once overlay these plume-generation zones at the core-mantle boundary.

Thermochemical structures beneath Africa and the Pacific Ocean

Large low-velocity seismic anomalies have been detected in the Earth's lower mantle beneath Africa and the Pacific Ocean that are not easily explained by temperature variations alone. The African anomaly has been interpreted to be a northwest-southeast-trending structure with a sharp-edged linear, ridge-like morphology. The Pacific anomaly, on the other hand, appears to be more rounded in shape. Mantle models with heterogeneous composition have related these structures to dense thermochemical piles or superplumes. It has not been shown, however, that such models can lead to thermochemical structures that satisfy the geometrical constraints, as inferred from seismological observations. Here we present numerical models of thermochemical convection in a three-dimensional spherical geometry using plate velocities inferred for the past 119 million years. We show that Earth's subduction history can lead to thermochemical structures similar in shape to the observed large, lower-mantle velocity anomalies. We find that subduction history tends to focus dense material into a ridge-like pile beneath Africa and a relatively more-rounded pile under the Pacific Ocean, consistent with seismic observations.