Palaeozoic–Mesozoic brittle normal faults onshore along the SW Barents Sea passive margin off northern Norway give valuable insight into fault and fluid flow processes from the lower brittle crust. Microstructural evidence suggests that Late Permian–Early Triassic faulting took place during multiple phases, with initial fault movement at minimum P–T conditions of c. 300 °C and c. 240 MPa (c. 10 km depth), followed by later fault movement at minimum P–T conditions of c. 275 °C and c. 220 MPa (c. 8.5 km depth). The study shows that pore pressures locally reached lithostatic levels (240 MPa) during faulting and that faulting came to a halt during early (deep) stages of rifting along the margin. Fault permeability has been controlled by healing and precipitation processes through time, which have sealed off the core zone and eventually the damage zones after faulting. A minimum average exhumation rate of c. 40 m Ma−1 since the Late Permian is estimated. It implies that the debated Late Cenozoic uplift of the margin may be explained by increased erosion rates in the coastal regions owing to climate detoriation, which caused subsequent isostatic recalibration and uplift of the marginal crust. The studied faults may be used as analogues of basement-involved fault complexes offshore, revealing details about the offshore nature of faulting, including past and present basement and fault zone permeability.
It is well known that the strength of the crust depends not only on the frictional strength of dry rocks, but also on pore pressure at depth (e.g. Hubbert & Rubey 1959). The consequence is that the brittle strength of the crust can be of the order of several hundred MPa under hydrostatic pore pressure conditions (depending on depth) or negligible as pore pressures approach lithostatic values (Sibson 1973; Byerlee 1978; Brace & Kohlstedt 1980; Jaeger et al. 2009). The importance of this effect on crustal strength was first fully appreciated when attempts to model earthquakes had to involve fluids to replicate the strength of the brittle crust (e.g. Sibson et al. 1975; Sibson 1977, 1989; Byerlee 1990, 1993; Zoback 1991; Rice 1992; Blanpied et al. 1992; Bruhn et al. 1994). These models implied that pore pressures within fault zones were commonly elevated, also at deeper levels of the seismogenic zone.
Several studies have since discovered that faults act as important pathways for fluids in the upper crust as they represent high-permeability zones owing to fracturing and cataclasis (e.g. Sibson 1973, 1977; Hickman et al. 1995; Seront et al. 1998; Gudmundsson et al. 2001; Zoback & Townend 2001; Crampin et al. 2002; Ganerød et al. 2008). Experimental data have shown that fluid-involved processes, which control fault strength and permeability, include fault healing through pressure solution and hydrothermal precipitation of minerals (Karner et al. 1997; Olsen et al. 1998; Bos & Spiers 2000; Kanagawa et al. 2000; Nakatini & Scholz 2004). However, observations and analysis of fault rock, fluid interactions and fluid behaviour (including fluid pore pressure) from natural examples are relatively scarce in the literature (e.g. Bruhn et al. 1994; Eichhubl & Boles 2000; Caine et al. 2010; Faulkner et al. 2010). In particular, this appears to be the case for preserved examples from the deeper parts of the seismogenic zone (5–12 km), which are crucial to understand fault and fault–fluid behaviour at depth (see Faulkner et al. 2010).
In the present study, we present observations from naturally occurring fault rocks, which show evidence for elevated pore pressure and fluid flow during faulting in the deeper parts of the seismogenic zone. The fault zones crop out as a network of Late Permian–Early Triassic brittle normal faults in Troms, northern Norway (Fig. 1; Andresen & Forslund 1987; Forslund 1988; Opheim & Andresen 1989; Olesen et al. 1997; Davids et al. 2013; Indrevær et al. 2013) and constitute major fault complexes that run partly onshore, partly offshore along the SW Barents Sea passive margin (Indrevær et al. 2013). The fault zones display mostly Permian–Early Triassic ages of faulting (Davids et al. 2013) and varying amounts of displacement (>3 km to <100 m) (Forslund 1988; Opheim & Andresen 1989; Olesen et al. 1997; Indrevær et al. 2013).
We describe and characterize features indicating fault–fluid interactions and high pore pressures in fault rocks from selected onshore fault zone outcrops in the region (Fig. 2), and we aim to estimate P–T conditions and fluid pressures during faulting from mineral assemblages and microstructures. This approach may yield insight into important processes operating during faulting in the intermediate to deeper levels of the brittle crust (<15 km). The implications of syn- and post-deformation fault rock permeability will be discussed in the context of the potential for hydrocarbon migration and entrapment within basement rocks in the region. Lastly, the results will be set into context with the current understanding of evolution of the SW Barents Sea passive margin (see Gabrielsen et al. 1990; Faleide et al. 2008; Smelror et al. 2009).
