Constraints on mantle structure from geochemistry, seismology, and mixing studies: A numerical investigation of the compositional heterogeneity in the Earth's mantle

Geochemical and seismological evidence indicate that the Earth's mantle is heterogeneous on all scales. Isotopic data indicate that the recycled components have a mean age of 1–2 Gyr, and that there may be a reservoir containing trace elements inferred to be missing from the shallow mantle. Constraints from argon and helium isotopes require that ∼50% of the Earth's mantle is outgassed, and that some OIBs have high 3He/4He. There is no consensus about the identification of this high helium ratio material, although the favorite explanation is primordial, undegassed material. Seismic tomography identifies the existence of long-wavelength strong heterogeneity at the base of the mantle and in the lithosphere. The existence of lateral heterogeneities requires that different isotopic domains be maintained for around 2 Gyr.; To test whether a 2-D mantle convection model with self-consistent plate tectonics is able to reproduce the observed isotopic data, and to study how geochemical reservoirs evolve, suites of numerical experiments have been performed to investigate different aspects of the problem. The study focuses on the sensitivity of the results to uncertain physical properties, such as the density of subducted oceanic crust in the deepest mantle, and elemental partition coefficients. The main findings of the thesis are as follows (the caveat is that these need to be verified using 3D calculations): (1) The system self-consistently evolves the observed range of 3He/ 4He, but the exact distribution of ratios depends strongly on physical parameters and sampling method. Some parameter combinations simultaneously lead to MORB-like distributions of 3He/4He ratios in erupted material, and ∼50% outgassing of radiogenic 40Ar, consistent with geochemical constraints. (2) The system self-consistently evolves regions of HIMU (high 238U/204Pb) and low 143Nd/144Nd by segregation of subducted crust at the core-mantle boundary (CMB). The Pb-Pb age is about 2 Gyr. The effect of large strain on erasing heterogeneities must be included in the model, otherwise old material may have a strong effect on the age distribution, and the mean age of mantle heterogeneity. Moreover, the HIMU reservoir obtained in the model may provide a storage for some of the “missing” heat-producing elements. (3) Oceanic crustal differentiation and segregation cause strong seismic heterogeneity. When the crust is dense at the CMB, the model reproduces long-wavelength (L = 2) heterogeneity in the lower mantle. The presence of oceanic crust in the model contributes to the heterogeneity in the lithosphere. (4) Plate tectonics and variable viscosity have a strong effect on the lateral mixing behavior. Although chaotic mixing is typically observed in the study, horizontal heterogeneities may persist for more than 2 Gyr. Lateral mixing efficiency is controlled by strain rate and convective vigor.