Faults represent a critical heterogeneity in basaltic sequences, yet few studies have focused on their architectural and hydromechanical evolution. We present a detailed, multi-scale characterization of passively exhumed fault zones from the layered basalts of the Faroe Islands, which reveals cyclic stages of fault evolution. Outcrop-scale structures and fault rock distribution within the fault zones were mapped in the field and in 3-D virtual outcrop models, with detailed characterization of fault rock microstructure obtained from optical and scanning electron microscopy. The fault zones record deformation localization from decameter-wide Riedel shear zones into meter-wide fault cores that contain multiple cataclastic shear bands and low-strain lenses organized around a central slip zone. Shear bands and the slip zone consist of (ultra-) cataclasites with a zeolite-smectite assemblage replacing the original plagioclase-pyroxene host rock composition. Low-strain lenses are breccias of weakly altered host rock or reworked fault rocks. Slip zone-proximal zones show significant late-stage dilatation in the form of hydrothermal breccias or tabular veins with up to decimeter apertures. We interpret these structures as evolving from alternating shear-compaction and dilation through hydrofracture. The fault core preserves slip zone reworking, which is interpreted to indicate repeated shear zone locking and migration. The alternating deformation styles of shear-compaction and dilatation suggest episodic changes in deformation mechanisms driven by transient overpressure and release. The fault zone mechanical properties are thus governed by the combined effects of permanent chemical weakening and transient fluid-mediated mechanical weakening, alternating with cementation and healing. We suggest that the model presented for fault evolution should apply widely to shallow, basalt-hosted fault zones.
Faults represent a critical heterogeneity in basaltic sequences, yet few studies have detailed their architectural and hydromechanical evolution. Much work has been dedicated to understanding the evolution of the surface expression of fault zones in basalts, with the main focus on fissure-type surface-breaching faults (e.g., Holland et al., 2006; Bubeck et al., 2018; von Hagke et al., 2019) and/or on fault growth in units with a strong mechanical anisotropy such as cooling joints (e.g., Walker et al., 2012; Soden and Shipton, 2013; Crider, 2015; Kettermann et al., 2019; Weismüller et al., 2019). Relatively few studies have focused on the internal architecture and evolution of fault rock development in basaltic systems at depth. In the subsurface, basalts undergo pervasive fluid-mediated alteration during fault evolution (Kristmannsdóttir, 1979; Walker et al., 2013a), with the new mineral assemblage drastically changing bulk-rock mechanical properties, such as replacement of frictionally strong feldspar-dominated assemblages by relatively weak zeolites (Yukselen-Aksoy, 2010; Frolova et al., 2014) or clay minerals (Shimamoto and Logan, 1981; Haines et al., 2013; Boulton et al., 2014; Carpenter et al., 2016). These mineralogic changes not only affect the mechanical structure but also the permeability structure of the fault (Walker et al., 2013a, 2013b), which may also affect fault strength and stability.
Although some studies have characterized the evolution and deformation mechanisms affecting faults in basalts that operate at small displacement scales (millimeter-to decameter-scale displacement: Walker et al., 2012, 2013a, 2013b) and initially described fault displacement versus fault-damage thickness scaling, these studies have not analyzed in detail the deformation mechanisms responsible for slip zone and fault zone growth at larger displacement scales, nor the processes that may lead to fault core thickening.
In this study, we seek to characterize the relationship between fluids and fault zone evolution in basaltic rocks based on exceptionally well-exposed field sites in the Faroe Islands, on the NE Atlantic Ocean margin. We focus on three exhumed, large-displacement (decameter-to potentially kilometer-scale) fault zones that were active at ~1–2 km depth (Fig. 1). We studied fault architecture in the field and in virtual outcrop models acquired using unoccupied airborne vehicle (UAV)-derived photogrammetric models, and microstructures were observed using petrographic microscopes and scanning electron microscope–backscattered electron (SEM-BSE) imaging. Mineral identification is based on X-ray diffraction (XRD) and scanning electron microscope–energy dispersive spectroscopy (SEM-EDS) in combination (data in Supplemental Material1) with petrographic microscopy. From these observations, we identify deformation mechanisms within a framework of cyclic fault zone evolution involving cataclastic deformation punctuated by episodic hydrofracture restrengthening and slip zone migration. Our results and conceptual model may have implications for, and application to, the mechanical and permeability evolution of large-displacement faults in basaltic rocks more broadly.
Fault zones that have accommodated multiple slip events typically display a complex internal architecture featuring weakly deformed fault rock lenses hosted in anastomosing, cataclastic shear bands (Faulkner et al., 2003). This anastomosing macrostructure can arise from shear zone locking and slip migration as a consequence of hardening caused by strain or mineralization (Faulkner et al., 2003; Caine et al., 2010; Callahan et al., 2020). Other faults show evidence for periodic roughening through slip zone segmentation, followed by progressive smoothing through frictional wear (Sagy and Brodsky, 2009; Brodsky et al., 2011). Alternatively, large-scale roughening could be the result of fault linkage and the formation of subsidiary faults (Childs et al., 1996; Candela et al., 2009; Candela and Renard, 2012). All of these models have a commonality in that they are characterized by episodic delocalization and fault zone dilatation followed by failure and frictional wear, which can be preserved in discrete fault rock assemblages. The spatial distribution of fault rock types is important to the present-day hydrologic properties of individual fault zones, which can be calculated in situ, whereas the mechanism of fault slip and the processes of fault growth have important implications for the fluid flow history, and previous transient fluid flow properties, of the fault.
Fault slip is recognized as a trigger for fluid pressure transients (e.g., Wibberley et al., 2008; Faulkner et al., 2010; Bense et al., 2013), which may destabilize the fault and lead to extensive failure (Byerlee, 1993). Equally, a progressive decrease in permeability and fluid pressurization during creep can promote fault slip. Recent studies by Callahan et al. (2020) and Proctor et al. (2020) highlight how fluid flow can become the dominant driver of fault evolution and stability as a complication beyond slip zone geometry or rate-and-state friction effects. For example, fluids can exert a direct control on stability by modulating the effective stress field acting on the fault zone, as has been well established elsewhere (e.g., Byerlee, 1990; Sleep and Blanpied, 1992; Ougier-Simonin and Zhu, 2015; Faulkner et al., 2018; Proctor et al., 2020). At the macroscopic scale, faults can trap fluids and maintain near lithostatic internal fluid pressure throughout the entire brittle domain (Byerlee, 1990). In complex fault zones, heterogenous permeability distribution may lead to fluid pressure compartmentalization within the fault zone (provided the strength of the seal is not exceeded), making them more or less prone to failure (Byerlee, 1993). Localized failure of one compartment may result in propagation of the fluid pressure transient through neighboring compartments, causing runaway failure of the fault, as well as allowing intermittent fluid flow through the fault zone in a process known as fault valve behavior (Sibson, 1990; Byerlee, 1993). On a smaller scale, low-permeability granular rocks such as cataclasites and gouges can experience peaks in fluid pressure during shear as they cannot drain fast enough to compensate for internal porosity destruction during deformation. The transient increase in fluid pressure that ensues can lead to dynamic weakening (Ikari et al., 2009; Faulkner et al., 2018).