The studied fault rock outcrops are located within the West Troms Basement Complex, a basement horst situated on the SW Barents Sea margin, northern Norway (Fig. 1). The horst is bounded by the Lofoten igneous and high-grade metamorphic province to the SW (Corfu 2004) and by largely Permian–Early Triassic high-angle normal faults, which down-drop Caledonian nappes, to the east (Andresen & Forslund 1987; Zwaan 1995; Davids et al. 2013; Indrevær et al. 2013). The West Troms Basement Complex is made up of various Meso- and Neoarchaean tonalitic, trondhjemitic and granitic gneisses and meta-supracrustal belts, intruded by felsic and mafic igneous rocks (Corfu et al. 2003; Bergh et al. 2010).
The Permian–Early Triassic rift-related activity in the region is manifested within the horst by widespread NNE–SSW- and ENE–WSW-trending brittle normal faults and fractures arranged in a zigzag pattern along the southeastern and northwestern limits of the West Troms Basement Complex. A subsidiary NW–SE-trending fracture system is also present (Fig. 1; see Indrevær et al. 2013). The onshore fault zones can be divided into the Vestfjorden–Vanna Fault Complex, which marks the southeastern boundary of the West Troms Basement Complex, down-dropping Caledonian nappes to the east by some 1–3 km (Forslund 1988; Opheim & Andresen 1989; Olesen et al. 1997), and a less prevalent, SE-dipping, segmented fault array that runs along the outer rim of the islands of the West Troms Basement Complex (Antonsdóttir 2006; Thorstensen 2011; Indrevær et al. 2013) with displacement of the order of hundreds of metres or less (Indrevær et al. 2013; Fig. 2).
The northwestern limit of the West Troms Basement Complex is identified as the Troms–Finnmark Fault Complex (Indrevær et al. 2013), which runs offshore, parallel to the West Troms Basement Complex, where basement is down-faulted more than 5 km in the Harstad Basin, to c. 10 km depth (Fig. 1; Indrevær et al. 2013). Preserved Caledonian nappes thrust on top of Precambrian basement, as observed in the mainland of northern Norway, are indicated to be present within large regions of the SW Barents Sea (Gernigon & Brönner 2012) and may be present offshore from the West Troms Basement Complex, overlying down-faulted Precambrian basement.
The Troms–Finnmark and Vestfjorden–Vanna fault complexes (Figs 1 and 2) can be traced for hundreds of kilometres along strike of the north Norwegian margin, linking up major structural elements in the south, such as the Lofoten and Nordland ridges, with offshore fault complexes in the north, such as the Ringvassøy–Loppa, Nysleppen and Måsøy fault complexes (Doré et al. 1997, 1999; Olesen et al. 1997; Indrevær et al. 2013). Because the Vestfjorden–Vanna Fault Complex, which runs partly onshore, links up with offshore fault complexes, the fault zones in western Troms act as a unique natural laboratory for studying offshore-correlative basement-seated fault zones and their characteristics.
Timing of faulting
Ages of onshore rocks and fault rocks derived using the 40Ar/39Ar and/or apatite fission-track methods are interpreted to indicate that faulting in western Troms largely occurred in the Permian to Early Triassic with insignificant fault displacement or reactivation later in the Mesozoic and Cenozoic (Hendriks et al. 2010; Davids et al. 2013). Instead, Mesozoic fault activity took place offshore, along the Troms–Finnmark Fault Complex (Gabrielsen et al. 1990), further north in Finnmark (Roberts & Lippard 2005; Torgersen et al. 2013) and to the south in Vesterålen and Andøya (Dalland 1981; Fürsich & Thomsen 2005; Eig 2008; Hansen 2009; Hendriks et al. 2010; Osmundsen et al. 2010; Davids et al. 2013). The onshore faults in Troms are therefore believed to have been abandoned after the Late Permian–Early Triassic rifting phase and thus reflect conditions of early stages of rifting (Davids et al. 2013; Indrevær et al. 2013). However, palaeomagnetic evidence for Permian as well as Cenozoic to recent phases of faulting has been obtained for fault segments within the Vestfjorden–Vanna Fault Complex in Troms (Olesen et al. 1997), suggesting that the major offshore fault activity triggered minor reactivation on some onshore faults.
Several fault segments with good outcrops of fault rocks in the West Troms Basement Complex have been investigated through structural mapping, including extensive sampling of rocks from both the core zones and damage zones. Fault rock studies using polarizing light microscopy and SEM analysis, including backscatter detection (BSD) and energy-dispersive spectrometry (EDS), have been conducted to identify mineral assemblages, which have been used in estimation of P–T conditions during faulting, and to identify and describe fault rocks to provide insight into fault and fluid flow characteristics.
