Throughout Earth’s mantle, several significant phase transitions occur, with the Ol→Wd and Rw→Brm+Pc reactions (exothermic and endothermic respectively) producing large discontinuities in Earth’s seismic velocity structure at 410 and 660km depth respectively (‘410’ & ‘660’). The equilibrium depth of these reactions is sensitive to temperature, and the resulting topography has been observed with various seismic phases. Numerical modelling from the 1980s onwards has suggested that the topography on endothermic phase transitions can stagnate downwellings and even layer mantle convection for extreme Clapeyron slopes or density changes. The thermodynamic properties of the post-spinel reaction make it unlikely that slabs would stagnate due to effects associated with phase transitions. At cooler temperatures the post-spinel reaction splits into two reactions (Rw + Ak → Ak + Pc → Brm + Pc) which seems to explain well aspects of the observed topography of the ‘660’ discontinuity. It has been suggested that this second reaction (which has a more extreme Clapeyron slope than the post-spinel reaction) could stagnate downwellings. Recently, Ishii et al (2023) suggested that the post-garnet reaction (Gt → Brm + Cor [+ St]) is in fact univariant, producing a sharp reaction that is endothermic for cooler temperatures and exothermic at higher temperatures – and that this may contribute to slab stagnation. Here, we test these slab stagnation mechanisms using realistic mineral physics and whole-mantle convection models (MCMs).

The lack of anti-correlation between the topography of the ‘410’ and ‘660’ discontinuities does not match simple theory if they are controlled solely by temperature variations across the post-olivine and post-spinel reactions respectively. Previous work has shown that the calculated topography on the discontinuities can be markedly different for various single-composition mantles generated from MCMs (Papanagnou et al, 2022). Here we will explore the impact of laterally varying chemistry generated in thermochemical MCMs on global discontinuity topography.