Focusing Fluids in Faults: Evidence From Stable Isotopic Studies of Dated Clay‐Rich Fault Gouge of the Alberta Rockies

Isotopic studies of Canadian Rocky Mountain thrust faults preserve the timing and identity of orogenic fluids and their fault zone pathways. Using previously dated samples, we measure the O‐ and H‐isotopic compositions of fault gouge. These nearly 100% neomineralized gouges and their associated damage zones act as primary orogenic fluid pathways. As such, they provide a specific and local look into the nature of the Late Jurassic to Early Eocene orogenic plumbing system in the Alberta Rockies. Considering clay polytype stability and regional temperature conditions, we obtain a range of geofluid isotopic compositions during Jurassic‐Eocene thrust faulting: δ18Ofluid ranged from ∼−3.3 to 9.2 ± 3.2‰; δDfluid ranged from −119 to −46 ± 13‰ VSMOW. The range of O‐ and H‐isotopic compositions reflects mixing of fluid sources, including the pervasive presence of surface‐sourced fluids (up to ∼90%). The interpreted prevalence of a surface fluid source in fault rocks is in agreement with regional isotopic trends previously observed in undated veins of fractured host rock. Our results confirm that thrust faults of the Alberta Rocky Mountains acted as major fluid‐focusing conduits during orogenic activity. We further show that these faults incorporated both deeply sourced and surface‐sourced fluids into zones of enhanced and dynamic permeability, heterogeneously distributing fluids along fault planes across the fold‐thrust belt, promoting the growth of fault‐zone weakening clay minerals.

However, the direct study of fault fluids has been difficult for several reasons, among them the lack of readily extractable and isolatable mineral phases in fault rock material. Advances in precision shallow fault-dating overcome this particular hurdle, providing insight into the timing of fault activation though 40 Ar/ 39 Ar-dating of secondary clay minerals separates that form during fluid flow in active fault zones (e.g., van der Pluijm et al., 2001). Building on the dating of fault-grown mineral studies, this article utilizes previously dated clays from the Alberta Rocky Mountains (Pană & van der Pluijm, 2015) to determine the isotopic composition and source of fluids that were channeled through fault zones during episodic fault slip and regional deformation. Using fault gouge samples as fluid proxies is complementary to vein-based studies, as they provide independent insight into the absolute timing of fluid flow through Ar-dating, and isotopic studies utilize regional temperature constraints gained from fluid inclusion analysis. The application of paired oxygen and hydrogen isotopic analysis of dated clays provides a multi-dimensional picture of the role and location of fault rock fluids in major orogenic settings. This article presents hydrogen and oxygen isotope data from the direct study of dated gouge, clarifying the relationship between regional deformation and localized faulting, associated fluid flow and fluid-driving forces in the southern Canadian Rocky Mountains.

Regional Geologic Context
Our study area is in the Alberta portion of the southern Canadian Rocky Mountain fold-and-thrust belt (RM-FTB), which is part of the Cordilleran Foreland belt of North America. Westerly from the Foreland belt, the southern Canada Cordillera is traditionally subdivided into the Omineca, Intermontane, Coast, and Insular morphogeological belts (e.g., Gabrielse et al., 1992; Figure 1). The Foreland belt comprises strata of North American origin, the Omineca Belt is the region of overlap between ancestral North America and allochthonous rocks, whereas belts to the west include a collage accreted of allochthonous and autochthonous terranes (e.g., Monger, 1984Monger, , 1989Price, 1986Price, , 1994. The upper-crustal tectonic elements (or allochthonous terranes) were juxtaposed over each other and over the western margin of the North American craton along a system of interleaved, northeast-and southwest-verging major thrust faults (Monger et al., 1982;Struik, 1988;Tempelman-Kluit, 1979).
Although the paleogeography and tectonic models of the southern Canadian Cordillera are somewhat controversial, it is widely accepted that Neoproterozoic rifting of Rodinia led to the onset of Windermere deposition and was followed by seafloor spreading and continental drift in the latest Neoproterozoic. By the earliest Cambrian, a persistent continental shelf-slope system was established between ancient North America and the newly opened ocean, a distant ancestor of the present Pacific Ocean. The paleogeography of the ancient continental margin evolved from a passive margin until Middle Devonian to a mainly convergent plate margin until the present (e.g., Monger, 1984Monger, , 1989Monger & Price, 2002).  Pană and van der Pluijm (2015). The locations of samples collected from the Canadian Cordillera fold-thrust belt in Alberta are shown on the map and positioned relatively on the cross-section. Italicized thrust names listed on the cross-section do not intersect the section. Sample A (a footwall shale sample) shares the same location (within 300 m) as Sample 9. Samples 13-15 and 17 from Pană and van der Pluijm (2015) were not available for use in this study.

