Report

Superplastic nanofibrous slip zones control seismogenic fault friction

See allHide authors and affiliations

Science  12 Dec 2014:
Vol. 346, Issue 6215, pp. 1342-1344
DOI: 10.1126/science.1259003

Abstract

Understanding the internal mechanisms controlling fault friction is crucial for understanding seismogenic slip on active faults. Displacement in such fault zones is frequently localized on highly reflective (mirrorlike) slip surfaces, coated with thin films of nanogranular fault rock. We show that mirror-slip surfaces developed in experimentally simulated calcite faults consist of aligned nanogranular chains or fibers that are ductile at room conditions. These microstructures and associated frictional data suggest a fault-slip mechanism resembling classical Ashby-Verrall superplasticity, capable of producing unstable fault slip. Diffusive mass transfer in nanocrystalline calcite gouge is shown to be fast enough for this mechanism to control seismogenesis in limestone terrains. With nanogranular fault surfaces becoming increasingly recognized in crustal faults, the proposed mechanism may be generally relevant to crustal seismogenesis.

Nanofibers involved in fault rupture

Changing fault properties during rupture dictates the size and extent of an earthquake. Faulting leads to well-known microstructures that may play a role in how natural faults slip during rupture. Verberne et al. investigated tiny, nanogranular fibers found in microstructures generated on simulated carbonate faults. A microphysical model was able to account for how the small and aligned fiber produced runaway fault slip, similar to that seen in natural faults. These small structures play a role in carbonate faulting and similar microstructures could control fault rupture in other types of rocks.

Science, this issue p. 1342

The cores of seismically active fault zones are frequently characterized by a narrow (millimeters to centimeters thick) principal slip zone (PSZ) composed of nanogranular fault rock (1, 2). Recent field studies demonstrate that fault-parallel exposures of such PSZs often form highly reflective (specular) surfaces, suggesting that these “mirrorlike” fault surfaces may be indicators of past seismic slip (37). Independently of fault plane morphology, for earthquake ruptures to nucleate, the internal slip zone wear product, or “fault gouge,” must show a decrease in frictional strength with increasing fault slip rate. In other words, it must show runaway velocity- (or v-) weakening behavior, as opposed to self-stabilizing, v-strengthening slip (8). Elucidating the mechanisms controlling v-weakening and v-strengthening slip, especially those involving the formation and slip behavior of mirrorlike nanogranular fault surfaces, is therefore crucial for understanding the processes leading to earthquake nucleation.

Low-velocity (v ≈ 10−6 m/s) friction experiments, designed to investigate the onset of seismogenesis in the upper crust, recently demonstrated a transition from v-strengthening to v-weakening slip in simulated calcite fault gouge at temperatures above 80° to 100°C (911). Microstructural studies of the sheared samples revealed the development of continuous 10- to 100-μm-wide nanocrystalline PSZs with a crystallographic preferred orientation (CPO) (10). When split along the shear plane, these zones become partially exposed to display multiple patches of striated, highly reflective slip surfaces (11). Similar though more continuous nanogranular mirror-slip surfaces have been reported to form in high-velocity friction experiments achieving coseismic slip rates (v = 0.1 to 3 m/s) (4, 5, 12). At these slip rates, nanograins are believed to facilitate powder lubrication, causing strong dynamic fault weakening (13, 14), but during slow, pre-seismic and interseismic slip their role remains enigmatic. Nanophase materials are well known to exhibit much faster diffusion and mass transport than their coarser-grained counterparts (15, 16). Moreover, a recently proposed microphysical model for (thermally activated) v-weakening shear of fault gouge undergoing granular flow demonstrates that the rate of interparticle diffusive mass transfer can play a central role in controlling gouge strength and its velocity dependence (supplementary text and fig. S1) (17). Therefore, the nanophysical deformation mechanisms that operate in shiny PSZs may hold the key to understanding the processes controlling the initiation of seismic slip.

We investigated the micro- and nanostructural properties of PSZs developed in simulated calcite fault gouge subjected to direct shear at low slip rates (fig. S2 and table S1). The experiments were conducted on nominally dry samples at v = 1 μm/s at a normal stress of 50 MPa, at temperatures of 18° and 140°C (18). These temperatures respectively correspond to nonseismogenic, v-strengthening conditions and seismogenic, v-weakening conditions (911). Upon disassembly after each experiment, PSZs formed near the sheared sample boundary split open to reveal multiple shiny patches within the ultra–fine-grained PSZ (Fig. 1A) (11). Surface roughness analysis using atomic force microscopy (AFM) showed that the shiny patches have a striated nanoscale surface topography similar to that of the more continuous surfaces formed at coseismic slip rates (Fig. 1B) (12). This suggests that co- and subseismic mirror-slip surfaces share a common origin, independently of their extent or of shearing velocity.

Fig. 1 Mirror-slip surfaces formed at subseismic sliding velocities (1 μm/s).

