UC Berkeley

Deviatoric stress generated within subducting slabs by the olivine-spinel transformation has been modeled by assuming the phases to be simply connected rather than comprising a mixture in which one phase is embedded within another. Here, we use a simplified model to explain how transformation strain is incorporated into a continuum model, and we then use the simplified model to explain quantitatively the origin of the unreasonably large deviatoric stresses predicted by existing slab models. We review experiments on the transformation of single-crystal samples and argue that they are consistent with the occurrence, at the grain scale, of deviatoric stresses comparable with those predicted (erroneously) to exist at the slab scale by those slab models. Using a simple example, we show that although large deviatoric stresses can exist at the grain scale, their average over a sample containing many grains can be hydrostatic. This leads us to the problem of modeling the microscale structure. We outline the thermodynamics needed for such nonhydrostatic systems, and we illustrate their use and implications with examples.