Compositional, mechanical, and hydrologic controls on fault slip behavior

Open Access
Ikari, Matt
Graduate Program:
Doctor of Philosophy
Document Type:
Date of Defense:
July 12, 2010
Committee Members:
  • Demian Saffer, Dissertation Advisor
  • Chris Marone, Committee Chair
  • Demian Saffer, Committee Chair
  • Eliza Richardson, Committee Member
  • Derek Elsworth, Committee Member
  • Andre Niemeijer, Committee Member
  • faults
  • friction
  • earthquakes
  • clay minerals
  • permeability
  • subduction zones
While many aspects of active fault processes have been well characterized, the mechanisms controlling the slip behavior of major faults remain elusive. Fault slip behavior may range from aseismic creep, to intermediate behavior such as slow slip events, to large and destructive earthquakes. Laboratory experiments are an essential component of fault studies, because they allow detailed investigation of processes operating at otherwise inaccessible timescales and locations in the earth. In order to examine the roles of a variety of factors that are likely important in regulating the occurrence or lack of seismic slip, I evaluate the results of numerous laboratory studies of fault behavior, focusing on the effects of fault mineralogy, mechanical effects, and interactions between fluids and faulting processes. More specifically, these experiments are designed to investigate the underlying mechanisms controlling the transition from aseismic slip at shallow levels in the crust to seismic slip at depth, known as the updip limit of the seismogenic zone. Results of laboratory experiments indicate that mineralogy of fault gouge is a major control on fault behavior. The clay mineral montmorillonite (smectite) has been noted for its potential effect on seismogenesis in subduction zones (as well as all faults in general) due to its ability to take up water in its crystal structure. Dehydration of montmorillonite tends to increase its frictional strength as well as increase its propensity for seismic slip, as documented by a decrease in the frictional velocity dependence parameter a-b. However, the observed decrease in a-b is assisted by both increasing relative quartz percentage and increasing normal stress, implying that the onset of seismic behavior with increasing depth should not be attributed solely to smectite dehydration. Furthermore, clay-rich gouges in general, including those consisting of montmorillonite, illite, and chlorite, are both frictionally weak (&#956; < 0.35) and velocity-strengthening (frictionally stable, a-b > 0) at fluid-saturated conditions and effective normal stresses up to ~60 MPa. Sheared gouges may also exhibit low fault-perpendicular permeability (k < 1x10-18), making them candidates to host high pore pressure. This indicates that faults containing granular, clay-rich gouges are unlikely to show seismic behavior, due their velocity-strengthening nature and stabilizing hydro-mechanical effects resulting from low permeability. Natural, clay-rich fault gouge from the Nankai subduction zone is consistent with these assessments, also showing low friction, low permeability, and velocity-strengthening slip behavior. This behavior is consistent over a variety of faulting systems, including the megasplay fault zone, the frontal thrust system, and the décollement zone. The velocity-dependence of the Nankai samples reveals a frictional stability minima in the range of slip rates that correspond to rates during slow slip events, indicating that these events may be prevalent in the shallow faulting environment. Friction experiments conducted to high shear strains and using a wide variety of mineralogic compositions as gouge show that weak, phyllosilicate gouges such as most clays and micas tend to be uniformly velocity-strengthening even at high shear strains. By contrast, minerals such as quartz and feldspar tend to become velocity-weakening after a critical amount of shear strain. An intriguing observation is that fault stability may be linked with overall fault strength, in that weak (&#956; < 0.5) gouges are velocity-strengthening, while strong gouges (&#956; > 0.5), even those composed of phyllosilicates, may be both velocity-strengthening and velocity-weakening. In comparing the frictional behavior of granular gouge and lithified fault rock as an analogue for cataclastic fault rocks at seismogenic depths, the lithification of fault rock is found to have a significant strengthening effect, however in phyllosilicate-rich rocks pre-existing foliation provides a weakening mechanism that offsets the strengthening due to lithification. This weakening depends on the intensity of foliation such that strongly foliated rocks, such as books of mica sheets, are significantly weaker than granular mica gouges. Very thick fault zones can exhibit a reduction in measured apparent friction, the magnitude of which may be related to the orientation of through-going R1 shears and internal structural complexity. Consistent velocity-strengthening behavior is observed for both lithified and granular phyllosilicate-rich samples despite the observation of slip localization features in microstructural analysis, suggesting that as an isolated parameter advanced lithification state of fault rock is also inadequate for allowing seismic slip nucleation. Collectively, the results of the experiments in this study have several important implications for fault slip behavior. Granular, unconsolidated phyllosilicate-rich gouges, such as those that are common at shallow depths in both subduction zones and strike-slip faults, will tend to be aseismic, a condition that may be related to their overall weakness. The transition from aseismic to seismic slip at the updip limit of the seismogenic zone should be driven by changes in pressure and temperature, due to the overall ambient conditions as well as inducing changes in the character of the fault material itself. These may include compositional changes and mechanical effects of the lithification process, such as consolidation and cementation. However, when tested as isolated variables, the dehydration of smectite, conversion of smectite to illite, and lithification of fault gouge were found to be insufficient in allowing unstable slip behavior. It is possible that these processes may still play a role but must be combined with other conditions such as high shear strain, localized deformation, and an increased proportion of intrinsically strong minerals in order to drive seismogenic behavior.