Paleoproterozoic
Tectonic events did not markedly affect ancestral North American rocks in Canada until the Middle Jurassic. Events leading to Cordilleran mountain building started in Middle Jurassic time, as a result of breakup of Pangea and North American plate motion toward subduction zones at its western margin, followed by collisions with eastward and northeastward drifting island arcs on the proto-Pacific lithosphere (e.g., Gabrielse et al., 1992;Monger, 1984Monger, , 1989Monger & Price, 2002;Monger et al., 1972Monger et al., , 1982. Between the Middle Jurassic and early Eocene, the Cordilleran realm was mainly under compression, accompanied at different times by sinistral and dextral transpression (e.g., Evenchick et al., 2007;Monger & Gibson, 2019).
The investigated RM-FTB formed as a thin-skinned accretionary wedge in a retroarc tectonic setting between the Middle Jurassic and early Eocene (Monger & Price, 2002;Pană & van der Pluijm, 2015). It is bounded to the east by the elusive eastern limit of Cordilleran deformation, and to the west by the Rocky Mountain trench. The detached and displaced supracrustal rocks comprise several broad tectono-stratigraphic assemblages, mostly of North American origin, deposited within the Western Canada sedimentary basin. The thick stack of east-vergent, generally downward-and eastward-younging thrust slices includes Proterozoic strata, locally overprinted by low-to medium-grade metamorphism, in the western parts of the RM-TFB, unmetamorphosed Paleozoic strata in the central and eastern parts, and Mesozoic to Cenozoic rocks in the frontal parts (Monger, 1989). The southernmost portion of the Canadian RM-FTB also includes strata of the Belt-Purcell Supergroup deposited in a controversial Mesoproterozoic tectonic setting (e.g., Ross & Villeneuve, 2003;Sears & Price, 2003).

Sample Location and Mineralogy
Fifteen (15) samples analyzed in this study were collected from the eastern, non-metamorphosed portion of the RM-FTB in Alberta, spanning the length of the belt from approximately 50°-54°N latitude ( Figure 1, Table 1). Twelve (12) samples of fault gouge and one (1) footwall shale sample were previously dated using Ar geochronology (Pană & van der Pluijm, 2015); two (2) additional fault gouge ages were reported by van der Pluijm et al. (2006). Using the combined illite ages from both studies, Pană and van der Pluijm (2015) identified four major pulses of contractional deformation between the Late Jurassic and Early Eocene, which preceded middle to late Eocene extensional collapse of the orogen. Authigenic illite shows that the growth of fault-related clay minerals occurred in the presence of ancient orogenic fluids, so their stable isotopic makeup reflects the stable isotopic composition of the deformational fluids. Though earlier work determined the polytypes of illite present in each gouge sample (required for Ar/Ar-dating), additional work was needed to fully characterize the clay mineralogy in order to extract the relevant isotopic signatures from authigenic illite.
The methods to process samples and characterize illitic materials are described in Pană and van der Pluijm (2015) and van der Pluijm et al. (2006). We completed additional clay mineral X-ray diffraction (XRD) characterization on the each of the four <2 μm size fractions through low-angle (2°-40°2θ) scanning of oriented mounts, which were prepared using the suspension method (Moore & Reynolds, 1997). We used a Cu-source Rigaku Ultima IV X-Ray Diffractometer equipped with a Ni foil k-beta filter, scanning at a speed of 1°/minute and a step size of 0.02°2θ. Though illite was the dominant clay mineralogy for all samples, we also identified the presence of minor quartz, calcite, kaolinite, and chlorite in some of the samples (Figure 2). DP10-1 (Sample 7) also contained a trace amount of gypsum. Using the mineral reference intensities (MRI) method (Moore & Reynolds, 1997), we quantify the proportions of clay minerals present in each sample (Table 2).