SEM micrographs are secondary electron images of the shear plane. Arrows indicate the shear sense of material above the shear plane. (A) Shiny patches formed under v-strengthening conditions (18°C). (B) AFM deflection image showing nanoscale surface roughness. (C) and (D) Nanofibers formed at 18°C. The inset in (D) shows a protruding fiber with a spherule at the tip. (E) Part of a shiny patch formed under v-weakening conditions (140°C). (F) Chaining of nanospherules at the edge of a fibrous surface. Note the gradation from the zone dominated by nanospherules into the fibrous zone.

We further investigated the surface morphology of the shiny patches formed in our experiments, using high-resolution scanning electron microscopy (SEM). We found them to consist of ultra–fine-grained films composed of ~100–nm-wide fibers aligned subparallel to the shear direction (Fig. 1, C to F, and fig. S3), embedded within granular zones of frequently clumped, ~100-nm spherules, constituting the bulk PSZ (fig. S4). Individual spherules are sometimes visible within the fibers, especially at their tips (Fig. 1D and fig. S4), whereas linear arrays of nanospherules grading into fibers are also evident (Fig. 1F and fig. S4). On this basis, we suggest that the fibers form by progressive nanospherule chaining. In shiny patches formed in v-strengthening samples (18°C), microcracks cutting the nanocoatings show marked extension and plastic bending of the fibers (Fig. 1, C and D). In view of the high confining stress used in our tests (50 MPa), such crack opening could only have occurred upon sample depressurization or disassembly at room conditions (fig. S3). The nanofibers affected often show extreme local extension without the formation of a neck before fracture (Fig. 1, C and D), which are classical characteristics of superplastic deformation (16). Shiny patches from v-weakening samples (140°C) also consist of ultra–fine-grained fibrous coatings (Fig. 1E) and show essentially the same SEM microstructures as the v-strengthening samples (18°C).

We used a focused ion beam (FIB) scanning electron microscope to mill cross sections normal to the shear plane and parallel to the shear direction, to investigate the internal structure of the PSZ and shiny patches (figs. S5 and S6). This showed that the PSZ is a 10- to 100-μm-wide, relatively porous (20 to 30%), nanogranular volume (10, 11), whereas the shiny patches consist of dense, 0.1- to 1-μm-thick, planar fibrous nanocoatings. Some samples show small amounts of an amorphous carbon-rich phase, locally occupying pores ~0.1 to 5 μm in diameter (fig. S5). These deposits show no correlation with mechanical behavior or microstructure, and no temperature changes were measured during shear. We therefore infer them to be derived from the polymer sleeve used to jacket our sample assembly rather than from mechanochemical decomposition (19). To resolve the internal structure of the fibrous coatings, we isolated a single nanofiber and milled a thin foil parallel to the fiber length for investigation using transmission electron microscopy (TEM) (Fig. 2A and fig. S7). Selected-area electron diffraction, applied to the nanofiber using a 300-nm aperture, showed ring-shaped diffraction patterns with arcs of greater intensity (Fig. 2B) that attest to a preferred orientation of the (104) or r-plane of calcite, consistent with the optical CPO previously reported for the nanocrystalline PSZs formed in experiments on simulated calcite gouge under the same conditions (supplementary text) (10). Bright-field and high-resolution TEM further revealed the presence of spherule-like domains, 100 to 200 nm in diameter, within the nanofiber. These are composed of 5- to 20-nm crystallites, with mutual misorientations of typically just 5° to 20° (Fig. 2, A and C).

Fig. 2 Individual nanofiber imaged using TEM.

(A) Bright-field electron micrograph of a single nanofiber recovered from a mirror-slip surface using FIB-SEM (fig. S7). (B) Selected-area electron diffraction pattern showing arcs of greater intensity in the diffraction ring corresponding with the (104) or r-plane of calcite (supplementary text). (C) High-resolution micrograph taken from a nanofiber. The dashed white circle highlights a single nanocrystallite, and the thick white lines indicate the orientation of the (104) plane (supplementary text).

