Structural controls and metallogenic model of polyphase uranium mineralization in the Kiggavik area (Nunavut, Canada)

The Kiggavik area is located on the eastern boundary of the Paleo- to Mesoproterozoic Thelon Basin (Nunavut, Canada) and hosts uranium mineralization in Archean basement rocks. The major fault/fracture network in the Kiggavik area is mainly oriented ENE-WSW and NE-SW, consisting of polyphased fault zones initiated during the Thelon and Trans-Hudsonian orogenies (ca. 1900–1800 Ma). These faults were subsequently mineralized in four stages referred to as U0, U1, U2, and U3. The first event U0 is inferred to be of magmatic origin and is related to microbrecciation and weak clay alteration under a WSW-ENE σ1. U0 is a ca. 1830 event which predates intense quartz brecciation (QB) and veining at ca. 1750 Ma. QB is associated with emplacement of the Kivalliq Igneous Suite and caused pervasive silicification of former fault zones, which in turn controlled subsequent fracture development and behaved as barriers for later U mineralizing fluids (U1 to U3). U1, U2, and U3 postdate deposition of the Thelon Basin. U1 and U2 occurred under a regional strike-slip stress regime, with the direction of σ1 evolving from WNW-ESE (U1) to NE-SW /ENE-WSW (U2); both formed at ~ 1500–1330 Ma and are related to circulation of Thelon-derived uranium-bearing basinal brines. A post U2, but pre-Mackenzie dykes (ca. 1270 Ma), extensional/transtensional stress regime with σ3 oriented NE-SW caused normal-dextral offset of the orebodies by reactivating NNW-SSE and E-W trending faults. This fracturing event triggered circulation of hot (~ 300 °C), probably acidic, fluids that dissolved quartz, and caused illitization and bleaching of the host rocks. Finally, U3 records remobilization of the previous mineralization along redox fronts through percolation of low-temperature meteoric fluids during two main tectonic events at ca. 550 and 350 Ma. This study provides evidences for the presence of a primary, pre-Thelon Basin uranium stock within the Kiggavik prospects, and a strong structural control on mineralization in the Kiggavik area. Our study also shows a nearly similar evolution of uranium mineralization in this area compared to the world-class uranium district of the Athabasca Basin (Saskatchewan, Canada).


Introduction
Numerous uranium occurrences have been discovered since the 1970s in the Thelon-Baker Lake area of Nunavut, Canada (Curtis and Miller 1980;Miller 1980Miller , 1982Miller , 1995Miller and LeCheminant 1985;Miller et al. 1986). In this area, the Paleoto Mesoproterozoic Thelon Basin is similar in terms of sedimentology, geology, structural evolution, and diagenetic history with the Athabasca Basin (Saskatchewan, Canada) that is located approximately 750 km south-west and hosts the largest number of high-grade unconformity-related uranium (URU) deposits in the world (Miller and LeCheminant 1985;Fuchs et al. 1986;Weyer et al. 1987;Friedrich et al. 1989;Fuchs and Hilger 1989;Jefferson et al. 2007). Among the prospective zones, the Kiggavik area, located on the eastern border of the Thelon's Aberdeen sub-basin ( Fig. 1), hosts several U orebodies of economic interest: Kiggavik (Main Zone, Center Zone, East Zone), Bong, Andrew Lake, and End prospects, in addition to the Granite grid, 85W, Sleek, Jane and Contact minor prospects (Fig. 2;Fuchs et al. 1986;Riegler et al. 2014;Chi et al. 2017;Roy et al. 2017).
Various scientific studies have been carried out on these orebodies in order to characterize the key parameters which controlled the formation of U mineralization, with the objective of comparing them with those of the world-class URU deposits in the Athabasca Basin (Jefferson et al. 2007;Cuney and Kyser 2009). Previous studies in the Kiggavik area mainly focused on the characterization of mineralizing fluids and their alteration products through analysis of fluid inclusions, geochemistry, and dating of U oxides and related clay minerals (Farkas 1984;Riegler et al. 2014;Sharpe et al., 2015;Chi et al. 2017;Shabaga et al. 2017). The tectonic history, the related structural controls, and the relative timing of U mineralization in the prospects have not been studied in detail, and are consequently poorly understood, despite being key parameters for the URU mineralization. A study of the Contact prospect (Grare et al. , 2018a has provided a detailed picture of the successive pre-and syn/post-Thelon fracturing/faulting events at different scales and their role in the formation of U mineralization. A major difference with the Athabasca Basin, where ductile to brittle tectonics involving reactivated Hudsonian graphitic-rich shear zones exerted a major structural and metallogenic control on the formation of the U mineralization (Martz et al. 2017;Pascal et al. 2015), is that the tectonic style of deformation in the Kiggavik area is dominantly brittle, as exemplified by prevailing cataclastic to ultracataclastic fault rocks (Grare et al. 2018a, b). Graphite is also almost absent from the basement lithologies and faults within the Kiggavik area.
Based on the study of fluid inclusions in quartz-carbonate veins spatially associated with U mineralization at the End prospect, Chi et al. (2017) proposed that U mineralization initially formed from high salinity (> 25 wt% eq NaCl + CaCl 2 ), 80-200°C Na-to Ca-rich basinal brines derived from the Thelon Basin. These Na-to Ca-rich basinal brines are slightly more saline than the original Na-dominated diagenetic fluids defined within the Thelon Basin (Renac et al. 2002) but are similar in terms of temperature and composition to the mineralizing fluids in the URU deposits of the Athabasca Basin (Pagel 1975;Kotzer and Kyser, 1995;Derome et al. 2005;Richard et al. 2010Richard et al. , 2012Richard et al. , 2016Chu and Chi 2016;Martz et al. 2018;Wang et al. 2018). For the Bong prospect, Sharpe et al. (2015) proposed a five-stage metallogenic model with primary U mineralization occurring at ca. 1500 Ma, followed by different U mineralization stages/remobilization almost until the present time. The nature of the mineralizing fluids are unconstrained for the first stage and those related to the second stage of mineralization (at 1100 Ma) are isotopically closer to meteoric water than to basinal brines. On the basis of chemical characterization and dating of alteration and U mineralization, Shabaga et al. (2017) proposed that three mineralization events (ca. 1030 Ma, 530 Ma and < 1 Ma) formed the Andrew Lake deposit and that fluids were initially dominantly meteoric and oxidizing and secondly became more acidic. Illite-chlorite-U oxide association is commonly observed (Riegler et al. 2014;Sharpe et al., 2015;Chi et al. 2017;Shabaga et al. 2017) for the primary stage of mineralization in the whole Kiggavik area. The hydrothermal chlorite is dominantly sudoitic in composition, as observed for the hydrothermal chlorite linked to U mineralization in the Athabasca Basin (Hoeve and Quirt 1984;Kotzer and Kyser 1995). Anand and Jefferson (2017) display a structural synthesis for the Kiggavik area established from satellite imagery, geophysical interpretations, and field observations in which a district-scale structural evolution from ca. 2760 to 447 Ma is proposed, encompassing multiple reactivations of a Riedel shear system. Their approach contrasts with the detailed sequence of fracturing/faulting events published by Grare et al. (2018a) for the Contact prospect that was reconstructed on the basis of both macroscopic (drill core and field data) and microscopic (optical microscopy, cathodoluminescence and SEM) paragenetic observations, and which includes precise chronological constraints.
This review of previous works highlights that the nature of the U mineralization at Kiggavik and its controls (structural, fluid, timing, and physico-chemical) are still under debate. Despite the work by Anand and Jefferson (2017) that provides insights into the succession of regional tectonic events, the link between the macroscale and the microscale fractures and U mineralization, and the 3D architecture of faults zones in the Kiggavik area remains unconstrained. These points need however to be carefully addressed in order to build a complete metallogenic model of U mineralization in the Kiggavik area.
The twofold aim of this contribution is (1) to depict the brittle tectonic stages and the spatial-temporal structural evolution that led to the formation of the fault/fracture network that controls U mineralization in the Kiggavik area, and (2) to constrain the nature of the paleo-fluids and the conditions and timing of their circulation through the identified fracture networks by the analysis of the resulting fluid-rock interactions (cements, clays, ore minerals), and U mineralization. As an outcome of this multi-source approach, we propose an integrated structural and metallogenic model for the formation of the U mineralization in the Kiggavik area and make comparisons with the world-class URU mineralization from the Athabasca Basin. This model includes both pre-Thelon and post-Thelon Basin U ore formation stages, and for the first time fully illustrates the genesis and evolution of U mineralization of the prospective Kiggavik area and related Thelon Basin.

