Lithosphere http://doi.org/jsj (2012)
A fragment of an ancient subducted slab is thought to have been left dangling in the upper mantle beneath Idaho, western USA, for over 40 million years. Numerical modelling demonstrates that slab buoyancy combined with high mantle viscosity helped stall the slab's descent into the mantle.
Erin Burkett and Michael Gurnis at the California Institute of Technology, USA, investigated the conditions that could allow a cold, dense slab fragment to persist in the upper mantle for so long, without detaching from the surface tectonic plate and sinking into the warm mantle. In their numerical simulations, they find that a short slab fragment will stall for longer than a long fragment because it is, comparatively, stronger and more buoyant. A high viscosity contrast between the upper and lower mantle also helps to slow the slab's sinking velocity.
Together, these factors can delay slab detachment and sinking for about 28 million years, but it is difficult to explain the presence of dangling slabs for much longer. It is therefore possible that the piece of lithosphere observed beneath Idaho is a delaminated part of the surface plate, rather than a dangling slab left over from subduction.