T43D-01 INVITED
On the Relative Importance of Foliation Development and Syntectonic Metamorphism on Shear Zone Formation
A theoretical description of the formation of ductile shear zones remains an elusive goal of geodynamics modeling. Standard approaches such as shear heating and grain size evolution seem unable to produce spontaneously shear zones out of an essentially homogeneous, slowly deforming, rock. Progress in this field hinges on the inclusion of geological information to guide further modeling effort. Interconnected layers of weak minerals, especially phyllosilicates, typically mark ductile shear zones in the continental middle to lower crust. Shear zone material is often richer in phyllosilicates than the original host rock. These characteristics need to be included into the next generation of shear zone development models. The authors of this abstract have proposed in independent publications mathematical descriptions of the rate at which phyllosilicate enrichment and layer development proceed. In this contribution, we determine to what extend these processes might contribute to the formation of localized shear zones. To achieve this goal, we apply a variety of simple theoretical analysis to these phenomena. First, we calculate how much faster a shear zone rock is expected to shear compared to the host rock (localization potential). Then, we quantify the rate at which weakening is occurring through the calculation of the effective stress exponent of each of these processes. Finally, we integrate numerically the ODEs that describe these processes to follow the strength evolution of rock undergoing layer development, phyllosilicate enrichment, or both. These analyses are repeated for a range of ambient temperature and starting material to explore under which circumstance one or the other of these phenomena is expected to dominate.
T43D-02 INVITED
Mechanisms of Ductile Shear Localization From Observations of Naturally Deformed Peridotites
Deformation in the oceanic lithosphere is largely confined to plate boundaries, indicating the importance of strain localization. To examine the role of ductile processes during shear localization, we present results from microstructural analyses of naturally deformed peridotites. We examine the roles of dislocation accommodated grain boundary sliding (DisGBS) and melt in promoting strain localization. Peridotite mylonites have been collected at many oceanic transform faults. In mylonites from the Shaka Fracture Zone, our observations indicate that olivine deformation was accommodated by a combination of DisGBS in coarser (10-100 μm) grained regions and diffusion creep in finer (1-10 μm) grained regions. Within olivine-rich domains, the preferred orientation of misorientation axes for misorientation angles <10° is consistent with [010] tilt walls resulting from olivine slip on (001)[100]. By contrast, at higher misorientation angles, misorientation axes do not exhibit strong crystallographic control, suggesting the influence of grain boundary sliding. DisGBS also leads to the mixing of different phases, providing a mechanism to limit grain growth. While multiple phases are not necessary for the transition to DisGBS and diffusion creep, the presence of multiple phases appears to promote permanent strain localization as a result of grain size pinning. For example, we observe that deformed dunites often have a coarser grain size than adjacent harzburgites. We suggest that a correlation exists between the modal abundance of secondary phases (pyroxenes and spinel) and the degree of grain size reduction. Another factor leading to strain localization is the presence of melt, as observed in a series of shear zones from the Josephine Peridotite in southwest Oregon. In one shear zone, strain gradients indicate a > 100-fold variation in effective viscosity over a distance of 25 meters, as inferred from the strain localization. Grain size variation across the shear zone is minor and cannot explain the viscosity variation. Instead, we suggest that focused melt transport in the form of a syn-deformational dunite, as inferred from field observations (Kelemen and Dick, 1995), promoted the strain localization. Based on experimental constraints for the effect of melt fraction on effective viscosity, these observations require melt contents within the shear zone of 10-20% during deformation.
T43D-03
Influence of Pre-existing LPO on Texture Development in Ultramafic Shear Zones
The Red Hills ultramafic massif, the South Island, New Zealand, contains a suite of cm-scale shear zones composed of dunite, pyroxenite, and olivine websterite. The shear zones cross-cut banded dunite and harzburgite host rock. Offset along the shear zones was measured using displacement of distinct compositional layers in the host rock, and is interpreted to occur by dominantly simple shear. While these zones contain a different mineralogy that the host rocks, they contain microstructures that are similar to the host rock, including coarse grain sizes (>1 mm) and dominantly polygonal grains. The olivine lattice preferred orientation (LPO) in the host rocks is consistent with the (010)[100] slip system, known to be active at high-temperature, upper mantle conditions. Within the shear zones, the olivine LPO, when plotted relative to the shear zone foliation and lineation, suggests (010)[001] slip. The shear zone LPO is typically more poorly clustered than in host rock samples, and double maxima are observed in some samples, with the second maxima suggesting (010)[100] slip. Fourier transform infra-red spectroscopy analysis of the shear zone olivines indicates that they are not fluid-rich relative to the host rock. When olivine LPO from both host rocks and shear zone samples is plotted within the same (geographic) reference frame, it is apparent that the shear zone LPO retains evidence for the dominant LPO of the host rocks. We suggest that the LPO in the shear zone rocks developed as a result of reactivation of a pre-exiting fabric; the poorly clustered data and double maxima reflect the change during deformation along the shear zone. These rocks demonstrate the influence of pre-existing LPO on texture development during deformation and the need to consider the LPO of deformed rocks within different reference frames.