Regional setting
The Thelon Basin and the Athabasca Basin are Proterozoic intracratonic basins (Gall et al. 1992;Creaser and Stasiuk, 2007;Davis et al. 2011;Hiatt et al. 2003;Jeanneret et al. 2017) hosted in the Churchill Province. The Churchill Province resulted from the collisional amalgamation of the Rae and Hearne cratons along the Snowbird Tectonic Zone (STZ), either in the Neoarchean or during the Snowbird  Fig. 1). These basins are located between the eroded remnants of the Trans-Hudson orogenic belt to the southeast (ca. 2070-1800 Ma, overall NW-SE shortening) and the Thelon-Taltson orogenic belt to the west (ca. 2020(ca. -1900. The Thelon Basin mainly consists of the Thelon Formation, a 1800-m-thick sedimentary pile of conglomerates and coarse-grained sandstones, overlain by the ca. 1540 Ma shoshonitic basalts of the Kuungmi Formation (Chamberlain et al. 2010) and marine dolomites from the Lookout Point Formation (Gall et al. 1992) of the Barrensland Group (Fig. 1).
The Thelon Formation unconformably overlies a complex set of sedimentary and bimodal volcanic-sedimentary rocks that infilled the Baker Lake Basin between 1850 and 1750 Ma (Rainbird et al. 2006;Rainbird and Davis 2007). The Baker Lake Basin formed as a result of retro-arc extensional to transtensional rifting tectonics related to the collision between the Churchill province and the Superior province (i.e., the Trans-Hudsonian orogeny). It was followed by uplifting, extensive erosional peneplanation, and regolith formation, then by deposition of the Thelon Formation in response to thermal subsidence (Rainbird et al. 2003;Rainbird and Davis 2007;Hadlari and Rainbird 2011). The Barrensland, Wharton, and Baker Lake groups are parts of the Dubawnt Supergroup (Peterson, 2006) (Fig. 1) which overlies the metamorphosed Archean basement. The latter comprises ca. 2870-Ma granitic gneisses  Group , and a distinctive package of 2620-2580-Ma felsic volcanic and related hypabyssal rocks known as the Snow Island Suite Peterson 2015;Johnstone et al. 2016). These groups are overlain by the Paleoproterozoic (2300-2150 Ma, Rainbird et al. 2010) orthoquartzite of the Ketyet River Group (Fig. 2). These lithologies, especially the Woodburn Lake Group, host the U mineralization in the Kiggavik area (Fig. 2). The Archean to Paleoproterozoic rocks of the Churchill province were intruded by three intrusive suites: (i) the late syn-orogenic (ca. 1830 Ma) Hudson suite (Peterson et al. 2002); (ii) the Dubawnt minette suite, contemporaneous of the Hudson suite, with ultrapotassic rocks, minette dykes, and lamprophyres; and (iii) the rapakivi-style Nueltin granite of the anorogenic (ca. 1750 Ma) Kivalliq igneous suite (Hoffman 1988;van Breemen et al., 2005;Peterson et al. 2015;Scott et al. 2015). Minor U mineralization is hosted in fractures and faults developed in Hudson granitoids at the 85W prospect. Dykes of the Mackenzie diabase swarm (1267 ± 2 Ma; LeCheminant and Heaman 1989) form prominent linear aeromagnetic features trending NNW-SSE (Tschirhart et al. 2013(Tschirhart et al. , 2017. They cut across all of the previous lithologies and represent the last magmatic-tectonic event in the region.

