Nucleation of Stick‐Slip Instability Within a Large‐Scale Experimental Fault: Effects of Stress Heterogeneities Due to Loading and Gouge Layer Compaction

Geodetic observations and large‐scale laboratory experiments show that seismic instability is preceded by slow slip within a finite nucleation zone. In laboratory experiments rupture nucleation is studied mostly using bare (rock) interfaces, whereas upper crustal faults are typically filled with gouge. To investigate effects of gouge on rupture nucleation, we performed a biaxial shearing experiment on a 350 mm long saw‐cut fault filled with gypsum gouge, at room temperature and a minimum horizontal stress σ2 = 0.3–5 MPa. The gouge layer was sandwiched between polymethylmethacrylate (PMMA) plates For reference also a fault without gouge was deformed. Strain gauges and Digital Image Correlation were used to monitor the deformation field along the fault zone margins. Stick‐slip behavior occurred on both the gouge‐filled fault and the PMMA fault. Nucleation of instability on the PMMA fault persistently occurred from one location 2/3 to 3/4 along the fault adjacent to a slow slip zone at the fault end, but nucleation on the gouge‐filled fault was more variable, nucleating at the ends and/or at approximately 2/3 along the fault, with precursory slip occurring over a large fraction of the fault. Nucleation correlated to regions of high average fault stress ratio τ/σn, which was more variable for the gouge‐filled fault due to small length scale variations in normal stress caused by heterogeneous gouge compaction. Rupture velocities and slip rates were lower for the gouge‐filled fault than for the bare PMMA fault. Stick‐slip persisted when σ2 was lowered and the nucleation zone length increased, expanding from the center to the sample ends before transitioning into instability. Key Points Nucleation of dynamic instability was observed along a 350 mm long experimental gouge‐filled fault sandwiched by compliant PMMA blocks Nucleation occurred in regions of high average stress ratio τ/σn but was more variable for a gouge‐filled fault than a bare PMMA fault The presence of gouge resulted in more heterogeneous fault stress, more precursory slip and lower rupture velocities