Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 Very-low-frequency earthquakes (VLFE) occur within accretionary prisms and near subduction plate boundaries at slip rates of 0.05–2 mm/s. However, the geological and frictional aspects of VLFE remain poorly understood. The thrusts in the Shimanto accretionary complex exhumed from source depths of VLFE are composed of quartz-rich fault rocks with or without clay foliations. We examined the frictional velocity dependence of thrust materials. At slip rates of 0.0028–0.28 mm/s, the powder sample from non-foliated fault rock shows velocity-weakening behavior, while that from foliated fault rock exhibits velocity-strengthening behavior. Both samples show velocity-strengthening behavior at slip rates of 0.28–2.8 mm/s. Microstructural analysis reveals that the velocity-weakening sample shows a shear localization, while the velocity-strengthening sample is marked by clay foliations oblique and parallel to shear zone boundaries. Our results imply that frictional instability generates along quartz-rich fault rock but slip becomes stable at higher slip rates, regardless of mineral composition. This is favorable for the occurrence of VLFE in subduction zones.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Recent observations by on-land seismic networks and broadband ocean-bottom seismometers in the Nankai and Ryukyu subduction zones have identified the occurrence of shallow very-low-frequency earthquakes (VLFE) along thrusts in accretionary prisms and near subduction plate boundary at depths of ~10 km [e.g., Ito and Obara, 2006a; Ando et al., 2012; Sugioka et al., 2012] (Figure 1a). The focal mechanism of VLFE indicates reverse faulting, and the duration of VLFE (30–100 s) is much longer than that of ordinary earthquakes with comparable magnitudes [Sugioka et al., 2012]. The estimated VLFE slip rates range from 0.05 to 2 mm/s [Ito and Obara, 2006b]. These features suggest that the slip along thrusts in accretionary prisms is slow yet unstable and seismic. On the basis of the constitutive relationships between P-wave velocity, porosity, and effective stress, Kitajima and Saffer  have shown that the source region of VLFE in the Nankai accretionary prism coincides with the region having elevated pore pressure. Sugioka et al.  found that VLFE are accompanied by high-frequency waves, which are interpreted as representing the simultaneous occurrence of shear failure and hyrdrofracturing at an elevated fluid pressure. However, seismological observations are unable to ascertain the processes involved in thrusts during VLFE.
 Previous friction experiments have revealed that clay-rich accreted sediments recovered by ocean drilling generally exhibit velocity-strengthening frictional behavior, which is consistent with frictional velocity dependence of clay-rich gouges [Morrow et al., 1992; Saffer and Marone, 2003; Ikari and Saffer, 2011]. Thus, unstable slip is unlikely to be initiated in clay-rich fault zone materials.
 The Shimanto accretionary complex distributed along the Pacific side of southwest Japan and the Ryukyu Arc is considered to represent an ancient analogue of the Nankai accretionary prism [Taira et al., 1988] (Figure 1a). The Eocene Kayo Formation in the Shimanto accretionary complex of Okinawa Island, Ryukyu Arc consists of coherent turbidites that were deformed by folds and thrusts [Ujiie, 1997]. Structural and thermal histories recorded in the Kayo Formation suggest that folds and thrusts developed when trench turbidites were scraped off and accreted to the overriding plate by the subduction of the young oceanic plate [Ujiie, 1997]. It is noted that there was a high paleo-geothermal gradient in the Eocene Shimanto accretionary complex (50 °C/km) [Sakaguchi, 1996], and that the buried depths of the Kayo Formation lie at 5–6 km. Recently, Ando et al.  reported that VLFE occur in the accretionary prism along the Ryukyu subduction zone, which includes offshore Okinawa Island (Figure 1a). Therefore, the thrusts in the Kayo Formation could be useful in investigating the actual conditions in the source region of VLFE. In this study, we report the characteristics of thrusts in the Kayo Formation and the frictional velocity dependence of thrust materials. We then present the geological and frictional aspects of VLFE in accretionary prisms.
