In situ observation of faulting in olivine at high pressures and high temperatures using high-flux synchrotron X-rays

The mechanisms of intraslab earthquakes at depths of > 40 km are fundamentally different from those of shallow earthquakes because the frictional strength of silicate rocks is proportional to the confining pressure. To understand the process triggering intraslab earthquakes, many experimental studies on faulting of slab-forming rocks have been conducted at upper mantle pressures. Previous studies have revealed that shear localization induced by dehydration of hydrous minerals (e.g., Okazaki & Hirth, 2016) or adiabatic shear heating (e.g., Kelemen & Hirth, 2007) is essential for the occurrence of faulting at high pressures. Although acoustic emission (AE) monitoring technique for D-DIA apparatuses enabled us to discuss the process of microcracking at high pressures, mechanical behavior at the onset of faulting is still unclear due to low time-resolution stress/strain measurements using synchrotron X-rays. The cause of bottleneck in stress/strain measurements is a long exposure time required for the acquisition of a two-dimensional X-ray diffraction pattern of minerals. Considering that the timescale of stress drop associating faulting is on the order of 0.01 sec (e.g., Okazaki & Katayama, 2015), a significant improvement for time resolution of stress/strain measurements is required. To improve the time resolution of stress/strain measurements, we installed a series of new devices at BL15XU, SPring-8.

We conducted in situ triaxial deformation experiments on olivine aggregates at pressures of 1-3 GPa and temperatures of 700-1250 K under nominally dry conditions using a D-DIA apparatus, installed at BL15XU, SPring-8. Two-dimensional radial X-ray diffraction patterns and radiographic images were alternately acquired by adjusting sizes of the incident slit and operating a flatpanel detector and a CCD camera using a high-flux pink beam (energy 100 keV) from an undulator source (0.2 s of exposure time for both ones). Pressure and differential stress were determined from the d-spacing of olivine. Strains of deforming samples were evaluated from the distance between platinum strain markers. AEs were recorded continuously on six sensors glued on the rear side of the 2^nd-stage anvils, and three-dimensional AE source location were determined.

Stress increased with strain at the beginning of sample deformation, and it reached the yielding point at strains of ~0.1 or less. AEs from the deforming sample were detected when stress exceeded ~1 GPa and the amplitude of AE is positively correlated with the magnitude of stress. At strains higher than 0.1 (i.e., beyond the yielding point), both softening (i.e., decrease in stress and/or increase in strain rate) and a decrease in AE rate were observed prior to the occurrence of faulting. Faulting was observed at 880-1150 K. Most of unstable slips proceeded within 1 s and associated a sudden stress drop (~0.5 GPa) and temporal radiation of large AEs. In contrast, neither stress drops nor AEs were associated with a few “aseismic” unstable slips. Differential stress continuously increased when stable slip proceeded and the stable slip was terminated by the occurrence of another unstable slip. Our observations suggest that unstable slips can be divided into two types (i.e., seismic and aseismic ones) under the P-T conditions of shallow subducting slabs.