The evolution of slip surface roughness during earthquake propagation in carbonate faults

Monday, 15 December 2014: 3:25 PM
Bojing Zhu, Nicola De Paola, Edward W Llewellin and Robert Holdsworth, University of Durham, Durham, DH1, United Kingdom
Slip surface roughness is understood to control the dynamics of earthquake propagation. Quantifying the micro- and nano-scale roughness of slip surfaces can give insight into the grain-scale processes controlling the strength of faults during earthquake propagation.

Friction experiments were performed on fine-grained calcite gouges, at speed 1 ms-1, normal stress 18 MPa, displacements 0.009–1.46 m, and room temperature and humidity. Results show a two stage-evolution (S1-2) of the fault strength, with an initial increase up to peak value 0.82 (S1), followed by a sudden decrease to a low, steady-state value 0.18 (S2). Samples retrieved at the end of S1 show the development of a cohesive slip zone (SZ), made of micron-scale, angular clasts formed by brittle fracturing and cataclasis. The SZ of samples deformed up to S2, is composed of nanograin aggregates which exhibit polygonal grain boundaries indicating high temperature grain boundary sliding creep deformation. In both cases, the SZ is bounded by a sharply defined slip surface.

The 3-D geometry of seven experimental slip surfaces (40μm×40μm) has been reconstructed by digital processing of sets of 1800 images of SZ cross sections acquired at 20 nm intervals perpendicular to the slip direction, using a slicing (Focussed Ion Beam) and viewing (Field Emission Scanning Electron Microscope) technique. Spectrum power density analyses show that nano- and micron-scale slip surface roughness is anisotropic for both S1 and S2 slip surfaces. At the nano- and micron-scale, root mean square values decrease with length for S1 slip surfaces, but only slightly for S2 surfaces, and are anisotropic in the slip-normal and slip-parallel directions. The anisotropy is reduced at the nano-scale, although S2 slip surfaces are still smoother parallel to slip than normal to slip. Hurst exponents vary through scales, and are anisotropic in the directions parallel and normal to slip.

Variable Hurst exponents indicate that slip surface roughness is scale-dependent with anisotropic, not self-affine behaviour at the micro/nano-scale, in contrast to the self-affine behaviour inferred at the mm to km scales. Dynamic weakening and creep deformation, observed during S2, coincide with an evolution towards less anisotropic and scale-dependent slip surface roughness at the nanoscale.