Title: Nucleation of frictional slip: A yielding or a fracture process?

Abstract: The onset of frictional sliding between contacting bodies under shear load is nucleated by the quasi-static growth of localized slip patches. After reaching a certain critical size, these patches become unstable and continue growing dynamically, eventually causing the sliding of the entire interface. Two different theories have been used to compute the nucleation length of such patches depending on the dominant process driving their growth. If it is only the yielding of contact asperities (large-scale yielding), a stress criterion is applied, based on linear stability analysis, whereas if fracture dominates, an energy criterion is applied, based on fracture mechanics and classical nucleation theory. Both approaches contain important underlying assumptions that are well-suited to describe either one situation or the other. However, what happens in-between is not captured by any of them. In this work, we use numerical simulations to study what is the dominant underlying process driving nucleation for different conditions of heterogeneity and what are the implications for nucleation dynamics and the onset of frictional sliding. We show that large frictional heterogeneities enable a transition from a yielding-driven nucleation phase to a fracture-driven one. This transition occurs only above a certain level of heterogeneity and can either be quasi-static (stable) or dynamic (unstable), depending on the correlation length of frictional strength along the interface and the difference in strength between the strongest and the weakest point (the amplitude). Unstable transitions generate localized dynamic slip events, whose magnitude increases with higher correlation length and decreases with larger amplitude. Our work sheds new light on the role of heterogeneity and fracture in the nucleation of frictional slip, bridging the gap between the two main governing theories for nucleation.