Discriminating lower mantle composition

Abstract

Constraining Earth's bulk composition is fundamental to understanding our planet's formation and evolution. While the lower mantle accounts for a majority of the bulk silicate Earth, it is also the least accessible. As experimental and theoretical mineral physics constraints on mineral elasticity at lower mantle temperatures and pressures have improved, comparisons between predicted seismic velocity and density profiles for hypothesized bulk compositions and 1D seismic models have become commonplace. However, the degree to which a given composition is a better or worse fit than another composition is not always reported, nor are the influences of the assumed temperature profile and other uncertainties discussed. Here we compare seismic velocities and densities for perovskitite, pyrolite, and harzburgite bulk compositions calculated using advanced ab initio techniques to explore the extent to which the associated uncertainties affect our ability to distinguish between candidate compositions. We find that predicted differences between model compositions are often smaller than the influence of temperature uncertainties and therefore these comparisons lack discriminatory power. The inability to distinguish between compositions is largely due to the high sensitivity of seismic properties to temperature accompanied by uncertainties in the mantle geotherm, coupled with diminished sensitivity of seismic velocity to composition toward the base of the mantle. An important exception is the spin transition in (Mg,Fe)O-ferropericlase, which is predicted to give a distinct variation in compressional wave velocity that should distinguish between relatively ferro-magnesian and silica-rich compositions. However, the absence of an apparent spin transition signature in global 1D seismic profiles is a significant unresolved issue in geophysics, and it has important geochemical implications. The approach we present here for establishing discriminatory power for such comparisons can be applied to any estimate of seismic velocities and associated uncertainties, and offers a straightforward tool to evaluate the robustness of model comparisons.