Local setting, main fault structures, and tectonic evolution
In the Kiggavik area, the basement lithologies are intruded by the Schultz Lake Intrusive Complex (SLIC) . The SLIC are comprised from two intrusive suites as described by Scott et al. (2015): (i) the "Hudson granite," nonfoliated granitoid sills, syenite, and lamprophyre dykes of the late syn-orogenic Hudson suite; and (ii) the "Nueltin granite," anorogenic granite to rhyolite of the Kivalliq Igneous Suite (KIS; Peterson et al. 2015). Uranium and thorium primary enrichment were described to be associated with bostonite dykes  and pegmatites attributed to emplacement of the Dubawnt minette (ca. 1830 Ma) and KIS, respectively. Such bostonite dykes were observed near the Kiggavik Main Zone (Anand and Jefferson 2017). These rocks are unconformably overlain by the Thelon Formation, which crops out in the northern part of the Kiggavik property (Fig. 2). Several dykes of the Mackenzie swarm form NNW-SSE lineaments and are also observed in the area.
The main structural features in the Kiggavik area are the ENE-trending Thelon Fault, the ENE-trending Main Zone fault in the northern part of the property, the ENE-trending Judge Sissons Fault in the central part, and the NE-trending Andrew Lake Fault in the southwestern part (Fig. 2). The Main Zone fault hosts the 85W, Granite grid, and Kiggavik (main, central, and east zones) occurrences. The End prospect is hosted by the Judge Sissons Fault, while Andrew Lake, Jane, and Contact occurrences lie along the Andrew Lake Fault (Fig. 2).
Little is known of the long tectonic history predating the Trans-Hudsonian orogeny; however, it is probable that the tectonic initiation of the main faults in the Kiggavik area dates back to the convergence of cratonic blocks during the latest stage of Trans-Hudsonian orogeny. The Thelon Fault constitutes the contact between the siliciclastic sedimentary rocks of the Thelon Formation to the north and the metamorphosed basement rocks to the south (Fig. 2). South of the Thelon Fault, magnetic maps show that the SLIC is cross-cut by numerous ENE-trending parallel and sub-parallel faults with apparent right-lateral displacement (Tschirhart et al. 2017). The steep northward dip of the Judge Sissons Fault was inferred from observations in discontinuous outcrops and on drill cores. The Andrew Lake Fault constitutes the mapped boundary between the Hudson granite to the west and the metamorphosed basement rocks to the east (Fig. 2). The Andrew Lake Fault is delineated from interpretation of aeromagnetic and ground gravity maps (Fig. 2) (Roy et al. 2017;Tschirhart et al. 2017), as outcrops are nearly absent.
The study of the Contact prospect (Grare et al. 2018a) provided new insights into the structural evolution of the Andrew Lake Fault, which also included a post-Thelon U mineralization event. Grare et al. (2018a) highlighted the presence of an extensive silicification event along the Andrew Lake Fault, characterized by a quartz-healed breccia, termed the Quartz Breccia (QB, Grare et al. 2018b). This structure is a main feature of both the Judge Sissons Fault and the Andrew Lake Fault: it is observable on outcrops (Anand and Jefferson 2017) and is systematically intersected by drill holes (Grare et al. 2018b). Finally, the main stages of fracturing and mineralization recognized by Grare et al. (2017;2018a, summarized in Fig. 3) at Contact include brittle tectonic activity along the Andrew Lake Fault (fracturing stage 1 or f1), development of the QB (f2), the first stage of U mineralization (f5), the second stage of U mineralization (f6), faulting associated with strong clay alteration and bleaching of the host rock (f8), and late U mineralization remobilization linked to the circulation of meteoric fluids (f9).

Methodology
Collection of structural data from field and drill cores One hundred and forty samples were collected from drill cores obtained during 2014 and 2015 exploration campaigns by Areva Resources Canada (now Orano Canada); 28 samples came from Andrew Lake, 5 from End, 14 from Kiggavik Center Zone, 66 from Contact, and 27 from 85W (Fig. 2). In addition to classical drilling carried out during the exploration campaigns, two double tubed exploration drill holes (CZ-15-01, at Kiggavik Center Zone, and AND-15-01 at Andrew Lake), allowing for a better preservation of the drill cores, were examined. Several drill holes (~50) from previous exploration campaigns were also described in detail (fracture orientations, cross-cutting relationships, and characterization of cements, alteration, and mineralization).
Field observations of faults, joints, and veins were made on outcrops and combined with previous field observations. Fault core zones were characterized in drill cores by the identification of breccias or gouges. Fault damage zones (Chester and Logan 1986;Wibberley et al. 2008;Faulkner et al. 2010 Fig. 3 Orientations of faults and breccias for each fracturing stage and for the main prospects of the Kiggavik area. The sequence of faulting/ fracturing events and cross-cutting relationships are displayed on the first and second columns for reference. The presence of each fracturing stage is relative to the same fracture set or to the same prospect for the whole Kiggavik area. The chronology of the fracturing events was presented by Grare et al. (2018a). The qualitative frequency of the fracture sets for each prospect is shown by colored symbols (e.g., f7, white clay-altered fault zones, is dominant at Bong). Schmidt's lower hemisphere stereoplots "undifferentiated fracture" in this case relates to a fracture plane which cannot be unambiguously classified as vein, joint, or fault/microfault (no kinematic indicator). Fracture corridors and isolated veins, joints, and undifferentiated fractures were also systematically reported. Oriented data from Orano Canada's Kiggavik area database were collected and processed. Orientations of~2000 fractures and foliation planes were measured from oriented drill cores. Acoustic televiewer probing ABI40 (Williams and Johnson 2004) was run through several holes providing accurate oriented data in faulted core intervals. The mesostructural data were processed to their true orientation and plotted with Dips 6.0 software by Rocscience (see ESM 1 for detailed methodology about processing oriented data). Uncertainty on orientation measurements is usually about 10°as estimated from the comparison between oriented core-measurements and acoustic televiewer data.

