The formation and preservation of large-scale primordial heterogeneity in the Earth's mantle

Seismic imaging of subducted Farallon and Tethys lithosphere in the lower mantle has been taken as evidence for whole-mantle convection, and efficient mantle mixing. However, chondritic bulk Earth compositions are better explained by significant silica enrichment in the lower mantle relative to upper-mantle pyrolite. Moreover, ¹⁸²W anomalies in Phanerozoic igneous rocks provide direct evidence for the long-term preservation of primordial reservoirs somewhere in the mantle. Here, we present yet unpublished numerical models of global-scale mantle convection in 2D spherical-annulus geometry. We find that lateral variations in intrinsic rock viscosity, such as due to heterogeneity in Mg/Si, strongly affect the mixing timescales of the mantle. Intrinsically strong rocks may remain unmixed through the age of the Earth, and persist as large-scale (1000 1500 km diameter) blobs in the mid-mantle due to focusing of whole-mantle convection and related deformation along weak conveyor belts between blobs. We show in particular that physical properties of blobs that correspond to slightly Fe-enriched bridgmanitic compositions can best reconcile the dynamical requirements for long-term stability. Accordingly, we refer to the blobs as "bridgmanite-enriched ancient mantle structures" (BEAMS). Seismic evidence for blobs of anomalous composition in the present-day mid mantle comes from a global interrogation of seismic reflectors through SS and PP precursors. For example, we find large-scale coherent seismic reflectors in the NW Pacific as well as northern Eurasia at 1,000 km depth. The depths and lateral extents of these reflectors are consistent with the predictions of numerical models. In terms of scenario of origin, we favor that BEAMS material has been formed as cumulates from a Fe-enriched basal-magma ocean. If a basal magma ocean has been already separated from the rest of the mantle before the Moon-forming giant impact, BEAMS may indeed provide a viable reservoir for ancient ¹⁸²W signatures. Along these lines, detailed seismic studies of unmixed primordial reservoirs in the mid mantle may inform about proto-Earth composition and structure.