Coal maturation by frictional heat during rapid fault slip


Corresponding author: M. Kitamura, Graduate School of Science, Hiroshima University, 1-3-1 Kagamiyama, Higashi-Hiroshima, Hiroshima 739-8526, Japan. (


[1] The detection of frictional heating effects along faults provides key insight into the dynamics of earthquakes and faulting. Thermal maturity of organic matter has been considered a possible fault-thermometer that records the frictional heat signature of ancient earthquakes. However, whether or not organic matter can mature on the order of seconds, typical earthquake rise time, remains uncertain. Here we present the results of experiments aimed at revealing coal maturation by frictional heat generated at slip velocities representative of natural earthquakes of up to 1.3 m/s. Our results show that coal can mature coseismically in ∼11 seconds at temperatures induced by frictional heat ranging from 26 to 266°C. Even with a temperature rise to only 28.7°C over 15 m displacement in ∼3.2 hours, coal can slightly mature within a shear localized zone. The commonly used kinetic model of vitrinite maturation cannot predict the experimental results. A kinetic model involving the effect of flash temperature at grain contacts and mechanochemical effects on reaction kinetics is necessary to better estimate heat generation along a fault.

1. Introduction

[2] Quantitative assessment of heat generation along a fault during coseismic faulting is of primary importance in understanding the dynamics of earthquakes [e.g., Brodsky et al., 2010]. Evidence of substantial frictional heating along a fault is also a reliable indicator determining whether a fault has slipped at high velocity in the past, which is crucial for assessing earthquake and tsunami hazard. For such reasons, over the last several years various analyses of fault zone materials have been used to infer evidence of localized heating in fault zones, such as: ferromagnetic resonance signal [Fukuchi et al., 2005], trace element and isotope anomalies [Ishikawa et al., 2008], mineralogical changes of clay [e.g., Hirono et al., 2008; Yamaguchi et al., 2011] and vitrinite reflectance [e.g., O'Hara, 2004; Sakaguchi et al., 2011]. Among these, the reflectance measurement of vitrinite (one of the primary components of coals) has been considered a possible geothermometer of fault zones, especially in accretionary wedges where vitrinite fragments are common. Under normal burial conditions, vitrinite reflectance (Ro) increases by irreversible maturation reaction as temperature is elevated and thus sensitively records the maximum temperature to which the vitrinite is subjected. However, the commonly used kinetic models of vitrinite maturation [e.g., Sweeney and Burnham, 1990] may not yield accurate estimates of the peak temperature in a fault zone resulting from fast frictional heating rates [Fulton and Harris, 2012]. Whether or not vitrinite can mature in typical earthquake rise time (e.g., ∼10 seconds) remains uncertain.

[3] In this study, we sheared a mixture of quartz and coal grains at conditions representative of earthquakes, especially in terms of slip velocity and displacement, using a rotary-shear friction apparatus to examine whether coal reflectance (R) increases during heating durations on the order of seconds.O'Hara et al. [2006] conducted similar experiments on coal and reported ∼0.3% increase of R by frictional heating. However, their experiments were conducted exposed to air and eventually the coal reacted with oxygen to carbon dioxide. Such oxidization by frictional heat has been often observed on carbonaceous materials [Oohashi et al., 2011]. As a result, O'Hara et al. [2006]could not well constrain the R increase associated by frictional heating alone. To prevent oxidization in this study, the experiments were conducted under oxygen-free, nitrogen atmosphere. We also measured temperature of the sliding zone during the experiment with thermocouples to better determine a correlation between R and peak temperature for rapid heating over short durations on the order of seconds.

