Frictional Melt: Fault Lubrication or Brake?

Slip weakening is an essential process for large earthquakes to occur. Frictional melting has been known to be one of the effective slip weakening mechanisms, although there are some suggestions that frictional melt may strengthen fault during seismic slip. Our high-velocity friction experiments on a siltstone imply that frictional melt may act as a brake rather than lubricant. In the high-velocity rotary shear tests on siltstone at slip rates of 1.14-1.18 m/s and normal stresses of 9.8- 20.9 MPa, three cycles of strengthening and weakening occur before the simulated faults exhibit a steady- state friction with an average friction coefficient (μss) of 0.27. After the final weakening, the simulated fault zone consists of a molten layer (0.30-0.75 mm thick) mantled by `damage' layer (0.1-0.2 mm thick). The microstructures of the fault zone at the second peak friction (μp = 0.66) after the second strengthening but before the second weakening) include fragments of melt mixed with gouge. In contrast, the fault zone immediately after the second weakening (before the final strengthening) consists of a continuous molten layer (0.2-0.6 mm thick) with a high fraction of clasts. At the final strengthening to the third peak friction (μp = 0.7-0.8), the molten layer is fragmented and mixed with gouge again. The surface temperature of the fault zone measured by a radiation thermometer is 650-700°C before reaching the final peak friction. In contrast, the measured temperature of the fault zone at the final peak friction is 800-850°C. High-velocity friction of gabbro and tonalite also shows similar behavior to that of siltstone. However, the high- velocity friction behavior of peridotite is somewhat different from that of siltstone, tonalite and gabbro, in that it does not involve a significant strengthening except first, instant strengthening (presumably due to lower viscosity of melt). Our experimental results from siltstone indicate that a continuous molten layer forms at the initial stage and that fault strengthening due to high viscosity of melt can result in the breakage of the molten layer. At least two cycles of the molten layer fragmentation before the final stress drop suggest that frictional melt may arrest further slip unless melt viscosity is much lowered with a significant rise in temperature and decrease in the fraction of clasts.