Influence of phyllosilicates on fault strength in the brittle–ductile transition: insights from rock analogue experiments

Despite the fact that phyllosilicates are ubiquitous in mature fault and shear zones, little is known about the strength of phyllosilicate-bearing fault rocks under brittle–ductile transitional conditions where cataclasis and solution-transfer processes are active. In this study we explored steady-state strength behaviour of a simulated fault rock, consisting of muscovite and halite, using brine as pore fluid. Samples were deformed in a rotary shear apparatus under conditions where cataclasis and solution transfer are known to dominate the deformation behaviour of the halite. It was found that the steady-state strength of these mixtures is dependent on normal stress and sliding velocity. At low velocities (,0.5 mm s) the strength increases with velocity and normal stress, and a strong foliation develops. Comparison with previous microphysical models shows that this is a result of the serial operation of pressure solution in the halite grains accommodating frictional sliding over the phyllosilicate foliation. At high velocities (.1 mm s), velocityweakening frictional behaviour occurs along with the development of a structureless cataclastic microstructure. Revision of previous models for the low-velocity behaviour results in a physically realistic description that fits our data well. This is extended to include the possibility of plastic flow in the phyllosilicates and applied to predict steady-state strength profiles for continental fault zones containing foliated quartz–mica fault rocks. The results predict a significant reduction of strength at mid-crustal depths and may have important implications for crustal dynamics and seismogenesis. Classical models for the steady-state strength of the crust consist of a two-mechanism brittle– ductile strength profile, based on Byerlee’s law plus a dislocation creep law for quartz (e.g. Sibson 1977; Schmid & Handy 1991; Scholz 2002). However, grain size sensitive processes, such as pressure solution, and the production of weak phyllosilicates known to be important under mid-crustal, brittle–ductile transitional conditions (e.g. Rutter & Mainprice 1979; Passchier & Trouw 1996; Imber et al. 2001; Holdsworth et al. 2001) are neglected, as are the effects of phyllosilicate foliation development. Such processes have long been anticipated to lead to some form of hybrid frictional–viscous rheological behaviour in the brittle–ductile transition (Sibson 1977; Rutter & Mainprice 1979; Lehner & Bataille 1984/85; Schmid & Handy 1991; Wintsch et al. 1995; Handy et al. 1999). The steady-state stress levels at which midcrustal fault rocks deform may, therefore, be much lower than those predicted using a classical two-mechanism strength profile (see Fig. 1) (Sibson 1977; Byerlee 1978; Schmid & Handy 1991; Scholz 2002). Moreover, phyllosilicate foliation development and processes such as pressure solution can be expected to play an important role in controlling transient healing, cementation and strength recovery of fault rocks, thus influencing the rateand statedependent frictional and seismogenic behaviour (Fredrich & Evans 1992; Beeler et al. 1994; Karner et al. 1997; Bos & Spiers 2000, 2002a; Beeler & Hickman 2001; Saffer & Marone, 2003). Numerous authors have considered the possible weakening effects of phyllosilicates, foliation development, pressure solution and cataclasis within faults and shear zones in the brittle–ductile transition, using both theoretical and experimental approaches (Rutter & Mainprice 1979; Lehner & Bataille 1984/85; Logan & Rauenzahn 1987; Kronenberg et al. 1990; Shea & Kronenberg 1992, 1993; Mares & Kronenberg, 1993; Chester 1995; Blanpied et al. 1998; Gueydan et al. 2001, 2003). Such studies have produced a consensus that fault rocks containing a contiguous and From: BRUHN, D. & BURLINI, L. (eds) 2005. High-Strain Zones: Structure and Physical Properties. Geological Society, London, Special Publications, 245, 303–327. 0305-8719/05/$15.00 # The Geological Society of London 2005. well-developed phyllosilicate (mica) foliation can potentially be as weak as the frictional strength of the phyllosilicate, at high crustal levels (Logan & Rauenzahn 1987; Shea & Kronenberg 1992), or as weak as the crystalplastic flow strength of the phyllosilicate basal plane, at deeper levels (Hickman et al. 1995; Wintsch et al. 1995). However, the time and technical limitations of experiments on silicate fault rocks (e.g. Blanpied et al. 1991, 1995; Chester & Higgs 1992; Chester 1994; Kanagawa et al. 2000) have precluded systematic studies of large strain sliding behaviour, with associated foliation development, under hydrothermal, brittle–ductile conditions where the relevant processes are active. Large strain, steady-state rheological laws for faults incorporating the effects of cataclasis, pressure solution and foliation development in the brittle–ductile transition are therefore not available. Recently, Bos and co-workers (Bos et al. 2000a, b; Bos & Spiers 2001) performed ultrahigh strain rotary shear experiments on simulated fault gouges consisting of mixtures of halite (rock salt) and kaolinite. The experiments were carried out under room temperature conditions where pressure solution and cataclasis are known to dominate over dislocation creep in the halite, thus modelling the brittle–ductile transition. In wet samples with .10 wt% clay brittle failure was followed by strain weakening towards a steady-state shear strength that was dependent on both sliding rate and normal stress (frictional–viscous flow). Strongly foliated microstructures were produced in these experiments, closely ressembling natural mylonite or phyllonite microstructures, without the operation of dislocation creep. Bos & Spiers (2002b) developed a microphysical model to explain the observed steady-state behaviour, based on the steady-state microstructure and corresponding mechanical analogue diagram of Figure 2. In this model, the shear strength of the gouge is determined by the combined resistance to shear offered by frictional sliding on the phyllosilicate foliae, pressure solution in the halite and dilatation on the foliation (work against the normal stress). The model accordingly predicts three velocity regimes, namely: (1) a low-velocity regime, where pressure solution is so easy that the strength of the gouge is determined by sliding friction on the foliation; (2) an intermediate velocity regime, where the strength of the gouge is determined by accommodation through pressure solution; and (3) a highvelocity regime, where pressure solution is too slow to accommodate geometric incompatibilities, so that dilatation occurs. Bos & Spiers (2002b) reported good agreement between their experimental data and model, and went on to apply the model to predict crustal strength profiles for quartz–mica fault rocks. These showed a major weakening (2–5 times) of crustal fault zones around the brittle–ductile transition (5–15 km, see Bos & Spiers 2002b), in qualitative agreement with inferences drawn from numerous geological and geophysical studies (Lachenbruch & Sass 1980; Schwarz & Stöckhert 1996; Imber et al. 2001; Stewart et al. 2000; Zoback 2000; Townend & Zoback 2001). However, there are several aspects of the Bos–Spiers model that have not been tested or are not physically realistic. First, the model is based on experiments in which the phyllosilicate phase (ultrafine kaolinite) was unrealistically fine in relation to the halite grains and to natural faultrock microstructures (e.g Imber et al. 2001). Second, the model has not been tested in the low-velocity regime (Regime 1), nor adequately at high velocities (Regime 3) where it must eventually break down due to fault-rock failure. Third, the model employs an unnecessary and physically unrealistic approximation to couple the mechanical effects of pressure solution and dilatation, and deals only with a single-valued grain size. Finally, the model is restricted to frictional behaviour of the phyllosilicate foliae, whereas under the conditions of the Fig. 1. Schematic of classical crustal strength profile, showing brittle–frictional behaviour dominating at upper crustal levels, and dislocation creep determining crustal strength at deeper levels (solid lines). The dashed line represents the widely accepted effects of fluidassisted deformation mechanisms on crustal strength (After Bos et al. 2000b.) A. R. NIEMEIJER & C. J. SPIERS 304