2013 年 16 巻 p. 23-29
Because the frictional properties of fault gouge are important for understanding earthquake slip behavior, laboratory studies of high velocity rock friction have been conducted. A sleeve of polytetrafluoroethylene (PTFE) is used around the specimen to maintain pressure and prevent leaks during the experiment, and sometimes appeared to be worn and material from it was mixed into the specimen. However, the effect of PTFE contamination in the specimen is uncertain, although PTFE is known for its extremely low frictional coefficient. Here we reported new quantitative measurements of the amount of contamination by worn PTFE sleeve material in the specimen after experiments by a calorimetric technique using simultaneous thermogravimetry and differential scanning calorimetry. We also performed the friction experiments using mixtures of illite-rich shale and PTFE powder, and demonstrated that high PTFE contamination can affect the friction values of specimens.
Fault gouge, a major component in active faults, is a granular material generated during seismic slip on faults (e.g., Scholz, 1987). Because the frictional properties of fault gouge are important for understanding earthquake slip behavior, laboratory studies of rock friction have been conducted for decades. Early experiments used triaxial or biaxial apparatus at low slip rates (<0.01 m/s) and small displacements (<1 mm) (e.g., Dieterich, 1972; Scholz, 1990; Shimamoto and Logan, 1981; Marone, 1998). However, coseismic slip on natural faults during earthquake has high slip rates (0.01–1 m/s) and large displacements (∼10 m), and high-velocity rotary-shear friction (HVR) experiments like those developed by Tsutsumi and Shimamoto (1997) have recently been adopted to study the behavior and mechanism of more realistic slip (e.g., Hirose and Shimamoto, 2005; Mizoguchi et al., 2007; Han et al., 2007).
For our HVR experiments we used the apparatus shown in Fig. 1(a) to rotate the ends of two cylindrical specimens against each other. One specimen is held fixed and the other is rotated at rates up to 1500 rpm. From measured data on torque, axial force, shear displacement, and axial displacement (change in thickness of the specimen) during the experiments, we can derive the changes of shear stress and frictional coefficient as functions of time and displacement. Whereas rigid specimens such as plutonic rocks may easily be tested without sleeves (Tsutsumi and Shimamoto, 1997; Hirose and Shimamoto, 2005), our tests of noncohesive powder specimens replicating natural fault gouge required sleeves around the interface between the cylinders. We put a 1 mm-thick layer of our experimental material between a pair of gabbro cylinders with rough end surfaces, then sheared it under applied normal stress (Fig. 1(b)). We used a sleeve of polytetrafluoroethylene (PTFE; commercially named Teflon, DuPont; molecular formula, (C2F4)n) around the specimen to maintain pressure and prevent leaks during the experiments (Mizoguchi et al., 2007; Han et al., 2007; Brantut et al., 2008).
High-velocity friction testing apparatus. (a) Simplified sketch of the testing machine: 1, specimen; 2, motor; 3, torque limiter; 4, torque gauge; 5, electromagnetic clutch; 6, rotary encoder; 7, rotary column; 8, torque-axial force gauge; 9, spline; 10, axial force gauge; 11, displacement transducer. Modified from Hirose and Shimamoto (2005). (b) Detailed sketch of the specimen assembly.
After these experiments, the PTFE sleeve sometimes appeared to be worn and material from it was mixed into the specimen (Fig. 2). In actual the contamination within the specimen after HVR experiments was semi-quantitatively measured using by electron probe micro analyzer (Sawai et al., 2012). The presence of the PTFE sleeve has been shown to contribute to the measured shear traction by ∼18% and <8% at the initial peak and the steady-state stage, respectively (Oohashi et al., 2011). However, the effect of PTFE contamination in the specimen is uncertain, although PTFE is known for its extremely low frictional coefficient of approximately 0.1–0.3 (Zhang et al., 1997; Xiang and Gu, 2006). This paper reports our quantitative measurements of the amount of contamination by worn PTFE sleeve material in the specimen after HVR experiments by a calorimetric technique using simultaneous thermogravimetry (TG) and differential scanning calorimetry (DSC). We also investigated the influence of this contamination on the frictional coefficient of a specimen.
(a) Photomicrograph of wear on PTFE sleeve and (b) scanning electron micrograph of sleeve fragment in specimen after HVR experiment (run number, 2366).
The experimental material was prepared from chips of Rochester Shale (New York state, USA), which has been reported to contain dominantly illite (59%), quartz, (23%), kaolinite/dickite (9%), and plagioclase (4%) by X-ray diffraction analysis (Saffer and Marone, 2003). The rock chips were crushed into powder in an agate mill, and the fine fraction with grain size less than 150 μm was used for the HVR experiments.