Optical microscopy and scanning electron microscope
Thin sections made on veins and fault rocks were studied using a Motic BA310 POL trinocular, using transmitted and reflected lights, and a HIROX SH-3000 scanning electron microscope (SEM) equipped with a back-scattered electron detector and a nitrogen free energy dispersive spectrometer (EDS) BRK D351-10 with digital mapping capabilities at Orano (Paris, France). The SEM was operated at low accelerating voltage (10 kV), 100-nA filament current, and 600-Å beam width for a working distance between 8 and 39 mm.
Complementary observations on mineralogical observations and U mineralization were performed at Service Commun de Microscopie Electronique et de Microanalyses (SCMEM) of GeoRessources lab (Vandoeuvre-lès-Nancy, France), using a JEOL J7600F scanning electron microscope equipped with an EDS.

Cathodoluminescence
The different generations of carbonate vein cements were characterized using a Technosyn cold cathodoluminescence instrument (model 8200 MkII), operating between 10-and 12-kV gun potential and between 150-and 350-μA beam current. Observations were carried out at the University of Barcelona, Spain.

Electron microprobe analysis and geothermometry of clay minerals
The chemical composition for major elements in U oxides and clay minerals (mainly chlorite and illite) were measured using a CAMECA SX-100 Electron Microprobe Analyser also at SCMEM (see ESM 1 for detailed methodology).
Complementary measurements on U oxides and clay minerals (chlorite and white mica) were made on 6 thin sections with a CAMECA SX-50 electron microprobe and conducted at the Camparis service in Sorbonne Université (Paris), based on the same analytical conditions as used at SCMEM. Temperatures were calculated from chlorite composition following the equation of Cathelineau (1988), Zang and Fyfe (1995), and Kranidiotis and MacLean (1987). White mica crystals were selected from the main altering and/or mineralization stages for electron microprobe analysis, and major element compositions (site occupancy and end-member mineral data) were used to calculate the precipitation temperatures. Temperatures were calculated following the equation of Cathelineau (1988).
Secondary ion mass spectrometry (SIMS) and LA-ICP-MS U-Pb dating and geochemical tracing of U oxides The Pb/Pb and U-Pb isotopic compositions of U oxides was determined (see ESM 1 for detailed methodology) using a CAMECA ims 1280-HR Secondary Ion Mass Spectrometer (SIMS) at CRPG-CNRS (Nancy, France). The standard used was a uraninite sample from Zambia (concordant age of 540 ± 4 Ma; Cathelineau et al. 1990), analyzed before and after each sample for sample bracketing calibration. The 204 Pb/ 206 Pb ratio were low (< 0.00001) for the standard and unknowns, indicating that common Pb was not incorporated at the time of crystallization, except for sample 9850 (0.000019-0.004619). A correction for common Pb was however made for each analytical spot for all samples by precisely measuring the 204 Pb amount and by calculating the composition of the common Pb at the time of crystallization, based on the 207 Pb/ 206 Pb measured age and using the Pb isotopic composition calculated from Stacey and Kramers (1975) model. Ages and error correlations were calculated using the ISOPLOT flowsheet of Ludwig (Ludwig 2007). Uncertainties in the ages are reported at the 1σ level. Rare earth element (REE: La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu) concentrations in the different U oxides were quantified using a LA-ICP-MS system (GeoLas excimer laser (ArF,193 nm,Microlas) coupled to Agilent 7500c quadrupole ICP-MS at GeoRessources lab) with a 16-24-μm spot size. The details regarding instrumentation and methodology are described in Lach et al. (2013).

Fluid inclusions and fluid inclusion planes
Secondary fluid inclusions (FIs) organized in fluid inclusion planes (FIPs) were studied in primary magmatic quartz (primary quartz in granite and granitic gneiss, predating all other quartz generations) from 85W and Contact using a similar approach to Mercadier et al. (2010), Wang et al. (2015), and Martz et al. (2017). The objective was to define the properties of the successive fluids that circulated through those lithologies and the related tectonic stress, as FIPs are formed during healing of mode I microfractures. FIPs have been proposed to form perpendicular to the least principal stress axis σ3, i.e., in the plane that favors the maximal decrease of the total energy of the system (Lespinasse and Pêcher 1986;Gueguen et al. 1995;Lespinasse 1999;Lespinasse et al. 2005). A selection of four oriented samples was collected at 85W and Contact (only locations where oriented samples were available) and attention was paid to take samples away from main fault zones when possible in order to avoid local perturbation of the stress field. FIP dipping higher than 70°represent the majority of the FIPs in selected samples, and the study hence focuses on this group. Thus, a thin section and a wafer were prepared along a horizontal plane from selected zones of the oriented drill core in order to have a "map view" of the subvertical (dip > 70°) FIPs. FIP orientations were measured under transmitted light microscope using the AnIma software (Lespinasse et al. 2005). More than 100 orientation measurements were done per sample, on a restrained zone (~25 mm 2 ), to generate statistically robust data.
Microthermometry was carried out on FIs from the FIPs using a Linkam® MDS600 heating-cooling stage, adapted to an Olympus® microscope at the GeoRessources lab. A total of 61 secondary fluid inclusions from FIPs were studied for microthermometry (see ESM 1 for detailed methodology).
Gas species in the vapor phase of fluid inclusions were determined at room temperature with a Labram Raman microspectrometer equipped with an Edge filter, a holographic grating with 1800 grooves per millimeter, and a liquid nitrogen-cooled CCD detector at GeoRessources (Nancy, France) (Dubessy et al. 1989).

