Amorphization of quartz by friction: Implication to silica-gel lubrication of fault surfaces



[1] To understand physico-chemical processes at real contacts (asperities) on fault surfaces, we conducted pin-on-disk friction experiments at room temperature, using single crystalline quartz disks and quartz pins. Velocity weakening from friction coefficientμ ∼ 0.6 to 0.4 was observed under apparent normal stresses of 8–19 (18 > 19), when the slip rate was increased from 0.003 to 2.6 m/s. Frictional surfaces revealed ductile deformation of wear materials. The Raman spectra of frictional tracks showed blue shifts and broadening of quartz main bands, and appearance of new peaks at 490–520 and 610 cm−1. All these features are indicative of pressure- and strain-induced amorphization of quartz. The mapping analyses of Fourier transform infrared (FT-IR) spectroscopy at room dry conditions suggest selective hydration of wear materials. It is possible that the strained Si-O-Si bridges in amorphous silica preferentially react with water to form silica-gel. In natural fault systems, amorphous materials would be produced at real fault contacts and accumulate over the fault surfaces with displacements. Subsequent hydration would lead to significant reduction of fault strength during slip.

1. Introduction

[2] Slip– and velocity–weakening of faults is a fundamental process leading to the generation of earthquakes. Recent high speed friction experiments using rotary shear apparatus revealed various mechanisms of dynamic fault weakening: melt lubrication [Hirose and Shimamoto, 2005; Di Toro et al., 2006], powder lubrication [Reches and Lockner, 2010; Han et al., 2011], silica-gel lubrication [Goldsby and Tullis, 2002; Di Toro et al., 2004; Hayashi and Tsutsumi, 2010]. Among them, silica-gel lubrication is clearly distinguished from others because the weakening has been observed even at relatively low slip velocities (<0.03 m/s) under low normal stress (<10 MPa). Using a micro-FT-IR spectrometer and a powder X-ray diffractometer,Hayashi and Tsutsumi [2010]identified hydrated amorphous silica in the gouge materials produced from quartz and chert blocks, and discussed that weakening was related with the formation of a silica-gel layer. However, details of silica-gel lubrication are still poorly understood because of the lack of unequivocal evidence for hydrated amorphous silica formation on slip surfaces of quartz.

[3] Herein we conducted pin-on-disk type friction experiments to simulate a real contact (asperity) on a fault surface (Figure 1). The changes of atomic bonding structures in contact areas due to frictional sliding were analyzed by micro-Raman spectroscopy. Micro-FT-IR spectroscopy was also utilized to determine the degree of hydration on friction surfaces. Based on the results of spectroscopic analysis and mechanical data, we propose a model of fault lubrication due to amorphization and hydration of quartz.

Figure 1.

Schematic drawing of pin-on-disk apparatus.

2. Methods

[4] The pin-on-disk apparatus (Figure 1) [Miura et al., 2004; Muto et al., 2007] simulates large slip (>10 m) at a real contact on a fault plane. Friction occurred at the contact between the pins and rotating disk. Single crystals of natural Brazilian quartz, cut normal to the c-axis, were used for the disks. The thicknesses of disks were 1 and 7 mm, and the radius is 30–35 mm. The surfaces of disks were mirror polished. The tip of the pin is a spherical crystal of natural quartz with a curvature radiusr = 1, 1.5 or 3 mm. The normal load F was applied by a weight of 1 N. The real contact stresses were estimated by two methods (Table 1): Elastic Hertzian stress σh at the contact area is calculated from a curvature radius r, load F, and elastic moduli of the disks and pins [e.g., Tabor, 1951]. Contact stress σc is also calculated from the CCD camera image of the contact area (Figure 1), or estimated from the widths of frictional tracks observed under a scanning electron microscope (SEM) after experiments. The sliding velocity V was kept constant (Run Nos. 1–3) or changed stepwise between 0.0032 and 2.6 m/s (Run No. 4). In the latter case, rotation of the quartz disk was stopped for several minutes between each velocity step. Slip displacements ranged from 20 to 1150 m. All experimental runs (Table 1) were performed under a room temperature. The shear force τ between disk and pin were measured by a force gauge on the horizontal arm. The friction coefficient is calculated as μ = τ / F.

Table 1. Summary of Experimental Runs
RunPinLoad (N)σh a (MPa)σr b (MPa)Velocity (m/s)Displacement (m)
  • a

    Hertz contact stress calculated by the curvature radius and the load of a pin.

  • b

    Real contact stress calculated from the contact images observed by an optical microscope with a CCD camera.

Synthetic Quartz
1quartz (φ = 3 mm)
Brazilian Quartz
2quartz (φ = 6 mm)1.011.8-0.075319.5
3quartz (φ = 6 mm)1.011.8-0.031695.4
4quartz (φ = 6 mm)1.011.8-0.0032 ∼ 2.61149.5

[5] After friction experiments, scratched surfaces of the disks and worn materials were coated with a graphite or gold and observed by SEM. Raman spectra of quartz disks were acquired by a micro–Raman spectrometer (RAMAN-11, Nanophotone Inc.) equipped with a 532 nm laser source. Water-related species on frictional surfaces were investigated by a Nicolet iN-10 infrared microscope (Thermo Fisher Scientific Inc.). A transmission spectrum of a non-deformed quartz disk was taken as a background.

