Grain Size Sensitive Creep During Simulated Seismic Slip in Nanogranular Fault Gouges: Constraints From Transmission Kikuchi Diffraction (TKD)

Nanograins (≪1 μm) are common in the principal slip zones of natural and experimental faults, but their formation and influence on fault mechanical behavior are poorly understood. We performed transmission Kikuchi diffraction (spatial resolution 20–50 nm) on the principal slip zone of an experimental carbonate gouge (50 wt% calcite, 50 wt% dolomite) that was deformed at a maximum slip rate of 1.2 m/s for 0.4 m displacement. The principal slip zone (PSZ) consists of nanogranular aggregates of calcite, Mg‐calcite, dolomite and periclase, dominated by grain sizes in the range of 100–300 nm. Nanograins in the ultrafine (< 800 nm) PSZ matrix have negligible internal lattice distortion, while grains > 800 nm in size contain subgrains. A weak crystallographic preferred orientation is observed as a clustering of calcite c‐axes within the PSZ. The high‐resolution microstructural observations from transmission Kikuchi diffraction, in combination with published flow laws for calcite, are compatible with high‐velocity slip in the PSZ having been accommodated by a combination of grain size sensitive creep in the ultrafine matrix, and grain size insensitive creep in the larger grains, with the former process likely controlling the bulk rheology of the PSZ after dynamic weakening. If the activation energy for creep is lowered by the nanogranular nature of the aggregates, this could facilitate grain size sensitive creep at high (coseismic) strain rates and only moderate bulk temperatures of approximately 600 °C, although temperatures up to 1000 °C could be locally achieved due to processes such as flash heating.

Nanograins have been generated in rock deformation experiments simulating both aseismic and coseismic deformation conditions (e.g., Han et al., 2010;Tisato et al., 2012;Verberne et al., 2013;Spagnuolo et al., 2015;Smeraglia, Bettucci, et al., 2017, Pozzi et al., 2018, 2019. In the low-velocity, aseismic regime (v ≤ 100 μm/s), microstructural observations of calcite gouges indicated that nanograins can be generated by plastic deformation, fracturing, and abrasion, and that they subsequently deform by grain boundary sliding aided by either dislocation or diffusion creep (Verberne et al., 2014). Verberne et al. (2014) documented sintering between the nanograins leading to the development of nanofibers that likely deformed by a granular flow mechanism, with competition between dilation caused by grain-neighbor swapping and compaction due to active diffusive mass transfer. In their case, the presence of a crystallographic preferred orientation (CPO) within shear bands cutting the gouge was interpreted to be the result of cataclastic flow aided by shear-induced dislocation glide at the grain contacts (Verberne et al., 2013).
High-velocity (v > 0.1 m/s) rock friction experiments aimed at reproducing seismic deformation conditions introduced the term "powder lubrication" to describe the production of nanograins in slip zones and the concomitant dynamic fault weakening (e.g., Han et al., 2010;Han et al., 2011;Reches & Lockner, 2010;Tisato et al., 2012). However, dynamic weakening may not be directly related to the formation of nanograins. More recent experimental studies have explored the possibility that nanograins could influence fault rheological properties due to the activation of "viscous" deformation mechanisms at high slip velocity (De Paola et al., 2015;Pozzi et al., 2018Pozzi et al., , 2019. In particular, the production of nanograins coupled with temperature rise due to frictional heating could promote grain size sensitive deformation mechanisms such as grain boundary sliding accompanied by dislocation or diffusion creep (i.e., superplastic flow in the broad sense; Schmid et al., 1977), which may lead to efficient strain accommodation and dynamic weakening at high velocities (De Paola et al., 2015;Green et al., 2015;Spagnuolo et al., 2015;Pozzi et al., 2018Pozzi et al., , 2019. Microstructural investigation of experimental slip zones showed the presence of equant nanograins (from 10 to 600 nm in size) forming a foam-like fabric with abundant grain triple junctions and a low dislocation density (De Paola et al., 2015;Green et al., 2015;Spagnuolo et al., 2015). The occurrence of similar microstructures has been associated with superplastic flow, defined as the ability of materials to accommodate unusually high strains without complete necking (Boullier & Gueguen, 1975;Fliervoet et al., 1999;Poirier, 1985;Walker et al., 1990).