Results
Macroscopic texture, geometry, and kinematics of fracturing/faulting events The approach and methods adopted in the present work are similar to those of Grare et al. (2018a). Figure 3 shows the oriented mesostructural data collected in the prospects of the Kiggavik area (except for Jane and Sleek). A detail of oriented data displaying kinematic indicators is presented in ESM 2. Together with orientation data, a statistically significant number of kinematic indicators on veins and faults were also gathered mainly from outcrops located in different parts of the Kiggavik area, usually close to a known prospect, and also from drill cores; data are synthetized in Fig. 3. The aim is to establish whether the sequence of events recognized at Contact is also valid in the other prospects. Grare et al. (2018a) recognized 10 fracturing events at Contact among which three resulted in fracture-controlled U mineralization (Fig. 3).
Fracture stage 0 (f0) and oldest stage of U mineralization (U0, Fig. 3) This stage is named f0 and U0 in order to respect the annotation from f1 to f10 presented in Grare et al. (2018a) as it predates stage f1/U1 observed at Contact. The f0 event, unrecognized at Contact, has been identified at End, Andrew Lake, Kiggavik Main Zone, and Bong. This fracturing event is characterized by U-mineralized microbreccia usually displaying a dense network of millimeter-wide greenish microfractures (ESM 2), in breccia zones up to a 10 m thick (e.g., in End). Uranium mineralization is observed within the microfractures and is weakly disseminated within the host rock. The mineralization is usually weak but can reach several thousands of counts per second (CPS, AREVA SPPɣ®; i.e., 1.25 %U 3 O 8 over 1 m) in some breccia zones displaying more intense fracturing. Irrespective of the grade, the brecciated and mineralized host rock is weakly clay altered. This mineralized event is cross-cut and partially overprinted by the QB (ESM 2), precluding collection of any reliable oriented data in most places. Thus, oriented data for the mineralized microbreccia are rare; however, data from End display a NE-SW trend and a steep dip to the NW. Numerous microstructures similar to solution seams are observed at the microscale and display a NW-SE to NNW-SSE orientation (data not displayed in Fig.  3).
Stage f1, pre-Thelon fracturing event (Fig. 3) At Contact, this stage is marked by faults with proto-to ultra-cataclastic fault rocks along the Andrew Lake Fault (Grare et al. 2018a), showing evidence of later silicification. These fault rocks are crosscut by white quartz veins and mosaic breccia of QB. Because Si-bearing fluids that formed QB (f2; Fig. 3) reused preexisting main fault zones in the Kiggavik area (e.g., ENE-WSW trending Judge Sissons Fault; Grare et al. 2018b), the related faults/fractures display variable orientation. The QB shows a NE-SW trend at Contact and Andrew Lake. Secondary ENE-WSW fracture directions are also observed at Jane and Andrew Lake. The QB displays a main ENE-WSW trend with a steep dip to the north at End and Kiggavik Main Zone. At Bong, the QB is characterized by a NNE-SSW direction and a steep dip mainly to the west and a secondary E-W direction with a steep dip to the north. At 85W, mainly small white quartz veins were observed, displaying dominant E-W directions with steep dip to the north. Quartz veins (from QB, f2) display stepped veins, associated with (Fig. 4a)  Stage f2 (Fig. 3) This stage is observed in all prospects and generated the QB (Grare et al. 2018b), with its characteristic mosaic white quartz-sealed veins and breccia. Clasts of the QB are observed in the Thelon Formation, indicating its later reworking during deposition of the Thelon Formation. The QB is recognized along major fault zones at Bong and Kiggavik Main Zone along the ENE-WSW trending Main Zone fault, at Jane along the NE-SW trending Andrew Lake Fault and at End along the ENE-WSW trending Judge Sissons Fault. Uranium mineralization (all stages) occurs in the hanging wall and footwall of the QB but not within the breccia core. The f1 and f2 fracturing stages, and the structural control exerted by the QB on later U mineralization, were described in detail by Grare et al. (2018a, b).
Stage f3 (Fig. 3) This stage is represented by centimeter-thick dolomite veins and microbreccias, and was observed in all prospects of the Kiggavik area. The microstructures related to this stage cross-cuts the quartz veins of the QB.
Stage f4 (Fig. 3) This stage is marked by calcite-cemented (Ca1) veins and microbreccias that cross-cut the quartz veins of the QB. The veins were described by Grare et al. (2018a) as mixed mode-I mode-II shear fractures. They are observed in all prospects of the Kiggavik area, spatially associated with Umineralized fault zones (f5). Ca1 calcite displays a yellowishorange color under cathodoluminescence and a darker tint when closer to U oxides. Ca1 is observed as cementing microbreccias in samples from 85W and End, where U oxides coat the edges of calcite crystals. U oxides also occur as subhedral colloform shapes intergrown with the Ca1 cement, thus indicating synchronous precipitation of calcite and U oxides. Ca1 veins occur in both E-W and NW-SE directions, with conjugate dip directions. Kinematic indicators are rare but stepped veins cemented with Ca1 (f4) and U oxides (U1) were observed in drill cores (Fig. 4b). These veins display ENE-WSW and NW-SE directions.
Stage f5 (Fig. 3) This stage corresponds to the second U mineralization event (also called U1, see ESM 2). It is characterized by gray-greenish-colored, clay-altered narrow fault zones which were also observed in all prospects of the Kiggavik area. Clay alteration is characterized by illite mainly with the illitization of feldspar and chlorite through chloritization of biotite and retrograde metamorphic chlorite. This fracturing event developed strongly mineralized fault zones (e.g., 4.76 %U 3 O 8 over 50 cm) with protocataclastic to cataclastic fault rocks and reopened, microfractured quartz veins from which the mineralization leaks into the foliation (see also Chi et al. 2017). At 85W, the U mineralization is completely hosted in Hudsonian granites. Due to the absence of foliation and to the coarser grain size of the minerals, the control of fractures on mineralization is stronger and easier to observe. In strongly mineralized fault zones, where the mineralization is disseminated in the host rock, biotite is coated with pitchblende, which expands along the mineral cleavage. Like at Contact (Grare et al. 2018a), clay alteration (illitization mainly, but also chloritization) of the host rock is weak to moderate for this mineralization stage. The f5/U1 fractures follow ENE-WSW to ESE-WNW faults at Contact (Fig. 3) and are dipping to the north. At Andrew Lake, mineralization is found within N-S and E-W faults, but also within reopened and/or microfractured quartz veins trending NE-SW (f2, QB stage). At 85W, faults and minor fractures show an ESE-WNW direction, with a steep dip mainly to the north.
Stage f6 (Fig. 3 . 3) are well observed throughout the Kiggavik area. At Contact, and Andrew Lake, they display N-S to NE-SW directions (i.e., the direction of the Andrew Lake Fault). Contact is located on a W-dipping segment of the Andrew Lake Fault, while Jane and Andrew Lake are located on an Edipping segment of the Andrew Lake Fault. The map of Fig. 2 suggests that End is likely hosted within an important relay zone of the Judge Sissons Fault where oxidized fault zones are trending mainly NNW-SSE to NE-SW (Fig. 3). At Sleek, the three main faults trends are N-S, E-W, and NW-SE, only the latter two being visible on regional maps. At Bong, 85W and Kiggavik Main Zone, the fault trends are similar to those of the QB. However, at Kiggavik Main Zone, like in Contact, an E-W trend with a shallow dip to the south is also recognized. Stage f6 fractures were observed in the field to be associated with radioactive anomalies. Mineralized fractures of f6 stage (Fig. 3) are better represented at Contact, Andrew Lake, and Kiggavik Main Zone. At Andrew Lake, U2 mineralization is characterized by ESE-WNW to NW-SE microfaults and by NE-SW millimeter-wide Ca2 veins. Ca2 veinlets are usually too thin to be measured, so reliable oriented data were obtained only at Andrew Lake. At End, U2 mineralization was guided by NW-SE to NNW-SSE faults that mainly dip to the east. At 85W, this mineralization stage was controlled mainly by W-dipping NNE-SSW faults and by NW-SE faults as a second direction. At Kiggavik Main Zone, f6c faults trend either ENE-WSW or NW-SE to WNW-ESE. Kinematic indicators were observed in the field and in drill cores and are presented in Fig. 4. Study of relay zones in the St-Tropez area, which is located 20 km to the NNE of Kiggavik Main Zone, allowed for the interpretation of the kinematics of the fault zone (Fig.  4c). Faults with dominant E-W to ENE-WSW orientation show evidence for reverse and/or sinistral motion. Other faults with dextral indicators were also observed at Contact. The Judge Sissons Fault, trending ENE-WSW, displays a reverse-sinistral slip component (ESM 2), which postdates the main dextral normal slip component responsible of the offset of the Hudsonian granite (SLIC, Fig. 2, see also Anand and Jefferson 2017).
Stages f7-f8 (Fig. 3) These stages represent the strongest clay alteration events in the Kiggavik area, characterized by illitization, dissolution of quartz, and bleaching of the host rock. They are observed in all prospects. Numerous, un-mineralized, illitized fault zones were observed at Bong (ESM 2). These faults cross-cut mineralized fault zones (Fig. 4e) and reworked clasts bearing U oxides are observed within cataclastic fault rocks. This fracturing stage, recognized for each prospect, was extensively observed at Bong. The greenish color of some fault zones is due to the presence of reworked retrograde metamorphic chlorite and therefore does not reflect a different fracturing process; additionally, white clay fault zones (f8) are much more frequent compared to greenish clay fault zones (f7). On the basis of these new observations, we propose to gather these two fracturing events that were first presented as distinct in Grare et al. (2018a). The strongest clay alteration event is definitively termed f7. f7 fractures were recognized ( Fig. 3) to be associated with NW-SE and E-W trending fault zones (e.g., 85W and End). At Bong, it is characterized by E-W trending, N-steeply dipping fault zones. These post-ore fractures display NNW-SSE direction with evidence of strike-slip sinistral motion in End and 85W areas; these faults display E-W to WNW-ESE orientation, steeply dipping to the north, with a dip-slip normal kinematics. At 85W, microfaults offset U mineralized f6 fractures (Fig. 4d). At Bong, the orebody is offset by~E-W striking faults (Fig. 4e). The meter to tens of meters dipslip displacement of the U orebodies appears to be greater than the strike-slip component (cm-to m-scale).
The Mackenzie diabase dykes trend NNW-SSE in the Kiggavik area. Mackenzie dykes were observed only at Kiggavik Main Zone; however, these dykes provide strong constraints on the timing of the fracturing events. The diabase dykes cross-cut the altered and mineralized fault zones (ESM 2) and are rarely fractured and altered; they can therefore be considered having emplaced after the main tectonic and mineralization history.
Stage f8 (Fig. 3) This stage is represented by a weak reactivation of the fracture network and U redox fronts (U3) as described at Bong (Sharpe et al., 2015), Andrew Lake (Shabaga et al. 2017), and Contact (Grare et al. 2018a); they have also been recognized at End, Kiggavik Main Zone, and 85W (ESM 2). Even after careful review of drill cores throughout the Kiggavik area, it was not possible to precisely link redox fronts to specific fault zones. These remobilization fronts display the typical following succession: an oxidized (goethite), un-mineralized zone; a thin black layer where U oxides are concentrated; and a gray, reduced mineralized zone. These redox fronts are sometimes bleached by a later event described at the Contact prospect (f10), which removed iron oxides from the goethite-rich zone. It was not possible to link this final event to a specific set of fractures.
Macroscopic re-examination of prospects in the Kiggavik area therefore confirms the structural stages previously defined by Grare et al. (2018a) at Contact. As such, the sequence of fracturing events can be considered representative of the whole Kiggavik area, although changes in lithologies between the various prospects may have induced differences in the mineralogy of fault rock cements and alteration products. In addition, our approach allowed for the identification of a new stage of deformation and mineralization (f0 and U0). This initial mineralization stage is hereinafter characterized more in detail with the help of microscopic observations and geochemical data.
Mineralogy, ages of U oxides, and chemical compositions of associated clays

