Can a subducted sedimentary layer survive into the deep mantle? Maybe, if it is strong or thin.

Geochemical analyses of ocean island basalts (OIB) reveal a heterogeneous mantle. The origins of enriched endmembers often are ascribed to recycled components in the OIB mantle source. For example, EM2 is classically attributed to ancient subducted sediment. However, prior geodynamic modeling suggests that sediment may not be dense enough to endure into the upper mantle and beyond. We use a combination of geochemical mass balance constraints, geodynamic models, and new seismic observations to assess the degree to which sediments survive at depth, and whether they are geophysically resolvable.

To explore the physical parameters required to subduct sediments, we utilize simple numerical models. We vary thickness, density and viscosity of a sediment layer in a 2-D section of a downgoing "slab" (sediment + crust). Preliminary results show that sediment survival into the deep mantle is a function primarily of the viscosity contrast between the surrounding mantle and slab, and initial sediment layer thickness. While density is important, the range of likely values is sufficiently narrow that the other parameter variations dominate. Assuming that some sediment survives beyond the "subduction factory", sediment layers that are thinner or have greater viscosity are more likely to be carried deeper into the mantle.

We also utilize deep Wadati-Benioff zone earthquakes below dense seismic arrays in Peru and Chile to analyse waveforms traveling through the Nazca slab. The fine-scale structure of low velocity layers at the top of subducting slabs can potentially be detected by high frequency body waves from deep events that travel up the slab interface; these layers may be subducted sediments.

Additionally, recent SS precursor work finds evidence for significant reflectors in the mid-mantle. Sporadic and varied reflectors in regions associated with downwelling are linked to heterogeneous slab material. We analyze complementary large SS and PP datasets to search for seismic structures corresponding to potential slab lithologies, incorporating synthetic resolution testing using estimated velocity models calculated for realistic compositions.