Introduction

Japan is subject to frequent earthquakes, and active faults are abundant not only along offshore plate subduction boundaries but also inland (Fig. 1a). The earthquake risk to nuclear power plants in Japan was a matter of debate before the 2011 Tohoku-oki earthquake, notably after the 2007 Mw 6.6 Niigata Chuetsu-oki earthquake caused minor damage to the Kashiwazaki-Kariwa power plant and the release of a negligible amount of radiation1. The earthquake risk of the Chubu active fault system was also considered in 20002. However, since the Fukushima Daiichi nuclear power plant disaster in 2011, most nuclear power plants in Japan have been shut down and their seismic safety is being reassessed. A prime concern is whether mapped faults near a plant are active; for example, the Tsuruga plant’s reactors are 250 m from a known active fault, and several minor faults may run beneath the reactors3,4.

Figure 1
figure 1

Locations of nuclear power plants and active faults in Japan.

(a) Tectonic setting of the Japanese islands with locations of nuclear plants, major active faults, and features named in the text. EUR, Eurasia plate; NAM, North America plate; PHS, Philippine Sea plate; PAC, Pacific plate. This map was modified by the authors from open-source data53 using Adobe Illustrator graphic software. (b) Geologic map around the ATTL, modified by the authors from published data10 using Adobe Illustrator. (c) Photo of study site showing displacement between a Quaternary terrace deposit and Cretaceous granite along the ATTL. (d) Close-up of the ATTL fault zone showing sampling points. HW, hanging wall; FB, fault breccia; PSZ, primary slip zone; FG, fault gouge; FW, footwall.

The debate centres on the recent activity of faults near the plants, as recognized in the criteria for formally designating an “Active Fault”. The currently preferred evaluation standard is the displacement by faulting of Quaternary strata of known sedimentary age5, but it is not feasible where suitable material was removed during plant construction and no nearby outcrops of the fault are available. Another potential proxy for recent fault activity was proposed after the discovery of ultrafine particles (partly amorphous) on the Chelungpu fault in the slip zone of the 1999 Chi-Chi earthquake in Taiwan6. As such particles, produced by comminution during the earthquake6, are easily dissolved into the interstitial fluid7,8, their preservation in a slip zone would mean that ruptures on the fault are recent. Because more investigations in various types of fault rock are needed to confirm the proxy6, it was noteworthy when another example of ultrafine particles in an active fault was discovered in fault gouge on the Median Tectonic Line (MTL), Japan9. The Chelungpu fault and the MTL intersect sedimentary and metamorphic rocks, respectively. However, another crucial analytical task is to assess the durability of these chemically unstable ultrafine particles. Thus we chose to focus on the lifetime of ultrafine particles under the near-surface environmental conditions found within faults.

The ultrafine particles in the slip zone may be expected to dissolve in interstitial fluids during the interseismic period. To explore this process, we targeted the ultrafine particles in the Chelungpu fault and in the Arima-Takatsuki Tectonic Line (ATTL, Fig. 1b), which last slipped during the 1596 Keicho-Fushimi earthquake (M ~ 7.5)10. We first quantified the amount of amorphous ultrafine particles in these active slip zones, and then established the mineralogical and crystallographic nature of these particles. We next evaluated the duration of these particles, in both the ATTL and the Chelungpu fault, by using the chemical kinetics of the dissolution reaction to establish an indicator for recent fault activity.

Methodological procedure

We tested a new approach consisting of the following four steps: (1) Determine the amount of the amorphous component both in and around the fault and conduct microscopic observations of the ultrafine particles. (2) Determine whether the ultrafine particles are crystalline or amorphous and analyse their atomic composition. (3) Characterize the environment in which the ultrafine particles formed (near-surface or hydrothermal) on the basis of their nanometric structure and in-situ environmental conditions such as temperature and pH. (4) Perform chemical kinetic modelling of the dissolution process to estimate the lifetime of amorphous ultrafine particles, thus constraining the most recent time that the fault slipped.

