Episodic Slow Slip Hosted by Talc‐Bearing Metasomatic Rocks: High Strain Rates and Stress Amplification in a Chemically Reacting Shear Zone

Episodic tremor and slow slip (ETS) downdip of the subduction seismogenic zone are poorly understood slip behaviors of the seismic cycle. Talc, a common metasomatic mineral at the subduction interface, is suggested to host slow slip but this hypothesis has not been tested in the rock record. We investigate actinolite microstructures from talc‐bearing and talc‐free rocks exhumed from the depths of modern ETS (Pimu'nga/Santa Catalina Island, California). Actinolite deformed by dissolution‐reprecipitation creep in the talc‐free rock and dislocation creep ± cataclasis in the talc‐bearing rock. This contrast results from stress amplification in the talc‐bearing rock produced by high strain rates in surrounding weak talc. We hypothesize that higher strain rates in the talc‐bearing sample represent episodic slow slip, while lower strain rates in the talc‐free sample represent intervening aseismic creep. This work highlights the need to consider fluid‐mediated chemical change in studies of subduction zone deformation and seismicity.

• Contrasting actinolite microstructures in talc-free and talc-bearing subduction interface metasomatic rocks record stress variations • Stress amplification in actinolite from talc-actinolite schist results from high strain rate deformation of surrounding talc • High strain rates reflect episodic slow slip localized in talc-actinolite schist under high pore fluid pressures during metasomatism

Supporting Information:
Supporting Information may be found in the online version of this article.

Citation:
Hoover, W. F., Condit, C. B., Lindquist, P. C., Moser, A. C., & Guevara, V. E. (2022). Episodic slow slip hosted by talc-bearing metasomatic rocks: High strain rates and stress amplification in a chemically reacting shear zone. Geophysical Research Letters, 49, e2022GL101083. https://doi. org/10.1029/2022GL101083 rocks are common in exhumed paleosubduction interface mélanges where they are interpreted as the result of fluid-mediated reactions between chemically contrasting strong block and weak matrix lithologies (e.g., Bebout & Barton, 2002;Bebout & Penniston-Dorland, 2016;Gyomlai et al., 2021;King et al., 2003;Spandler et al., 2008). Talc has a low coefficient of friction and is velocity-strengthening to -neutral, making it a prime candidate to host slow slip if frictional deformation can be activated, and emphasizing the potential rheological impact of metasomatic minerals (Chen et al., 2017;Hirauchi et al., 2020;Misra et al., 2014;Moore & Lockner, 2004. Here, we investigate the role of talc in hosting ETS through a microstructural study of talc-bearing and talc-free subduction interface rocks from a terrane exhumed from the depths of ETS in modern subduction zones (Figures 1a and 1b). These rocks display contrasting amphibole microstructures that are best explained by stress amplification resulting from higher strain rates in talc-bearing rocks. These observations are consistent with episodic slow slip accommodated by frictional deformation in talc-bearing metasomatic rocks during periods of elevated pore fluid pressure.

Field Context
The sampled mélange zone and adjacent LBS and EA were mapped to contextualize the microstructural observations ( Figure 1c). Both the EA and LBS are composed of coherent seafloor lithostratigraphy containing meta-(pillow) basalts, metachert (garnet quartzite), and metasedimentary schists (Platt, 1976). The intervening mélange zone, ∼100 m thick, contains blocks of EA metasedimentary rocks (metachert, graphitic, and mica schists) and variably altered garnet hornblendite in a metasomatized matrix ( Figure 1c). No blocks of EA metabasalt or any LBS lithology were found in the mélange zone. The metasomatic mélange matrix is dominated by layers of talc ± actinolite and chlorite-actinolite schists and actinolitite from 10 to 100 s of cm thick (Figures 1d-1f). Talc ± actinolite schist is in contact with a serpentinite block and chlorite-actinolite schist is in contact with blocks of graphitic schist and metachert (Figures 1e and 1f). Coherent layering in the mélange matrix is traceable up to ∼100 m. Quartz veins are commonly observed cross-cutting actinolitite and chlorite-actinolite schist (Figures 1g and 1h). Representative samples of the chlorite-and talc-bearing mélange matrix were collected from a single outcrop to investigate their microstructures ( Figure 1d).