2.2. HVR experimentsWe used the HVR testing apparatus in Kochi Core Center, details of which were described by Hirose and Shimamoto (2005). A dry powder specimen of approximately 500 mg was put between the ends of two cylinders of gabbro 24.98 ± 0.01 mm in diameter and 20.00 ± 0.50 mm long (Fig. 1(b)). A PTFE sleeve over the cylinders prevents leakage of the specimen as the mobile cylinder is rotated against the fixed cylinder. Inner surface of the sleeve was carefully polished with #2000 emery paper, and was shaped to be 24.99 ± 0.02 mm in inner diameter. As the slip rate varies within the specimen as a function of distance from the axis of rotation, an equivalent slip velocity is defined as a representative value to divide the area of the slip surface into the rate of total frictional work assuming constant shear stress over the surface (e.g., Mizoguchi et al., 2007).
Seven HVR experiments were performed ranging from 1.4 to 2.0 MPa of normal stress and from 3.14 to 6.28 m of slip displacement at ambient room temperature and humidity (Table 1). After the experiments, samples were collected from the slip surface in the outer portion at 12.5–7.0 mm radius and the inner portion at 7.0–0 mm radius (Fig. 3). Each sample was gently crumbled to homogenize it and equalize its particle size before TG-DSC analysis.
Powdered rock specimen after high-velocity friction experiment, showing outer and inner portions sampled for TG-DSC analysis.
Run number | Velocity (m s-1) | Normal stress (MPa) | Total displacement (m) | ΔHapp for solidification of PTFE in samples (J g-1) | Fraction of PTFE in samples (wt%) | ||
---|---|---|---|---|---|---|---|
Outer | Inner | Outer | Inner | ||||
HVR2367 | 1.31 | 1.4 | 5.23 | 2.486 | 0.000 | 7.60 | 0.00 |
HVR2368 | 1.31 | 1.6 | 5.23 | 2.769 | 0.000 | 8.47 | 0.00 |
HVR2365 | 1.31 | 1.8 | 5.23 | 5.342 | 0.000 | 16.34 | 0.00 |
HVR2369 | 1.31 | 2.0 | 3.14 | 0.987 | 0.000 | 3.02 | 0.00 |
HVR2370 | 1.31 | 2.0 | 4.19 | 1.307 | 0.000 | 4.00 | 0.00 |
HVR2366 | 1.31 | 2.0 | 5.23 | 3.321 | 0.000 | 10.16 | 0.00 |
HVR2371 | 1.31 | 2.0 | 6.28 | 4.826 | 0.000 | 14.76 | 0.00 |
We performed TG and DSC analysis of the specimens simultaneously using a Netzsch STA 449 C Jupiter balance. The TG curve, showing the weight lost by a specimen during heating, can be used to detect reactions accompanying mass loss, such as dehydration of clay minerals. The DSC curve allows us to estimate the amount of energy required for the reaction during heating. The resolutions of TG, DSC, and temperature signals were 1 μg, 1.25 μW, and 0.01°C, respectively.
The TG-DSC curve for the Rochester Shale specimen before the HVR experiment from 50°C to 1000°C is shown in Fig. 4(a). A decrease of ∼5 wt% in the specimen weight, with a large endothermic valley around 400–800°C, corresponds to dehydroxylation of illite and kaolinite/dickite in the specimen. A slight endothermic valley around 570°C without a weight-loss peak is probably related to the α–β phase transition of quartz. In contrast, the TG-DSC curve of the material after the HVR experiment (HVR2366: slip velocity 1.31 m s-1, normal stress 1.4 MPa, slip displacement 5.23 m) shows an extreme exothermic peak accompanying a weight loss of approximately 10 wt% (Fig. 4(b)). This signal corresponds to fluoridation of silicate minerals by fluoride gas released from thermally decomposed PTFE, and an endothermic valley without a weight-loss peak observed around 327°C corresponds to melting of PTFE (e.g., Androsch et al., 2005; Lage et al., 2004). These results imply that the specimen became contaminated with PTFE from the sleeve during the experiment. The same slight endothermic valley around 570°C, related to the α–β phase transition of quartz, is recognized in the TG-DSC curve of the post-experiment sample.
(a) Relationships of heat flux and total weight loss to temperature of the specimen (a) before the high-velocity friction experiment over the temperature range 50–1000°C and (b) collected from the outer portion after HVR experiment HVR2366 (slip velocity 1.31 m s-1, normal stress 1.4 MPa, slip displacement 5.23 m). DSC, differential scanning calorimetry (left axis); TG, thermogravimetry (right axis).
The amount of PTFE contamination in the specimen was evaluated from the TG-DSC data under the assumption that the solid-liquid phase change of PTFE is reversible and that the latent heat required for the reaction in the bulk specimen corresponds to the amount of PTFE it contains. The latent heat for the solid-liquid transition, ΔH, on the TG-DSC curve is expressed as
(1) |
(2) |
Approximately 30 mg of the post-experiment samples was separated in a covered Pt90Rh10 crucible, heated from 50°C to 350°C, and then cooled from 350°C to 250°C at a rate of 10°C min-1 under a flow of argon gas (50 mL min-1). Because the sample weight during the heating stage decreased owing to dehydration of clay minerals, its DSC curve was not appropriate for determination of the PTFE content. In contrast, the weight during cooling stage was almost constant, and we used this DSC curve. The peak 0.7 area between 260°C and 321.5°C relative to a linear baseline on the cooling curve was determined under constant value of K, and ΔHapp was then calculated by dividing by the initial mass of the bulk specimen before heating. Finally, the amount of contaminated PTFE in the specimen from the outer and inner portions after the HVR experiments was evaluated using equation (2).