Despite this recent progress in understanding the formation and role of nanograins during faulting, the deformation mechanisms active in nanogranular slip zones are still poorly understood, partly because of the difficulty in characterizing the microstructural and crystallographic properties of nanogranular materials. Microstructural studies and interpretation of deformation mechanisms within nanograin aggregates have relied mainly on scanning (SEM) and transmission electron microscope (TEM) analyses. Limitations with the spatial resolution of conventional electron backscatter diffraction (EBSD) on geological materials (approximately 50 to 250 nm depending on the material; Humphreys & Brough, 1999;Prior et al., 1999Prior et al., , 2009) has hindered application of EBSD to the study of nanograins. Recently, Pozzi et al. (2019) successfully applied EBSD to the study of nanogranular slip zones in experimentally deformed calcite gouges, and were able to resolve grains as small as 500 nm in size (step size of the analysis 200 nm). However, natural and experimental slip zones often contain grains that are tens to hundreds of nanometers in size, hence too small to be properly resolved using conventional EBSD (e.g., De Paola et al., 2015;Green et al., 2015). Although TEM has a spatial resolution of ≪ 1 nm, the study area with TEM is often less than a few tens of square micrometers, and therefore, crystallographic data sets are often limited to a few grains. As a consequence, the collection of statistically significant data sets concerning the crystallography of grain aggregates is time consuming and challenging using TEM.
Transmission Kikuchi diffraction (TKD) is a SEM-based technique that can provide similar crystallographic data to conventional EBSD, but at a significantly higher spatial resolution (i.e., up to 1-2 nm in metal alloys; Keller & Geiss, 2012;Trimby, 2012;Trimby et al., 2014). Application of TKD to geological samples is still in its infancy, but recent studies have shown that TKD can be applied successfully to study nanogranular geological materials including diamonds and sheared travertines (i.e., Delle Piane et al., 2018;Piazolo et al., 2016). In this study, we applied TKD to a nanogranular principal slip zone (PSZ) that was produced in a high-velocity shear experiment on calcite-dolomite gouges. We use the nanostructural observations to test current hypotheses regarding deformation mechanisms that are potentially active during seismic faulting in carbonates.

Materials and Methods
The TKD analysis was performed on a sample of deformed gouge composed of 50 wt% calcite and 50 wt% dolomite ( Figure 1). The calcite-dolomite mixture was obtained by crushing Carrara marble (99.  (Figure 1a). Following the experiment, the principal slip surface (PSS) and adjacent PSZ were extracted and embedded in low-viscosity epoxy. Estimates of temperature rise during shearing were obtained by applying the mathematics of Rice (2006), assuming that the principal slip surface can be treated as a plane of zero thickness (supporting information).
Sample preparation for TKD was carried out by Argon Ion Slicing following the methods outlined in Stojic and Brenker (2010; Figure 2). The Argon Ion Slicing procedure produces a large (2-4 × 10 4 μm 2 ) electrontransparent area along the leading edge of the sample, which in our case corresponds to the region containing the PSS and PSZ that formed during high-velocity shearing ( Figure 2).
The TKD data were collected on a Zeiss Sigma VP Field-Emission-Gun SEM equipped with a NordlysF EBSD camera from Oxford Instruments, located at the Otago Centre for Electron Microscopy, University of Otago (New Zealand). The TKD patterns were indexed and processed using Oxford Instruments AZtec software. The sample was mounted on a microclamp SEM substage fixed to a standard SEM sample holder and back-tilted 3.3°from horizontal (supporting information). The sample was kept at a working distance of 8 mm and the SEM operated at an accelerating voltage of 30 kV and a beam aperture of 120 μm. Analysis was performed under variable pressure conditions and charging was avoided by maintaining a few pascals of nitrogen inside the chamber. Given the sufficiently similar geometry of the TKD analysis to EBSD, the estimated angular resolution for TKD analysis was of ± 0.5°(measured on a Si standard).