Isotopic Measurement
Stable isotopic measurements of hydrogen and oxygen were completed at the Institute of Earth Surface Dynamics (IDYST) at the University of Lausanne (UNIL). Approximately 1.5-2 mg of duplicate sample separates were encapsulated in silver foil packets and kept under vacuum for at least 12 hr prior to analysis. Samples were then quickly transferred to a helium-flushed zero-blank autosampler connected to a Thermo Finnigan Delta Plus XL thermochemical elemental analyzer (TC/EA). A helium carrier gas transferred the reduced hydrogen gas to the mass spectrometer, which measured the ratios of H 2 and DH gases, and the weight percent water for each sample. Results are reported using δ-notation relative to standard mean ocean water (SMOW) and are reproducible to ±3‰ across duplicate sample aliquots.
Prior to oxygen analyses, samples were loaded onto a platinum sample plate and heated in an oven at 150°C for at least 12 hr. Oxygen gas was isolated from silicate samples for isotopic measurements with laser fluorination (e.g., Sharp, 1990), using a vacuum of approximately 10 −4 Pa prior to fluorination. Extracted oxygen gas was collected on a zeolite molecular sieve and transferred to a Finnigan MAT 253 Mass Spectrometer for measurement. As with hydrogen, results are reported using δ-notation relative to SMOW and are reproducible to ±0.2‰. We were unable to measure one sample (16: KKF-91-1A) for oxygen isotopic composition due to its reaction with F 2 gas at room temperature.

Hydrogen Isotopic Results
Two aliquots of each sample size fraction were measured for hydrogen isotopic composition. In nearly all cases, measurements are reproducible to ≤3‰, with the maximum error on duplicate measurements of 3.7‰ (sample 12: DP11-112MC) ( Table 3).
A York-style bivariate linear regression analysis of hydrogen isotopic compositions and authigenic (1M d ) illite quantifications allows the extrapolation to 100% authigenic material and therefore, the determination of the hydrogen isotopic composition of deformation-related illite of the gouge (Table 4; Boles et al., 2015;Lynch et al., 2019;York, 1968). During the preparation of one sample (sample 4:DP10-166D, Brule Thrust), hydrogen-rich organic material was concentrated by centrifugation into the fine fraction, which we discard to obtain a regression value of −136.5 ± 22.4‰ δD.

Oxygen Isotopic Results
Oxygen measurements were completed for the finest fraction of each sample. Unlike hydrogen, oxygen isotopic values are not affected by the presence of hydrocarbons that may concentrate into the finer fractions. Instead, oxygen isotopic values are affected by the presence of other rock-forming minerals, including silicates, oxides, and carbonates. Non-clay silicate minerals are absent in any of the finest fractions, except trace amounts of quartz and gypsum in sample 7 (DP10-1). Minor (<5 wt%) calcite was removed prior to oxygen isotopic analysis by reaction with 10% HCl (Table 4). The variable presence of the 2M 1 , high-temperature detrital illite polymorph, which was seen in the fine fractions in concentrations up to 18% ± 2% (Lewis Thrust, sample 18), with an average of 8% ± 2%, is an irreducible source of error. Though we are unable to Figure 2. Two representative series of oriented XRD patterns. In both diagrams, the coarsest fraction is on the top, the finest fraction on the bottom. The left patterns (DP10-2) are representative of the several samples whose clay mineralogy contain only illite. The right patterns (DP11-107) are more representative of samples that have two clay minerals present, in this case, illite and chlorite. Both samples also indicate the presence of quartz, particularly in the coarser fractions (peaks at 20.8° and 26.5°2θ). The right sample also shows evidence of calcite, present in the two finer fractions (peak at 29.4°2θ).  separate the authigenic from detrital illite in the finest fraction, we use the δO fine values as representative of near-authigenic values. We find no systematic variation between the percentage of detrital illite and the δO fine values.