Mineralogy and microscopic textures of U mineralization
Samples of the U-mineralized microbreccia (f0, U0) display multiple irregular microstructures with interlocking pegs and sockets (Fig. 5a) similar to solution seams and usually surrounded by a millimeter-scale halo of quartz dissolution. These microstructures connect to microbreccias, within which millimeter clasts display irregular boundaries reflecting dissolution patterns (Fig. 5b). These microstructures are cross-cut by the QB veins ( Fig. 5c and e). Electron microbeam imaging and analyses show that microfractures are cemented with ironrich clinochlore, pitchblende (containing up to several percent of thorium), brannerite, titanium oxides, and sulfide minerals (mainly pyrite). These minerals are subhedral to anhedral mixed phases ( Fig. 5d and f). Microprobe analyses show that the iron-rich clinochlore is altered into an Mg-rich chlorite (Fig. 5e), close to sudoite (ESM 3). Subhedral crystals of brannerite were observed in the same type of fractures/ breccias in samples from Kiggavik Main Zone, associated with pitchblende-cemented microbreccia (Fig. 5f, g). Rutile with micro-inclusions of U oxide and coated with anhedral U-Ti phases were also observed on samples from Andrew Lake and 85W.  Stage f5 and associated U1 is characterized by polymetallic mineralization (few f5 faults are associated with monometallic mineralization). This stage consists of pitchblende and uraninite in fractures often cemented by calcite (Fig. 5h). The U1 event is spatially associated with open quartz veins displaying colloform pitchblende along the edges, with U oxides precipitated in voids created by quartz dissolution. Bravoite-pyrite coated with pitchblende ( Fig. 5i) was observed on samples from Jane, a mineralogical association also described at Contact (Grare et al. 2018a). Silver, gold, and Mo-Ni-As-Co compounds were observed in samples from Andrew Lake, End, and Kiggavik Main Zone. Anhedral Ag-U phases coating galena appear to cement microfractures; native gold is usually present as micro-inclusions within pitchblende. Aluminum phosphate sulfate (APS) minerals were observed associated with pyrite and pitchblende in a gray-greenish fault zone at 85W (Fig. 5j).
Compared to U1, f6 and associated U2 displays monometallic mineralization, characterized by pitchblende and rare sulfides (mainly pyrite). U oxides are associated with APS in oxidized rocks at 85W. Illite and sudoite were observed for both U1 and U2 mineralization stages. Anastomosing microfractures cemented by anhedral pitchblende are characteristic of this mineralization stage and were observed at Andrew Lake (Fig. 5k).