The electron transparent area in the thinned sample is located between 1.2 and 27.75 μm from the PSS (Figures 2a-2c). Microstructural observations and compositional analyses (i.e., X-ray powder diffraction analysis) suggest that this part of the gouge layer accommodated most of the strain during deformation, and that the dolomite within this layer experienced thermal decomposition (Figures 1c, 1d, and 3). For X-ray powder diffraction analysis, intact pieces of the cohesive PSZ were recovered after the experiment and the analysis was performed directly on the polished PSS, without any grinding or separation, at the Department of Geosciences, University of Padova (Italy). Four adjacent TKD maps were collected at a step size of 20-50 nm, covering a total area of 477 μm 2 (Figures 2b and 2c; approximately 22 × 22 μm). Although this area is small compared to the total dimensions of the sample, SEM analysis indicate that the microstructures targeted with TKD are representative of the general microstructure of the PSZ (i.e., ultrafine-grained recrystallized layer adjacent to the PSS; see Figures 1c and 1d and section 3.2 below). Therefore, our TKD observations and data can be interpreted as representative of similar areas within the PSZ. The TKD patterns were stored, allowing the quality of indexing to be improved in the postprocessing stage (see supporting information for details on the data postprocessing). Grains were constructed from raw TKD data using the Voronoi decomposition in the open-source MTEX toolbox for MATLAB (Bachmann et al., 2011). Grains were defined as being surrounded by boundaries with a misorientation angle of > 10°. Grains composed of fewer than 10 pixels were ignored to avoid artifacts. Orientation data were plotted as stereoplots and contoured using the Orientation Density Function (after Bunge, 1982;Bachmann et al., 2010). The Orientation Density Function was calculated using a de la Vallee Poussin kernel, a half-width of 10°, and accounting for one point per grain. All pole figures were plotted as equal area and upper hemisphere.
For analysis of grain sizes and shapes, grain boundaries were hand-traced from band contrast images ( Figure 2c). The analysis was performed in Image SXM, excluding grains touching the border of the maps (Figure 4; Barrett, 2015). Grain size was calculated as the diameter of an area-equivalent circle. Grain shape-preferred orientation was calculated for the best fit ellipses of grains with an aspect ratio ≥ 1.5. Angles in this paper are counterclockwise from the PSS.

Mechanical Behavior and Temperature Rise in the Gouge Layer
The mechanical behavior of the calcite-dolomite gouge is consistent with previous studies (e.g., Smith et al., 2017). The gouge shows an initial increase of the apparent friction coefficient (μ = 0.68) followed by dynamic weakening to reach a steady state friction coefficient of μ = 0.28 ( Figure 1a). During deceleration at the end of the experiment, the friction coefficient increases to μ = 0.56 ( Figure 1a). Calculations of the temperature rise suggest that the PSS reached a temperature of approximately 620°C, which is consistent with thermocouple measurements of temperature increase in the adjacent PSZ (supporting information). Additionally, X-ray powder diffraction analysis performed on the principal slip surface shows the presence of periclase, one of the products of the thermal decomposition of dolomite above temperatures of approximately 550°C (i.e., CaMg(CO 3 ) 2 → CaCO 3 +MgO+CO 2 ; Samtani et al., 2002), along with dolomite and Mg-calcite ( Figure 3).