2. Experimental Methods

[4] We conducted friction experiments on a mixture of 90 wt% quartz powder (average grain size ∼100 μm, obtained from Wako Pure Chemical Industries, Ltd.) and 10 wt% coal grains (grain size ranging between 0.15 and 0.5 mm) for simulated fault gouge using a rotary-shear, high velocity apparatus [Shimamoto and Tsutsumi, 1994; Hirose and Shimamoto, 2005]. The coal grains were obtained from highly concentrated fragments of organic matter dispersed in sandstone collected from the Nonokawa Formation in Taisho Group in the Shimanto accretionary belt in Kochi Prefecture, Japan [Taira et al., 1980]. The coal fragments were separated by density separation of the crushed sandstone rock and consist of 72.9 vol% vitrinite, 19.9 vol% inertinite and 7.2 vol% liptinite. Although liptinite was easy to identify microscopically, it was difficult to discriminate between vitrinite and inertinite grains as their grain size became less than 50 μm. Thus, we measured mean random coal reflectance (R) of vitrinite and inertinite grains under thin section with an oil immersion microscope with a micro spot lighting (spot size 1.6 μm) system without the polarizer. The R value of starting materials was 1.34 ± 0.30%.

[5] One gram of the simulated gouge was pressed between a pair of solid-cylindrical sandstone cores with 25 mm diameter for each experiment. A hollow cylindrical Teflon (polytetrafluoroethylene) sleeve was used to confine the gouge between the two sandstone cores (see method inMizoguchi et al. [2009]). High-velocity slip on a fault was obtained by keeping one specimen stationary while rotating the other at high speed by a servomotor. As the slip rate varies within the slip surface as a function of distance from the center of rotation axis, we used an “equivalent slip velocityVeq” to characterize the experiments, defined such that τ·Veq·S gives the rate of total frictional work on the fault area S, assuming that shear stress τ is constant over the slip surface [Shimamoto and Tsutsumi, 1994]. The equivalent slip velocity is referred simply as “velocity” hereafter. Experiments were conducted at three different velocities of 0.0013, 0.65, and 1.3 m/s, a constant normal stress of 1.0 MPa and displacement of ∼15 m under anoxic, dry nitrogen atmosphere in order to avoid oxidization reaction of vitrinite by frictional heat. Temperature in the gouge zone was measured by a K-type thermocouple that was placed on the surface between gouge zone and host rock in the stationary side and at 3 mm away from the periphery of the host rock.

[6] After the experiments thin sections were made perpendicular to the slip zone. Microstructural observation and R measurement were then conducted on the thin sections by optical microscope. The R values were measured on all vitrinite and inertinite grains with size ranging from ∼3 μm to 0.5 mm, with an average size of 30 μm.

3. Results

[7] The evolution of shear stress and temperature with displacement at three different slip velocities are shown in Figure 1. At a velocity of 0.0013 m/s shear stress is nearly constant at ∼0.65 ± 0.1 MPa. For this condition temperature only increases 2.8°C, even after ∼3 hours of sliding. In contrast, for the faster velocities shear stress decreases with displacement from 0.65 to 0.5 MPa and to 0.35 MPa at 0.65 m/s and 1.3 m/s, respectively. Temperatures rise nearly linearly with displacement and reach 211°C in 22 seconds at 0.65 m/s and 266°C in 11 seconds at 1.3 m/s (Table 1). After the fast sliding ceases, temperature returns to initial room temperature in less than 10 minutes.

Figure 1.

Evolution of shear stress and temperature in the simulated gouge zone as a function of displacement at slip velocities of (a) 0.0013 m/s, (b) 0.65 m/s and (c) 1.3 m/s under 1.0 MPa normal stress and dry conditions. Mechanical data were corrected by subtracting Teflon friction obtained from the experiments with air instead of the gouge between the host rocks. Temperature was measured at the surface between the gouge zone and host rock on the stationary side (see also Figure 2).

Table 1. Summary of Experimental Conditions and Results
Run NumberSlip Rate (m/s)Normal Stress (MPa)Displacement (m)Duration (sec)Initial Temperature (°C)Maximum Temperature (°C)Average Rate of Temperature Rise (°C/sec)Average R (%)
HVR 25870.00131.015.31081425.928.70.00023.59 ± 1.30
HVR 30000.651.014.822.918.3210.98.42.01 ± 0.59
HVR 25861.31.014.511.325.7266.421.361.35 ± 0.45

[8] Microstructural observation across the simulated gouge zones shows that deformation localized into a ∼0.15 mm thick zone close to the boundary of the rotation side of host rock for all experiments (Figure 2). The localized shear zones are characterized by grain size reduction to the nano-scale in both quartz and coal and by a grain-supported texture. Observed textures suggest that this zone acted as a zone of localized shear during sliding which would act as a localized heat source.