Fractionation Temperature Constraints
With constraints on fractionation temperature, the isotopic composition of mineralizing fluids is calculated from the isotopic composition of authigenic clay. Fractionation temperatures are constrained by the minimum formation temperature of 1Md illite, ∼90°C (e.g., Haines & van der Pluijm, 2012). Maximum fractionation temperatures are obtained from mineralogic and other geologic evidence. A geothermal gradient of ∼20-25°C/km has been estimated for the Canadian Rocky Mountain foreland fold-thrust belt region (e.g. England & Bustin, 1986;Hardebol et al., 2009;Osadetz et al., 2004). With a maximum thickness of ∼8 km for the deformed foreland wedge (Pană & Elgr, 2013;Price, 1981), this equates to temperatures less than 160°C-200°C. Additionally, Nesbitt and Muehlenbachs (1995) recorded fluid inclusion homogenization temperatures in calcite veins from the fold and thrust belt to be between 120°C and 200°C. These observations, along with maximum temperature estimates from organic maturity indicators (England & Bustin, 1986;Hardebol et al., 2009;Kalkreuth & Mc-Mechan, 1984) and conodont alteration indices (Symons & Cioppa, 2002) characterize the thermal history of the fold-thrust belt and suggest that the viable temperature range during deformation was 100°C-200°C. Since many of the exhumed thrusts likely formed at shallower depths and, noting that the upper stability of low-temperature 1Md illite of ∼180°C (Haines & van der Pluijm, 2012), we use an upper temperature of fault rock illite of 180°C, reflecting absolute maximum thrusting and fault rock formation at 7-8 kilometers depth.

Characteristics and Identity of Mineralizing Fluid
We calculate the composition of the fluid isotopic values for a large temperature window to capture any uncertainty related to local variations in geothermal gradient and fluid-mediated heat exchange along the faults. Water composition was calculated using the fractionation equations of Capuano (1992) and Sheppard and Gilg (1996) for O and H respectively (Table 5, Figure 3). The range of results produced show a broad overlap between mineralizing fluids and Alberta Basin fluids (Connolly et al., 1990;Hitchon & Friedman, 1969;Sheppard, 1986), regardless of the temperature used for the fractionation calculation. On the higher end of the temperature range, fluids have slightly more positive δ 18 O values and more negative δD values. This would imply more water-rock interaction and oxygen buffering (smaller water/rock ratio and/or longer fluid travel pathways through the fold thrust belt). However, the presence of very light hydrogen requires a high latitude or high elevation meteoric fluid as  an original fluid source. One sample (sample 7: DP10-1) yields a fluid composition that very closely resembles the isotopic composition of seawater, suggesting that isolated pockets of connate seawater may have persisted locally prior to their expulsion along thrust faults. Several of the calculated isotopic fluid values overlap with the magmatic/metamorphic field, illustrating that though meteoric fluids are a major component in geofluids in many of the fault zones, deeper fluids likely also play a role in deformation. The range in isotopic values of mineralizing fluids shows no systematic temporal or spatial pattern, indicating that fluid regimes did not vary with the timing of orogenic pulse activity nor along orogenic trend (Figure 4).

Percentage Approximations of Fluid Mixing
To allow for discussion of fluid regimes, we simplify major crustal fluid sources into two bins: surface-sourced and deeply sourced. Surface sources include meteoric fluids and meteorically charged, relatively unevolved basinal fluids. Deep sources include magmatic, metamorphic, and highly evolved (high-temperature) basinal fluids. We assume that these ideal end-member fluids were homogeneous through time and space, and that the illitic clay material crystallized at constant temperatures, and under equilibrium fractionation conditions. This is a large oversimplification and is not intended to result in precise measurements or calculations. Instead, we use this platform to spark a discussion around the heterogeneity that is observed in the samples through the lens of two ideal, end-member fluids. With this caveat, we present two fluids for consideration, which we refer to as "surface-sourced" and "deeply sourced" throughout the following discussion.
In order to define our surface-sourced end member, we applied a least squares regression to the data described in Table 5 (excluding the outlying sample 7, which was the only sample containing gypsum, and whose fluid equivalent closely resembles SMOW) for both the maximum and minimum temperature