. Microstructures of the Principal Slip Zone
The bulk gouge is characterized by a well-developed foliation defined by alternating calcite-and dolomiterich domains oriented at~40°to the PSS (Figure 1b; similar to Smith et al., 2017). The PSZ consists of a continuous 15-20 μm thick, ultrafine-grained layer (≪ 1 μm in size; Figures 1c-1e) composed of calcite, Mg-calcite, and periclase ( Figure 3). Adjacent to the PSZ, a~30-40 μm thick layer includes dolomite grains with clusters of small holes and vesicular rims, interpreted to form by degassing during dolomite decarbonation (Figures 1c-1e; Mitchell et al., 2015). Elongate patches of recrystallized calcite are observed adjacent to the PSZ (Figure 1d).
Band contrast maps derived from TKD show that grains in the PSZ are polygonal and roughly equant (typical aspect ratio < 2.5; Figures 2c and 4b). Areas characterized by high-quality data (i.e., lighter areas in Figure 2c) show a compact recrystallized PSZ with negligible porosity. Grain boundaries are commonly straight to slightly curved (Figure 2c). Grains meet at 120°triple junctions, and quadruple junctions are also common (Figure 2c). Figure 4a shows the results of grain size analysis. Grains <100 nm in size are not included in the data set due to the lower resolution of the TKD analysis (i.e., 20 to 50 nm). Overall, the grain size distribution shows a range of grain sizes up to 2068 nm and a median value of 252 nm. The largest grains are composed of dolomite and are embedded within an ultrafine calcite-rich recrystallized matrix (Figures 1c and 1d). The matrix shows one main population of grains with a peak centered at approximately 200-300 nm (Figure 4a). Analysis of the shape-preferred orientation of all grains shows two weak preferred orientations with grain long axes oriented at approximately 40°and approximately 150°from the PSS (Figure 4c).

Crystallographic Preferred Orientation and Intragranular Misorientation in Nanograins
Significant variation in the CPO is observed in the four different TKD maps. In maps 25 and 9, calcite shows two main clusters of (0001) oriented at 20-45°to the y-axis ( Figure 5). Multiple clustering is observed on 1120 and 1014 È É , whereas on 0118 È É , the CPO is similar, but weaker, than that observed on (0001). In map 13, caxes are organized into two clusters located near the circumference of the pole figure and aligned subparallel to the y-axis ( Figure 5). There is no preferred alignment of the other crystallographic planes and directions. Finally, map 15 is characterized by four weakly defined clusters of (0001) and multiple clustering on the other crystallographic planes and directions ( Figure 5).

Discussion
Microstructural data collected by TKD allow us to explore the deformation mechanisms that are potentially active during slip at high velocities, equivalent to those experienced during localized coseismic slip in natural faults (Rowe & Griffith, 2015). Recent work by De Paola et al. (2015) and Pozzi et al. (2018Pozzi et al. ( , 2019 on calcite gouges suggested a progressive transition from brittle deformation to viscous flow accompanying dynamic weakening and temperature increase at high slip velocities. During the initial slip-strengthening phase, the gouge deforms mainly by cataclasis aided by twinning, resulting in the progressive development of a comminuted and fine-grained shear band (Beeler et al., 1996;Smith et al., 2015). During dynamic weakening, the temperature increase in the principal slip zone activates both grain size sensitive (GSS) and insensitive (GSI) creep mechanisms, accompanying a dramatic decrease in shear stress. Once the shear stress on the fault has dropped to a steady state, GSS creep is dominant, and may weaken any CPO that had previously formed due to GSI creep.