The DSC curves during the cooling stage of the samples from the outer portion along with the latent heat ΔHapp for the solidification of PTFE are shown in Figure 5. One example from the inner portion (experiment HVR2371) is also exhibited. The values of ΔHapp determined from the DSC data and the resulting amounts of contaminated PTFE calculated using equation (2) are listed in Table 1.
Relationship of heat flux to specimen temperature during cooling from 350°C to 250°C (red curve) and latent heat (blue pattern) for solid-liquid transition of PTFE. (a) PTFE powder. (b) PTFE-contaminated specimens after HVR experiments.
PTFE was present in the outer samples from all experiments, ranging in abundance from 16.34 wt% (experiment HVR2365) to 3.02 wt% (experiment HVR2369). The amount of PTFE increased with displacement and with normal stress, excepting HVR2365. No contamination was detected in the inner samples (Table 1).
Because PTFE has a low friction coefficient (Zhang et al., 1997; Xiang and Gu, 2006), fragments of the PTFE sleeve may act as lubricating particles. Previous HVR experiments using mixtures of fault gouge (ultracataclasite from Punchbowl fault, San Andreas fault system) and PTFE under conditions of 0.6 MPa normal stress and 1.3 m s-1 slip velocity showed that the frictional coefficient was unaffected by PTFE contamination of less than 50 wt% (Kitajima et al., 2010). However, our series of evaluations used higher normal stresses and showed that the amount of contamination increases with not only slip displacement but also normal stress (1.4–2.0 MPa). Therefore, we performed four HVR experiments on mixtures of Rochester Shale and PTFE powder (average grain size of 25 μm) under conditions of 2.0 MPa normal stress and 1.31 m s-1 slip velocity as listed in Table 2. The resultant changes of frictional coefficient with slip displacement are shown in Fig. 6. Peak values of frictional coefficient decreased when 10 wt% PTFE was added to the shale, but the steady state values after weakening were similar. Specimens with 20 wt% and 30 wt% PTFE showed very small initial peak values, but steady state levels that were the same as in the other experiments. We have no explanation for the slight peak around 1.3 m displacement in the specimen with 20 wt% PTFE.
Frictional coefficient as a function of slip displacement for different mixtures of Rochester Shale and PTFE powder.
Run number |
Velocity (m s-1) |
Normal stress (MPa) |
Total displacement (m) |
Proportion of PTFE powder before experiment | |
---|---|---|---|---|---|
Outer | Inner | ||||
HVR2734 | 1.31 | 2.0 | 5.23 | 0.0 | 0.00 |
HVR2368 | 1.31 | 2.0 | 5.23 | 10.0 | 0.00 |
HVR2365 | 1.31 | 2.0 | 5.23 | 20.0 | 0.00 |
HVR2369 | 1.31 | 2.0 | 5.23 | 30.0 | 0.00 |
Our calorimetric technique using TG-DSC enabled us to evaluate the frictional effects of contamination by the PTFE sleeve during HVR experiments: high normal stresses and high total displacements accompany high contamination of as much as 16.34 wt% PTFE. Experiments using mixtures of shale and PTFE powder demonstrated that PTFE contamination strongly affects the initial peak friction values of specimens, but does not affect the steady state friction. However, the initial stage during HVR experiment does not have much contamination over 10 wt% (Table 1), so the affection on the value of frictional coefficient would not have to be considered practically. In addition, our experiments applied intrinsically low friction material of illite-rich shale powder (e.g., approximately 0.3 friction coefficient; Ikari et al., 2009). Small difference in friction between the shale sample and PTFE may result in same level of the steady state friction regardless of the amount of PTFE contamination. For comprehensive evaluation of such cause and affection, more experiment using various rock types are of importance.
Although PTFE has no substitute with equivalent properties of low friction, high wear resistance, and nonreactivity, PTFE contamination can be evaluated using our calorimetric technique, allowing the validation of the friction data. However, PTFE starts to melt around 310°C and to decompose around 400°C (Fig. 4(b)). The fluorine gas released by decomposition can react with most components of the rock specimens accompanying intense oxidization. Frictional heat during an HVR experiment that reaches the melting temperature may cause leakage of PTFE into the specimen, and above the decomposition temperature the specimen is chemically transformed. These changes may affect the frictional coefficient of specimens. For this evaluation, more experiment reaching higher temperature by friction is needed as future study.
We are grateful to Takehiro Hirose and Hiroko Kitajima for helpful comments and to Hideki Mukoyoshi for his technical support in the HVR experiments. We also thank Kohtaro Ujiie and an anonymous reviewer for their constructive reviews, and editor Weiren Lin for editing this report.