A suite of experiments was performed to investigate the effect of solution transfer processes on the frictional properties of faults. Triaxial experiments were conducted at an effective confining pressure of 175 MPa and temperature of 235°C on sawcut cylinders of Sioux quartzite which contained a layer of simulated quartz gouge. Some samples were "healed" at elevated confining and pore pressures and temperatures up to 800øC for varying periods of time prior to loading. Unhealed samples slide in a stable mode and approach a quasi steady-state strength at shear strains of 3-4%. Healed samples, slid at identical pressures and temperature, are stronger (i.e. higher ?.) and exhibit unstable stick-slip behavior. A quasi steady-state strength is eventually established after the initial stick-slip event in samples healed at 636°C; however, samples healed at 817øC stick-slip repeatedly. Experiments in which the healing time was varied indicate that the bulk of the strength recovery occurs during the first 60 minutes at elevated temperature. Unhealed samples for which the time of stationary contact was varied do not exhibit differences in either strength or stability. No strength recovery is observed in a sample healed without pore fluids at identical effective pressures. SEM of fracture surfaces of the healed gouge reveals a marked absence of fines as compared to the starting powder, smooth grain faces, rounded grain shapes, and grain interpenetration.
Seismologic observations suggest that faults may recover strength during the interseismic period of the earthquake cycle. Larger values of average stress drop are associated with crustal faults with longer repeat times (Kanamori & Allen 1986). Also, stiess drops associated with intraplate earthquakes are typically larger than those of interplate earthquakes, perhaps owing to the longer recurrence times associated with the former (Scholz et al. 1986; Kanamori & Anderson 1975). Although laboratory observations of the variation of stress drop amplitude with loading vel .ocity and the time of stationary contact at room temperature are in qualitative agreement with the seismological observations, the observed variations are significantly less than expected, thus suggesting that other healing mechanisms operate during the interseismic phase (Scholz et al. 1986; Wong & Zhou 1990). Field observations also suggest that the physical properties of fault zones change with time. For example, fluid flow localized along fault zones may cause variations in the density, permeability, and mineralogy of rocks and gouge in the fault zone which may, in turn, affect the mechanical properties of the fault (Angevine et al. 1982; Chester & Logan 1986; Parry et al. 1988).
In this paper we present results of a suite of triaxial experiments performed to investigate the effect of solution transfer processes on the mechanical properties of simulated fault gouge. To study such thermally activated processes in the laboratory, it is necessary to accelerate the kinetics by elevating temperature. We conducted frictional sliding experiments at confining pressures, Pc, of 250 MPa, pore pressures, Pp, of 75 MPa, and temperatures of 250°C on samples which were "healed" at temperatures up to 800°C for varying periods of time prior to axial loading. Our results indicate that the thermally activated processes which occur in the presence of aqueous pore fluids strongly affect both the frictional strength and stability of sliding.