Static grain growth during cooling at the end of this type of high-velocity experiment potentially modifies the grain size distribution and grain shapes (e.g., Covey-Crump, 1997). Estimates of the cooling rate for the PSZ at the end of our experiment (supporting information) suggest that the temperature fell below 150°C after a period of less than 1 s. Calculations of the original grain size distribution within the analyzed areas prior to annealing were made using the equation for normal grain growth (e.g., Covey-Crump, 1997): where d is the final grain size, d 0 is the initial grain size, k is a constant with an Arrhenius temperature dependence, n is a dimensionless constant that depends on the process controlling the grain growth, and t is the duration of the grain growth. Our calculations were done by assuming that the PSZ is composed of pure calcite (yellow bars in Figure 4a). Assuming 1 s of annealing, the calculations predict that individual grains may have grown by up to 30-50 nm. However, grain growth during annealing is hindered by the presence of impurities in calcite (i.e., partial exchange of Ca cations with Mg; Herwegh et al., 2003), as well as pinning of grain boundaries by second phases or pores (e.g., Covey-Crump, 1997;Davis et al., 2011). Grain growth rates in Mg-calcite and dolomite are lower than calcite by a factor of 10 3 -10 4 (Davis et al., 2011;Herwegh et al., 2003). We suggest therefore that grain growth in the calcite-dolomite gouges was negligible (on the order of a few nanometers or less), and that the microstructures and grain sizes preserved in our sample are similar to those that were active during the high-velocity experiment. It is possible that extreme comminution and thermal decomposition generated amorphous carbon and CO 2 (e.g., Delle Piane et al., 2018;Demurtas et al., 2019;Spagnuolo et al., 2015), which could have enhanced grain growth rates. However, amorphous carbon was not identified during TKD analysis. The relatively high porosity of the PSZ, and the unconfined nature of the gouge holder, also makes it likely that CO 2 produced, and measured, during the experiment was able to escape relatively easily (supporting information).
In our experiments, the PSZ contains grains that are dominantly 200-300 nm in size (Figure 4a). The aggregates have a foam-like microstructure (i.e., equant grains with relatively straight grain boundaries), triple junctions between roughly equant nanograins have approximately 120°dihedral angles, and quadruple junctions are common (Figures 2c and 6). These microstructures closely resemble those described by Schmid et al. (1977) and Walker et al. (1990), which formed during high-temperature deformation of calcite aggregates, interpreted to be the result of GSS creep during superplastic flow. The activation of grain boundary sliding aided by GSS creep in the ultrafine matrix of the PSZ is also supported by the presence of equant strain-free (i.e., low internal misorientation) grains in the matrix, usually confined to grain sizes of < 800 nm (Figure 6a; Poirier, 1985). However, GSI creep also seems to have been active during the experiment, because larger grains (> 800 nm) show irregular grain boundaries and contain subgrains (Figure 6b; Rutter, 1995). In addition, a weak CPO on (0001) is developed locally in some TKD maps ( Figure 5). The similarity between the subgrain size in the larger grains and the size of grains in the surrounding ultrafine matrix suggests that the latter may have formed (at least partly) by subgrain rotation recrystallization. Pozzi et al. (2019) interpreted the progressive weakening of the CPO in calcite nanograins as evidence for a transition to a diffusion creep regime enhanced by grain size reduction and concomitant temperature increase. Previous work has suggested that deformation by diffusion creep accompanied by grain rotation may cause weakeningbut not necessarily complete randomizationof any preexisting CPOs (Rutter et al., 1994;Taplin et al., 1979;Walker et al., 1990;Wheeler, 2009). However, deformation experiments in the GSS regime have also shown that CPO can develop under such conditions in fine-grained aggregates of calcite (Rutter et al., 1994), olivine (Fliervoet et al., 1999;Hansen et al., 2011), ice (Goldsby & Kohlstedt, 2001), anorthite (Barreiro et al., 2007), dolomite (Delle Piane et al., 2008), and calcite-dolomite mixtures (Delle Piane et al., 2009;Kushnir et al., 2015). In our experiments, the weak CPO may have been inherited from that developed in the bulk gouge as a result of early grain rotation and preferential fracturing along the r-rhomb calcite cleavage planes (Demurtas et al., 2019). Subsequently, extreme grain size reduction and temperature increase led to the activation of GSS creep, with little to negligible change in the overall CPO in the PSZ. Given that the matrix in our experiment is dominated (90-95 vol %) by strain-free nanograins < 800 nm in size (Figure 4a), we infer that GSS creep mechanisms likely controlled the bulk rheology of the PSZ following dynamic weakening. Figure 7 shows deformation mechanism maps calculated for calcite at temperatures of 600°C and 1000°C. The PSZ is assumed to be composed only of calcite, based on microstructural observations and mineralogical analysis showing a PSZ composed mainly of calcite and Mg-calcite with sparse dolomite grains. Flow laws for GSI and GSS creep were taken from Renner et al. (2002) and Herwegh et al. (2003), respectively. For GSI and GSS creep the stress exponent used were n = 2 and n = 1.1, respectively. The field boundary of de Bresser et al. (1998de Bresser et al. ( , 2002 corresponds to the stable grain sizes predicted to occur at the boundary between the diffusion and dislocation creep fields. The shear strain, γ, in the PSZ is calculated as γ ¼ rθ 2 h = , where r is the outer radius of the sample, θ is the angular displacement, and h is the average slip zone thickness (in our case h = 100 μm). The strain rate in the PSZ was calculated as _ γ ¼ Δγ Δt , where Δt is the time interval considered and Δγ is the shear strain accommodated in the PSZ during Δt. Calculated strain rates in the PSZ are _ γ ≈ 6 × 10 3 s -1 , in accordance with previous estimates for similar experiments (e.g., De Paola et al., 2015). At the calculated temperatures, both the ultrafine matrix grains (red areas in Figure 7) and the larger grains containing subgrains (yellow areas in Figure 7) lie within the GSS field. The strain rates calculated for the observed grain size range in the deformation mechanism map at 600°C are between 10 −1 and 10 −3 , which is a factor of 10 4 -10 6 lower than the strain rate estimated to have occurred in the experiment. At 1000°C, the strain rates are between 10 0 and 10 3 , and the higher strain rates are consistent with the sizes of the nanograin population showing low values of intracrystalline distortion (Figure 7b). K-type thermocouples in the gouge layer measured peak temperatures of approximately 620°C, consistent with our bulk temperature calculations (after Rice, 2006; see Text S2 in the supporting information for additional details). However, it is possible that significantly higher temperatures were reached locally in the PSZ in areas where heat dissipation was less efficient, or in areas where the temperature increase was enhanced by local processes such as flash heating along asperities. Temperatures of up to 1000°C have previously been estimated for the PSZ in calcite gouges deformed at similar deformation conditions (De Paola et al., 2015;Pozzi et al., 2018).
If the peak temperature of the PSZ was indeed limited to 600°C, the inconsistency between the estimated strain rate ( _ γ ¼6 × 10 3 s -1 ) in the PSZ and the calculated strain rates in the deformation mechanism map (Figure 7a) might be related to the larger grain size at which the flow laws of Renner et al. (2002) and  Renner et al. (2002) and Herwegh et al. (2003) were used for grain size insensitive (GSI) and grain size sensitive (GSS) creep, respectively. The blue line represents the field boundary after de Bresser et al. (1998de Bresser et al. ( , 2002. Strain rates are shown as labeled contours. The red contour represents the calculated strain rate of approximately 10 3 s for the PSZ in our experiment. The differential stress measured during steady state deformation is shown as dashed grey line. According to the deformation mechanism maps, in order to accommodate the estimated strain rates during deformation, the PSZ is supposed to reach a temperature of 1000°C.