Described fault segments with distinct fault rocks include the Rekvika, Bremneset, Tussøya and Kvaløysletta–Straumsbukta fault zones (Fig. 2; Forslund 1988; Opheim & Andresen 1989; Olesen et al. 1997; Antonsdóttir 2006; Thorstensen 2011; Indrevær et al. 2013). These and other fault rocks of the West Troms Basement Complex have been mapped and discussed in a regional context by Indrevær et al. (2013). The Rekvika, Bremeset and Tussøya faults have estimated amounts of displacement of the order of hundreds of metres or less (Indrevær et al. 2013). The Kvaløysletta–Straumsbukta fault zone is a fault segment of the Vestfjorden–Vanna Fault Complex with an estimated amount of displacement of 1–3 km (Forslund 1988). In the following, we aim to describe the meso- and microscopic petrological characteristics of the fault rocks from these fault zones as a basis for discussing P–T conditions, fault evolution, fluid flow characteristics, and fluid pressures during faulting.
Rekvika fault zone
The sinistral-normal Rekvika fault zone (Fig. 2, map A; Antonsdóttir 2006; Indrevær et al. 2013) dips SE and cuts through parts of a granitic intrusion. The damage zone ranges in width from 30 to 50 m into the footwall, consisting mainly of protocataclastic red stained granite. It is abundantly cut by irregular quartz veins and enriched in epidote (Fig. 3a and b). The process zone occurs as a c. 2 m wide, light green, ultracataclasite composed of dominantly quartz and minor hematite (Fig. 3c). The ultracataclasite is cut by later irregular quartz veins (Fig. 3c). Both the hanging wall and the footwall are characterized by a >400 m wide zone along strike of conspicuous red staining of the granite (Fig. 3d).
In thin section, at least three generations of cataclasites are observed within the light green core zone (Fig. 3e and f). The first generation cataclasite is preserved as sub-angular 0.2–1 mm clasts of very fine-grained quartz ultracataclasite (internal grain size of 5 µm to <1 µm) containing c. 99% quartz and 1% hematite. The first generation cataclastic aggregates are embedded within rounded to sub-rounded clasts of a second generation of quartz ultracataclasite (Fig. 3e and f), which is composed of c. 80% quartz and c. 20% hematite. The sub-rounded clasts of the second generation are, again, embedded within a third generation, composed of 99% quartz and less than 1% hematite with a grain size in the range from 50 µm to <1 µm. The fine-grained domains show microstructural characteristics of healing and grain growth, evident from the development of cuspate grain boundaries (Fig. 3g), indicating that the initial clast grain size was less than what is observed in thin section.
Later extensional fractures with infill of quartz cut both generations of cataclasites (Fig. 3e and f). These fractures are cut first by fractures containing both hematite and quartz, and subsequently by fractures showing only hematite. A lens of mafic gneiss that shows an amphibolite-facies mineral assemblage away from the fault zone (Fig. 2, map A) is statically retrograded to chlorite (60%), quartz (30%), epidote (5%), albite (5%), and accessory titanite and muscovite within the damage zone of the fault zone (Fig. 3h).
Tussøya fault zone
This sinistral-normal fault zone dips moderately ESE and separates granite in the footwall from banded mafic and felsic gneisses in the hanging wall (Fig. 2, map B and Fig. 4a; Indrevær et al. 2013). The fault crops out as a 1–3 m thick proto- to ultracataclastic zone (Fig. 4b). The granite in the footwall is stained red, similar to the Rekvika fault zone, within a 200 m wide zone and is commonly cut by dark bands of ultracataclasite (Fig. 4c). Altered granite also occurs in the hanging wall in the southern parts of the outcrop area. Subsidiary, ENE–WSW-trending, dextral normal faults interact with the overall main NNE–SSW fault trend and are displaced by the main fault. In general, the footwall is more deformed than the hanging wall.
The dark bands of ultracataclasite are injected into fractures in the sidewall of the faults (Fig. 4c and d). These injections show a lack of internal foliation, indicating no significant amount of displacement along the fracture, which would be needed to produce the observed cataclasite in situ, and a lack of any significant damage to the side-walls of the fracture. Also, the fractures that are filled with cataclasites are commonly at high angles to the main fault zone. However, the fractures do have some displacement along them, as the side-walls of the fracture do not match up properly (Fig. 4d).
Within the core zone, at least three generations of cataclasites are observed in thin sections (Fig. 4e and f). The oldest generation consists of angular to sub-rounded clasts, 5 mm to <1 µm in size, made up of quartz (90%) and feldspar (10%) within a matrix composed of quartz (80%), epidote (20%) and accessory hematite (Fig. 4e and f). The second generation is characterized by sub-angular to rounded clasts of quartz (90%) and feldspar (20%) within a matrix of epidote (40%), hematite (30%), quartz (20%) and chlorite (10%) (Fig. 4e and f). The third generation of cataclasite is observed in association with microfaults, which cut the two previous generations (Fig. 4e and f) and consist of quartz (60%), albite (30%), pumpellyite (10%) and hematite (1%).
