Friction during earthquakes: 25 years of experimental studies

Abstract

Dynamic fault strength t and its evolution with seismic slip and slip rate are among the most relevant parameters in earthquake mechanics. Given the large slip rate (up to 10 m/s), displacement (up to tens of meters), effective stress (tens of MPas), typical of seismic faulting, thermo-mechanical effects are outstanding: dynamic fault strength is affected by phase changes, extreme grain size reduction, and the production of amorphous and unstable materials in the slipping zone. In the first earthquake rupture models it was proposed that fault strength decreases inversely with slip rate (t ∼ 1/V, Burridge & Knopoff, 1967). Given the lack of determination of dynamic fault strength through seismological methods, elucidating constraints may arise from experimental studies. However, the experimental evidence for the strong dependence of fault friction with slip rate has been lacking for decades. After the preliminary studies of friction in rocks at seismic slip rates (Bridgman, 1936; Spray, 1987), it has been only with the installation of dedicated machines (Tullis & Weeks, 1986; Shimamoto & Tsutsumi, 1994) that the extreme deformation conditions achieved during earthquakes have been approached in the laboratory and systematically investigated. In 1997, Tsutsumi and Shimamoto reported the first data of dynamic fault strength in cohesive rocks at seismic slip rates: this study opened a new era in experimental rock deformation. Several fault dynamic weakening mechanisms first proposed theoretically (i.e., frictional melting lubrication and, more recently, fluid thermo-mechanical pressurization and elastohydrodynamic lubrication), were eventually reproduced in the laboratory (Cornelio et al., 2019; Aretusini et al., 2021). These experiments allowed us also to explore new dynamic fault weakening mechanisms and other will be probably discovered in the future. Nevertheless, the t ∼ 1/V relation was systematically observed, independently of the rock composition and presence of fluids (e.g., Di Toro et al., 2011). Nowadays, theoretical thermo-mechanical models, calibrated by experimental measures, enable us to construct constitutive equations for dynamic fault strength which find application into numerical modeling of earthquakes ruptures for both natural and human-induced seismicity (Murphy et al., 2018). Despite these recent advancements, the activation of several proposed dynamic fault weakening mechanisms and their occurrence in natural faults are debated. In this contribution, we will focus on the technical challenges and the main limitations (i.e., poor estimate of the temperature in the slipping zone, lack of in-situ observations during rock sliding, determination of the loading conditions that lead to the transition between concurrent dynamic weakening mechanisms, etc.) of the current experimental approach to unravel earthquake-related deformation mechanisms, and their application to the study of earthquakes in nature.