Geological background

The ATTL has a predominantly right-lateral strike-slip sense of displacement10. A fault outcrop near Arima exposes a fluvial terrace with a 14C age of 7870 ± 14 years before present (BP) and Cretaceous granite (Fig. 1c), both of which display a vertical offset of 2.1 m along the ATTL10. Displacement at this locality during the 1596 earthquake is supported by the similar displacement of fluvial deposits with a 14C age of 580 ± 40 years BP at a nearby outcrop10. The fault outcrop displays a fault zone consisting of a fault breccia 7 cm thick and a layer of very fragile fault gouge 5 cm thick that includes a sharply defined surface, identified as the primary slip zone (PSZ) (Fig. 1d). Fluid seeping from the fault zone probably originated from the local meteoric water. We collected 12 samples from the PSZ and the surrounding material for the following analyses and observations.

A borehole core sample of the Chelungpu fault was recovered from the Chinshui shale, of early Pliocene age, by the Taiwan Chelungpu-fault Drilling Project (Supplementary Figure 1). Three major zones of faulting within the Chinshui shale have been interpreted as segments of the Chelungpu fault11,12, and the shallowest of these is most likely the one that slipped during the Chi-Chi earthquake, because recent heating and a major anomaly in stress orientation were observed in that zone13,14,15. The PSZ of the Chi-Chi earthquake was identified as a millimetre-thick shear zone within black gouge, because newly formed magnetite in the gouge carried a stable palaeomagnetic component consistent with the 1999 reference geomagnetic field16. Fluid flow along the fault zone was detected by borehole hydraulic measurements17,18, and isotopic evidence suggested that the fluid originates from surface meteoric water and a deep crustal source19. For this study we examined six samples from the PSZ and the surrounding materials that were used in a previous analysis6.

Quantification of amorphous ultrafine particles

Mineral phases and their modal amounts were determined by X-ray diffraction (XRD) using the RockJock program20 to fit the diffraction data to the patterns of standard minerals (Fig. 2a,b; see Methods). The fault gouge and PSZ samples from the ATTL were identified as predominantly clay minerals and unfitted components, along with relatively small amounts of non-clay minerals such as quartz. The unfitted components were attributed to amorphous ultrafine materials rather than to minerals that were not identified in RockJock. In previous analyses of the PSZ and surrounding host materials of the Chelungpu fault (Fig. 3a,b; also see Fig. 2 in ref. 6), the unfitted components were also attributed to amorphous ultrafine materials6. The standard deviation of total mineral amounts (and unfitted components) from these RockJock results was reported as ±5.9 wt.%6.

Figure 2
figure 2

Data and images of amorphous ultrafine particles preserved in the PSZ of the ATTL.

(a) X-ray diffraction patterns of all samples (see Method). (b) Weight fractions of clay minerals, non-clay minerals and component not fitted by RockJock program. (c) Weight fraction of amorphous materials determined by standard curve for the integral XRD intensity versus the amount. (d–f) Scanning electron micrographs of material from the hanging wall, footwall and PSZ, respectively. (g) Bright-field transmission electron micrograph of ultrafine particles from the PSZ. (h) Electron diffraction pattern of the particle within the circle in g. (i) Energy dispersive X-ray signals from the particle from the PSZ.

Figure 3
figure 3

Data of amorphous ultrafine particles preserved in the PSZ of the Chelungpu fault.

(a) Photo of the borehole core sample. Small boxes in the photo indicate the sampling points, except for the sample at 1139.08 m depth. (b) Weight fractions of clay minerals, non-clay minerals and component not fitted by RockJock program (data from ref. 6). (c) Weight fraction of amorphous materials determined by standard curve for the integral XRD intensity versus the amount.

To test this attribution and attempt to quantify the amount of amorphous material in the samples, we obtained XRD spectra of prepared mixtures of amorphous silica (Kanto Chemical, Japan) and quartz powder (Wako Pure Chemical Industries, Japan) by using a zero-diffraction plate (made from a single silicon crystal). Sharp peaks in the XRD spectra of these mixtures were attributed to crystalline quartz and the internal standard (α-alumina), but wider, more subdued peaks were observed around 20–30° 2θ, indicating amorphous material21,22, in samples with smaller quartz contents (Supplementary Figure 2a). Most silicate minerals, such as feldspar, show similar broad bump in their amorphous state23. After subtracting the crystalline peaks and the background curve from the XRD patterns by using multivariate analysis (with expressions for position, height and width of crystalline peaks and the background curve), we found the integral intensities of the amorphous peaks to be inversely proportional to the weight fraction of crystalline components (Fig. 2b) and were thereby able to quantify the complementary fraction of amorphous components in the samples. Triplicate multivariate analyses of the XRD spectra from two separate samples (six analyses in total) showed that determinations of the amount of the amorphous component had a standard deviation of ±1.2 wt.%.