Alternatively, fluids may contribute to fault strengthening by catalyzing the dissolution-precipitation reactions of fault healing (Gratier, 2011). In natural fault zones, these processes lead to cyclic activity that is recorded in overprinting generations of fault rocks. The faults commonly have a high inherent structural heterogeneity with anastomosing slip zones enclosing breccia lenses, which promotes compartmentalization (Faulkner et al., 2003; Caine et al., 2010). Fluid pressure transients trigger failure and slip (Sutherland et al., 2012), followed by dilatancy and mineralization (Caine et al., 2010; Sutherland et al., 2012; Callahan et al., 2020). Restrengthening through cementation may facilitate slip migration into adjacent, relatively weaker structures but also primes the fault for a subsequent failure (Caine et al., 2010; Gratier, 2011; Sutherland et al., 2012; Callahan et al., 2020).
Geology and Faulting in the Faroe Islands
The Faroe Islands are an archipelago situated between the Shetland Islands and Iceland, on the European continental shelf of the NE Atlantic Ocean (Fig. 1A). The Faroe Islands are a subaerial remnant of the Paleogene North Atlantic large igneous province that resulted from the initial outburst of the Icelandic plume before the onset of Atlantic rifting (Ritchie and Hitchen, 1996). The islands are composed mainly of basaltic rocks of the Faroe Islands Basalt Group, which extends for over 200 km to the E and SE of the archipelago (Passey and Bell, 2007; Passey and Jolley, 2008). The Faroe Islands Basalt Group has a maximum stratigraphic thickness of ~6.6 km, which is inferred from the slightly dipping surface exposure and the Lopra-1/1A borehole on Suðuroy (Waagstein et al., 1984; Chalmers and Waagstein, 2006; Passey and Bell, 2007). However, more recent seismic studies point toward a lower maximum vertical thickness of only ~3–4 km under the islands (Ólavsdóttir et al., 2017). The Faroe Islands Basalt Group is subdivided into seven formations that record several periods of volcanic activity with a transition from submarine hyaloclastites to subaerial lavas, and interspersed with volcaniclastic deposits recording periods of volcanic quiescence (Passey and Jolley, 2008). A detailed characterization of the stratigraphy of the Faroe Islands is provided by Passey and Jolley (2008). Initiation of volcanic activity is placed between 60.1 Ma and 55.8 Ma (Hansen et al., 2002; Waagstein et al., 2002; Storey et al., 2007; Passey and Jolley, 2008), with volcanic cessation before 54.5 Ma (measured in correlated units in Greenland and the Faroe-Shetland Basin: Larsen et al., 1999; Hansen et al., 2002; Larsen and Tegner, 2006).
The structural evolution of the Faroe Islands is dominantly influenced by the opening of the NE Atlantic Ocean. Pre-breakup to post-breakup deformation was characterized previously as a sequence separated into six tectonic stages that are recorded by cross-cutting brittle structures on the islands, and which accommodated a progressive vertical axis rotation of the horizontal extension vector (Walker et al., 2011). This stress/strain rotation is interpreted to represent the growth of a continental relay zone between the NE-advancing Reykjanes and SW-advancing Aegir Ridges, with an eventual switch to shortening as a result of abandonment of the Aegir Ridge in favor of the Kolbeinsey Ridge north of Jan Mayen (Gaina et al., 2009; Walker et al., 2011; Gernigon et al., 2012; Ellis and Stoker, 2014; Bubeck et al., 2017). Deformation was accommodated initially by dip-slip faults and conjugate dykes accommodating NE–SW extension (tectonic stages 1–2, respectively) and then by conjugate dykes and strike-slip faults accommodating N–S extension (tectonic stages 3–4, respectively), which was followed by NW–SE extension on strike-oblique slip faults (tectonic stage 5; Walker et al., 2011). Late-stage overprinting deformation (tectonic stage 6) is also evidenced by the reactivation of tectonic stage 1–5 structures, perhaps as a result of post-breakup passive margin uplift (Walker et al., 2011). Early deformation overlaps with the last stages of basalt emplacement as continental rifting began around 55–53 Ma, and tectonic activity continued at least into the mid Miocene (Roberts and Walker, 2016; Ólavsdóttir et al., 2017).
In many cases, faults and dykes in the Faroe Islands are expressed by deeply incised gullies. Most notably, this includes the multiple NW–SE-to N–S-trending fjords that separate each of the northern islands, which mark the positions of tectonic stage 1–2 faults and dykes (e.g., the Brynhild and Westray fault zones shown in Fig. 1B; Ellis et al., 2009). Tectonic stage 2 faults and dykes also feature prominently in the topography and bathymetry of the Faroe Islands, such as the ESE–WNW-striking Skopanafjorður fault (Fig. 1B; Passey, 2009), which separates the northern and southern islands, and an unnamed ENE–WSW structure that caps the northern extremity of the archipelago (Walker et al., 2011). We focus here on three fault zones in the Faroe Islands (Fig. 1B) that are inferred to represent three different displacement scales, from tens-of-meter–scale to potentially kilometer-scale displacement.
The inferred smallest displacement fault in the study is located at Í Botni on the southern island of Suðuroy (Fig. 1B). The 30 m displacement at Í Botni can be measured directly from the offset of stratigraphic horizons, and accounts for the plunge of slickenlines on the surfaces of exposed slip zones. Í Botni is a N–S-striking fault associated with tectonic stage 1 (Walker et al., 2011) and is hosted in the extensive sheet lobes of the Beinisvørð Formation.
The inferred middle-displacement fault is at Gøtugjógv, Eysturoy (Fig. 1B), an ESE–WNW-striking sinistral strike-slip fault associated with tectonic stage 4. The Gøtugjógv fault zone forms a prominent lineament that can be traced across the island of Eysturoy and westwards across the island of Streymoy. The fjord between these two islands marks the position of the Brynhild fault, which is offset across the trace of the Gøtugjógv fault (Fig. 1C), from which we estimate a sinistral displacement of up to ~500 m. At the position of exposure studied here, the Gøtugjógv fault is hosted in thin compound lavas of the Malinstindur Formation.