T43D-04
Inferring the kinetics of dynamic recrystallization from microstructural observations, with implications for the deformation of peridotite
The processes of dynamic recrystallization have important influences on the deformation of rocks. We have revisited sheared-lherzolite xenoliths using modern SEM-based analytical techniques. In this manner, we demonstrate that olivine and orthopyroxene recrystallized grains deformed by different deformation mechanisms. There are two additional contrasting features of the olivine and orthopyroxene deformation microstructures: First, the olivine recrystallized grain-size is at least a factor of five larger than the orthopyroxene recrystallized grain size. Second, the olivine recrystallization is complete, whereas the orthopyroxene recrystallization is only partially complete. To understand the origin of these differences, we have employed the Avrami theory of phase transitions (or recrystallization). In this theory, two observable parameters, i.e., grain-size (the Avrami length) and the characteristic time for recrystallization (the Avrami time) are expressed in terms of two microscopic parameters, i.e., the nucleation and the growth rate. We determined the nucleation and growth rates for olivine and orthopyroxene from the observed grain-size and the inferred characteristic time for recrystallization. The results indicate that the both the nucleation and growth rates in orthopyroxene are slower than those of olivine, although the difference in growth rate is much larger. We interpret the large contrast in growth rates to be a consequence of impurity atoms segregated to grain-boundaries of orthopyroxene, as demonstrated by the experimental study of Skemer and Karato (2007). The inferred sluggish kinetics of grain-boundary migration in orthopyroxene will help maintain the small grain-size of orthopyroxene, making it possible for orthopyroxene to deform by grain-size sensitive creep for a geologically significant length of time.
T43D-05
Localization in rate-dependent shearing deformation, with application to torsion testing
In discussions on the localization of plastic deformation in situations such as the formation of "ductile" shear zones, it is commonly conjectured that strain softening, as evidenced by a falling stress-strain curve at constant strain rate, is likely to lead to strain localization in simple shear. Yet observations in torsion tests at constant twist rate on calcite and olivine aggregates fail to show such an effect. This apparent discrepancy is resolved by going to a continuum mechanics analysis using the theory of Fressengeas and Molinari (J. Mech. Phys. Solids 1987, 35, 185-211) for material showing strain-rate dependence of the flow stress. Their treatment of simple shear shows that, in the absence of material changes, whether localization occurs or not can depend on the nature of the boundary conditions. Thus, when the boundary conditions are specified in terms of displacements, no localization is predicted in case of strain softening, in agreement with the torsion test observations. In contrast, if the boundary conditions are set in terms of forces, localization can be expected for a strain-softening material. This prediction needs experimental testing in torsion creep tests at constant torque.
T43D-06 INVITED
Strain Localization and Ductile Failure of Anorthite-Diopside Aggregates
We investigate the effect of varying phase distribution on strength and strain localization of rocks during ductile deformation. The effect of phase distribution was examined by high-strain torsion experiments on synthetic fine- grained anorthite-diopside aggregates. Starting materials were hot-isostatically pressed fine-grained glass- powders of anorthite (An) and diopside (Di). Two kinds of samples were fabricated: homogeneous mixtures of 50vol% An and Di and layered composites composed of 2-3 alternating layers of An and Di, each 1.5-3 mm thick. The samples contain up to a few percent residual glass (melt). Mean grain size is about 3μm and the average water content is about 0.1wt%. Cylindrical samples of 10 mm diameter and length between 3.5 and 9 mm were deformed in torsion using a Paterson-type deformation apparatus. Most tests were conducted at 400 MPa confining pressure, temperatures of 950-1100°C, and shear strain rates between 2×10-4 and 2×10-5 s-1 to shear strains of 1.2-6.7. Creep behavior was linear viscous at stresses below about 80 MPa. Continuous strain hardening was observed in most cases, except for the lowest temperature and highest strain rate, possibly owing to dilatancy and grain growth. All layered samples and mixtures deformed at low T failed abruptly at high shear strain, coupled with the formation of thin shear zones and cavity stringers. Presumably nucleation, growth, and coalescence of cavities combined with grain rotation and sliding and with redistribution of melt leads by ductile failure and small scale partitioning. Below the equi-viscous point of the two phases the strength of homogeneous mixtures is higher than that of layered aggregates where strain is partitioned into weaker An-layers. In contrast to the behavior of the pure end-members, the homogeneous mixtures show bulging and necking on the sample surface. The intensity of localization increases with decreasing strain rate and increasing temperature, i.e. decreasing stress. At high strain a moderate phase mixing of two-phase mixtures, which were slightly clustered before deformation, indicate grain boundary sliding. However, no phase mixing occurred at the boundaries of layered composites. The results show that localization evolves with strain and critically depends on the presence and distribution of a second phase.