Figure 2.

Photomicrographs of the quartz and vitrinite mixture gouge sheared at slip velocities of (a) 0.0013 m/s, (b) 0.65 m/s and (c) 1.3 m/s for 15 m of displacement. Location of the thermocouple is shown as a highlight in grey color in the bottom of each image. Reflectance, R, of all vitrinite and inertinite grains were measured and represented in the images as six color differences with 1% steps. R distributions are also shown as a histogram. Average R values are 1.35 ± 0.45% in Figure 2a, 2.01 ± 0.53% in Figure 2b and 3.59 ± 1.30% in Figure 2c, whereas R of the starting coal material is 1.34 ± 0.30%. The photographs were taken under planed-polarized light. SLZ: shear localized zone.

[9] Figure 2 shows the locations of coal grains and their R indicated as six color differences with 1% steps. The R tends to increase toward the shear localized zone. This trend is observed characteristically in the gouge sheared at 1.3 m/s in which the highest temperature of 266.4°C was measured by thermocouple at the end of the experiment (Figure 2c). Even for the small temperature rise to 28.7°C in the experiment at 0.0013 m/s for 3 hours, R values of some coal grains in the shear localized zone change from 1.3% to >2.0% (Figure 2a). Thus, the average R over the gouge zones increases with measured maximum temperature associated with ∼15 m displacement in each experiment (Figure 3a).

Figure 3.

Average values of R for the entire thin section from each experiment (Figure 2) as a function of (a) temperature measured by a thermocouple, and (b) average rate of observed temperature rise. The first standard deviations are shown as a bracket through the dots. The increase in scatter of R at higher temperature results from the larger thermal gradient in the fault zones at higher slip velocity. The measured temperature indicates a minimum value within the gouge zone. The R of the starting coal material is 1.34 ± 0.30%, shown as a highlight in grey color. The kinetic model of vitrinite maturation by Sweeney and Burnham [1990] was applied to the observed experimental temperature histories. The kinetic model predicts no change in R for the experimental results (solid diamonds).

4. Discussion and Conclusion

[10] The experiments at coseismic velocities clearly indicate that coal can mature during fault slip, even over a short time scale. O'Hara et al. [2006]have conducted high-velocity experiments on coal at nearly the same conditions as this study. They reported the increase of R from 0.5 to 0.8% with elevation of temperature up to ∼900°C that is more than two times higher than in this study. However, the R rise of Δ0.3% in their experiments is much smaller than that of Δ2.5% in this study. This is probably because their experiments were conducted in an oxygen-existing atmosphere. Carbonaceous materials react easily with oxygen to carbon dioxide during rapid sliding [Oohashi et al., 2011]. Gas emission from the gouge zone was observed in their experiment [O'Hara et al., 2006], implying the evaporation of coal with high R as a carbon dioxide gas. In contrast, in this study the anoxic, dry nitrogen atmosphere prohibited such reaction as evidenced by the area proportion of coal in the gouge zone remaining nearly unchanged before and after the experiments. However, most natural faults likely contain water within their pore space. It is likely that the existence of water containing dissolved oxygen may further influence the reaction kinetics of coal. In addition, earthquakes in nature occur at normal stress of at least an order of magnitude higher than our experiments. An increase in normal stress linearly increases heat production for a certain period of time so that higher normal stress potentially has a much larger effect on the coal maturation. The effects of water and normal stress on coal maturation during frictional sliding will be examined in subsequent studies.

[11] In our experimental configuration, measured temperature gives a minimum value within the gouge zone. Temperature at a heat source localized in the shear zone must be much higher than the temperature measured by thermocouple at the edge of the gouge zone (Figure 2). Higher temperatures might be reached and thus lead to the higher R values as compared with those reported in O'Hara et al. [2006]. To further estimate the range of temperatures experienced within the gouge zone we performed calculations using a two-dimension finite element method. We assumed that all frictional work during our experiments converts into heat and that a heat source of 0.14 mm thick zone is placed between gouge and host rock within the 1.4 mm-thick gouge zone. Heat generation from the source was determined from the time varying shear stress and velocity data. Thermal properties used in the calculation are: thermal conductivity of 1.0 and 3.9 W/mK, specific heat capacity of 1000 and 800 J/kgK, and density of 2000 and 2350 kg/m3, for the gouge and sandstone, respectively.

