Moisture-related Weakening and Strengthening of Faults for Quartz-poor Gabbro and Quartz-rich Granite

Significance of moisture-related mechanisms has been pointed out from the slide-hold-slide tests on gabbro (quartz-poor rock) under dry and room-humid conditions using high-speed rotary-shear apparatus [Mizoguchi et al., 2006, GRL]. In this paper, we conducted similar experiments using granite (quartz-rich rock) and discuss the results from the viewpoints of the quartz content. In the experiments, one of the two solid-cylindrical specimens (diameter ~ 25 mm) was rotated with a speed of 100 rpm (equivalent slip velocity is ~ 85 mm/s) under a constant normal stress of 0.62MPa. We conducted the tests using initially bare rock sample (1) and rock sample initially containing wear materials produced during sliding (2). A typical result for the gabbro case (1) is that friction coefficient decreased from more than 1.0 (we call initial friction) to less than 0.4 when the amount of slip was between 10 and 30 m then it became a steady state. The layer of wear materials along the fault was about 50 μm thick at steady state. Since the fault was not covered with a jacket, wear materials were squeezed out from the fault during sliding. The result for the gabbro case (2) showed a similar slip-weakening curve to that of the gabbro case (1), it represented a remarkable dependence of the initial friction on hold time and humidity. Under a room-humidity condition, the initial friction increased from 0.2 to 1.2 as hold time increased as 200 ~ 400 s. In contrast, under the dry condition, the initial friction did not increase with hold time and kept the level of 0.2. To explain this time- and humidity- dependent frictional behavior of gabbro, a mechanism of moisture-drained weakening due to frictional heating and moisture-absorbed strengthening due to cooling is proposed. The result for the granite case (1) showed that the friction decreased from more than 0.9 down to less than 0.2 when the slip was between 50 and 75 m. The dynamic slip weakening behavior is similar to that of gabbro case (1). However, the time- and humidity-dependence of initial friction of the granite case (2) was different from that of the gabbro case (2). The initial friction did not fully recover when the hold time was two orders of magnitude larger than the full recovery time (~ 400 s) for the gabbro case (2). Another mechanism should be working, which prevented the granite from increasing the initial friction. From the X-ray diffraction analysis, no crystal structure was found in the wear materials of gabbro. The wear materials of granite had only a crystal peak pattern of quartz. This difference might be related to the presence of quartz in the host material. Furthermore, the sliding surface of granite after slip became shiny slickenside. Goldsby and Tullis [2002, GRL] and Di Toro et al. [2004, Nature] proposed a silica gel lubrication as a weakening mechanism at higher slip rates for quartz-rich rocks. Both the gel lubrication and moisture- absorbed strengthening seem to work during the slide-hold-slide tests on the granite case (2). However the initial friction of the granite was kept low with increasing the hold time. This is due to the gel lubrication, rather than the moisture-absorbed strengthening because the frictional strength is determined by a weakening mechanism, rather than a strengthening mechanism.