3. Frictional Strength

[6] Figure 2a shows the first (V = 0.1 m/s) and the third (V∼ 1 m/s) step of the velocity-step test (Run No. 4). When the pin started to slide on the mirror-polished surface of the quartz disk, the friction coefficient gradually increased. In the subsequent steps, steady-state values were immediately attained without showing significant slip weakening or hardening. The change of steady state friction coefficientsμss is shown in Figure 2b. Although data scattering is large, μss decreased from 0.6 to 0.4 when V was increased from 0.1 to 2.5 m/s. After showing an instantaneous increase at the highest V, μss gradually increased with decreasing V.

Figure 2.

Results of a velocity step test (Run No. 4). (a) Representative friction data at slip velocities of 0.1 and 1 m/s. (b) The change in steady state friction coefficients. Red and blue arrows indicate the paths of increasing and decreasing slip rates, respectively.

4. Microstructures of Wear Materials

[7] Several wear scars and grooves are observed in a friction track of a single experimental run (Figure 3). The surface of the tracks becomes fragmented and scraped from the disks (Figures 3a–3c). Mixtures of fine-grained particles and cylindrical particles with ‘flowing’ structures were accumulated on the disk (Figure 3c). The cylindrical particles were aligned perpendicular to the sliding direction (Figure 3a). Their lengths are less than 20 μm and diameters are ∼200 nm (Figures 3a and 3d). Particles that are similar in shapes and sizes have been reported from friction surfaces of chert in rotary shear experiments [Hayashi and Tsutsumi, 2010]. Wear particles of this kind are termed ‘rolls’ in tribology of industrial materials [e.g., Danyluk et al., 1994]. Surface structures and profiles of rolls show that they were originated from off-scraped fragments (Figure 3c) and very fine-grained debris (Figure 3d).

Figure 3.

Secondary electron images of (a–c) friction surfaces (Run No. 1) and (d) roll-like wear debris acquired by a Hitachi S-3400N SEM (Figures 3a–3c), and by a Hitachi S-5500 field emission (FE) SEM (Figure 3d) at acceleration voltage of 15 kV and 1 kV, respectively. Arrows indicate the slip direction.

5. Raman Microspectroscopy

[8] The Raman spectra of quartz disks outside the sliding tracks show clear peaks of α-quartz at 128, 204, and 464 cm−1 (Figure 4e) [Kingma and Hemley, 1994]. Within the friction track, peak shifts and broadening of the main bands of quartz were observed (Figures 4a–4d). The most intense band at 464 cm−1due to Si-O-Si bending vibration gradually broadens to show a shoulder on the higher wavenumber side (Figure 4a), and new peaks appeared around 490 ∼ 520 cm−1 (Figures 4b–4d). In addition, a weak broad peak was observed at 600 ∼ 630 cm−1. These peaks reflect transformation of SiO4ring structures from ordinary six-membered one to others; the bands at 490 and 606 cm−1can be assigned to the symmetric stretching vibration of Si-O in a four-membered SiO4 ring (D1band) and a planar three-membered SiO4 ring (D2 band) of amorphous silica, respectively [Awazu and Kawazoe, 2003]. The peaks at 500 cm−1 and 520 cm−1corresponds to the strongest bands of moganite and coesite, respectively, both of which consist of four-membered rings of corner-sharing SiO4 tetrahedra [Kingma and Hemley, 1994].

Figure 4.

(a–d) Raman spectra measured at friction tracks of a quartz pin with a grating of 1200 lines/mm. (e) The Raman spectrum of intact quartz is also presented. The peak positions of quartz are indicated by dotted lines. Run No. 2 (Figure 4a), Run No. 3 (Figure 4b), and Run No. 4 (Figures 4c and 4d).

[9] The distribution of amorphosed materials was investigated by a micro-Raman imaging technique (Figure 5). To calculate the peak areas, the baseline was drawn at the intensity level of 750–770 cm−1. The results clearly show localized occurrence of amorphous silica within the wear grooves and wear materials on the quartz surface.

Figure 5.

Raman mapping images of a friction track overlapped with optical micrographs (Run No. 3). The exposure time was 30 s for each point. White squares indicate mapped area (36 × 45 μm2). Each map shows the integrated intensities around (a) 490 cm−1, (b) 500 cm−1, (c) 520 cm−1, and (d) 610 cm−1, indicated by the color bands in Figure 3. Arrows indicate the slip direction.

6. Infrared Spectroscopic Analysis

[10] The FT-IR absorption spectra of frictional tracks show the presence of broad peaks around 3400 cm−1, corresponding to –OH symmetric stretching of H2O molecules (Figure 6a) [Aines and Rossman, 1984]. The distribution of water on quartz disks was visualized by micro-FT-IR mapping analyses. The integrated absorbance between 2900 cm−1 and 3700 cm−1 above the dotted line in Figure 6a was mapped in Figure 6b. Hydration was only observed along friction tracks with visible grooves on friction surfaces.