Table 4
Isotopic Composition of Authigenic Illite constraints ( Figure 5; Lynch & van der Pluijm, 2021). We extrapolate these regression lines to the intersection with the global meteoric water line (δD = 8*δ 18 O + 10‰) (Sheppard, 1986). Fractionation at 90°C corresponds to a meteoric fluid with the isotopic signature of −13.5‰ δ 18 O and −98 ‰ δD; fractionation at 180°C corresponds to a meteoric fluid with the isotopic signature of −21.0‰ δ 18 O and −158‰ δD. These intersection points constrain the composition of the meteoric fluid input during fault activity, notably overlapping with penecontemporaneous surface fluids (Bowen & Revenaugh, 2003;Chamberlain et al., 2012;Longstaffe & Ayalon, 1990). The close correlation between the derived isotopic composition of surface fluids and the end-member meteoric fluid composition from fault gouge clays confirms that ancient meteoric fluids were a likely major fluid source in the evolving fold-thrust belt, variably mixing with heavier fluids ( Figure 5). For our ideal surface-sourced fluid end member approximation, we define δDsurface as −128‰ and δ 18 O surface as −15‰.
Previous studies emphasized the role of migrating, hot, metamorphic fluids during orogenesis (Cooley et al., 2011;Machel & Cavell, 1999;Nesbitt & Muehlenbachs, 1989). Because hydrogen isotopic signatures preserve the original fluid source-as they are not easily reset by water-rock interaction-we use fluid δD values from this study to estimate the relative proportions of end-member fluid input into fault zones. Based on the δD values of fluid inclusions and hydrous silicates in veins collected from greenschist facies in the fold-thrust belt-as high as −20‰- Nesbitt and Muehlenbachs (1989) suggested that most fluids involved in the Rocky Mountain thrusting originated from metamorphic devolatilization. Using this end-member δD values as representative of deeply sourced water (δD deep = −20‰) and the midpoint of our calculated MWL intersections as representative of meteoric water (δD surface = −128‰), we estimate the relative proportion of each fluid source. The average of our calculated fluid δD values (−78 ± 12‰, Table 5) would result from an approximately equal mixture of deeply sourced and surface-sourced fluids (46%/54%), whereas the minimum fluid value (δD = −119 ± 13‰) is just over 90% surface derived, and the maximum (δD = −46 ± 13‰) from a 76%/24% deep/surface fluid mixture. The observed range of values indicates that mixing of fluid was not constant through time and space.

