Scaling of the dynamic slip weakening distance with acceleration during laboratory earthquake-like slip events

During earthquakes faults rapidly accelerate and weaken, facilitating the propagation of dynamic rupture and radiation of seismic energy. To date, numerous laboratory studies have shown that faults weaken through a combination of thermally driven processes. However, the majority of these studies are conducted at constant velocity, neglecting the fact that on natural faults we expect complex slip histories that are compatible with source time functions and elastodynamics (e.g. Gaussian or Yoffe slip functions). To this end, we will present results of a recent set of experiments performed with the rotary-shear apparatus SHIVA by imposing simulated Yoffe-like slip functions (i.e. impulsive initial slip acceleration and deceleration followed by slower deceleration), with a large range of initial acceleration rates (A~1 to 75 m/s^2). The strength response of the experimental faults made of calcitic marbles and basalts is strongly velocity-dependent, showing little dependence on the slip history. The velocity-dependent strength results in the weakening distance, D_w ≈ A^-0.5 (typically 0.05-0.5m). Additionally, and in contrast to standard constant velocity experiments, we observe a long restrengthening phase where the sample shear strength often recovers to the initial static shear strength. These results have important feedbacks on the earthquake energy balance, and may be at odds with a self-similar model of earthquake rupture. From a laboratory perspective, the scaling of weakening distance with acceleration may be able to reconcile the observations of very short weakening distances in stick-slip experiments, and possibly (very tentatively), in rate-and-state friction experiments.