Rare earth element composition
The rare earth element (REE) concentrations for U1, U2, and U3 are available in ESM 3. Figure 6 a displays the chondritenormalized patterns for samples from Andrew Lake and End. U0 oxides could not be analyzed by LA-ICP-MS due to their small size, anhedral shapes and mixing with Ti-U phases. Both U1 and U2 samples display bell-shaped chondrite-normalized patterns centered on Tb with a small anomaly in Eu (Fig. 6a) Sample 85W-10-04 (Fig. 6a) displays a different pattern compared to the other samples of the Kiggavik area with a much higher concentration of REE (up to a factor 10). The pattern displays a positive slope from La to Sm, and then a negative slope to Lu. Such a signature has never been described before in the Kiggavik area.

Composition and crystallization temperatures of chlorite and illite
Chlorite Chlorite from the Kiggavik area falls into three types: retrograde metamorphic chlorite, chlorite (altered and unaltered) associated with U0 mineralization, and chlorite associated with U1 and U2 mineralization. The compositions are presented in a chlorite ternary diagram in Fig. 7 a (Velde, 1985). Composition of chlorite associated with U0, U1, and U2 are presented in ESM 3.
Representative compositions of chlorite analyzed from U mineralized microbreccia samples (f0, U0) are plotted in Fig. 9 and chlorite cement analyzed from the microbreccia are shown in Fig. 7 c. Representative cation proportions (based on 14 oxygens) of unaltered chlorite associated with U0, together with calculated temperatures, are in ESM 3, and indicate an iron-rich species of trioctahedral chlorite. The octahedral occupancy is close to 6 atoms when the structural formulae are calculated with total Fe in the ferrous state and the Fe/(Fe+Mg) ratio is generally close to 0.5. Structural formulae of chlorite linked to U1 and U2 plot toward the sudoitic end member (di-trioctahedral chlorite, Fig. 7b, in ESM 3), characterized by an Al-Mg-rich composition and an octahedral occupancy close to 5 cations. Sudoite was also characterized through XRD by Pacquet (1993a, b, c). Numerous analyses of chlorite associated with U0 reveal a mixture of variable amounts of iron-rich chlorite and sudoite, showing that neoformed dioctahedral chlorite, co-genetic with U0, was altered to di-trioctahedral chlorite. This is consistent with petrographic observations showing dark green chlorite surrounded by light green chlorite (Fig. 7c).
Crystallization temperature were calculated using several methods, only for the Fe-rich chlorite variety linked to U0, as the geothermometers cannot be applied on low-T°ditrioctahedral chlorite such as sudoite. The Cathelineau (1988) and Jowett (1991) thermometers returned temperatures ranging from 295 to 333°C whereas Kranidiotis and MacLean (1987) and Zang and Fyfe (1995) thermometers returned temperatures ranging from 130 to 159°C.