[12] The calculation for the experiment at 1.3 m/s shows the highest temperature of 573°C at the heat source and lowest value of 270°C at the edge of the gouge zone, consistent with the measured value of 266°C by thermocouple. Thus, there is a difference of ∼300°C in peak temperature across the gouge zone, although this gradient depends strongly on the thickness of heat source, which is a difficult parameter to determine, as it evolves with displacement [e.g., Mizoguchi et al., 2009]. The peak temperature estimation, however, is still lower than the temperature reported in O'Hara et al. [2006]. In summary, our experiment at coseismic velocity of 1.3 m/s indicates that R increases from ∼1.3 up to ∼5.9% with elevation of temperature up to 573°C over a very short time scale.

[13] An interesting observation is that R of some coal grains in the shear localized zone increases by more than 1.0% even without significant temperature rise, from 25.9 to 28.7°C over 15 m displacement in ∼3.2 hours (Figure 2a). The modeled temperature calculation for this experiment's conditions also indicates temperature rise of <3°C even in the zone of the heat source. Observed high R values likely result from a combination of two possible effects: flash temperature at asperity contacts between quartz and coal grains [Archard, 1959] and mechanochemical reaction associated with deformation increasing defect density in particles and greatly enhancing the reaction kinetics [e.g., Balaz, 2008]. Such deformation-induced reaction was reported on serpentinite sheared at coseismic velocities [Hirose and Bystricky, 2007]. They found that the antigorite serpentine dehydrated in 0.2 s after the onset of rapid sliding, even when the fault surface temperature was more than 400°C lower than the dehydration temperature. Observation by transmission electron microscopy suggested that the rapid dehydration reaction was attributed to flash heating at asperity contacts and mechanochemical effects [Viti and Hirose, 2010]. Thus, when a fault slips fast, chemical reactions may occur in a fault zone much more efficiently than expected based on reaction temperatures in mineral phase diagrams.

[14] According to Levine and Davis [1984], deformation and/or rotation of the coals generate anisotropy within the coal grain, leading to a high reflectance value at a specific crystallographic orientation of the coals. Such deformation effects on the intensity of reflectance could be more significant in a narrow fault zone. Our preliminary observation using a Raman microscope and an optical microscope with the polarizer implies that the reflectance anisotropy is perhaps not developed within sheared coal grains in our experiments. Whether and/or how the deformation generates the anisotropy of coal will be evaluated appropriately through subsequent experimental studies in order to detect heat signal in a fault zone more properly.

[15] To test if a kinetic model of vitrinite maturation can predict our experimental results, we calculated the R increase using the temperatures observed in the experiments (Table 1) and a commonly used kinetic model described by a series of first order Arrhenius equations with activation energies for the parallel reactions within vitrinite [Sweeney and Burnham, 1990]. The calculation indicates no change of R for the short heating duration and temperatures observed in the experiments (Figure 3b). Even for a scenario with heating up to 573°C over 11 seconds, as suggested by thermal modeling, no significant change from the initial R value is expected. Local increase of vitrinite reflectance above the observed regional trend has been reported within fault zones in nature [e.g., O'Hara, 2004; Sakaguchi et al., 2007, 2011]. Those reflectance anomalies were converted to fault temperatures during seismic faulting using the same kinetic model of vitrinite maturation. Our results, as illustrated in Figure 3, may imply that those conversions may overestimate fault temperatures. To better estimate heat generation at a fault during an earthquake, it is necessary to establish a kinetic model of vitrinite maturation that involves the effects of flash temperature and mechanochemical reaction at fast sliding conditions.


[16] We thank A. Sakaguchi for his assistance to install an optical microscope with a vitrinite measurement system. We also thank Michael Underwood and Thomas Mitchell for careful reviews and suggestions. This work was supported by the Japan Society for the Promotion of Science (22740334) and MEXT (21107004).

[17] The Editor thanks Michael Underwood and Thomas Mitchell for assisting in the evaluation of this paper.