10.1029/2019JB018071
Journal of Geophysical Research: Solid Earth Herwegh et al. (2003) were calculated (i.e., grains > 5-10 μm). In the case of nanoparticles, it has been demonstrated that grain size can have a major control on lowering the activation energy (E A ) required to activate creep and diffusion mechanisms (e.g., Cai et al., 2000;Guisbiers & Buchaillot, 2008;Li et al., 2004) or decomposition reactions (Yue et al., 1999). This grain size effect is a consequence of the excess energy stored on the nanograin surface (e.g., Asaro & Suresh, 2005;Yue et al., 1999) and the larger number of grain boundaries compared to a fine-grained aggregate. For nanoparticles produced by mechanical milling, the widespread presence of defects and internal plastic deformation in the nanoparticles is expected to result in a further lowering of E A (Fisher, 1988). In Figure 8, we test two different scenarios where the E A for diffusion (E A diff ) and dislocation creep (E A disl ) is reduced by 25% (E A = 150 kJ/mol; Figure 8a) and 50% (E A = 100 kJ/mol; Figure 8b) compared to the values of  and Herwegh et al. (2003). Such reductions in E A are consistent with previous observations of a decrease of E A for thermal decomposition of calcite of 70-80 kJ/mol, corresponding to 35-40% of the reference value (Yue et al., 1999). By comparing the deformation mechanism maps in Figures 7 and 8, it is clear that a reduction in E A due to the presence of nanograins could facilitate accommodation of very high strain rates ( _ γ ¼6 × 10 3 s -1 ) by grain size sensitive flow, without necessarily having to invoke temperatures higher than approximately 600°C. If this is the case, the activation of viscous deformation mechanisms during coseismic shearing could be expected to control the rheology of our experimental PSZ (also see Pozzi et al., 2018). Notably, the decomposition temperature is also expected to decrease in the presence of nanoparticles (Yue et al., 1999). As a consequence, particular care should be exercised when making peak temperature estimates based on the occurrence of decomposition products.
Collectively, our TKD observations provide quantitative microstructural and crystallographic data sets that are compatible with other observations of experimental carbonate slip zones (Verberne et al., 2013(Verberne et al., , 2014De Paola et al., 2015;Green et al., 2015;Delle Piane et al., 2018;Pozzi et al., 2018Pozzi et al., , 2019. Our observations suggest that during high-velocity shearing the PSZ deforms mainly by GSS creep, possibly facilitated by a reduction in E A within nanogranular aggregates, which allows high strain rates (_ γ ¼6 × 10 3 s -1 ) to be accommodated at moderate bulk temperatures (approximately 600°C). Additionally, the presence of dolomite in the gouge mixture appears to have hindered significant grain growth after the experiment, resulting in final grain sizes and shapes that are likely to be similar to those active during high-velocity shearing. TKD allows quantitative investigation of the microstructure and crystallographic properties of nanograins in fault zones, and provides information that is complementary to TEM observations and EBSD data from coarser grain populations.  Bresser et al. (1998Bresser et al. ( , 2002. Strain rates are shown as labeled contours. The red contour represents the calculated strain rate of approximately 10 3 s for the PSZ in our experiment. The differential stress measured during steady state deformation is shown as dashed grey line. By accounting for the nature of the nanoparticles, the reduction of E A to creep allows the PSZ to deform at the calculated strain rates at temperatures of 600°C.

10.1029/2019JB018071
Journal of Geophysical Research: Solid Earth

Conclusions
TKD was performed on the nanogranular principal slip zone of a calcite-dolomite gouge sheared experimentally at a maximum slip velocity of 1.2 m/s for 0.4 m displacement. X-ray powder diffraction shows that the slip zone consists of calcite, Mg-calcite, dolomite and periclase. Calculations of grain growth during annealing suggest that the post-mortem microstructures are representative of those active during high-velocity shearing. The TKD analysis reveals that the principal slip zone is recrystallized and composed of a population of equant grains approximately 200-300 nm in size with negligible intracrystalline misorientation, together with larger grains (d > 800 nm) that typically contain subgrains. Weak CPOs are defined by clustering of the calcite c axes in some areas of the PSZ. Together with published flow laws for calcite, these microstructural observations suggest that grain size sensitive creep in the ultrafine-grained PSZ matrix was the main deformation mechanism controlling slip zone rheology during high-velocity shearing. Grain size insensitive creep was also active and resulted in the formation of subgrains via subgrain rotation recrystallization inside the larger grains. A reduction of the activation energy (E A ) for creep in calcite nanograins would allow high (coseismic) strain rates to be accommodated at moderate bulk temperatures of approximately 600°C.