Results

Amounts of amorphous ultrafine particles in active faults

In the case of the ATTL, the PSZ samples contained the largest proportion of inferred amorphous material (18.2 wt.%), and the surrounding rock samples averaged 11.1 wt.% amorphous material (Fig. 2c). Likewise for the Chelungpu fault, the PSZ samples contained the largest proportion of inferred amorphous material (22.1 wt.%), and the surrounding rock samples averaged 13.6 wt.% amorphous material (Fig. 3c).

Under the electron microscope, the host rocks of the ATTL exhibited structures typical of sedimentary rock and consisted mainly of intact quartz and feldspar particles several to several tens of micrometres in diameter (Fig. 2d,e), whereas the PSZ of the ATTL contained single and aggregated, well-rounded ultrafine particles several tens to several hundreds of nanometres in size (Fig. 2f,g; see Methods). The diffuse scattering exhibited by selected portions of individual particles in one of these aggregates (Fig. 2g) suggests a noncrystalline structure (Fig. 2h), and the energy dispersive X-ray spectrum of the aggregate is consistent with that of clay minerals (Fig. 2i). Similar amorphous ultrafine particles were observed in the PSZ of the Chelungpu fault (Figs 4 and 5 in ref. 6).

The transformation of crystalline mineral phases into amorphous material and the formation of ultrafine particles may result from lattice distortion and granulation by coseismic slip, as has been demonstrated by friction and milling experiments on rock samples24,25. Thus we interpreted the ultrafine particles in the PSZ of both the ATTL and the Chelungpu fault as amorphous material. In contrast, the amorphous component in the surrounding intact rock and breccia around the ATTL, in which no ultrafine particles were observed (Fig. 2d,e), may be attributed to noncrystalline weathering products or damaged noncrystalline portions of large grains, because pulverization was reported within the host rocks near this outcrop26. The amorphous component in the intact rock around the Chelungpu fault may be attributed to volcanic glass, because the fault passes through sedimentary rocks including small amount of tuffaceous components. Although we could not definitively classify the amorphous component in the surrounding rocks, it is nevertheless significant that amorphous material and ultrafine particles were far more abundant in the PSZ of both the ATTL and the Chelungpu fault.

Chemical kinetic evaluation of dissolution process

The central hypothesis of this study is that the ultrafine particles, produced by earthquake slip, in the PSZ may be expected to dissolve in interstitial fluids (which commonly flow along fault zones27,28) during the interseismic period, given the high solubility of mineral particles smaller than 0.1 μm7 and the high dissolution rate and solubility of amorphous material7,8. In the case of the Chelungpu fault, three major zones of faulting were observed in borehole samples at depths of 1136, 1194 and 1243 m11,12, but only in the shallowest of these, corresponding to the slip zone of the 1999 Chi-Chi earthquake13,14,15, were ultrafine particles present6. However, the time scale of these processes has not previously been quantified.

The dissolution rate, k (mol m−2 s−1), of spherical mineral particles into fluid29 is

where M and M0 are respectively the mass and initial mass of the component, Vm is the molar volume of the component (m3 mol−1), t is time and r0 is the initial radius of the particles (m) (see Methods). The reaction rate k has a temperature dependency expressed by the Arrhenius equation. The kinetic data for the dissolution of representative minerals (quartz, muscovite, kaolinite and montmorillonite)30,31,32 and amorphous silica8 were experimentally determined by taking into consideration the reverse reaction (precipitation) and equilibrium state (see Methods). From these data, we evaluated the dissolution of ultrafine particles in the fault as a function of time.