2 Characteristics of Thrusts in the Exhumed Accretionary Prism
 The examined thrusts in the Kayo Formation are accompanied by a hanging wall anticline (Figures 1b and 1c). The kinematic indicators such as slickenlines, slickensteps, mesoscopic duplex structures, and composite-planar fabric show top-to-the-southeast sense of shear, representing a reverse slip orthogonal to the general trend (NE-SW) of the Kayo Formation. The host rock of the thrusts is quartz arenite that constitutes sandy turbidites. The thickness of the fault zones varies both along the strike and dip directions, ranging from 5 to 75 cm.
 The fault zone structure is complex and is characterized by foliated fault rocks, non-foliated fault rocks, and discrete slip surfaces. The foliated fault rocks typically comprise fragments of sandstone and quartz veins surrounded by clay foliations (Figure 1d). The clay foliations are mostly composed of illite, and they commonly show anastomosing surfaces or composite-planar fabrics such as the P-Y fabric. The non-foliated fault rocks comprise randomly oriented brecciated sandstone and quartz veins. The discrete slip surfaces are indicative of the relatively localized deformation. Some discrete slip surfaces are continuous and straight (Figure 1e), while others bifurcate into several strands. Slip surfaces often are filled with quartz veins or are amalgamated with the surrounding rocks without grain size-reduction associated with cataclasis. The lithology or the orientations of sedimentary structures such as parallel lamination in sandy turbidites commonly change across the discrete slip surface. Where several discrete slip surfaces are observed, the brecciated rocks and the laminated sandstone come in contact with each other across these surfaces (Supporting Information). Extensional quartz veins and the sandstone cemented by quartz are also cut and displaced by the discrete slip surfaces.
3 Friction Experiments on Thrust Materials
3.1 Experimental Methods
 Samples were taken from foliated and non-foliated fault rocks in the thrusts and from the host rock (quartz arenite) of the fault rocks in the Kayo Formation. The quantitative rock composition was determined from X-ray diffraction data using the RockJock computer program [Eberl, 2003] (Supporting Information). The samples taken from the foliated fault rock have higher clay content and lower quartz content compared to those from the non-foliated fault rock and the host rock (Table 1). The quartz content of the sample from the non-foliated fault rock is higher than that of the host rock; this is consistent with the concentration of quartz veins and the cementation in the thrust. The clay minerals are mostly composed of illite. We tested samples as powder gouge: samples were disaggregated by pestle and then sieved to a grain size of <0.17 mm to remove the remaining large clasts.
Table 1. Mineral composition of experimental samples
HR: sample from host rock
N-FR: sample from non-foliated fault rock
FR: sample from foliated fault rock
 We conducted friction experiments using the rotary shear, an intermediate- to high-velocity frictional testing apparatus in Kyoto University [Ujiie and Tsutsumi, 2010; Tsutsumi et al., 2011]. This apparatus can control the input rotation speed by using a servomotor and reduction gears, which permits velocities as low as 0.0028 mm/s. In each experiment, 1 g of gouge was placed between a pair of solid-cylindrical gabbro specimens. The gabbro specimens were 25 mm in diameter, and the end surfaces were ground with #80 SiC powders to prevent slip along the interfaces between the gouge and gabbro specimens. Since the thrusts are well preserved fluid-related phenomena (e.g., quartz veins and cemented sandstone), 0.6–0.8 mL of distilled water was added to the gouge to fulfill wet conditions. A hollow-cylindrical Teflon sleeve surrounded the gouge to prevent gouge extrusion during rotary shearing. The assembled gouge was consolidated for 30 min at the same normal stress (σn = 1.0 MPa) as that during tests. The initial gouge thickness was 1 mm. Experiments at low normal stress under wet conditions could characterize the frictional slip at low effective stress associated with elevated pore pressure.
 Experiments were conducted at 0.28 mm/s until the displacement of ~150–200 mm, over which steady-state friction was achieved. Thereafter, we imposed velocity steps in the range of 0.0028–2.8 mm/s to examine the velocity dependence of friction. The coefficient of friction (μ) is calculated as μ = τ/σn where τ is the steady-state shear stress. We estimated the velocity dependence of the steady-state friction (μss) as the friction rate parameter (a – b) = Δμss/ΔlnV where V is the slip velocity. The up-step velocity is defined as the higher of the two slip velocities in each velocity step experiment, and we used the first several hundred microns of displacement after the up-step velocity changes for estimating (a – b) at slip rates of 0.0028–0.28 mm/s. Positive values of (a – b) indicate velocity-strengthening behavior, resulting in stable slip. Negative values of (a – b) indicate velocity-weakening behavior, potentially resulting in unstable slip and earthquake nucleation [Marone, 1998]. Concerning velocity steps between 0.28 and 2.8 mm/s, the friction apparatus is difficult to measure the transient friction response to velocity steps. Thus, velocity dependence was evaluated from the steady state portion of the friction curve over several millimeters of displacement.