Chlorite-Actinolite Schist
The chlorite-actinolite schist has a strong foliation defined by mm-to cm-scale layering of chlorite-and actinolite-rich domains. Microstructure and modal mineralogy are highly variable, but the dominant assemblage is chlorite, actinolite, and graphite and minor rutile, titanite, apatite, and talc (<1%). A cm-thick chlorite layer is made up of large (∼750 μm) interlocking chlorite plates that are radially oriented and exhibit kinking and undulose extinction (Figure 2a). The boundaries of this radial chlorite layer are sharp with chlorite plates truncated against highly aligned actinolite ± chlorite rather than rotated into the foliation (Figure 2a). A large (1.5 mm) actinolite grain growing across this boundary is broken and rotated into the neighboring actinolite ± chlorite foliation ( Figure 2b). Adjacent to the radial-chlorite layer are highly deformed actinolite ± chlorite domains.

Talc-Actinolite Schist
The talc-actinolite schist is composed of talc and actinolite with minor chlorite and inclusions of Cr-spinel in actinolite. This sample has weak compositional layering of talc-and actinolite-rich domains and a strong foliation defined by alignment of talc and actinolite. In talc-rich domains, foliation is defined by near-uniformly oriented talc cleavage planes and aligned actinolite porphyroclasts and aggregates ( Figure 3a). In actinolite-rich domains, actinolite ranges from porphyroclastic to acicular in texture. Actinolite porphyroclasts and aggregates show distinct fracturing and disaggregation along basal cleavage planes in all domains ( Figure 3b).

Methods
A Thermo-Fisher/FEI Apreo-S scanning electron microscope equipped with an Oxford Instruments Symmetry electron backscatter diffraction detector and Ultim-Max 100 energy dispersive spectroscopy detector was used to map actinolite composition and crystallographic orientation. Analytical conditions and full data set are provided in the Supplementary information S1.

Actinolite Composition and Zonation
Amphibole compositions overlap in both samples and fall within the fields of actinolite, tremolite, and magnesio-hornblende (Supplementary Data S1; Leake et al., 1997). In the chlorite-actinolite schist, oscillatory zoning of Al, Fe, and Na in actinolite is often truncated by grain boundaries (Figures 2e and 2f). In the talc-actinolite schist, Cr and Al are enriched in actinolite porphyroclast rims, along fractures, and in domains of grain size reduction (Figures 3c and 3d).

Subduction Interface Mélange Setting
The mapped mélange zone is characterized by a block-in-matrix texture and occurrence of hybrid lithologies comparable to other subduction interface mélanges of the Catalina Schist (e.g., Bebout & Barton, 1989;Sorensen & Barton, 1987). The absence of LBS lithologies in the mélange suggests limited deformation after underplating of the EA, including during juxtaposition with the LBS and exhumation, and that mélange deformation reflects EA peak metamorphic conditions (Figure 1c; Harvey, Walker, et al., 2021). The position of the studied mélange stratigraphically above a coherent EA slab section corroborates its origin as a paleosubduction interface (Figure 1c). Field relationships suggest that the talc-actinolite and chlorite-actinolite schist formed from ultramafic and sedimentary protoliths, respectively, likely representing the reaction between overriding mantle wedge and downgoing slab sediments (Figures 1e and 1f). Truncation of actinolite zonation at deformed grain boundaries and chemical modification of actinolite along subgrain boundaries and in grain-size-reduced domains indicate that metasomatism and deformation of the mélange were concurrent (Figures 2e and 2f, 3c and 3d). Together, the pressure-temperature conditions of this coherent slab section, subduction interface setting, and co-occurring deformation and metasomatism of this mélange are ideal for investigating processes at the depths of ETS in modern subduction zones like Shikoku, Mexico, and Cascadia (Figure 1b; e.g., Condit et al., 2020).