We tested the time scale of the dissolution of ultrafine particles for both the ATTL and the Chelungpu fault. For the ATTL, we adopted a temperature condition of 13.8 °C (the annual mean temperature during 1979–2014 at the Sanda observatory33 near the outcrop) together with climatic fluctuations of ±5 °C in mean annual temperature since 47 000 years BP as reconstructed from the pollen profile from Lake Mikata34. Although the value of k also depends on thermodynamic activities of hydrogen and hydroxyl ions (that is, pH)29,30, the pH of interstitial fluid in the ATTL fault gouge was 6.0 and we therefore assumed neutral conditions (see Methods). For the Chelungpu fault, we adopted the borehole temperature of 46.5 °C determined by direct measurement13. Although the pH of interstitial fluid along the slip zone was not measured, it was reported that pyrite was absent in the slip zone35. Because pyrite dissolution would lower the pH of the immediately surrounding sediment36, we used pH values of 3.0 and 6.0.

By solving equation (1) and the Arrhenius equation, we determined mass fractions of each component as a function of time. The results showed that nanometric particles (100 nm radius) of kaolinite, amorphous silica and muscovite would disappear within approximately 1000 years whereas quartz and montmorillonite particles would survive longer (Fig. 4a) and that micrometric particles (10 μm radius) of all components would hardly dissolve at all within 1000 years (Fig. 4b). Given that 419 years elapsed between the 1596 Keicho-Fushimi earthquake and our sampling, some preservation of ultrafine amorphous silica particles in the ATTL fault gouge is likely, consistent with our analytical results (Fig. 2). However, the amorphous aggregates in the ATTL fault gouge were not pure silica, but also contained other atoms typical of silicate minerals (Fig. 2i). The dissolution rate at 25 °C for various silicate glasses37 ranges from 1.91 × 10−9 to 8.13 × 10−10 mol m−2 s−1, which is much faster than the dissolution rate of 5.59 × 10−13 mol m−2 s−1 for amorphous silica8. Silica gel, an amorphous phase found in some fault gouges38, has a similar dissolution rate of 9.88 × 10−13 mol m−2 s−1 at 18 °C39, and it also transforms into microcrystalline quartz by dehydration and recrystallization40. In the case of the Chelungpu fault, most of the nanometric particles could survive 5 years whether the pH is 3.0 or 6.0 (Fig. 4c), but micrometric particles of all components would hardly dissolve at all over 5 years (Fig. 4d). Therefore, the presence of relatively abundant ultrafine amorphous silica or silicate particles in fault material strongly suggests that the fault slipped some time in the last 1000 years.

Figure 4
figure 4

Changes in mass fraction and size of spherical particles with time.

(a,b) Change in mass fractions of single particles of 100 nm and 10 μm initial radii, respectively, for the ATTL. (c,d) Change in mass fractions of single particles of 100 nm and 10 μm initial radii, respectively, for the Chelungpu fault. (e) Change with time in the particle size of amorphous silica at different temperatures. Initial radius of the particle is 100 nm. (f) Change with time in the particle size of amorphous silica at pH 6 and 13.

Discussion

Our proposed proxy requires careful consideration of certain factors. Nanometric particles are generated by breaking of grains and subsequent granulation (that is, comminution) during slip under a wide range of conditions (e.g., slip rates of 10−9.5 to 1.3 m s−1 and confining stresses of 1 to 75 MPa24,25,41, corresponding to depths of approximately 0.1 to 4.8 km under hydrostatic assumption with 2.6 g cm−3 mineral density and 1.0 g cm−3 pore-fluid density). Besides, generation of nanometric particles during earthquake could trigger low friction on the fault (so-called, powder lubrication)41,42. Low friction induces a large slip and a short slip weakening distance which increases both the amount of seismic energy released and the extent of the rupture area43.