The inferred largest fault is an E–W-to ESE–WNW-striking fault zone at Selatrað, Eysturoy, which is also associated with tectonic stage 4. As with the Gøtugjógv fault, the Selatrað fault is associated with an offset of the bathymetric lineament of the Brynhild fault, between Eysturoy and Streymoy, from which we estimate a sinistral displacement of up to ~1.5 km. This would make the Selatrað fault one of the largest faults exposed on the islands. At the present level of outcrop, the Selatrað fault is hosted in thin compound lavas of the Malinstindur Formation and thick simple lavas of the Enni Formation.
Each of the three fault zones exhibit large amounts of smectite and heulandite with minor analcime and chlorite within the fault rock, which places them at a likely depth of around 1–2 km while they were active (Frey et al., 1991; Neuhoff et al., 2000; Jørgensen, 2006). The timing of activity for tectonic stage 4 faults (e.g., Selatrað) is constrained by U–Pb dating of calcite cements (Roberts and Walker, 2016), which returned mid Eocene dates (44.8–40.1 Ma). A tentative, and significantly younger, early Miocene (19.4 ± 3.0 Ma) measurement for Gøtugjógv (see supplementary data from Roberts and Walker, 2016) could indicate either long-term permeability preservation or hint at long-lived and/or periodic fault activity. Tectonic stage 1 faults (including Í Botni) did not return U–Pb ages, but stratigraphic thickness variations across the fault zones indicate syn-volcanic deformation (Passey and Bell, 2007; Ellis et al., 2009; Walker et al., 2011).
FAULT ZONE ARCHITECTURE
The fault zone at Í Botni (IBO, Fig. 2) accommodates ~30 m of E-down displacement and is composed of multiple arrays of steep (60–80°) faults distributed over a ~50-m-thick zone. As highlighted by a red clay marker horizon (blue line in Fig. 2C), the majority of the displacement is accommodated across the easternmost fault strand. The steep topography of the outcrop and its proximity to the shoreline renders most of the fault zone inaccessible. Fortunately, the lower few tens of meters of the easternmost fault core have some accessible sections. Here, the fault core is ≤50 cm thick and consists of gray to light brown, chaotic breccia, gouge, and some cataclasite (fault rock nomenclature is from Woodcock and Mort, 2008) bounded by striated and polished black slip surfaces (Fig. 2D). These structures are bound by similar but cohesive fine-grained, chaotic breccia and cataclasite, which grade to coarser breccias and eventually undeformed host rock to either side.
The Gøtugjógv fault zone (GOT, Fig. 3) is exposed in a 50-m-long and up to 10-m-high rock face that reveals a section through the 2–3-m-thick fault core, from the hanging wall damage zone in the west to the footwall in the east (Fig. 3). The exposed damage zone consists of crackle breccia with clasts that gradually coarsen away from the core and is ~10 m thick in the hanging wall and 3 m thick in the footwall. Veins or cement in the damage zone are rare and limited to thin zeolite veins and calcite or zeolite mineralization on slip surfaces. As with the damage zone, the ~3-m-thick core (comprised of multiple slip zones) has an asymmetric structure in the exposed section. The most prominent slip zone cuts through the fault core in direct contact with the hanging wall damage zone (Figs. 3C and 4B) and consists of foliated and unfoliated cataclasites hosting a smooth slip surface with an average orientation of 194/56° (dip azimuth/dip angle) and slickenlines plunging 104/08° (azimuth/plunge angle) (Fig. 4E). This slip surface is corrugated at the meter scale, with corrugations exposed over the 10-m-high outcrop. As indicated in Figure 4, the corrugations represent individual lenses of fault rock separated by networks of thin cataclastic shear bands and/or discrete slip surfaces. Lenses contain either breccias or cataclasite, the latter being more prevalent along the prominent slip surface. The transition from breccias and coarse protocataclasites to highly comminuted (ultra-) cataclasites is usually distinct (across a surface) but in places is also gradual. Farther away from the slip surface, the fault rock lenses increase in size and potentially cover the entire exposed rock face. Thicknesses of cataclastic shear bands range within the centimeter scale. The core also hosts abundant zeolite and calcite veins, with tabular veins along cataclastic shear bands, either following the contacts or within the bands. Veins are, in places, stacked as 5–10-cm-thick composite structures, especially closer to the footwall cut-off (Fig. 3D), or as swarms of millimeter-aperture veins. The thick composite veins feature well-developed slickensides between individual veins and on their contacts with the wall rock (Fig. 3D). Tabular veins are dominantly parallel, or orthogonal, to the main fault orientation.
The Selatrað fault zone (SEL) is exposed in a <30-m-deep and 600-m-long gorge separated into two tiers—deeper in the west and shallower in the east—by a waterfall ~200 m from the western end. Most of the ~7-m-thick fault core lies underneath the river, carving the gorge. Several anastomosing slip zones enclosing breccia lenses are exposed (Fig. 5D). The most prominent slip zone dips northward and can be traced along the southern edge of the gorge (i.e., the footwall) in the upper eastern tier of the gorge and then curves toward the northern edge (i.e., the hanging wall) as the gorge deepens into the lower western tier at the waterfall seen in Figure 5E. This very well-defined smooth surface exhibits centimeter-to decimeter-wavelength, subhorizontal slickenlines along its entire exposure, which represents a slip zone that can be traced along much of the gorge. On the upper tier, the slip surface bounds a 5–10-cm-thick ultracataclasite. In the lower tier exposures, the slip surface either bounds a similar ultracataclasite layer or a coarser and less cohesive cataclasite that grades into a chaotic breccia a few centimeters away from the slip surface. An anastomosing network of large secondary faults runs parallel to the slip zone and compartmentalizes the fault core into a network of fault-bounded lenses (Figs. 5E–5F). These lenses are much less altered (e.g., some plagioclase is preserved, as identified in optical microscopy) and less fragmented than the bounding cataclastic shear bands. Figure 5F is a map view of one of these lenses and shows internal fragmentation by fractures in Riedel configuration, dominated by R shears (see Fig. 6 for illustration of Riedel shears). Similarly, large secondary faults are abundant in the damage zone and are usually defined by a distinct and continuous corrugated slip surface sandwiched between a centimeter-thick layer of cataclasite to either side. Veins and cements are abundant in the core but virtually absent from the damage zone. The average main fault orientation measured along the gorge floor is 000/71° with slickenlines plunging 088/08°. However, the slightly less steep topographic orientation of the gorge (~000/60°) may be more indicative of the overall fault zone orientation.
