Influence of subduction zone conditions and gouge composition on frictional slip stability of megathrust faults

To understand the temperature/depth distribution of destructive earthquakes in subduction megathrusts, and the mechanisms of nucleation of these events, data on the frictional behaviour of phyllosilicate/quartz-rich megathrust fault gouges under in-situ conditions are needed. We performed rotary shear friction experiments at effective normal stresses of 25-200. MPa, pore fluid pressures of 50-200. MPa, at 140-600. °C and sliding velocities of 1-100. μm/s, using gouge mixtures with an illite:quartz ratio between 65:35 and zero. Experiments on 65:35 mixtures, deformed at an effective normal stress (. σ_n^eff) of 170. MPa, a pore fluid pressure (. P_f) of 100. MPa and 150-500. °C provided a reference dataset. This showed three temperature-dependent slip stability regimes with potentially unstable, velocity-weakening behaviour at 250-400. °C and velocity-strengthening at lower and higher temperatures. The velocity-weakening regime was found to shift towards higher temperatures with decreasing σ_n^eff, being located at ~. 350-600. °C at 50. MPa. Increasing quartz content and decreasing sliding velocity also displaced the velocity-weakening regime towards lower temperatures. Increasing P_f increased (. a-b) at all temperatures, narrowing the temperature extent of the velocity-weakening regime. We explain our results qualitatively in terms of a microphysical model in which changes in friction coefficient and (. a-b) with velocity and temperature are brought about by changes in the relative importance of deformation of the clast phase, by thermally activated stress corrosion cracking and pressure solution, versus athermal granular flow of the mixture accompanied by dilatation. Our results imply that the depth extent of the seismogenic zone on subduction megathrusts depends not only on temperature and that to predict its location, it is essential to have well-constrained depth profiles for pore fluid pressure and effective normal stress.