Microstructural study of the inferred injected ultracataclasites confirms the lack of any high-strain zone within the cataclasite, supporting its injected nature inferred from outcrop observations. The ultracataclasites have formed elsewhere (probably along larger fault segments) and been transported into fractures (Fig. 4g and h), resulting in the dark bands of ultracataclasites observed in outcrops (Fig. 4d).
Within the damage zone, biotite is locally fractured and dilated by the precipitation of quartz (Fig. 5a). Larger clasts of quartz show undulatory extinction (Fig. 5b) and both quartz and feldspar show healed fractures, with nucleation of new grains along the healed fractures (Fig. 5b and c). Comparing unaltered and altered granite under a polarizing microscope reveals that the red staining is limited to grains of plagioclase (Fig. 5d–g). Micro-grains of iron oxides (<1 µm in size) in larger plagioclase grains produce the red staining (Fig. 5f and g).
Bremneset fault zone
This normal fault (Fig. 2, map C) dips ESE and occurs as a 0–3 m thick cataclastic zone that can be traced along the shore for c. 200 m, cutting migmatitic, banded gneisses (Fig. 6a and b). The gneiss foliation is in general at a moderate angle to the fault zone. Fault and fracture surfaces commonly show epidote precipitation, and they are locally cut by younger faults or fractures with hematite staining. The core zone is composed of epidote-rich cataclasite with minor quartz and chlorite (Fig. 6b).
A fractured amphibolite gneiss from the damage zone of the Bremneset fault zone consists of albite (40%), quartz (30%), chlorite (5%), epidote (5%), apatite (1%) and accessory titanite and calcite (Fig. 6c and d). Clasts of fractured epidote (Fig. 6e) show typical grain growth features including the formation of idiomorphic grain boundaries. Later faulting has fractured the annealed epidote grains. EDS analysis shows that pumpellyite occurs along these fractures (Fig. 6f and g). Chlorite grains are observed being fractured and dilated by titanite infill (Fig. 6h).
Plagioclase grains are commonly stained red, but the staining is not present along healed fractures when they cut plagioclase grains (see Fig. 5f and g). Preserved larger quartz and feldspar grains show traces of healed fractures similar to those in the Tussøya fault zone (Fig. 5c and d). The larger quartz grains show undulatory extinction. Late-stage microfaults cut the cataclasite without forming new cataclasite along the faults (Fig. 7a and b). Within healed fractures, fluid inclusions are observed, indicating the presence of fluids during fracturing.
Away from the Bremneset fault zone, viscously deformed amphibolites are preserved outside the fault zones. Foliation-parallel plastically boudinaged quartz veins are commonly found within the gneisses. As they are boudinaged, they clearly predate the brittle deformation (Fig. 7c). Approaching the fault zone, the amphibolite-facies ductile microstructure is progressively retrograded to a greenschist-facies mineral assemblage through static recrystallization, except in zones around the boudinaged quartz veins (Fig. 7d and e). As static retrogression is indicative of the percolation of fluids through the rock, the preserved amphibolite-facies metamorphic mineral assemblage along the outer rim of the quartz veins indicates that the quartz veins have prevented fluid flow across these zones.
Kvaløysletta–Straumsbukta fault zone
This is an oblique-dextral normal fault zone (Fig. 2, map D; Forslund 1988; Olesen et al. 1997; Indrevær et al. 2013), which dips SE and runs along the eastern rim of Kvaløya, juxtaposing Precambrian gneisses in the footwall and Caledonian nappes in the hanging wall. Near Straumsbukta, parts of the footwall damage zone crop out within foliated tonalitic and amphibolitic gneisses (Fig. 8a and b). Fault surfaces commonly trend north–south and locally NE–SW, and are parallel to a moderately east-dipping foliation in the gneisses. The footwall outcrop is increasingly deformed towards the east. The tonalitic gneisses are commonly stained red, as observed along the other studied fault zones. In outcrops, related fault and fracture surfaces are coated with chlorite and quartz, and systematically are cut by fractures coated with hematite.
At least two generations of cataclasites are observed in mafic rocks of the damage zone in Straumsbukta. The first generation consists of very fine-grained (<1 µm) epidote (40%), chlorite (40%) and quartz (20%) aggregates, which appear as clasts within a second generation of cataclasite composed of quartz (50%), chlorite (30%) and epidote (20%) in the matrix (Fig. 8c and d). Chlorite within the second generation of cataclasite commonly shows radial growth (Fig. 8e and f). Both generations of the fault rocks are cut by quartz and chlorite veins, in successive order.
Plagioclase grains from the tonalitic gneisses of the damage zone are commonly red stained, similar to the other studied fault zones. Notably, the plagioclase within the amphibolite gneisses is retrograded to sericitic albite with overgrowth of white mica and epidote, which is also present within amphibolitic gneisses from Bremneset (Fig. 7d).