Fault and Fracture Orientation
Different measuring techniques were required to extract fault orientation data at different scales, including: in the field, virtual outcrop models, and orthophotos. This approach could introduce scale- and method-dependent biases; hence, we only compare the three fault zones qualitatively. In Gøtugjógv and Selatrað, fault orientation was measured in the field and reconstructed from the UAV-derived, 3-D virtual outcrop models (Fig. 7). The orientation of an assumed uniform fault plane can be estimated from its intersection with the topography of the 3-D outcrop surface. For this, fault traces were picked in 3-D photogrammetric models of the fault zones, and a plane was fitted to these traces so that it would contain a maximum of vertices from the fault trace line selected. This method incurs uncertainty because: (1) the intersected rock faces generally have little relief, and fitting planes to fault traces is therefore dependent on the accuracy of the modeled surface and the position of the picked lineament; and (2) minor displacement (millimeter–centimeter-scale) fault surfaces may be composite structures that are below the resolution of the models. For instance, Walker et al. (2012) showed that surfaces of millimeter–centimeter-scale displacement faults associated with tectonic stage 4 comprised linked faults and fractures that ranged in strike from ~060°–115° on a fault with an overall strike of ~060° (e.g., their figure 4). We also report a lineament analysis on orthophotos of a close-up of the Selatrað gully floor (Figs. 5F–5G), which is almost flat; hence, we are confident that fault traces are representative of fault strike and accurate to within ~5°.
The large-scale structure of the fault core and damage zone is strongly dominated by faults in Y to R orientation (see Fig. 6 for an illustration of Riedel shears) as well as subordinate P faults (Fig. 7). High-angle shears (R′ and X) are much less common and, in the case of the Selatrað fault core (Fig. 5F), occur mostly as relatively late-stage structures that either fragment large clasts bounded by low-angle faults or transect all other structures. We see similar high-angle shears transecting the Gøtugjógv slip zone. This pattern is repeated on a smaller scale in centimeter-to decimeter-sized hand specimens, especially in foliated cataclasite (Fig. 8B). Here, R and P shears define a well-developed foliation bounded by fault-parallel Y shears separating different types of fault rock (e.g., ultracataclasite from cataclasite). Late-stage, high-angle R′- and X-parallel shears cut through these boundaries as well as the foliation. Notably, the sense of slip along some of these high-angle shears is opposite to the sense of slip that would be expected for R and X shears developed in simple shear (Fig. 6), which possibly indicates a pure shear (i.e., flattening) component of deformation.
FAULT ROCK CHARACTERIZATION
Damage concentration changes drastically over a very narrow zone in the ~30 m displacement fault zone of Í Botni. The slip zone consists of a centimeter-wide swarm of anastomosing, sub-millimeter thick, black, polished slip surfaces with clayey cataclasite in between (Fig. 2D). The slip zone is directly adjacent to crackle and chaotic breccias that comprise largely unaltered basalt clasts in an authigenic cataclastic matrix (Fig. 8A; Fig. 9, IBO i–ii; and Fig. 10A) on one side, and cataclasite or gouge with few relict clasts on the other side (Fig. 9, IBO iv). XRD analyses (Supplemental Material, see footnote 1) show plagioclase and pyroxene to be preserved throughout most of the fault zone.
Breccias and Veins
Breccia veins in the Gøtugjógv fault core are mostly composed of angular and poorly sorted clasts suspended in a mineral cement of either calcite or, more commonly, zeolite (Figs. 8E–8G; Fig. 9, GOT iv–x; and Fig. 10E). Clast concentration in these breccia veins varies considerably, from domains that are potentially clast-supported to domains that are pure mineral vein (i.e., without clasts, Fig. 11). Angular clasts ranging from sub-millimeter–to centimeter-scale are hosted in a cement of <50-μm-long acicular to bladed zeolite, or equant calcite, measuring a few millimeters across. The zeolite cement has a weak, apparent shape-preferred orientation (SPO) that forms a cross-shaped texture with axes at a 45° angle to the vein boundaries (Fig. 10E). A similar apparent SPO can be seen in the matrix of some cataclasites (Fig. 10F and paragraph below). The veins also contain numerous millimeter-sized pockets of irregular shape, with idiomorphic zeolite crystals protruding from the cement, and commonly filled with a single crystal of calcite (Figs. 9, GOT viii and x, and 10E). Similarly, clasts of wall rock in calcite-cemented breccias are commonly lined with syntaxial idiomorphic zeolite, and the cement contains millimeter-sized pockets of zeolite. The clast population includes altered host rock (Fig. 9, GOT iv–v), reworked fault rock (Fig. 9, GOT ix–x), and in some cases larger fragments of older veins or cement (Fig. 9, GOT vii).
Similarly, small, non-tabular veins can be observed in the domains farther away from the most prominent slip zone and especially in thicker cataclasite bodies. These zones are predominantly cemented by syntaxial bladed or blocky zeolite with a wide range of crystal sizes, and equant calcite mineralization is also common. Their shape and orientation is irregular, frequently splaying, merging, and diverging, with variable aperture and orientation, and the wider segments host small fragments of wall rock (Figs. 8G and 9, GOT xi). We interpret the clast angularity paired with the dominance of mineral cement over matrix as strong evidence for breccia veins originating from hydrofracture and implosion (see Jébrak, 1997).
Some outcrop-scale breccia pods are characterized by a more chaotic texture than the previously described breccia veins. Their clasts are sub-angular to rounded with less distinct edges. Clast concentration varies greatly, and they are hosted in a mixture of zeolite and clays. Clay matrix concentration in the cement is high but varies across centimeter-scale patches (Figs. 8D and 9, GOT vi).
Many areas, especially along the contacts of different fault rock units, host centimeter-wide, multiphase swarms of thin zeolite and calcite veins. Veins appear to exploit the vein walls or wall rock directly adjacent to previous veins rather than forming within each other, though examples of cross-cutting veins are observed. Veins and breccia cements record either extensive or no internal strain from re-shearing of the fault, as evidenced by twinning, inter- and intracrystalline cracking, and grain-size reduction. Where interfaces host slickensides, the outer few millimeters of cement are comminuted and separated from the unstrained inner layer by Y shears, as seen in Figure 9, GOT viii.
