MCMC inversion of the transient and steady-state creep flow law parameters of dunite under dry and wet conditions
AbstractThe rheology of the upper mantle impacts a variety of geodynamic processes, including postseismic deformation following great earthquakes and post-glacial rebound. The deformation of upper mantle rocks is controlled by the rheology of olivine, the most abundant upper mantle mineral. The mechanical properties of olivine at steady state are well constrained. However, the physical mechanism underlying transient creep, an evolutionary, hardening phase converging to steady state asymptotically, is still poorly understood. Here, we constrain a constitutive framework that captures transient creep and steady state creep consistently using the mechanical data from laboratory experiments on natural dunites containing at least 94% olivine under both hydrous and anhydrous conditions. The constitutive framework represents a Burgers assembly with a thermally activated nonlinear stress-versus-strain-rate relationship for the dashpots. Work hardening is obtained by the evolution of a state variable that represents internal stress. We determine the flow law parameters for dunites using a Markov chain Monte Carlo method. We find the activation energy $$430\pm 20$$ 430 ± 20 and $$250\pm 10$$ 250 ± 10 kJ/mol for dry and wet conditions, respectively, and the stress exponent $$2.0\pm 0.1$$ 2.0 ± 0.1 for both the dry and wet cases for transient creep, consistently lower than those of steady-state creep, suggesting a separate physical mechanism. For wet dunites in the grain-boundary sliding regime, the grain-size dependence is similar for transient creep and steady-state creep. The lower activation energy of transient creep could be due to a higher jog density of the corresponding soft-slip system. More experimental data are required to estimate the activation volume and water content exponent of transient creep. The constitutive relation used and its associated flow law parameters provide useful constraints for geodynamics applications. Graphical Abstract