Figure 6.

FT-IR analysis of a friction track (Run No. 1). (a) A representative FT-IR spectrum in arbitrarily unit (a.u.). Spectral resolution is 2 cm−1. (b) Optical microscope image of friction tracks and their IR mapping (c). A red squire indicates the mapped area (300 × 380 μm2). The arrow indicates the sliding direction. Absorption spectra were obtained at 128 accumulations for each point. The xy step size was 20 μm and the aperture size was 20 × 20 μm2.

7. Discussions

[11] The change in the vibration modes from those of six-membered rings to four- and three-membered ones observed in the Raman spectra reflects distortion and densification of SiO4network structures, which is accompanied with decrease in the Si-O-Si angle. The results are consistent with previous observation of synthetic or natural fault gouges using transmission electron microscopy (TEM) [Yund et al., 1990; Danyluk et al., 1994; Ozawa and Takizawa, 2007; Janssen et al., 2010]; the lack of distinct diffraction patterns in the worn materials indicates that long-range order was destroyed by comminution or frictional heating. In our experiments, the temperature increase at the tip of quartz pin due to flash heating [Ashby et al., 1991] is estimated to be less than 100°C for a low V experiment (Run No. 3), at which clear spectral changes were observed (Figure 3b). Thus, frictional heating had a negligible effect on quartz amorphization, Moreover, if the amorphous silica had been produced by melting, expansion of the atomic structure rather than densification would be observed.

[12] Blue shifts of the Raman bands and appearance of new peaks on the higher wavenumber side have been reported for solid-state amorphization of quartz in ultra-high pressure experiments using diamond anvil cells [Schmidt and Ziemann, 2000], and indentation hardness tests [Masuda et al., 2011]. In the present experiments, the apparent stresses σh and σrat the pin-and-disk contacts are around 10 MPa (Table 1); however, if we assume that the real contact areas were restricted within the colored areas in Figure 4, the real contact stress under a normal load of 1 N is estimated to be 5 GPa, which well exceeds the quartz-coesite transition pressure of 1.8 GPa at room temperature.Dieterich and Kilgore [1994, 1996] estimated even larger stress (∼10 GPa) for real contacts of quartz. The shear force imposed at the frictional surface would also assist distortion and breaking up of the SiO4 network structure.

[13] Although the hydration reaction requires large activation energy, strained Si-O-Si bonds are highly reactive especially for water, as predicted byAb Initio molecular dynamic simulation [Gibbs et al., 2003]. Therefore, strained rings with narrow Si-O-Si angles would preferentially react with water by the hydrolysis reaction: ≡Si-O-Si≡+ H2O → ≡Si-OH … OH-Si≡. It is possible that the above reaction has proceeded at the highly stressed contact areas, although the experiments were conducted at ambient temperature with room humidity. Extremely soft and ductile deformation of wear materials and formation of rolls might have been assisted by silica hydration.

8. Summary

[14] Because quartz is one of the most abundant minerals in the Earth's crust, the amorphization-and-hydration process described above may commonly occur in natural fault zones. As we have shown, even under a small load, SiO4 networks are disordered and undergo amorphization if the slip distance is large. At higher stress levels at the great depth, the same mechanism could operate in a smaller displacement. Because amorphization proceeds even at low to intermediate slip rates, glassy materials would accumulate over fault surfaces during aseismic fault creep or slow slip events. Subsequent hydration in the presence of water would lead to drastic reduction of frictional strengths. The lubrication of fault surfaces by amorphous silica is also important in earthquake nucleation processes because whether a small slip on the fault surface will grow to a larger event or not depends critically on its velocity dependence at low to intermediate ranges [Shibazaki et al., 2011]. Recently Pec et al. [2012] reported amorphization of feldspars at small bulk strains (γ∼ 2) under the conditions of brittle-plastic transition. It is possible that amorphization and hydration are commonly occurred in the crustal rocks and influence the mechanical behavior of fault zones in various stages of earthquake cycles.


[15] We would like to thank K. Morikawa and O. Kuwano for their help in pin-on-disk experiments, M. Nakamura and S. Okumura for technical assistance in FTIR measurements and FESEM observations, and N. Tsuchiya and A. Okamoto for their help in preliminary Raman spectroscopic measurements. We wish to thank J. Fukuda, T. Nagase, and K. Otsuki for valuable comments. We are also thankful for the thoughtful review of D. Saffer. This study was supported by Tohoku University Global COE Program “Earth Planetary Dynamics” and grants from the Japan Society for the Promotion of Science (22340148 and 70377985), and Observation and Research Program for Prediction of Earthquakes and Volcanic Eruptions from the Ministry of Education, Culture, Sports, Science and Technology (MEXT).

[16] The Editor thanks Demian Saffer for his assistance in evaluating this paper.