The textural transition from breccia to cataclasite and ultracataclasite is gradual (Fig. 11). On the coarse end of the scale, cataclasites resemble the unorganized breccias described above. They have sub-angular to rounded clasts of fault rock and altered host rock hosted in a comminuted and slightly altered authigenic matrix with sparse, unconnected, and irregularly shaped patches of cement (Figs. 8B and 9, GOT ii). Finer cataclasites host fewer clasts, and the composition of those clasts is dominated by polycrystalline zeolite aggregates without clay, in a matrix of homogeneous zeolite, pyroxene, and clay (Figs. 8C and 8G; Figs. 9, GOTs iii and xi; and Figs. 10D, 10G, and 10H). In some cases, the zeolites have an acicular habit that defines a lattice-shaped apparent SPO inclined to the slip plane, similar to zeolite cement in veins (Fig. 10F). In the hanging wall slip zone, the clay minerals show a distinct foliation. Mesoscopic foliation is defined by millimeter-thick clay bands parallel to the R, P, and Y orientations (Figs. 8B and 9, GOT ii) and only developed in the coarser cataclasite farther from the slip surface. Microscopic foliation is defined by very thin and continuous clay bands at a higher angle to the slip surface (40–55°), and the foliation is only developed in the ultracataclasite (Figs. 10G–10H). Late-stage, high-angle fractures (red in Fig. 8B) commonly transect and offset all other structures, layer boundaries (i.e., ultracataclasite to cataclasite), foliation, and other shear bands/fractures. They are parallel to the R′ orientation, but their sense of shear can be both synthetic and antithetic, which we interpret as a combination of potential compactional/pure shear (synthetic) and rotational/simple shear (antithetic) components of deformation (Fig. 6). Further, SEM-BSE imaging shows that the cataclasites maintain a significant porosity even where highly strained (Fig. 10I).
The sampled ultracataclasite that constitutes the slip surface (Fig. 9, SEL iii) is texturally very similar to the abandoned red clay ultracataclasite from the Gøtugjógv fault core (Fig. 9, GOT xi). It has not been brecciated, nor cut by zeolite veins, but the texture is nearly identical and shows a red clay-dominated matrix with very fine-grained pyroxene fragments and larger polymineralic clasts. Many of these clasts contain calcite, and calcite vein fragments can be identified, indicating prior stages of veining within the fault. The weakly consolidated cataclasite that can be found along parts of the well-exposed slip surface in the lower tier is dominated by a red clay matrix and contains polymineralic clasts that can be up to 5 cm across but are more commonly sized at the millimeter scale. Clast concentration gradually increases away from the slip surface, and the rock can be classified as mosaic breccia ~30 cm from the slip surface. Consequently, the internal structure of fault rock lenses is dominated by mosaic and crackle breccias with a systematic fragmentation along R to R′.
As mentioned above, secondary faults in the damage zone are separated by cataclasite layers on either side. The cataclasites are usually only a few centimeters wide, even for extensive and polished slip surfaces. Boundaries along the undeformed wall rock can be sharp or gradual. The cataclasite layer is strongly altered. In the case of the fault zone-oblique narrow cataclastic band in Figure 9 (SEL i), the assemblage has completely changed to zeolite and remnant pyroxene, with an undulating boundary back to unaltered wall rock expressed by linkage of the R and Y shear planes. The fault zone-parallel and slightly wider slip zone in Figure 9 (SEL ii) grades from a zeolite- and clay-dominated cataclasite on the left through a brecciated zone with angular and weakly altered wall rock clasts in a zeolite cement. Throughout the damage zone, basalt wall rocks show only minor or no alteration.
Fluid Flow and Chemical Alteration
Pervasive alteration in the fault zone is limited to fractured rocks (Fig. 9, IBO i–ii, SEL i, GOT i, and Fig. 10A), which we infer is the result of damage-enhanced fluid flow. The fault rocks from the low displacement fault in Í Botni preserve an original plagioclase-pyroxene assemblage in most samples, with the only exception being the heavily altered and reworked footwall breccia (Fig. 9, IBO iii). Hanging wall breccias in direct contact with the slip surface show alteration penetrating no more than 0.1–0.5 mm into clasts (Fig. 9, IBO i–ii). Likewise, in the Gøtugjógv and Selatrað damage zones, alteration is limited to fracture and vein walls (Fig. 9, GOT i, GOT xii, SEL i, and Fig. 10B), which may reflect the very low permeability (<10−20 m2) of intact Faroe Islands basalts (Walker et al., 2013a, 2013b). This is contrasted by the complete textural and mineralogical breakdown observed in the Gøtugjógv and Selatrað fault cores, which are composed of a clay-zeolite assemblage with remnant pyroxene (e.g., Fig. 9, GOT ii, GOT vi, GOT xi, and SEL iii). Wall-rock clasts entrained in the core consistently show strong alteration, even though primary texture may be preserved (Fig. 9, GOT iv, and Fig. 10C). The inferred alteration sequence is in line with permeability variation measured along basaltic faults in the Faroe Islands. Walker et al. (2013a, 2013b) recorded elevated permeability in the fault cores, which are enclosed laterally by low-permeability damage zones; this configuration promotes fault-parallel fluid channeling.
Fault core permeability has been shown to be strongly linked to the interplay between comminution and clay generation (permeability reduction) and the establishment of interconnected and porous (zeolite) vein networks and microdamage (permeability increase) (Sutherland et al., 2012; Walker et al., 2013a), but also to local strain concentration and petrographic heterogeneity along the fault (Caine et al., 2010; Sutherland et al., 2012). For example, low-porosity, clay-dominated cataclasite or gouge is probably more readily sealed than zeolite-dominated chaotic breccia (Walker et al., 2013a). The integrity, continuity, and abundance of veins, mineralized breccias, and clay-dominated shear bands varies strongly throughout the mature fault zones in Gøtugjógv and Selatrað, which should lead to significant permeability variations within the fault zones. Fault rock assemblages in the larger displacement faults are lens-shaped and hence have the potential to be fully enclosed by low-permeability, clay-dominated slip systems. This geometry of heterogeneous fault rock could therefore promote compartmentalization and fluid pressure variations along the fault zones, depending on the volume of entrapped fluid and the degree of local compaction (Byerlee, 1993; Caine et al., 2010; Sutherland et al., 2012). Such a configuration should lead to local overpressurization and hydrofracture and could explain the mutual overprinting of structures recording cataclasis and hydrofracture in Gøtugjógv. Hydrofracture is unlikely to happen simultaneously along the entire fault, and therefore it could cause fluid pumping and considerable transient effective stress variations, potentially triggering failure, as noted elsewhere (Byerlee, 1993; Eichhubl and Boles, 2000; Caine et al., 2010; Warren-Smith et al., 2019; Callahan et al., 2020). As with compartmentalization leading to fluid pressurization on a large scale, low-permeability fault rocks can be susceptible to dynamic internal pressurization on a much smaller scale. Their low permeability can prevent fluids from draining from the rocks to equilibrate pressure increases related to porosity destruction during shear-compaction (Sleep and Blanpied, 1992; Ikari et al., 2009; Faulkner et al., 2018). Combined, fluid trapping and pressurization may provide a mechanism for fluid pressure cycling and repeated mineralization events in the fault zones without the need for a maintained external fluid supply for pressure transients. In this scenario, mineral alteration and replacement would be largely driven by fault-internal redistribution, without considerable net loss or gain of chemical species.
