Laboratory study of fault healing and lithification in simulated fault gouge under hydrothermal conditions

Seismic data, geologic observations and laboratory friction studies suggest that faults lithify and strengthen (heal) during the interseismic period of the earthquake cycle. We report on experiments to investigate the influence of healing duration and temperature on the strength and healing rate of simulated faults. Layers of μm-sized quartz powder were used to simulate granular fault gouge. Gouge layers were sheared within Sioux quartzite in a triaxial pressure vessel at elevated pressures (P_c = 250 MPa), elevated temperatures (230-636°C), and in the presence of water (P_H2O = 75 MPa). We performed 'hold-slide' experiments in which samples were subjected to a period of healing under hydrostatic load, followed by shear deformation. Healing times ranged from a few hundred s to 10⁵ s. To isolate the effects of lithification and temperature on friction, we ran two types of hold-slide experiments: (1) samples were subjected to healing and shear deformation at elevated temperature; and (2) following healing at elevated temperature, temperature was reduced prior to shear. We also ran slide-hold-slide experiments in which samples were healed under shear load at elevated temperature (636°C). Experiments in which deformation was carried out at a lower temperature than healing show that both the static and sliding coefficient of friction increase with heal time. Samples healed and deformed at higher temperature showed lower peak strengths, and within the scatter of our data, did not exhibit time-dependent strengthening. These data are interpreted to result from enhanced lithification rates at elevated temperature. Samples healed under shear load exhibited a log-linear decrease in friction with hold time. Modeling using rate- and state-dependent friction laws indicates velocity-strengthening behavior, with friction a - b values of 0.05 to 0.08. Our data indicate that at high temperature, the effective state-evolution term b is negative. Moreover, the data indicate the rate of frictional strengthening varies significantly for hydrostatic and non-hydrostatic load conditions. Our measurements of healing are consistent with seismic estimates of fault-healing rates, if stress drop is a fraction of the total shear stress and effective fault normal stress is 60 to 100 MPa.