All samples investigated using SEM showed evidence for neck growth (sintering) between nanospherules and within nanofibers (Fig. 1D and figs. S4 to S6), implying that diffusive mass transport occurred throughout the PSZ. At the same time, from the porous spherular grain structure observed in the body of the nanogranular PSZ (Fig. 1F and figs. S4 to S6), dilatant granular flow involving spherule rolling and neighbor swapping (Fig. 3A) must also have occurred, whereas the fibrous films suggest cooperative sliding of sintered linear chains of nanospherules (Figs. 1, C to F, and 3, B and C). The Niemeijer-Spiers mechanism of granular flow (17) involves competition between dilatation caused by grain-neighbor swapping and compaction by diffusive mass transfer. The original model considers diffusive mass transfer to occur through thin grain boundary fluid films, but it is also valid for solid-state diffusion. For this mechanism to produce v-weakening, or a transition thereto, the diffusion-controlled compaction strain rate Embedded Image must be of similar order to the granular dilatation rate Embedded Image. To test whether this is the case for our samples, we estimated the dilatation rate from the shear rate Embedded Image imposed on the PSZ by assuming a dilatation angle of ~35° for granular flow [i.e., using Embedded Image (supplementary text) (17)] and we compared Embedded Image with the compaction rate Embedded Image predicted by equations describing compaction of wet calcite by diffusive mass transport (20, 21). We chose wet calcite because thermogravimetric analyses on nominally dry nanocrystalline calcite powder (particle size ~80 nm) demonstrates a water content of 2 to 3 weight % (fig. S8). This implies that an adsorbed water film a few nanometers thick must coat each ~100-nm particle in the PSZ, which is similar to the amount needed for water-assisted diffusive mass transfer to operate (22). Taking the width of the slipping zone to be 1 to 50 μm, Embedded Image in the shear bands formed in our experiments is on the order of Embedded Image, whereas, for a particle size of 100 nm, compaction creep is predicted to occur at a rate Embedded Image (18). This confirms that the Niemeijer-Spiers model is capable of explaining the v-weakening behavior seen in our experiments at temperatures above 80° to 100°C (fig. S2) (911), suggesting that this may be caused by enhanced compaction rates due to faster diffusive transport at these temperatures. Application of the model assuming solid-state grain boundary diffusion (23) demonstrates that this is 10 to 20 orders of magnitude too slow to produce v-weakening, except perhaps at temperatures of 500° to 700°C reached in experiments at coseismic slip velocities (4, 5, 14, 19).

Fig. 3 Two-dimensional representation of envisaged mechanism of nanofiber formation during nanogranular flow with diffusive mass transfer.

(A) Nanospherule rolling and grain-neighbor swapping with (B) local sintering attachment (neck growth) at preferred high-energy interfaces, leading to (C) formation of cooperatively sliding nanofibers (linear arrays or chains of sintered nanospherules). Note the cavitation (porosity development) in (A) to (C), compared with (D), which represents a zero-porosity grain boundary sliding mechanism that is fully diffusion-accommodated (30).

The internal polycrystalline substructure of the nanospherules and nanofibers that we observed bears a striking similarity to microstructures found in shocked ductile metals (24). As in metals, the well-known ductility of calcite (25) may allow the ~5- to 20-nm substructure to form by progressive development of nano–cell walls from dense dislocation networks generated by crystal plasticity. Plastic deformation, fracturing, and abrasion presumably generated the observed nanospherules from the starting “gouge.” To explain the chaining of nanospherules, producing the observed fiber structure and CPO, we note that oriented attachment at coherent nanoparticle interfaces is widely reported as a mechanism by which nanocrystallites can rapidly coalesce to form single crystals (26, 27), also in calcite (28). On this basis, we suggest that the strong anisotropy in the surface energy of calcite produced similar preferred sintering (neck growth) at high-energy crystallographic interfaces between neighboring spherules (Fig. 3, B and C), leading to dynamic chaining and alignment of the lowest-energy (104) plane (29) parallel to the shear plane, and thus to the observed fibrous structure and CPO.

In principle, the mechanism of frictional slip that we propose (Fig. 3, A to C, and fig. S1) is similar to the Ashby-Verrall model for superplasticity by diffusion-accommodated grain boundary sliding (GBS) (Fig. 3D) (30), but allows for frictional GBS and for intergranular cavitation (porosity generation by dilatation) when diffusive mass transport is too slow to accommodate GBS. Our findings imply that nanocrystalline PSZs developed in calcite faults can produce v-weakening, and hence seismogenic fault friction, by a mechanism of cooperative nanogranular or nanofiber flow plus diffusive mass transfer (Fig. 3, A to C), even in the upper crust where temperatures are generally considered too low to support diffusion or superplasticity at active fault slip rates. The reason that these processes are observed in our experiments is because diffusive mass transfer is dramatically accelerated by the nanogranular nature of the slip-zone rock that forms, and by water-enhanced grain boundary diffusion. A similar mechanism can also be envisaged to operate at coseismic slip rates, where the high temperatures generated will promote solid-state diffusion. Given the abundant recent observations of nanogranular fault surfaces in tectonically active terrains (17), and the anomalously high rates of diffusion found in nanomaterials (15, 16), the proposed mechanism may be relevant not only to faults cutting calcite-rich rocks such as limestones, but to crustal seismogenesis in general.

Supplementary Materials

www.sciencemag.org/content/346/6215/1342/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S8

Tables S1 to S2

References (3140)

References and Notes

  1. Acknowledgments: We thank A. Niemeijer, J. Chen, V. Toy, and H. de Bresser. H. King is thanked for the AFM measurements and P. van Krieken for the thermogravimetric analysis. B.A.V. was supported by grant 2011-75, awarded by the Netherlands Research Centre for Integrated Solid Earth Sciences; O.P. by Veni grant 863.13.006, awarded by the Netherlands Organisation for Scientific Research (NWO); and D.A.M.D.W. by ISES grant 2011-74. NWO funded the FIB-SEM. All data are available in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article