Most of the cataclasites studied have a very high zeolite concentration in their matrix as well as abundant zeolite and calcite vein fragments. It might be difficult to account for such high zeolite concentrations from a mass balance perspective without the influx of external fluids. A significant portion of matrix zeolite could derive from secondary mineralization and not just from the reworking of zeolitized host rock. In that case, zeolite would be progressively added to the fault zone, either through precipitation in veins and breccias over several mineralization cycles or by continuous precipitation into microporosity opened during dilatant cataclastic flow (Stel, 1981; Hadizadeh and Foit, 2000). Both processes for secondary mineralization could contribute to the high zeolite concentration and would necessitate an active fluid circulation.
Fault Core Widening
Fault core thickness of the three faults studied here increases from ~0.5–1.0 m in Í Botni (30 m displacement), to 2–3 m in Gøtugjógv (up to ~500 m displacement), and to 7 m in Selatrað (up to ~1.5 km displacement). These values consistently range on the higher end of fault core thickness relative to displacement compared to scaling relationships for other faults (Shipton et al., 2006; McKay et al., 2021). The displacement-thickness ratios (D/T) for faults in this study follow T = 0.1D0.56 compared to T = 0.025D0.65 calculated by McKay et al. (2021). However, it should be noted that (1) our data set is based on only three faults, (2) thicknesses from our study may not represent the maximum thickness, and (3) there is significant scatter in D/T plots based on large data sets. Hence, we do not seek to quantitatively explore this relationship further here. Observations from the three faults in the Faroe Islands do, however, suggest that once fault zones in basalts have accommodated displacements at the tens-of-meter scale (e.g., Í Botni), further displacement may be accommodated by fault core thickening, or by reworking of existing fault rock, potentially involving no thickness change or even a reduction in thickness. Individual slip zones are a relatively consistent thickness (5–10 cm). In contrast, the preserved fault core thickness at a given locality is expected to be highly variable, dependent on several factors including (but not limited to): (1) the number of slip events recorded; (2) the position of those events relative to previous events; (3) the nature of those slip events (shear zone thickening or thinning); and (4) fluid-mediated dilatation, alteration, and mineralization. The three faults studied here show a complex evolution for individual faults but are controlled by common processes that apparently become dominant beyond tens-of-meter–scale displacements.
The complex internal geometry, with strong strain partitioning between brecciated lenses and cataclastic shear bands, suggests repeated assimilation of less deformed wall rock into the fault core. Callahan et al. (2020) recently proposed, as widening processes for the Dixie Comstock fault, combined fluid-driven cyclic dilatation and mineralization, alternating with embrittlement and shear deformation, as well as damage zone assimilation during fault core stepping. This process has also been inferred for the Stillwater fault, which is part of the same system (Caine et al., 2010). We see evidence of both processes in the Faroe Islands fault zones. For instance, a mineralization-deformation cycle is supported by thick veins and widespread mineralization in the Gøtugjógv core and abundant fragments of reworked older veins in fault rocks. Wall-rock ablation may be recorded by relatively unaltered wall-rock fragments with preserved textures found in low-strain lenses. Following integration into the fault core, these lenses were probably fragmented and assimilated through frictional wear processes (as described in Sagy et al., 2007; Childs et al., 2009; Brodsky et al., 2011). The widespread occurrence of heavily altered fault rock fragments derived from frictional shear within mineralized portions of the Gøtugjógv fault most likely indicates that alteration weakening did not promote the development of a stable permanent slip zone, in contrast to the inferred development of the Dixie Comstock fault (Callahan et al., 2020). There are several potential mechanisms for internally driven fault core migration, which we will discuss in more detail below. However, wall-rock ablation can also be driven externally through the linkage of fault segments and secondary faults, cutting off large lenses of damage zone rock as well as introducing slip surface corrugation of the same geometric scale as the wall-rock lenses (Childs et al., 2009; Candela and Renard, 2012). This process appears to be dominant at earlier stages of fault zone evolution, as seen in the wide and low displacement fault zone at Í Botni.
Slip Zone Locking and Migration
Reworked clasts of cataclasites within the fault cores and the multitude of distinct, variably reworked slip zones require episodic principal slip zone (PSZ) migration. This implies that previously active slip zones became unfavorable for continued slip, despite the development of relatively weak clay assemblages: i.e., they became locked. The fault rock (micro-) structure provides evidence of several processes that might have contributed to slip zone hardening and locking, which are outlined here:
Principal slip zone and slip-surface dissection: High-angle shears dissecting strain-concentrating structures, juxtaposing low-strength and high-strength structures, e.g., clay foliation and zeolite matrix or gouge and cataclasite (Fig. 8B), could increase the bulk strength of the slip zone and result in PSZ locking. Should these high-angle shears form during slip, they would have to accumulate their small displacements at a faster rate than the slip rate accommodated across the slip zone to prevent bypassing or progressive removal and incorporation of any emerging asperities into the adjacent gouge/cataclasite layer. More likely, we infer that the observed high-angle shears developed in one of two scenarios: (1) during post-slip creep, forcing subsequent slip to relocate into another structure, or (2) during fragmentation of an embrittled and abandoned slip zone in the vicinity of a new PSZ. Sagy and Brodsky (2009) identified similar layer-transecting structures as an important process for re-roughening in mature slip zones and proposed that they are a type of boudinage structure emerging from the rheological contrast that developed as a consequence of strain-hardening in the cataclasite/gouge layer. Applied to the structures observed in the Faroe Islands, the Riedel shears would record segmentation into shearband boudins by simple shear, while the opposite-sense shears would record compaction of boudins in pure shear.
Zeolite recrystallization: The zeolite apparent SPO recorded in some cataclasite, which resembles the apparent SPO observed in veins (Figs. 9E–9F), could imply a similar process of formation. It is unlikely that zeolites in the cataclasites could maintain an elongated habit during cataclasis, as their shape would be expected to inhibit grain rotation necessary for granular flow in cataclasite and gouge. The apparent SPO is more likely a result of dynamic or static recrystallization, which could also have affected zeolite veins. Material science studies have shown that many natural and synthetic zeolites readily dissolve and reprecipitate in alkaline conditions under atmospheric pressure and temperatures below ~110 °C (e.g., Ivanova and Knyazeva, 2013). This is similar to the conditions thought necessary to favor zeolite over calcite precipitation in Icelandic basalts (Neuhoff et al., 2000). If reactivity in the PSZ is high enough for recrystallization to outpace comminution, it is also likely to allow for healing of the slip surface during aseismic creep (Renard et al., 2012), leading to overall strengthening of the slip system. However, the observed apparent SPO could also be a result of post-shear static recrystallization; in which case, it would only contribute to strength recovery after a slip event.