Contrasting Talc and Chlorite Deformation
Talc and chlorite are both considered "weak" minerals but exhibit contrasting microstructures in these samples (e.g., Moore & Lockner, 2008;Okamoto et al., 2019). Talc cleavage planes are near-uniformly aligned in the talc-actinolite schist, suggesting talc may have accommodated significant strain by frictional sliding along [001] (e.g., Moore & Lockner, 2008, Figure 3a). In contrast, the ability of chlorite to accommodate high strain appears to have been highly dependent on orientation and phase mixing. In the radial-chlorite layer in the chlorite-actinolite schist, large radial plates of chlorite exhibit kinking and undulose extinction (Figure 2a). Given that dislocation glide and frictional sliding in chlorite are both reliant on the alignment of [001] planes, these microstructures are consistent with strain hardening due to unfavorable mineral alignment and a lack of recovery mechanisms (e.g., Okamoto et al., 2019). These microstructures suggest that the chlorite layer acted as a strong body that localized deformation into adjacent actinolite and actinolite + chlorite domains. Only in mixed actinolite + chlorite domains did chlorite grain size remain small, potentially facilitating reorientation and alignment of [001] to accommodate significant strain (Figure 2c). These microstructural contrasts between talc and chlorite highlight the complexity of sheet silicate deformation and the importance of orientation, texture, and grain size in determining their strength, often in contrast with constitutive relations (e.g., chlorite; Okamoto et al., 2019).

Stress Variations in Actinolite
Actinolite shares similar compositional ranges and textures in both rocks yet exhibits strongly contrasting microstructures between rocks with or without volumetrically significant talc (>1%; Figures 2 and 3). In the talc-free sample, limited intragrain deformation and the truncation of chemical zoning at grain boundaries suggests dissolution-reprecipitation creep as the dominant deformation mechanism in actinolite ( . Dissolution-reprecipitation creep is a commonly observed deformation mechanism in natural and experimental amphibole (e.g., Berger & Stünitz, 1996;Condit & Mahan, 2018;Getsinger & Hirth, 2014;Imon et al., 2004;Lee et al., 2022). This mechanism has been reported in a range of amphibole compositions and pressure-temperature conditions and would be expected given the low stresses and abundant fluids in this subduction interface setting (e.g., Bebout & Penniston-Dorland, 2016;Behr & Platt, 2013;Condit et al., 2022;Wassmann & Stöckhert, 2013).
In contrast, in the talc-bearing sample, intragrain deformation of actinolite is abundant with the development of misorientation gradients and subgrains, a strong CPO, and petrographically observed fracturing and disaggregation of actinolite porphyroclasts (Figure 3). These microstructures are scale-invariant between single actinolite crystals surrounded by talc, mm-scale actinolite aggregates in talc, and cm-scale actinolite-rich domains. Based on these microstructures, actinolite in the talc-bearing sample was dominantly deformed via dislocation creep with or without cataclasis (e.g., Babaie & La Tour, 1994;Imon et al., 2004;Soret et al., 2019). While cataclasis has been reported in many previous studies of amphibole deformation, dislocation creep is rare and has been associated with dry conditions, high temperatures, and high stresses and strain rates (e.g., Brückner & Trepmann, 2021;Diaz Aspiroz et al., 2007;Hacker & Christie, 1990;Soret et al., 2019). Thus, the actinolite microstructures observed in the talc-bearing sample are unusual because of the activation of dislocation creep at relatively low temperature, under fluid-rich conditions, and in the presence of weak talc.
For minerals with well-mapped deformation mechanisms such as quartz, olivine, and feldspar, the shift from dissolution-reprecipitation creep to dislocation creep is produced by an increase in stress, holding other variables constant (e.g., Goetze, 1978;Rutter & Brodie, 2004;Rybacki & Dresen, 2004). Amphibole deformation mechanisms are poorly constrained in pressure-temperature-stress-fluid-grain size-composition space and clearly complex, but the little published experimental deformation of amphibole revealed plastic deformation was activated at high stresses and strain rates (e.g., Brückner & Trepmann, 2021;Hacker & Christie, 1990;Rooney et al., 1975). In this framework, the observed actinolite microstructures are best explained by higher stresses experienced by actinolite in the talc-bearing sample compared to the talc-free sample. Given the overlapping amphibole composition, grain size, and a shared pressure-temperature-fluid history for these adjacent mélange matrix samples, local stress variations are the most likely explanation for the microstructural contrast.