In summary, a common feature to all of the studied fault zones and related fractures is the red staining of plagioclase within the fault damage zone and hydrothermal precipitation of epidote, quartz, chlorite, calcite and/or hematite along fracture surfaces and in the matrix of cataclasites. Cross-cutting relationships demonstrate that epidote, quartz and chlorite veins formed first and were later cut by veins composed dominantly of hematite. At least two generations of cataclasites were formed during faulting, accompanied by mineral recrystallization and/or growth. Both the cataclasites and the statically recrystallized mafic gneisses show the mineral assemblage epidote + quartz + chlorite + albite + titanite ± apatite ± calcite ± muscovite. Subsequent fracturing and cataclasis introduced pumpellyite to the mineral assemblage.
P–T conditions during faulting
The mineral assemblage epidote + quartz + chlorite + albite + titanite ± apatite ± calcite ± muscovite is identified in the first generations of cataclasites in mafic gneisses of the Bremneset and Tussøya fault zones. This mineral assemblage indicates that greenschist-facies metamorphic conditions were present during faulting. The same mineral assemblage is observed in statically recrystallized mafic gneisses from the damage zones of the Rekvik and Bremneset fault zones. The formation of pumpellyite in later-stage fractures in cataclastic mafic gneisses at Bremneset and in granitoid fault rocks at Tussøya further suggests decreasing P–T conditions of faulting through time, as pumpellyite is stable at sub-greenschist-facies conditions.
The presence of greenschist-facies mineral assemblage (epidote + quartz + albite + chlorite ± white mica) is indicative of a minimum temperature during the formation of the early generations of cataclasites at c. 300 °C (Fig. 9; Bucher & Grapes 2011). Fractured grains of epidote show grain growth and the development of idiomorphic grain boundaries (Fig. 6e). As grains of epidote are fractured, this implies that epidote and thus, greenschist-facies metamorphic conditions, were reached prior to fracturing. The post-fracturing grain growth and the development of idiomorphic grain boundaries of the fractured epidote indicate that greenschist-facies metamorphic conditions persisted after the early event(s) of brittle faulting. Later fault movement fractured the idiomorphic grains, and fine-grained pumpellyite formed along these fractures. Pumpellyite forms at temperatures ranging from c. 250 to 300 °C (Fig. 9; Bucher & Grapes 2011). Thus, the presence of pumpellyite suggests that the temperature decreased below c. 300 °C during later stages of faulting.
The observed microstructures and greenschist-facies mineral assemblages do not allow for a precise estimate of confining pressure during deformation owing to the relatively wide pressure range of stability for greenschist-facies assemblages. The occurrence of pumpellyite, however, yields a minimum pressure for the later-stage faulting of c. 220 MPa corresponding to a minimum depth of c. 8.5 km, assuming an average meta-MORB (mid-ocean ridge basalt) composition of the amphibolites and an upper lithospheric density of 2600 kg m−3 (Fig. 9). From these estimates, it is possible to infer the highest possible geothermal gradient during the time of faulting as c. 30 °C km−1.
A minimum estimate for pressures during the earlier greenschist-facies fault activity may be obtained by extrapolating this geothermal gradient up to c. 300 °C (the minimum temperature estimate for the greenschist faulting conditions). This yields a minimum pressure of the early stage faulting at c. 245 MPa, corresponding to c. 10 km depth (Fig. 9).
Fault rocks and fluid interactions
These reactions are favoured in meteoric water-dominated granitic fault systems (Wintsch et al. 1995).
The occurrence of the quartzitic ultracataclasite at Rekvika may thus be preserved early stages of faulting, which, in contrast to the Wibberley (1999) study, are preserved where these fault rocks form relict clasts within phyllonites. If so, at least three generations of such cataclasites can be observed in Rekvika and suggest that the fault zone was abandoned after early stages of rifting so that subsequent fault activity and progressive formation of phyllosilicates (inducing fault weakening) did not occur.
These reactions are more consistent with the observed mineral compositions of the Rekvika and Tussøya fault zones and suggest that the fluid may originate from a mafic source. A potential mafic source of the infiltrating fluids does not preclude a meteoric origin, but may favour a metamorphic origin of the fluids. One way to generate metamorphic fluids is by prograde reactions during down-faulting of a mafic rock to a deeper crustal level (Caine et al. 1996). Mafic rocks are abundantly found close to all of the studied fault zones. Prograde metamorphism, however, is not consistent with the overall retrograding conditions inferred for the studied fault rocks, but may have been the case for rocks situated at deeper levels during faulting.