However, nanometric particles are produced not only by slip during earthquakes but also by precipitation from cooling supersaturated fluid under geothermal conditions44 and by chemical reactions such as thermal decomposition of carbonate minerals (under frictional heating to approximately 600–800 °C)45. Surface coatings of calcite nanoparticles on fault slickensides have been also reported46, and these could be preserved for long periods because sintering of calcite nanoparticles occurs at room temperature by solution transfer mechanisms via adsorbed water47. However, such particles could be distinguished by analyzing the fluid chemistry, temperature and mineralogy of the fault material. If faulting at deeper levels caused sintering (or recrystallization) of quartz at high temperatures48 and the fault surface was later exposed by erosion, the sintered ultrafine quartz particles might be found in outcrops, although such particles may coalesce49 and recrystallise, making them distinguishable from the dispersed amorphous ultrafine particles formed near the ground surface. In addition, whereas permeable fault gouge may react during fluid flow along the fault, the dissolution reaction could be assessed by chemical equilibrium and kinetic constraints that depend on temperature, pH and solved ions (both species and strengths8,37). For amorphous silica, for example, higher temperature enhances its dissolution (Fig. 4e), and its solubility and dissolution rate are greater in alkaline conditions than in neutral or acid settings50,51 (Fig. 4f). Therefore, detailed mineralogical and chemical analyses of the fault material should also include its nanometric structure and in-situ environmental conditions.

The use of our proposed new proxy for assessing the recent activity of a fault may help not only to estimate the elapsed time since the most recent earthquake on a major fault, but also to characterize the activity of minor faults that may threaten important structures such as nuclear power plants, chemical plants and petroleum facilities.

Methods

XRD analysis

XRD spectra of the PSZ samples from the ATTL and the Chelungpu fault and their surrounding rocks were obtained with a Spectris PANalytical X’Pert PRO MPD spectrometer with monochromatized Cu Kα radiation operated at 45 kV and 40 mA, using a step width of 0.004° (∆2θ), 0.25° divergence and anti-scattering slits, and a high-speed semiconductor array detector. In addition, the samples were blended with α-alumina (25 wt.%) as an internal standard and mounted on a plate by the side-load method to minimize preferred alignment of the phyllosilicates.

SEM and TEM observations and X-ray spectrometry

ATTL fault gouge and the surrounding hanging wall and footwall rocks were examined by a SEM (JSM-7600F, JEOL, Japan), operated at an acceleration voltage of 15 kV, and a TEM (JSM-2100, JEOL, Japan), operated at 200 kV. Elemental analyses in the SEM used energy dispersive X-ray spectrometry (Noran System 7 X-ray Microanalysis System, Thermo Fisher Scientific, USA).

PH analysis of fault-zone fluid

Three samples of wet fault gouge along the ATTL were obtained from the outcrop (Fig. 1c). Subsamples weighing 10 g were first dried under 50 °C in an oven and then soaked in 25 mL or 50 mL of pure water. Each sample was stirred and allowed to settle for 0.5, 2 or 24 hours, and then the pH of the suspension was measured by an electrometric method52. The measured pH data are listed in Supplementary Table 1.

Chemical kinetics of dissolution reactions

On the basis of the Lasaga equation expansion29, the amount n (mol) of a dissolved spherical particle is expressed as

where k is the dissolution rate (mol m−2 s−1), S is the surface area (m2) of the particle and r is its radius (m). The volume decrease with time is given by

where V is the volume of the particle and Vm is the molar volume of the component (m3 mol−1, Supplementary Table 2). In terms of the radius of the sphere, the volume is also expressed as

From equations (3) and (4), the decrease in size of the particle is

and thus,

where r0 is the initial radius of the particle. Thus, the volume V can be expressed as

Considering the density d of the particle (g m−3) to be constant, the mass M of the particle is thus

The mass fraction M/M0 of the particle with time is finally obtained as

where M0 is the initial mass of the particle, and this equation is also shown as equation (1) in the Results.

The Arrhenius equation is

where A is the pre-exponential factor (mol m−2 s−1), Ea is the activation energy (kJ mol−1) and R is the gas constant (8.314 J K−1 mol−1). The kinetic data for dissolution of minerals (quartz, muscovite, kaolinite and montmorillonite)30,31,32, representative of the fault gouge, and amorphous silica8, experimentally determined by taking into consideration the reverse reaction (precipitation) and equilibrium state, are summarised in Supplementary Table 2.

Additional Information

How to cite this article: Hirono, T. et al. Preservation of amorphous ultrafine material: A proposed proxy for slip during recent earthquakes on active faults. Sci. Rep. 6, 36536; doi: 10.1038/srep36536 (2016).

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.