Strain Rate Variations and Slow Slip
Local stress variations can occur in two-phase mixtures depending on the strength contrast between phases and the abundance of strong clasts versus weak matrix (e.g., Handy, 1990). For moderate clast fractions (>0.3), "jamming" of clasts creates load-bearing force chains that amplify clast stresses by 2-14×, depending on clast-matrix strength contrast (Beall et al., 2019a(Beall et al., , 2019bCates et al., 1998;Webber et al., 2018). Even at low clast fractions (<0.1) and in the absence of clast-clast interaction, clast stress shadows, and the matrix itself, can impose stress on clasts resulting in stress amplification of 2-5× that on the matrix (Ioannidi et al., 2022;Ladd & Reber, 2020). These numerical models, experiments and geologic studies are scale independent, and are consistent with the microstructures observed here reflecting stress amplification in strong actinolite due to force-chain formation and/or stress transfer from the weak talc matrix.
Actinolite stress amplification is absent from the chlorite-actinolite schist despite similar or greater actinolite abundances, suggesting stress amplification is controlled by matrix mineralogy and its orientation. Actinolite grains are juxtaposed in force chains by matrix strain, and stress transferred from weak to strong phase undergoes time-dependent relaxation, so in both cases higher strain rates result in increased efficiency of stress amplification (e.g., Beall et al., 2019a;Ladd & Reber, 2020). Clast stress amplification is also dependent on clast-matrix viscosity contrast, so lower viscosity of talc relative to chlorite (or misalignment of chlorite [001]) could explain stress amplification only in the talc-bearing rock and would yield higher strain rates in talc (e.g., Beall et al., 2019a;Fagereng & Sibson, 2010;Moore & Lockner, 2008;Okamoto et al., 2019). Thus, multiple lines of reasoning suggest stress amplification in the talc-bearing sample and its absence in the talc-free sample reflects higher strain rates accommodated by weaker talc.
In this rheological model, variations in stresses experienced by actinolite act as a key microstructural recorder of higher strain rates accommodated by talc deformation relative to chlorite or actinolite in the chlorite-actinolite schist. This interpretation is consistent with the prediction that talc will partition high-strain-rate deformation via frictional sliding during slow slip at high pore fluid pressures (French & Condit, 2019). The uniform alignment of talc cleavage planes observed in the talc-bearing sample would enable this frictional mechanism, though diagnostic microstructures for frictional sliding in talc are poorly constrained. In this model, lower stress and strain rate deformation of actinolite and chlorite in talc-free rocks occurred during periods of slower deformation (aseismic creep), while high strain rate deformation of talc-rich rocks occurred during slow slip events (Figure 4).