The red staining of feldspars, as observed within the damage zones of the Rekvika, Tussøya and Kvaløysletta–Straumsbukta fault zones (Figs 3d and 5f, g), is yet another indication of fault-related fluid infiltration; the red staining may in general be attributed to subsolvus alteration of feldspar minerals through hydrothermal alteration (Boone 1969; Taylor 1977; Smith & Brown 1988; Nakano et al. 2005; Drake et al. 2008), where plagioclase interacts with fluids to form pseudomorphs after plagioclase occupied by albite and K-feldspar. The red staining itself has been demonstrated to occur as a result of the precipitation of hematite within feldspar pores (Putnis et al. 2007). The iron is suggested to originate from either fluids (Putnis et al. 2007) or iron-rich plagioclase, where albitization of plagioclase leads to exsolution of iron from the plagioclase crystal at lower temperatures, and its reaction and oxidization in contact with fluids to form distributed hematite (Tegner 1997). Our microstructural studies also suggest that some of the iron may originate from biotite, which has retrograded to chlorite.
Healed fractures cutting plagioclase pseudomorphs are observed to have narrow zones without red staining along them. This indicates that, during later stages of fluid flow, fluids permeated through the fractures as they opened and removed hematite along the now healed fractures (Fig. 5f and g). This may explain the enrichment of hematite within some of the cataclasites and the wide occurrence of later stage precipitation of hematite on fault and fracture surfaces. The ubiquitous epidote in addition to the abundant hematite suggests that conditions were oxidizing during faulting.
Pore pressure and stress conditions
Pore pressure is a critical controlling factor of crustal strength. High pore pressure will effectively decrease the confining pressure and, depending on the differential stress, fracturing will occur as pore pressure approaches lithostatic values (e.g. Hubbert & Rubey 1959; Haimson & Fairhurst 1967). In the following, we will discuss observations that are consistent with fluids having dilated or created open space within the fault rocks, which is indicative of lithostatic pore pressures during faulting. For simplicity, we define lithostatic pressure as equal to σ3.
The microstructural observations include the following.
(1) Mineral precipitation on fault or fracture surfaces. For fluids to hold fractures open (at least by very small amounts) to allow precipitation of minerals in the dilatant sites, the fluid pressure must exceed the normal stress acting on a fracture plane and as a minimum, be equal to σ3 (Delaney et al. 1986).
(2) Fracturing and dilatation. Microstructural evidence for fracturing and dilatation of the fault rocks is observed and interpreted in the form of fractured biotite with injection of fluids and precipitation of quartz (Fig. 5g), and fractured chlorite with injection of fluids and precipitation of titanite (Fig. 6h). For fluid injection and crack dilatation, the pore pressure needs to equal or exceed σ3.
(3) Radial growth of chlorite within cataclasites. Undeformed (and thus post-faulting) radial aggregates of chlorite are observed in thin section (Fig. 8f). The radial shape suggests that it has grown into an open pore space. For pore space to exist at depth, the pore pressure must, as a minimum, have been equal to σ3.
(4) Injection of ultracataclasite into the host rock. As pore pressures were sufficient for opening veins for cataclasites to be injected during deformation (Figs 4c, d, g and h), the pore pressure must, as a minimum, have reached σ3.
Owing to the above-mentioned indications for pore pressures reaching σ3, a minimum estimate of the pore pressure during faulting is thus c. 240 MPa, corresponding to the pressure of deformation obtained from estimations based on mineral assemblages from earlier sections. It follows that the differential stress must have been relatively low for the fluid pressure to have reached lithostatic values without causing hydraulic fracturing during earlier phases of pore pressure build-up. Assuming a typical Mohr–Coloumb failure criterion (see Goodman 1989), the differential stress needed to fracture a rock that is subjected to lithostatic pore pressure is of the order of tens of MPa at the most (Fig. 10). The inferred lithostatic pore pressure thus implies that the middle crust was relatively weak during faulting.
The high pore pressure within the fault zones may be explained by episodic flow of high-pressure fluids, also known as seismic pumping (e.g. Sibson et al. 1975; Byerlee 1993). The cyclic nature of seismic pumping fits with the inference of episodic rupture in the studied fault zones as suggested by the range of mineral precipitates and several generations of cataclasites.
Fluid flow velocity
The presence of injected ultracataclasites into veins of the fault rocks (Fig. 4g and h) provides an opportunity to roughly estimate the co-seismic fluid velocity during faulting and injection. For cataclasites to be injected, they need to be fluidized.
where ρs and ρf are the density of a sphere (clast) and fluid, respectively, g is the gravitational constant, R is the radius of a sphere (clast), here set to 100 µm as observed in thin section (Fig. 5g and h), and µ is the dynamic viscosity, set to 10–4 kg m−1 s−1 for water (for simplicity) at 300 °C (Schmidt & Mayinger 1963). We obtain a value in the order of 10-1 m s–1 using the parameters derived from this study.
Eichhubl & Boles (2000) calculated palaeo-fluid flow rates during co-seismic fluid expulsion events of a minimum of 0.01 m s–1. The value obtained in the present study suggests that co-seismic rates of fluid flow may exceed those estimated by Eichhubl & Boles (2000) by one order of magnitude.