Cementation: The prevalence of fresh and reworked veins, hydraulic breccias, and the almost bimodal strain distribution in cements—either intact or highly strained—evidence periodic fluid overpressure and mineralization in the fault core. Cyclic fluid pressurization and hydrofracture is commonly observed in both seismogenic (e.g., Byerlee, 1993; Eichhubl and Boles, 2000; Caine et al., 2010; Callahan et al., 2020) as well as slow slip faults (Warren-Smith et al., 2019). As noted above, permeability can drop significantly within high-strain gouge and cataclasite, especially in clay-rich assemblages, through porosity compaction. To allow for repeated pressurization, porosity could be reset by fluidization during high-velocity slip events (Sleep and Blanpied, 1992; Ikari et al., 2009; Faulkner et al., 2018). While we observed no strong evidence for gouge fluidization (Monzawa and Otsuki, 2003; Otsuki et al., 2003), the reworking of fault rocks by granular flow after PSZ migration may offer a viable alternative process for reintroducing porosity. In any case, cataclasites clearly show open pore space in SEM-BSE images (Fig. 10I) as well as fabric elements most likely recording compaction (e.g., Fig. 8B). It is evident that fluids in the PSZ were overpressured and led to extensive hydrofracture. This might have caused a transient decrease of frictional strength, triggering rapid and potentially seismogenic slip (Byerlee, 1993; Faulkner et al., 2018). Ensuing precipitation of zeolites and calcite could restrengthen the slip zone and potentially shut it down (Caine et al., 2010; Wästeby et al., 2014; Callahan et al., 2020). Even though recent thermodynamic calculations (Williams, 2019) show that pressure-boiling induced by dilation is an unviable process for rapid mineral precipitation, post-earthquake hydrochemical recovery times of 8–10 years measured in Iceland (Wästeby et al., 2014) indicate that rapid precipitation sealing must be accommodated by other, unknown processes. This is in line with our observation of highly angular clasts in the breccia veins, which could only be preserved if transport-related wear were minimal.
Strain-hardening: Smectite-rich gouges have been shown to be dominantly strain-hardening (Saffer and Marone, 2003; Tembe et al., 2010; Boulton et al., 2014; Morrow et al., 2017), which could also significantly contribute to the cyclicity of fault rock production and PSZ migration. Shutdown and migration of cataclastic slip zones as a result of strain-hardening has been proposed, for example, in the Carboneras fault zone in Spain (Faulkner et al., 2003) as well as the Flowers Pit fault zone in the USA (Sagy and Brodsky, 2009; Brodsky et al., 2011). In this scenario, progressive strain wears down asperities, producing fine-grained material and thickening the cataclastic layer. Smoothing of the slip surface, lubrication by gouge, and strain-hardening in the cataclastic layer lead to slip localization onto a principal slip surface. Localization is counteracted by removal of asperities and boudinage-like transection of the embrittled cataclasite layer, deforming and re-roughening the adjacent slip surface and thus allowing continuous reworking and generation of gouge/cataclasite (Candela et al., 2009; Sagy and Brodsky, 2009; Brodsky et al., 2011). Such a process could explain gouge/cataclasite production in fault sections without extensive mineralization, such as those exposed in Selatrað.
Fault strength is governed by the interplay between fault structure and rock mechanics, chemical influences, and episodic effective stress variations (Faulkner et al., 2010). Local structure is a first-order control as the distribution and geometry of slip surfaces and fault rocks influence strain partitioning within the fault zone (e.g., Moore and Byerlee, 1992; Faulkner et al., 2003; Caine et al., 2010; Gratier, 2011; Renard and Candela, 2017). The generation of thick layers of cataclasite around a relatively smooth principal slip surface provides evidence for shear localization within the PSZ. Strain is most likely transferred gradually from distributed granular flow within the cataclasite (or even breccias) in a newly established slip zone and onto a localized smooth slip surface later in the cycle. This would imply a general strain-hardening behavior of the cataclasites (Sagy and Brodsky, 2009), which is probably due to progressively reduced grain size (Eberhardt et al., 1999) and increasing clay concentration (Tembe et al., 2010; Haines et al., 2013).
The chemical control on fault rock strength is defined by the opposing influences of alteration-weakening and cementation-strengthening (Caine et al., 2010; Callahan et al., 2020). The introduction of fluids into the fault zone most likely leads to the observed chemical breakdown of strong plagioclase into clays and zeolite, which is characterized by a significantly reduced mechanical strength (e.g., Saffer and Marone, 2003; Tembe et al., 2010; Yukselen-Aksoy, 2010; Frolova et al., 2014; Morrow et al., 2017) and thus permanently weakens the rock. In the evolved fault rocks described above, the chemical influence on strength is probably linked mainly to episodic mineralization with zeolite and calcite, both in (breccia) veins and disseminated throughout the fault rock. This most likely results in a transient strengthening through the direct effect of cementation (e.g., Caine et al., 2010; Callahan et al., 2020) but could also cause a long-term increase in strength if the overall composition progressively shifts toward more zeolite and calcite relative to clay minerals. Zeolite and calcite veins in the faults studied preferentially exploited pre-existing vein walls or the wall rock directly adjacent to veins. This indicates that cements are stronger than the clay-dominated wall rock or the vein-wall rock interface (Virgo et al., 2013). This is also supported by the presence of well-developed slickensides on the vein interfaces (Fig. 3D), which show that shear preferentially localizes here before disrupting the veins. It has been shown that the overall efficiency of healing processes active on slip surfaces is promoted by the highly reactive minerals on these surfaces, such as zeolites and calcite in our case, and the catalyzing effect of fluids, but is impeded by the less reactive clay minerals (Gratier, 2011; Renard et al., 2012). Analysis of post-seismic fluid chemistry on the analogous Húsavík-Flatey fault in Iceland shows full hydrochemical recovery, which is assumed to mirror fracture healing by mineral precipitation, after ~8–10 years even though it is delayed by minor refracturing events (Wästeby et al., 2014). It should be noted, though, that mechanical recovery by mineralization does not necessarily imply full sealing of fluid pathways (Aben et al., 2017).
Immediately before and during slip, the effect of fluid pressure changes imposed by compaction and dilation can exceed the intrinsic frictional characteristics of the fault rock to become the dominant control on fault stability (Proctor et al., 2020). As described previously, compartmentalization could result in considerable local fluid pressure variation in fault zones. If one such compartment fails, rapid fluid flow may create a fluid pressure transient that would migrate through the fault zone (Byerlee, 1993; Hill et al., 1993; Caine et al., 2010). The associated effective stress drop is recorded by numerous implosion breccias throughout Gøtugjógv, and centimeter-scale calcite crystals in veins, requiring fractures to be maintained open over a substantial period of time, potentially hundreds to thousands of years (Lee and Morse, 1999; Frery et al., 2015). Even in the absence of an external fluid pressure transient, creep of saturated granular fault rock (i.e., gouge or cataclasite) can result in internal overpressure through shear-compaction and ineffective fluid drainage from the low-permeability rocks (Sleep and Blanpied, 1992; Ikari et al., 2009; Faulkner et al., 2018). In both scenarios, the effective pressure reduction entails mechanical weakening (Ikari et al., 2009; Morrow et al., 2017; Faulkner et al., 2018) that might eventually lead to fluid overpressure and hydrofracture. Besides the direct effects on fault shear strength by effective pressure reduction, episodic fluid flow can mediate many of the permanent and transient chemical processes previously described, such as alteration-weakening, mineralization-strengthening, and healing.
