Of particular interest (and controversy) in geophysics and seismology are the mechanisms controlling brittle-like failure at depths of several hundred km or more where deep earthquakes occur. At these depths frictional sliding, the fundamental failure mechanism operating under the lower confinements present at shallower depths, is nominally suppressed. We propose new experiments and modeling aimed at elucidating the characteristics and the fundamental physical mechanism(s) that underlie localized brittle compressive failure under high confinement.
Among the mechanisms that have been proposed to explain brittle-like failure under high confinement, dehydration embrittlement and transformational faulting are the two most often cited. They act to reduce confinement locally, in effect allowing low-confinement frictional faulting to continue into the high-confinement regime. However, changes in earthquake seismic character at greater depths are possibly consistent with a change in the physical mechanism of rupture for the deepest earthquakes. An alternative mechanism, and the one we propose to investigate, is plastic faulting. This mechanism is fundamentally different from the other two: it is a non-frictional shear instability aided by adiabatic heating.
Spirited by past success in using ice as a model material for rock--Kirby's [1987] discovery of transformational faulting, and our own discovery [Renshaw and Schulson, 2001] of what appears to be a universal mechanism of low-confinement brittle failure--and encouraged by its mineralogical simplicity and by its marked similarity to the behavior of rocks and minerals, we propose a series of systematic compression experiments on granular ice to determine the effects, if any, of confinement, grain size and strain rate on both the terminal failure stress and the failure mode, with attention to microstructural detail. Preliminary experiments demonstrate a transition in brittle-like failure mode with increasing confinement and that the high confinement faults appear not to be friction controlled. Nor do they appear to be related to either mode II cracking, dehydration embrittlement, or phase transformations. Our working hypothesis is that plastic faulting is at play. In addition, we propose to develop quantitative models of high confinement compressive failure, using as inputs direct observations of the physical processes.