Fault zone permeability
The above discussion supports that the studied fault zones have acted as fluid conduits during faulting. In addition to their conduit properties such as grain growth and mineral precipitation, other microstructural observations have been made, which shed light on pre- and syn-faulting basement rock permeability.
First, the observed static recrystallization of pre-existing, Archaean–Proterozoic amphibolite-facies ductile fabrics, which are retrograded to greenschist-facies minerals, indicates that fluid infiltration has taken place in the adjacent gneisses. Consequently, permeability within the basement rocks must have been sufficiently large to allow for adequate fluid flow through the rock. Furthermore, basement permeability prior to the brittle deformation has been highly anisotropic, as illustrated by the impregnating properties of pre-existing, boudinaged, foliation-parallel quartz veins (Fig. 7c–e), which have protected its immediate surroundings from static retrogradation, hence limiting permeability across the vein. Thus, pre-existing, extensive, foliation-parallel quartz veining will allow for fluid flow parallel to foliation, but impede across-foliation fluid flow.
Second, the large amount of precipitation of hydrothermal minerals on fault and fracture surfaces in the damage and process zones implies that the fault zones have acted as fluid transport paths. The red staining of feldspar-rich host rocks may also influence fault zone permeability. Drake et al. (2008) have shown that the red staining process of plagioclase pseudomorphs increases the porosity compared with unaltered samples. This implies that, as fluid infiltration and red staining occurs in a plagioclase-rich rock, the permeability within these hydrothermally altered zones will increase with time, as more plagioclase grains are pseudomorphed, giving a positive feedback mechanism for the red staining process.
A model is proposed, using the terminology of fluid flow characteristics of Caine et al. (1996), describing faults as fluid conduits, fluid barriers, or fluid conduit–barriers (Fig. 11). The model summarizes the observed feedback mechanisms that may influence fault zone permeability by illustrating the dynamic evolution of three end-member fault zones: (1) a fault zone within granitic host rocks (the Rekvika fault zone); (2) a fault zone within foliated gneisses (the Bremneset fault zone); (3) a fault zone separating granitic rocks in the footwall from banded gneisses in the hanging wall (the Tussøya fault zone).
Prior to faulting, the permeability with granites is assumed to be isotropic. Within subhorizontal banded gneisses, the horizontal permeability is assumed similar to the permeability within granites, whereas the vertical permeability is strongly impeded owing to pre-existing, impermeable foliation-parallel layers, such as quartz veins. The observed microstructures suggest that, as faulting initiates, fracturing and cataclasis in the core zone and immediate damage zone leads to an increase in permeability and hence an increase in vertical fluid flow within the central parts of the fault (Fig. 11a).
As silica walls form owing to pressure release, the permeability within the core zone decreases rapidly and this zone probably becomes completely sealed. The trapped fluids are forced into the surrounding damage zone (Fig. 11b). Here, fluid flow is more restricted, but increases through time within feldspar-rich host rocks, as the red staining process increases permeability. Thus, the damage zone will become the preferred fluid pathway for along-fault fluid flow during periods after faulting (Fig. 11b). In the banded gneisses, however, any vertical (along-fault) fluid flow will be restricted by impermeable foliation-parallel layers and thus encourage foliation-parallel fluid flow away from the fault zone.
During longer periods of fault quiescence, a number of subsidiary factors, such as later grain growth, the sealing of fractures and secondary mineral growth–precipitation on fracture surfaces and within pore space, will completely seal the fault zones, impeding both vertical and horizontal fluid flow, independent of host-rock lithology (Fig. 11c). Keulen et al. (2008) have shown, based on microstructural criteria, that such healing may occur within a few months at 300 °C, through the processes of dissolution and grain growth. Any later reactivation of the fault zone, however, may restart the above-described cycle (Fig. 11a–c). The model implies that a fault zone formed at depths of 8–10 km may, depending on the host rock lithology, cyclically evolve from a fluid conduit, through a conduit–barrier, and finally a full fluid barrier.
Implications for passive margin evolution
The ultracataclasites observed in the Rekvika fault zone are suggested to be comparable with ultracataclasites described by Wibberley (1999) and thus reflect early stages of faulting. Their in situ presence indicates that the Late Permian–Early Triassic faulting came to a halt during early (deep) stages of faulting, to prevent subsequent formation of phyllosilicates within the fault zone. Thus, the preservation of these ultracataclasites may coincide with, and be a result of, the westward migration of fault activity to offshore fault complexes after Late Permian–Early Triassic (Davids et al. 2013; Indrevær et al. 2013).
