The Effect of Upper to Lower Mantle Viscosity Jump on the Regime Diagram of Slab Deformation in the Mantle Transition Zone

Abstract:

Slabs display a wide range of morphologies in the mantle transition zone. This slab transition zone deformation is likely caused by a barrier arising from a jump of viscosity between upper and lower mantle, and/or from the endothermic phase transitions at 660-km depth. We use 2-D thermo-mechanical models of a two-plate subduction system, modeled with the finite-element, adaptive-mesh code Fluidity, to investigate the influence of the viscosity jump on slab morphologies. We implement a temperature- and stress-dependent rheology, and variable viscosity increases from upper to lower mantle of 10, 30 or 100 (no mineral phase transitions). Various end-member subduction modes arise, ranging from vertical folding to horizontally deflected to retreating and penetrating slabs. For each viscosity contrast between upper and lower mantle, we build a regime diagram for subduction dynamics based on the initial subducting and overriding plate ages. Trench motion is facilitated by smaller upper-lower-mantle viscosity contrasts, and, for all but the oldest subducting plate cases, simulations with a 100-fold viscosity increase exhibit a stationary trench later in their evolution. Slower sinking rates also lead to weaker (lower-viscosity) slabs encountering the viscosity jump. These effects, together with the increased resistance to penetration associated with a more viscous lower mantle, produce increasingly deformed and stalling slabs at depth, as the viscosity contrast increases. Slab deformation in the transition zone leads to an alternation between phases of penetration into the lower mantle and stagnation phases, reflected in subducting plate velocity. The periodicity and amplitude of such oscillations is directly controlled by the magnitude of the viscosity jump. Hence, our dynamic models help to interpret present-day observations of slab morphologies, along with the time-evolution of plate surface velocities, in terms of Earth rheology. The range of observed slab morphologies is best reproduced by models with a 30-fold increase in viscosity.