Slow slip events (SSEs) and other forms of slow and transient fault slip are becoming increasingly recognized as important, due to their influence on seismicity and potential to provide information on fault zone properties. The majority of our current knowledge of slow fault slip has been obtained from geodetic and seismologic field measurements, but laboratory data of slow slip are comparatively scarce. Here, I present the results of laboratory friction experiments conducted on 11 natural samples from major fault zones around the world, all obtained by scientific drilling. The experiments are conducted water saturated, at room temperature and 10 MPa effective normal stress, representative of in-situ conditions on the shallow portions of fault zones from which the samples were recovered and where slow slip is known to occur. A key component of these experiments is shearing at realistically slow driving rates of cm/yr, accurately simulating tectonic driving rates. In most samples, these cm/yr driving rates produce laboratory SSEs, which are instances of accelerating slip accompanied by a stress drop. The peak slip velocities and stress drops measured in these laboratory SSEs are comparable with those of natural SSEs measured or estimated from geodetic data. A strong correlation is observed between reduced pre-SSE velocity and higher peak slip velocity for the entire laboratory SSE dataset. In contrast to the velocity data, significant scatter is observed in the % stress drop measurements. The source of this scatter can be attributed to samples with a significant expandable clay component, which tend to exhibit larger stress drops. Results of velocity-stepping tests at cm/yr rates show a tendency for velocity-weakening friction not observed at higher sliding velocities, and that the materials with lower values of the rate-dependent friction parameter a-b tend to produce faster SSEs. Critical stiffness analyses within the framework of rate-and-state friction laws show that most of the SSEs observed in this study do not satisfy the condition for slip instability. The SSEs are more consistent with accelerating stable slip, although the stiffness condition allowing such behavior is not always satisfied. Considering the laboratory SSEs to be accelerating stable slip, I present a conceptual model for their nucleation. Key elements of the model are a healing-dominated departure from steady-state causing partial locking and velocity decrease, followed by a transition to a velocity-dominated phase representing the actual slip event. The model is consistent with observations from geodetic measurements and the experimental observations in this study. In general, characteristics of SSE-producing fault portions such as the ability to strengthen and store elastic strain energy released as stress drops may be expected to enhance coseismic slip from remotely nucleating earthquakes, an effect which may be quite limited but should be investigated further.