Furthermore, an average geothermal gradient for the upper continental crust is in the range of 25–30 °C km−1. Within early stages of continental rifting, both palaeo- and present geothermal gradients tends to be elevated, especially within subsiding rift-basins, and may exceed 50 °C km−1 (Ru & Pigott 1986; Qiu & Wang 1998; Sandiford et al. 1998; Liu et al. 2001; Wang et al. 2011). The geothermal gradient of c. 30 °C km−1 calculated in this study suggest either that onshore faulting occurred during early phases of rifting along the SW Barents Sea margin, or that the faulting was located along the rift flanks, where geothermal gradients may have remained relatively unchanged.
An uplift of the coastal part of the SW Barents Sea margin has been suggested as part of the formation of the Scandes Mountains (see Corner 2005). The timing and nature of such uplift(s) have been widely discussed in the literature (see Doré et al. 2002). The depth estimates from the studied fault zones are used in conjunction with the Late Permian time constraints on faulting in the region (Davids et al. 2013) to calculate a minimum average exhumation rate of c. 40 m Ma−1 for the outer islands of Troms since the Late Permian. When we consider that continental average erosion rates typically range from c. 10 to 100 m Ma−1 (e.g. Stallard 1988; Schaller et al. 2001, 2002; Charreau et al. 2011) and that the SW Barents Sea region has experienced extensive glacial erosion in the last 2.7 Ma with an average erosion rate of c. 380 m Ma−1 (Laberg et al. 2012), the values suggest that erosion alone can explain the regional exhumation since Permian times: the Late Cenozoic tectonic uplift of the margin may be due to climate deterioration after the formation of the North Atlantic Ocean, which caused subsequent isostatic recalibration and greater uplift of the marginal crust owing to a faster rate of erosion along the margin than inland.
In terms of fault permeability and fluid flow, the studied Late Permian–Early Triassic fault zones may have acted as fluid barriers since this rifting phase. If we assume that the permeability of the basement fault rocks onshore is analogous to that of basement-seated faults offshore, then faults such as those present on the proximal shelf along the SW Barents Sea margin (i.e. the Finnmark Platform and the Loppa High) have the potential to act as hydrocarbon traps. In contrast, any later reactivation of faults after the emplacement of hydrocarbons would most probably increase along-fault permeability and allow for migration of hydrocarbons up through the fault zone.
(1) Late Permian–Early Triassic brittle normal faults make up a network of basement-seated faults that occur both onshore and offshore along the SW Barents Sea passive margin. These faults show varying amounts of displacement from >3 km to hundreds of metres or less, and they comprise distinctive cataclastic fault rocks.
(2) Minimum P–T conditions during early stages of faulting are estimated to be c. 300 °C and c. 240 MPa (c. 10 km depth) based on greenschist-facies mineral assemblages of the cataclasites. Later fault movement introduced pumpellyite, yielding minimum P–T conditions of c. 275 °C and c. 220 MPa (c. 8.5 km depth).
(3) The quartzitic ultracataclasites that occur within granitoid fault rocks are interpreted to be preserved fault rocks from early stages of faulting that formed as a result of the chemical breakdown of feldspar to epidote with the release of quartz. Fault activity is interpreted to have come to a halt during early (and deep) stages of rifting, as there is no subsequent formation of phyllosilicates within the process zone.
(4) Microstructural evidence indicates that pore pressures locally reached lithostatic levels (240 MPa) during faulting. Fluidized cataclasites allowed for the calculation of the minimum fluid velocity necessary, yielding a value in the order of 10-1 m s−1.
(5) The brittle faults within the basement rocks acted as important fluid conduits during rifting in the Late Permian–Early Triassic. A model is proposed, suggesting that fluid flow was concentrated in the core zone during faulting and in the damage zone during periods after faulting. Healing and precipitation processes probably sealed off the fault zones within a short time span after faulting.
(6) A maximum geothermal gradient during faulting in the Late Permian–Early Triassic of c. 30 °C km−1 is calculated based on the occurrence of pumpellyite. The un-elevated geothermal gradient suggests either that faulting occurred during early stages of continental rifting or that the studied fault zones were located along the rift flanks where little to no subsidence took place.
(7) A minimum average exhumation rate of c. 40 m Ma−1 since the Late Permian is estimated for the West Troms Basement Complex. When considering normal erosion rates, the proposed late Cenozoic uplift, which has been discussed widely in the literature, may be explained by climate deterioration after the formation of the North Atlantic Ocean, causing subsequent isostatic crustal recalibration and greater isostatic uplift of the margin crust owing to a greater amount of erosion along the margin than inland.
(8) As the studied fault zones are the onshore portions of large fault complexes that continue offshore, it is likely that the studied fault zones are analogous to basement-seated faults offshore. This implies that the conditions and nature of faulting observed onshore may be valid for offshore faults, including past and present fault zone permeability and potential for hydrocarbon entrapment.
This work was part of the Industrial PhD scheme organized by the Norwegian Research Council. The project is a collaboration between DONG E&P Norge AS and the University of Tromsø, financed by DONG E&P Norge AS and the Norwegian Research Council. We would like to express our gratitude to all those involved from these institutions.