Under relatively high temperatures and pressures, rock deformation is accommodated by crystalline plasticity and diffusive mass transfer processes. In contrast, brittle failure occurs under low temperatures and pressures, usually with the development of dilatancy and shear localization very early in the deformation. Brittle faulting has been intensively studied in the laboratory under upper crustal conditions of pressure and temperature. As shear localization develops, there is a concomitant degradation of the rock sampleÕs overall strength, and the faulting process becomes catastrophic unless the surplus of elastic energy stored in the test machine is quickly relieved. Therefore details of the shear localization process cannot be observed unless the ``post-failure'' deformation is stabilized. An important technical advance was made by Lockner et al. (1992), who successfully used acoustic emission (AE) rate as the feedback signal for the servo-control of deformation and the stabilization of stress drop in the post-failure stage. Adopting techniques from seismology, they also obtained the 3-dimensional location of the AE hypocenters, mapping out with high resolution the clustering and localization of microcracking activity. The data provide important constraints on the theoretical analysis of the onset of shear localization (e.g. Rudnicki and Rice, 1975) and on the energetics of faulting (e.g. Rice, 1980).
While AE measurements can be used to map the spatio-temporal complexity of the microcracking activities, details of the micromechanical process have primarily been inferred from microstructural observations of deformed samples. Using a computer automated system, Moore and Lockner (1994) characterized quantitatively the microcrack density and its anisotropy in granite samples which had been deformed stably through the post-failure stage. The micromechanical process involves the initially stable growth of stress-induced cracks subparallel to the maximum compression direction, and the subsequent development of strain softening as the microcracks interact and coalesce to form throughgoing shear band. While these recent observations on the micromechanics of failure are in qualitative agreement with previous microstructural studies (e.g. Wong, 1982; Myers et al., 1992) and theoretical models (e.g. Horii and Nemat-Nasser, 1985; Ashby and Hallam, 1986; Kemeny and Cook, 1991), important insights on the correspondence between microcracking and AE activities were gained by Lockner (1993), who synthesized the two types of data to conclude that the detected AE accounted for less than 1% of the total microcrack damage in granite, implying that the AE measurements may not be representative of the overall progress of failure. It highlights the complementary roles of AE and microstructural measurements in the micromechanical investigation of brittle failure.
New techniques such as laser scanning confocal microscopy (Fredrich et al., 1993) have been used to observe the 3-dimensional complexity of the pore space in rocks. Optical and transmission electron microscopy continue to be used for the elucidation of the complex interplay of microcracking, crystalline plasticity, diffusive mass transfer and pore collapse processes in the brittle-plastic transition. Recent observations demonstrate that the brittle-plastic transition in quartz aggregate involves at least three transitions in deformation mechanism (Hirth and Tullis, 1994). With increasing temperature and pressure, enhanced ductility is observed because microcrack propagation is readily arrested by dislocation glide and twin bands, and significant increase of hardening stress is required to extend the cracks to a critical length before crack coalescence and shear localization can occur. As the crack arrest process begins to dominate, the microcracks remain relatively short and shear localization is inhibited. However, microcrack density may continue to increase because dislocations themselves can pile up and induce local stress concentration sufficiently high to nucleate microcracks. The complete transition to plastic flow by dislocation creep occurs when the temperature and pressure are so high that microcrack growth is completely inhibited. Overall, the micromechanics seems to be adequately described by models on microcrack-dislocation interactions (Horii and Nemat-Nasser, 1986; Wong, 1990).
Tullis and Yund (1992) reported that the brittle-plastic transition in feldspar aggregates to be similar to the quartz aggregate with two important differences. In feldspar aggregates, there is a broad regime of cataclastic flow which is facilitated by the ease of cleavage cracking and the difficulty of dislocation glide and climb. Furthermore, shear localization can develop in highly complex patterns. In many respects, the micromechanical processes operative in the brittle-plastic transition in the quartzo-feldspathic aggregate are very similar those observed in Carrara marble (Fredrich et al., 1990). Although the volume change was not monitored, one suspects that significant dilatancy was involved in both the brittle faulting and cataclastic flow regimes in the low-porosity silicate rocks, analogous to porosity change in marble previously reported by Fredrich et al. (1990) and recently corroborated by Zhang et al. (1994). Although the micromechanical observations place important rheological constraints on crustal dynamics, little progress has been made in actually incorporating the laboratory data into constitutive equations appropriate for the transitional regimes. The relatively simple model (Brace and Kohlstedt, 1980) with an upper brittle layer which overlies a plastic substratum (deforming by dislocation creep) remains the standard model for the depth profile in tectonophysics. The laboratory data imply that this two-mechanisms model overestimates the strength of the lithosphere in the transition between the brittle and ductile (plastic) regimes.
Significant advances have also been made in our understanding of the transition from brittle faulting to cataclastic flow in porous sediments and sedimentary rocks. In such aggregates, significant stress concentration can develop at points where grains impinge on one another. Once nucleated, the tensile cracks tend to grow unstably across the grain, resulting in grain crushing and pore collapse. The micromechanical process can be modelled using elastic contact theory and linear elastic fracture mechanics (Zhang et al., 1990). Such a pore collapse process promotes strain hardening and inhibits the onset of shear localization, and the cataclastic flow is characterized by a yield stress which decreases with increasing pressure and by a yield envelope which expands with porosity reduction . The phenomenology seems to applicable to sandstone (Wong et al., 1992), carbonates (Teufel et al., 1991) and shale (Steiger and Leung, 1991), and it has been widely used in the modelling of accretionary prism tectonics, reservoir compaction and borehole instability.