3.2 Experimental Results
 Typical experimental results are shown in Figure 2. Friction is high and is almost independent of mineral composition; friction coefficients of powdered samples from foliated fault rock, non-foliated fault rock, and host rock are in the ranges of 0.55–0.66, 0.57–0.67, and 0.55–0.68, respectively.
 At slip rates of 0.0028–0.28 mm/s, the samples from non-foliated fault rock and host rock exhibit velocity-weakening frictional behavior, with (a – b) ranging from −0.0042 to −0.0007 and −0.0028 to −0.0007, respectively. In contrast, the samples from foliated fault rock commonly exhibit nearly velocity-neutral to velocity-strengthening behavior, with (a – b) ranging from −0.001 to 0.0047. For a sequence of velocity steps between 0.28 and 2.8 mm/s, all samples show velocity-strengthening behavior regardless of mineral composition.
3.3 Resulting Microstructures
 After experiments, the microstructures of the gouges were examined using thin sections cut tangential to the outer rims of the sample assembly (Figure 3). The microstructure of velocity-weakening samples is characterized by the development of a ~0.2-mm-thick localized shear zone along the boundary between the gouge and gabbro specimen (Figures 3a and 3b). The localized shear zone is marked by a dark band containing quartz and feldspar grains in the dark matrix. There is little evidence of grain size-reduction of quartz and feldspar grains; comminution is mostly accommodated by clay minerals (mostly illite), resulting in the dark matrix. The microstructure of the velocity-strengthening samples also shows a thin (~0.2 mm in thickness) shear zone along the gouge boundary (Figure 3c). However, the alignment of clay minerals oblique and parallel to gouge boundaries is observed in the thin shear zone, representing P- and Y-type shears, respectively (Figure 3d); these features are absent in the velocity-weakening samples. The clay foliations are neither penetrative nor continuous in the thin shear zone.
 High frictional strengths of samples are consistent with high quartz content, ranging from 51.5% to 75.8%, and exhibit no systematic change with clay content, changing from 4.9% to 23.5% (Table 1). On the other hand, there is a relationship between frictional velocity dependence and mineral composition. At slip rates of 0.0028–0.28 mm/s, the samples from non-foliated fault rock and the host rock with higher quartz content/lower clay content show velocity-weakening behavior, while those from foliated fault rock with lower quartz content/higher clay content exhibit nearly velocity-neutral to velocity-strengthening behavior (Figure 4). The difference in frictional velocity dependence between samples is consistent with the fact that the quartzofeldspathic-rich gouge tends to show velocity-weakening behavior, whereas phyllosilicate-rich gouge commonly shows velocity-strengthening behavior [Morrow et al., 1992; Saffer and Marone, 2003; Ikari et al., 2007; 2011].
 The examination of frictional velocity dependence using the rotary shear frictional testing apparatus can achieve a large displacement, and the microstructure of velocity-weakening samples is characterized by the development of a localized shear zone along gouge boundaries. Velocity-weakening behavior may represent an evolution from stable to unstable slip with increased displacement (shear strain), during which a deformation style in gouge could change from distributed to localized shear [Logan et al., 1992; Marone et al., 1992; Beeler et al., 1996; Scruggs and Tullis, 1998; Tsutsumi et al., 2011]. The clay foliations associated with P- and Y-type shears along thin shear zone only occurs in the velocity-strengthening samples. Recent friction experiments demonstrated that frictional sliding along foliations composed of phyllosilicates contributed to fault weakness [Collettini et al., 2009; Niemeijer et al., 2010]. The clay foliations in our velocity-strengthening samples do not develop as through-going foliations but are rather discontinuous. This could contribute to high frictional strength and stable frictional behavior of faults. When the slip rate is increased from 0.28 to 2.8 mm/s, all samples show velocity-strengthening behavior, regardless of mineral composition (Figure 4). This velocity-strengthening behavior may be a result of the dilatancy hardening due to the decrease in pore pressure [Segall and Rice, 1995; Samuelson et al., 2009]. Since the water-saturated powder gouge was sandwiched between a pair of impermeable gabbro specimens, the transient pore pressure effect may occur especially at high slip rates.