Implications for Ancient Fluid Flow in the Canadian Rockies
The presence of both deeply sourced and surface-sourced fluids in Rocky Mountain thrust faults allows us to explore the relative roles of fluid driving forces during deformation. The prevalence of surface fluids in the Canadian Rockies suggests that gravity-topographic head-is an important fluid driving force that promotes downward penetration of surface-sourced fluids across the mountain belts and foreland basins. This phenomenon has been proposed to explain meteoric isotopic signatures found in fluids of the Alpine fault of New Zealand, a major continental transform boundary (Koons & Craw, 1991;Menzies et al., 2014Menzies et al., , 2016 and various low-angle normal faults of the US southwest (Haines et al., 2016). Despite the seemingly unfavorable stress regime, gravity-driven downward penetration of surface fluids through the upper crust is likely a dominant fluid driver in compressional belts as well.
At the same time that deforming fold-thrust belts allow gravity-driven downward fluid penetration, they also present pathways for deeply sourced, hot, high-pressure fluid to be expelled to the surface in response to buoyancy and compressional forces. During continued orogenesis, one might expect Sample ID Hydrogen (δD‰) Oxygen ( Table 5. Major fluid reservoirs shown include metamorphic fluids (gray box), magmatic fluids (black box), Alberta/West Canada sedimentary basin fluids (blue shaded region), meteoric water (dark gray line) and standard mean ocean water (SMOW, black circle; Connolly et al., 1990;Hitchon & Friedman, 1969;Sheppard, 1986). Calculated fluid values largely overlap with basin fluids and partly with metamorphic/magmatic fluids. One fluid value corresponds with ocean water isotopic composition.
metamorphic fluid expulsion to peak during peak deformation, perhaps coinciding with the major deformation pulses described by Pană and van der Pluijm (2015). If this were the case, we would expect the earliest fault fluids to be dominated by surface sources, peak deformation to be accompanied by increased metamorphic/magmatic fluid release, and late stage orogenesis to be again dominated by surface-fluids. Looking at Figure 4a, we are tempted to conclude that this is the case for δ 18 O-oldest and youngest fluids, which seem to be more negative (similar to surface-sourced fluids), whereas those in the middle of the age range seem less negative (similar to metamorphic fluids). However, Figure 4b does not show the same pattern for δD, and considering our uncertainties the data does not permit us to take such a bold stand. Interestingly, Nesbitt and Muehlenbachs (1994) did observe a predominance of likely meteoric fluids in their study of postorogenic veins, implying that deepsourced fluids were not available for mineralization while deformation had nearly ceased. This could also have been due to the fact that the sampled veins were not long-lived conduits for flow, as the thrust faults likely were.
Regardless, though we find isotopic evidence for the involvement of deeply sourced fluids in the RM-FTB, we do not see any clear pattern in the isotopic composition of these fluids that can be easily attributed to either systematically changing fluid inputs or different source compositions through time or during orogenic pulses (Figure 4). Instead, the relative contribution of these deep fluids and surface fluids appears to be heterogeneous through time and across the mountain belt. This suggests one of two things: that local heterogeneities in stress regime and rock fracturing may be a driving factor promoting local to regional scale fluid infiltration to fault-depths, and/or that deeply sourced fluids are not released en masse during progressive deformation. As the two (or more) fluids migrate and mingle, they find crustal weaknesses in the thrust faults and promote new mineral growth therein, including friction-reducing clays. A more thorough sampling campaign, one that combines fault gouge studies with vein studies from the same outcrops, and places them in temperature-depth context on a palinspastically restored cross-section may, in the future, be able to further eliminate uncertainties in this study and continue to unlock the fluid mixing puzzle in fold-thrust belts.

Conclusions
Newly formed clays in fault rock that are found along major thrust faults in the Alberta Rockies allow us to determine the ancient sources and pathways of orogenic fluids during shallow crustal deformation. We examined fault fluids through isotopic analysis of secondarily formed, fault-grown, dated clays in fault gouge and explored the degree of fluid mixing in fault zones. Vein-based studies in older host rock have variably identified metaphorphic fluid as a significant contributor during compressional deformation (Cooley et al., 2011;Machel & Cavell, 1999;Nesbitt & Muehlenbachs, 1991, 1994, whereas our study of fault gouge shows that surface-sourced fluids often dominate and that they are efficiently channeled along fault block interfaces. Moreover, our study constrains the degree of mixing of fluid reservoirs and their relative volumetric contributions. Rather than painting a simple picture of single fluid activity, or homogenous fluid migration across or along faults, our results show that fluid systems in the RM-FTB are multi-dimensional and complex. Variable mixing of fluids implies that the fluid regime in the fold-thrust belt was an open system, allowing the introduction, movement, and mixing of different fluids throughout ongoing deformation, focused along active thrust faults toward the front and foreland of the mountain range.  (2015) and van der Pluijm et al. (2006). Plots C and D show isotopic composition with respect to latitude. Plots E and F compare isotopic composition to their relative positions in the FTB. The latter is expressed as a fractional distance across the belt, with 0 corresponding to the Southern Rocky Mountain Trench (SMRT) and 1 to the Approximate Limit of Cordilleraian Deformation (LCD) (both shown in Figure 1). The width is the measured distance from the SMRT to the LCD, perpendicular to the strike of the belt through each sample location. Error is estimated to be ±0.1. Considering error and uncertainty, there is no correlation between any of the variables explored and fluid isotopic composition.  (Bowen & Revenaugh, 2003;Longstaffe & Ayalon, 1990), which is shown as a black oval, and considered to be one of the end-member mixing fluids. It is likely that during orogenic activity, local meteoric waters would have been less negative than they are today, due both to the lower elevation and lower latitude during the early stages of mountain building. The stippled gray box shows the region of syn-to postorogenic fluids (Nesbitt & Muehlenbachs, 1994) interpreted from fluids inclusions in dolomite veins, which have a slight overlap with clay mineralizing fluids.