Stuck in the Mud? Earthquake Propagation Through Clay-rich Fault Zones (Invited)

Many major faults that are postulated to host earthquake ruptures contain a clay-rich fault core that shows a localized principal slip surface. Other clay-rich fault zones appear to show distributed deformation that might be related to strain/velocity hardening and fault creep. Indeed, the frictional behaviour of most clays at low slip velocities is velocity strengthening, supporting the notion that earthquake nucleation on clay-rich faults is difficult. We have performed high velocity frictional experiments on a range of common clays, both wet and dry, at a slip velocity of 1.3 m/s and normal stresses ranging from 0.8 to 3.25 MPa. In the dry tests, peak frictional coefficients are reached almost immediately and are comparable to those measured in low velocity, low normal stress tests (0.5 μm/s) on the same material. This is followed by prolonged weakening over slip distances of several metres to friction coefficients of ~0.2. In the wet tests, the peak friction is substantially reduced and the corresponding slip weakening distance reduces to very low values. This is despite the wet and dry frictional coefficients in low velocity, low normal stress tests being comparable. We infer from our results that thermal pressurization is responsible for the almost immediate weakening of the wet samples. These measurements suggest that earthquakes will propagate easily through clay-rich regions of the rupture plane, with minimal fracture energy. However, the nucleation of the rupture may have to occur on clay-poor regions of the fault plane where velocity weakening friction is dominant. However, these conclusions do not explain why earthquake ruptures appear to arrest in regions where fault creep dominates, for example, in the central portion of the San Andreas fault.