 Frictional velocity dependence of thrust materials in the accretionary prism exhumed from VLFE source depths provides an important implication for the occurrence of VLFE in accretionary prisms. Velocity-weakening behavior of quartz-rich fault rock derived from sandy turbidites at slip rates of 0.0028–0.28 mm/s could induce nucleation of unstable slip along thrusts, but the stabilizing property of the fault rocks at slip rates of 0.28–2.8 mm/s could inhibit rupture propagation, which would be favorable for slow earthquakes such as VLFE. In sediments-dominated convergent margins such as Nankai Trough, sandy turbidites tend to be accreted to the overriding plate, which can constitute one of the major host materials in source area of VLFE [Taira et al., 1992; Moore et al., 2001; 2009]. In the Nankai accretionary margin, the elevated pore pressure is linked to VLFE [Ito and Obara, 2006b; Sugioka et al., 2012; Kitajima and Saffer, 2012]. The thrusts in the Kayo Formation ubiquitously preserve fluid driven brittle processes such as slip surfaces filled with quartz veins. The amalgamated slip surfaces without grain size-reduction by cataclasis could also represent faulting under fluid-rich and nearly unlithified conditions. These features suggest that faulting occurs even in high friction material if the effective stress is low because of the elevated pore pressure (if fluid input is higher than fluid escape rate, high fluid pressure could occur even in permeable material).
 Quartz-rich materials also develop in fault rocks of subduction plate boundary (i.e., tectonic mélange) [Kimura et al., 2012]. For example, quartz fibrous shear veins and quartz cementation are commonly observed in tectonic mélanges in exhumed accretionary prisms [Ujiie, 2002; Fagereng et al., 2010]. Quartz-rich thrusts, derived not only from sandy turbidites but also from quartz veining and cementation in tectonic mélanges, are potentially sources of VLFE in accretionary prisms and subduction plate boundaries. However, when an authigenic growth of clay minerals occurs in thrusts (e.g., due to hydrothermal fluid flow and alteration), and clay foliations develop along the slip zone, the velocity-weakening behavior of quartz-rich thrusts could change to velocity-strengthening behavior, thereby arresting the generation of VLFE.
 Alternatively, in the heterogeneous fault zone where the competent lenses are surrounded by foliated phyllosilicates, fault creep may initially start along the weak and frictionally stable foliated phyllosilicates [Fagereng and Sibson, 2010; Collettini et al., 2011]. In such a case, fault creep could concentrate the stress within and around the strong and unstable competent lenses, potentially resulting in slip localization along the stronger competent lenses. This could be useful to describe the relation between aseismic slip and earthquake nucleation. In any case, the evolution of the internal structure of the thrusts can affect the spatial distribution of mineral composition and clay fabric, which in turn controls the frictional behavior of thrusts in accretionary prisms and subduction plate boundaries.
 Observations of the thrusts in the Eocene Kayo Formation of the Shimanto accretionary complex exhumed from the source depths of VLFE indicate that a frictional slip along quartz-rich material is not unusual in accretionary prisms. A frictional velocity dependence of the samples from quartz-rich thrust material, showing velocity weakening at 0.0028–0.28 mm/s but velocity-strengthening at 0.28–2.8 mm/s, is favorable for the occurrence of VLFE. However, when clay foliations develop in such thrust (e.g., owing to an authigenic growth of clay minerals), thrust becomes frictionally stable as the samples with clay foliations shows velocity-strengthening behavior at 0.0028–2.8 mm/s. These results suggest that the quartz content and development of clay foliations along thrusts may be factors in controlling the occurrence and spatial distribution of VLFE in accretionary prisms.
 We thank editor Andrew Newman and two anonymous reviewers for helpful suggestions and comments that have improved this paper. This research was supported by Grant-in-Aid for Scientific Research on Innovative Areas from MEXT (21107005).