Testing thermochemical mantle circulation with Mantle Transition Zone velocities and topography

Abstract:

Recent developments in our understanding of lower mantle mineralogy motivate an updated view of mantle circulation that may explain the absence of a correlation between topography on the 410 km and 660 km discontinuities. A common feature in both P and S wave tomography models is the lack of a definitive slab seismic signal in the lower mantle from 1500-2500 km depth, even though this depth range is well constrained by the data. Ab initio calculations of Wu and Wentzcovitch (PNAS, 2014) predict that (Mg,Fe)O will be seismically insensitive to temperature below 1500 km due to weakening of the bulk modulus during the high-spin to low-spin transition in iron. Yamazaki and Karato (AM, 2001) demonstrated that (Mg,Fe)O has a much lower viscosity than perovskite and could support strain weakening which would facilitate transport of slabs through the lower mantle. Since subducting oceanic lithosphere is dominantly harzburgite, it has a higher (Mg,Fe)O component than a pyrolitic or perovskitic lower mantle. When the mantle began solidification after the last giant impacts, the first solids to crystalize at depth were pure perovskite (the liquidus phase above ~35 GPa). When subduction initiated, the first slabs needed to work through this highly viscous, perovskitic lower mantle and form channels to the core mantle boundary. Tomography models show a column of seemingly warm, low P and S velocity material rising from the top of Large Low Shear Velocity Provinces and collecting in the upper 1000 km of the mantle. The dominant signature of the lower mantle is that it is seismically boring with little discernible tectonic structure. We conclude that the majority of the lower mantle is the sluggishly mixed remnant of its initial perovskitic crystallization. Although less viscous, the (Mg,Fe)O rich slabs have carved channels into the lowermost mantle and comprise the dominant material in the return flow to the upper mantle. This model implies slabs stagnate in the MTZ due to trench migration above silica rich domains. The behavior of upwelling harzburgitic material will be complicated by the high temperature perovskite and (Mg,Fe)O transition to majorite garnet. Thus, the lack of correlated patterns in the 410 and 660 km discontinuity topography outside of subduction zones could be due to partial chemical layering of the mantle.