Comparison to Rheological Modeling and ETS
Partitioning of slow slip deformation into talc is hypothesized from rheological modeling when pore fluid pressures are high enough to activate frictional sliding (French & Condit, 2019). The development of the metasomatic talc-rich lithologies studied here is the direct result of abundant fluids that catalyzed and mediated chemical reactions (e.g., Bebout & Barton, 2002). Quartz veins in adjacent metasomatic actinolitite suggest periods of high pore fluid pressures in this already fluid-rich environment, consistent with evidence of high fluid pressures in modern and exhumed subduction interfaces (Figures 1g and 1h; Audet et al., 2009;Condit & French, 2022;Kodaira et al., 2004;Raimbourg et al., 2022). Thus, in this mélange zone, the fluid pressure conditions of the rheological model for frictional deformation of talc are likely met, consistent with our interpretation that high strain rates in the talc-bearing sample may represent slow slip events.
Episodic slow slip in this suite of metasomatic rocks would result in alternating aseismic creep and slow slip consistent with our microstructural observations. In the chlorite-actinolite schist, lenses of metasomatic actinolite preserve relict fabrics recording earlier episodes of syn-reaction deformation ( Figure 4). Likewise, the growth, and later rotation, of a large actinolite crystal across the boundary between an actinolite layer and the radial-chlorite layer indicates a cessation and reinitiation of deformation at this boundary ( Figure 4). These textures provide direct evidence of intermittent static conditions in the chlorite-actinolite schist consistent with periodic partitioning of deformation into the talc-actinolite schist during slow slip controlled by pore fluid pressure magnitude and matrix strength.

Implications for ETS
We show here that talc-bearing rocks may have hosted episodic slow slip events in a paleosubduction zone. Constitutive relations, deformation experiments, and geologic studies suggest talc may be a ubiquitous host of deep slow slip (e.g., French & Condit, 2019;Moore & Lockner, 2008;Spandler et al., 2008). Talc is stable from seafloor to subarc conditions, is common in exhumed paleosubduction interfaces from a variety of depths and is formed by reaction between Si-and Mg-rich rocks (e.g., Bose & Ganguly, 1995;D'Orazio et al., 2004;Kim et al., 2010;Spandler et al., 2008). Juxtaposition of these rock types is most prevalent below the seismogenic zone (i.e., mantle wedge corner) where talc could host all deep episodic slow slip, but the broad stability, rheology, and ubiquity of metasomatic minerals, talc, and in its absence, chlorite, could make them important contributors to slow slip at a range of depths. Multiple lines of evidence for talc as the host of slow slip indicate a causal relationship between metasomatism and the rheological preconditions for slow slip. Metasomatism, and not closed system metamorphism, is required to produce these potential mineral hosts of deep slow slip. It is imperative that studies of deformation in subduction zones consider not only subduction zone inputs and their metamorphosed equivalents, but equally the lithologies produced by chemical reactions (e.g., Phillips et al., 2020). Ignoring these lithologies will result in spurious conclusions on the strength of the subduction interface and the host(s) of slow slip. During background aseismic creep deformation is partitioned into talc-free metasomatic rocks (I). During slow slip, fluid pressure increase (II) activates frictional sliding in talc (III) and high strain rates produce local stress amplification in actinolite. During slow slip events, the cessation of deformation allows the growth of static minerals in talc-free rocks (IVa) that are later rotated when fluid pressure drops (IVb and V), and aseismic creep resumes (I). Episodic deformation leads to the development of complex microstructures characterized by multiple overprinting fabrics in chlorite-actinolite schist (VI).

Conclusions
We document evidence of stress and strain rate variations in actinolite microstructures from subduction interface metasomatic rocks exhumed from the pressure-temperature conditions of modern episodic tremor and slow slip. Evidence for dislocation creep ± cataclasis in the talc-bearing rock and dissolution-reprecipitation creep in the talc-free rock reflects stress amplification in the talc-bearing sample during high strain rate deformation of the surrounding talc. These higher strain rates are interpreted to reflect episodic slow slip events in the talc-bearing sample with lower strain rates in the talc-free sample during intervening aseismic creep. This work demon strates the importance of considering metasomatism in studying subduction zone seismicity and its likely role in episodic tremor and slow slip.

Data Availability Statement
Plotted microstructural and geochemical data referenced in this paper are available in the Supplementary Information and the full dataset is available at https://doi.org/10.5281/zenodo.7236374. Data was processed using the open-source MTEX (v. 5.7.0) MATLAB toolbox.