Density profiles of oceanic slabs and surrounding mantle: Integrated thermodynamic and thermal modeling, and implications for the fate of slabs at the 660 km discontinuity

We have calculated the mineralogical properties of the Earth's mantle and the lithological units constituting the subducting oceanic slabs within a wide range of P-T conditions within the CaO-FeO-MgO-Al₂O₃-SiO₂ system, except for the basalt top-layer of a slab, for which the system is extended to include Na₂O. The mineralogical data are then converted, using the appropriate P-V-T relations, to bulk densities. The calculated adiabatic density vs. depth profile of the mantle between 200 and 725 km depths is in good agreement with geophysical and experimental data. The density data of the different compositional units are combined with calculated thermal structures for a variety of slab-mantle systems to construct equilibrium density profiles as a function of depth. The mean equilibrium densities of the slabs within the transition zone (400-660 km depth) are found to be ∼0.04-0.05 g/cm³ greater than those of the ambient mantle within the same depth interval. For the entire upper mantle, density differences between slabs and ambient mantle are slightly less, but the slabs still remain denser than the latter. At 670 km depth, slabs have lower density than the ambient lower mantle because of the commencement of perovskite forming reactions within the mantle, and displacements of these reaction boundaries to higher pressures within the slabs as a consequence of their negative P-T slopes. If perovskite forming reactions within slabs are hindered for kinetic sluggishness, then neutral buoyancy would be achieved when the slabs have penetrated ∼100 km into the lower mantle. However, using the available data on the kinetics of spinel to perovskite plus periclase reaction, we conclude that the reaction would go to completion in a Peru-type young slab (41 Myr), and very likely also in a Tonga-type old slab (110 Myr), before these penetrated 100 km into the lower mantle. Thus, slabs should always remain negatively buoyant, and therefore continue to subduct through the lower mantle once it penetrates through the 660 km discontinuity. Despite a negative buoyancy force, a slab could deflect at the top of the lower mantle (660 km) because of factors resisting subduction, namely viscosity jump, low dip angle, slab roll back, and metastable persistence of olivine in cold slabs. If published scale model experiments represent realistic approximations of the factors affecting plate subduction, then according to our density data, any slab with a dip angle of ≤40-50° would bend at the 660 km discontinuity if there is a viscosity jump of at least by a factor of ∼10 and trench migration. The basalt top-layer of a slab is denser than other slab components and the ambient mantle at all depths to 660 km, and therefore should continue to sink into the lower mantle, especially if a slab directly penetrates the 660 km barrier, instead of peeling off in the transition zone to form a "perched eclogite" or "piclogite" layer, as previously proposed. The harzburgite layer, which is sandwiched between denser basalt and lherzolite layers, faces greater resistance to subduction, especially in a young slab, and thus could significantly contribute to the deformation of a slab near the 660 km discontinuity.