T43D-07
Stress-Driven Melt Segregation and Shear Localization in Partially Molten Aggregates: Experiments in Torsion
Simple shear experiments of partially molten aggregates have demonstrated that shear induces organized patterns of melt distribution and strain localization [1,2]. New torsion experiments on partially molten aggregates of olivine + chromite + 4 vol.% mid-ocean ridge basalt provide additional insights into the interactions between deformation and melt segregation. Samples were deformed at constant strain rates ranging from 3x10-5 to 3x10-4 /s, corresponding to shear stresses of ~60 and ~100 MPa, respectively. When samples are sheared, melt segregates into distinct melt-rich bands oriented ~15° antithetic to the shear direction. The melt fraction in the bands ranges from 0.1 to 0.2. In samples deformed at higher stress, bands are narrower and more closely spaced. At a shear stress of ~60 MPa bands are ~21 μm wide and spaced ~104 μm apart. At a shear stress of ~100 MPa bands are ~15μm wide and spaced ~71 μm apart. Melt segregation occurs in both the dislocation creep and diffusion creep regimes. Our experiments demonstrate that melt-rich bands form at a strain of ~100% regardless of the stress at which the sample is deformed. Near this strain threshold, the behavior of the sample changes from strain hardening to strain softening. This observation indicates that, in addition to providing high-permeability pathways through which melt can travel, melt bands also become zones of localized deformation. Strain is partitioned and localized into melt-rich bands because of a viscosity reduction in regions of elevated melt fraction. This process of stress driven melt segregation has implications for melt transport in many geological settings, including beneath mid-ocean ridges, and for the formation of shear zones in partially molten rocks. [1] Zimmerman, M. et al., Geophys. Res. Lett., 26, 1999. [2]Holtzman, B. et al., Geochem. Geophys. Geosyst., 4, 2003.
T43D-08
The influence of rigid second phases on the mechanical evolution of calcite rocks, with implications for strain localization
Interactions between rheologically distinct phases may strongly influence both the microstructure and mechanical properties of shear zones, thus playing a key role in the development or inhibition of localized deformation. However, experimental, theoretical, and field observations have produced conflicting suggestions that either A) rigid phases may pin the grain-size of a dominant weak phase, driving deformation into the diffusion creep regime, and promoting weakening, or B) rigid phases may be load supporting, thus increasing the strength of the aggregate. To investigate the influence of chemically inert, rigid second phases on the strength, microstructural evolution, and dominant deformation mechanism of calcite rocks, we have performed conventional triaxial experiments (MIT), and high strain torsion experiments (GFZ-Potsdam) on synthetic calcite aggregates with no second phases added, 1% (by vol.) 4 μm carbon spheres, 10% (by vol.) 4 μm carbon spheres, and 10% (by vol.) 5 μm carbon splinters. Glassy carbon spheres and splinters were chosen as they are chemically inert under the experimental conditions, are mechanically strong compared to the calcite, and can be obtained with a narrow grain size distribution. Prior to deformation, samples were hot isostatically pressed (HIPed) at 1023K and 300 MPa, to control the starting grain-size. The presence of carbon spheres or splinters has a dramatic influence on the grain-size after HIPing: samples with 10% spheres or splinters have an initial grain-size that is a factor of three finer than pure samples after 2.25 hrs while samples with 1% spheres are ~1.5 times finer than pure samples, again after 2.25 hrs. Deformation experiments were performed at 1023K, at a confining pressure of 300 MPa (trixial) or 400 MPa (torsion), at constant equivalent strain rates between 5e-6 s-1 and 1e-3 s-1. In both triaxial compression and torsion experiments, samples containing 10% splinters are stronger than pure samples and those containing 1% or 10% spheres. Further, in torsion experiments, samples with 10% splinters show weakening, whereas all other samples show progressive hardening. As a result, the strengths of the four materials converge at high strains (γ>16). In triaxial experiments, pure samples and samples containing 1% or 10% spheres have similar strengths, whereas in torsion, samples containing 10% spheres are significantly weaker than pure samples, or those containing 1% spheres. These results indicate that, while the microstructure of the dominant (in this case weak) phase is most important, that of the volumetrically minor (rigid) phases will also significantly influence the strength and mechanical evolution of the aggregate.