Fault Zone Evolution
The three fault zones analyzed share features of fault damage growth and evolution in basaltic rocks that are captured at three different displacement scales. We infer that the main features of these faults can be treated conceptually as progressive stages in the evolution of individual faults in the Faroe Islands basalts in a space-for-time substitution, and displacement ranges from the tens-of-meters scale to potentially hundreds-of-meters-or kilometers scale.
Based on our observations, we propose a model for cyclic fault evolution in which localized cataclastic flow, probably during subseismic slip or creep, alternates with episodic core embrittlement and extensive hydrofracture and cementation. Our model may be subdivided into seven stages (Fig. 12), which are described below. The combined effects of transient changes to permeability and fault strength are illustrated in Figure 13.
Stage 1: Fault zones initiate in arrays of small faults in Riedel configuration—as exemplified in the damage zone of Selatrað (Figs. 3 and 7)—that link to create a network of continuously larger faults localizing strain (Peacock, 1991; Peacock and Sanderson, 1995; Walsh et al., 2003; Childs et al., 2009; Candela and Renard, 2012; Crider, 2015; Rotevatn et al., 2019).
Stage 2: Continued localization results in the formation of a cataclastic PSZ that accommodates the majority of displacement across the fault. Damage concentration around the PSZ leads to the progressive widening of the fault core (as seen in Í Botni, Fig. 2).
Stage 3: Formation of fine-grained fault rocks in the PSZ and pervasively fractured lenses of damage zone rocks, which comprise a host-rock mineralogy dominated by plagioclase and pyroxene, are almost completely replaced by a clay and zeolite assemblage. The timing of fluid-mediated alteration is constrained by the weakly altered Í Botni fault core and local alteration around damage zone fractures in Gøtugjógv and Selatrað (Figs. 9–10). All three faults should show extensively altered damage zones if alteration had preceded damage concentration.
Stage 4: The reduction in grain size and increase in clay content within the fault core reduces permeability and fluid flow within the fault. Compaction-related increases in fluid pressure cannot be drained from the fault core, which leads to hydrofracture, both as discrete veins and pervasive brecciation, followed by rapid mineralization (keeping clasts in suspension; Jébrak, 1997) by zeolite and calcite (e.g., Gøtugjógv; Fig. 9, GOT iv–xi), potentially leading to permanent fault core thickening.
Stage 5: Following extensive mineralization and cementation, renewed slip may relocate into a less cemented domain to form a new PSZ. Refracturing of abandoned older slip zones and subsequent entrainment into the newly forming PSZ results in the reworking of previous generations of fault rock and the formation of clast-within-clast textures (e.g., Figs. 8G and 9, GOT xi).
Stage 6: Fragmentation in the abandoned core follows Riedel orientations, as can be seen in Figure 8B, transecting and offsetting older shear planes. These newly brecciated fault rocks are progressively comminuted and homogenized as they are incorporated into the creeping slip zone.
Stage 7: The slip zone may lock due to a combination of the four processes discussed above: slip zone transection, zeolite recrystallisation, cementation, and strain-hardening. In the fine-grained and homogeneous slip zone, the rock texture resembles that from stage 3 again. Fluid pressure rises, resetting the cycle (stages 3–7).
Implications for Basalt-Hosted Fault Zones
The processes we infer to play a significant role during fault evolution in the Faroe Islands have all been shown to occur commonly in the shallow crust. Basalts typically break down into zeolites and smectites under hydrous conditions (Seyfried and Bischoff, 1979; Utada, 2001; Gysi and Stefánsson, 2012; Frolova et al., 2014). Cataclastic slip zones commonly develop in crystalline rocks of all kinds (Sagy and Brodsky, 2009; Caine et al., 2010) and are prone to episodic locking and reworking (Faulkner et al., 2003; Sagy et al., 2007; Caine et al., 2010; Callahan et al., 2020) as well as strain hardening if they contain smectites (Saffer and Marone, 2003; Tembe et al., 2010; Morrow et al., 2017). Similarly, hydrofracture as a consequence of permeability reduction and fluid trapping has been proposed in several fault zones and settings (Byerlee, 1993; Caine et al., 2010) and can develop intrinsically in fault zones without an external fluid pressure source (Byerlee, 1993; Faulkner et al., 2018; Proctor et al., 2020). Furthermore, the simple geological history of the Faroe Islands precludes any strong influence from structural inheritance or late overprinting of the fault zones (Neuhoff et al., 2000; Walker et al., 2011; Ólavsdóttir et al., 2017). Hence, we propose that our model for fault evolution in the Faroe Islands could apply more widely to shallow, basalt-hosted fault zones in general. Our model expands the displacement scaling for faults in basaltic rocks described in Walker et al. (2012, 2013b) by up to two orders of magnitude and provides new insights into the evolutionary stages and processes that might occur within mature, basalt-hosted fault zones.
Fault zones in the Faroe Islands are interpreted to display repeated principal slip zone migration and fault rock reworking. After an initial phase of permanent chemical weakening promotes strain localization into a principal slip zone, phases of stable cataclastic shear alternate with hydrofracture followed by pervasive cementation of the principal slip zone. The mechanical properties are governed by an interplay between coeval and successive weakening and strengthening processes. Comminution and increasing clay content in the principal slip zone lead to mechanical weakening through fluid pressure increases, while strengthening processes are either mechanical (in the form of strain hardening of the granular fault rocks) or chemical (in the case of fluid-mediated cementation and healing). Mechanical and chemical strengthening culminate in locking of the slip zone and force the fault to reorganize around a weaker structure, restarting a new cycle of cataclastic shear. Fluids play a critical role in the structural and mechanical evolution of these faults, through chemical alteration of the fault rocks, as a transport medium in dissolution-precipitation processes cementing the fault zone, and in modulating effective stress within the fault zone. We propose that similar processes could govern the evolution of shallow-level, basalt-hosted fault zones in general.
This work was funded by the College of Science and Engineering of the University of Leicester, UK. We thank the editors, David Fastovsky and Mike Williams, as well as two anonymous reviewers for their suggestions, which greatly improved this manuscript.