Previous friction experiments on rock analogue experiments of mixtures of salt and
phyllosilicates, demonstrated the possibility of producing mylonitic fault rocks through the
simultaneous operation of pressure solution and frictional sliding. This frictional-viscous flow
process produces a strong velocity-dependence of friction, with friction values dropping from
0.8 to ~0.2-0.3 over just one order of magnitude decrease in sliding velocity. Here, we
present the results of rotary shear experiments on simulated fault gouges of 80 wt% quartz
and 20 wt% muscovite. Sliding experiments using a four orders of magnitude range of
constant velocities (0.03 - 300 μm/s) to a displacement of 30 mm were done at 500 Ë C, 120
MPa effective normal stress and 80 MPa fluid pressure to verify the mechanism at
hydrothermal conditions and to link the produced microstructure to the observed
strength.
At the lowest sliding velocity tested, final friction reached a value of ~0.3, which is lower
than that of pure muscovite under similar conditions. With increasing sliding velocity, friction
increases, reaching a maximum of ~0.9 at 3 μm/s after which it decreases mildly to ~0.8 at
300 μm/s.
The bulk microstructure of the sample sheared at 0.03 μm/s shows an anastomosing foliation
of muscovite grain intervened by asymmetrical quartz clasts, with an average grain size of
about 20 μm, slightly lower than the median starting size (~49 μm). In contrast, the grains of
the sample deformed at 300 micron/s are very small, many of them smaller than
distinguishable in the light microscope (i.e. < 1 μm). In addition, the microstructure is
characterized by clear bands of strong uniform extinction in P- and B-shear orientations,
possibly indicating a Crystallographic Preferred Orientation. These zones of uniform
extinction can be found in all samples and their thickness decreases monotonically with
decreasing sliding velocity.
The microstructure observed at low velocity, in the frictional-viscous regime, is similar to
numerous examples from natural fault rocks (e.g. the Median Tectonic Line and the Zuccale
Fault). The slowest sliding velocity employed here corresponds to a shear strain rate of
~3 * 10-5 s-1, still several orders of magnitude higher than tectonic plate rates
(~10-10to 10-8 for fault thicknesses of 1 to 0.01 m). At natural, lower strain rates, the
frictional-viscous flow regime, where friction is low, is predicted to be operative down to
temperatures as low as 250 Ë C and possibly even lower for other minerals than
quartz.
In contrast to the low velocity regime, microstructures similar to those observed here at
high velocity, have not been reported for natural fault rocks, implying that either these do not
survive exhumation (possibly due to the very fine grain size), get overprinted by later,
slow deformation, or are not formed in the first place. The strain rates here are still
well below the values reached during seismic slip and are probably not common
values in nature, nor will they be long-lived and thus not impose a large shear strain.
Dynamic or static grain growth after a transient, faster slip pulse will most likely
obliterate any evidence of slip rates fluctuating between aseismic and seismic. Clearly,
more hydrothermal experiments aimed at understanding the link between the fault
microstructure and its strength and the variation of these with sliding velocity, are needed. |