Illite and mica
Representative compositions of white micas associated with pre-Umin alteration (pre-U0), with U1-f5 and U2-f6 mineralization stages and late clay alteration micas (f7) are detailed in ESM 3 and plotted on ternary diagrams representing the average compositional fields of dioctahedral mica-like phases (Dubacq et al. 2010; Fig. 7d). Pre-U mineralization (f1) alteration is characterized by interlayered illite/smectite and smectite. Most of the white mica alteration characteristic of U1 and late ore clay alteration consist of illite. White mica characterizing U2 is represented by illite and interlayered illite/smectite. The distribution of the data points in the diagram likely indicates inheritance of illite from muscovite/sericite. Crystallization temperatures calculated using the equation of Cathelineau (1988) return an average of 224°C for pre-U mineralization alteration, 289°C for U1, 282°C for U2, and 305°C for f7 (Fig. 7d). Calculation of temperatures using the Battaglia (2004) method yielded similar results. The temperature differences between the two methods for a given sample are generally within 10°C.

Composition of fluid inclusions
Two types of fluid inclusion planes (FIPs) were identified based on the nature of constitutive fluid inclusions (FIs): type 1 and type 2 FIPs (Fig. 8a). Rare FIPs present FIs that characterize type 1 and type 2 FIPs; they are labeled "mixed FIP" (Fig. 8a) and were not studied further.
Type 1 FIPs are represented by monophase, vapor-only, and dark fluid inclusions, rounded, with rare "negative crystal" shape (Fig. 8b). Their size varies from 3 to 10 μm. No phase change was observed for these FIs when  (Velde 1985 (Fig. 8c). They have variable shapes with a size varying from 3 to 20 μm. H 2 , O 2 , CH 4 , and N 2 were occasionally detected in trace amounts in the vapor phase of the biphase fluid inclusions by Raman spectroscopy. Triphase FIs are characterized by a dominant liquid aqueous phase with a vapor phase representing in most cases ca. 10% and display a cube of halite (Fig. 8d). They have variable shapes and their size range from 6 to 20 μm. H 2 and O 2 were occasionally detected in trace amounts in the vapor phase of the triphase inclusions using Raman spectroscopy. Type 1 FIPs are cross-cut by type 2 FIPs (Fig. 8e).
A summary of microthermometric data obtained for the analysis of biphase and triphase FIs (n = 61) is given in ESM 3. The FIs have been classified according to the nomenclature of Derome et al. (2005), also used by Richard et al. (2016) and Martz et al. (2017) in order to allow for a better comparison (refer to ESM 1). Over two third of FIs belong to the Lw′ or Lwh′ types that are typical of the so-called CaCl 2rich brine (Derome et al. 2005). They show high salinity (25.3 to 39.3 wt% eq. NaCl + CaCl 2 ), T h from 62 to > 215°C (no further heating to avoid decrepitation) and Ts NaCl generally higher than T h (120 to 235°C) for Lwh′. About 15% of the FIs belong to the Lw1 or Lwh types that are typical of the socalled NaCl-rich brine (Derome et al. 2005). They show slightly lower salinities (22.2 to 31.3 wt% eq. NaCl + CaCl 2 ), T h of 126°C and Ts NaCl close to T h (129°C). The remaining 15% of FIs belong to the Lw″ type that is typical of the so-called low-salinity fluid (Derome et al. 2005  eq. NaCl + CaCl 2 ), and T h systematically over 220°C, possibly related to the frequently high volume of vapor phase, up to 60% (no further heating above 220°C to avoid decrepitation).
The concentrations of several major, minor, and trace elements in aqueous fluid inclusions (mainly Lw′ and Lwh′ inclusions) were measured by LA-ICP-MS. The absolute contents of every element analyzed display a large range of values (Fig. 9), the highest being represented by Ca (31,000 to 80,000 ppm), Na (5600 to 38,000 ppm), K (3500 to 17,000 ppm), Mg (1200 to 14,000 ppm), Fe (from 5900 up to 11,200 ppm), and Sr (1000 to 6700 ppm). Metals such as Ba (590 to 14,000 ppm) and Pb (270 to 11,000 ppm) display moderate concentrations. Zinc (270 to 800 ppm), U (6 to 350 ppm), Cu (< 80 to 400 ppm), Rb (37 to 140 ppm), and Cs (23 to 40 ppm) display lower concentrations compared to other elements. The FIs are characterized by high concentrations in Ca compared to Na, indicating that FIs studied are dominated by a CaCl 2 -rich brines.
At Contact, type 1 FIPs were also the most represented type (75 FIP measurements out of 96); these FIPs display a main N130-150 direction, secondary N110-130 direction, and minor direction of N000-020. Aqueous biphase and triphase FIPs overprint type 1 FIPs like at 85W. Type 2 FIPs display two directions: N110-120 and N80-N100. The FIP directions were also compared with reference to the sub-constant N10/ 10°E orientation of the granitic gneiss foliation at Contact. Comparison is made with data from the Athabasca Basin; Richard et al. 2010Richard et al. , 2016 used. The evaluation of iron valence, thus the amount of Fe 3+ , directly impacts the calculation of the temperature and has always been an analytical challenge. However, Bourdelle and Cathelineau (2015) demonstrated that the lower the temperature, the lower the impact of Fe 3+ and a model based solely on Fe 2+ is suitable in terms of practicability and simplicity while still giving reliable results. In our study, the high temperatures (295-333°C) obtained using Cathelineau (1988) and Jowett (1991) methods for dioctahedral chlorite of this stage appear consistent with the presence of brannerite and the high Th contents of pitchblende in U0. A similar high Th content was also described by Weyer et al. (1987) at Kiggavik Main Zone. The incorporation of thorium (Th 4+ ) into the uraninite structure is indeed correlated with the temperature and is commonly observed in metamorphic or magmatic uraninite, which form at T > 300°C, contrary to hydrothermal U oxides which form at T < 300°C (Depiné et al. 2013;Mercadier et al. 2013;Frimmel et al. 2014;Cuney et al. 2015). The pre-Thelon high-temperature dioctahedral chlorite cementing f0 fractures would have likely been altered to sudoite at a later time, probably after the deposition of the Thelon Formation, by the circulation of Thelon-derived basinal brines thus crystallizing lower temperature di-trioctahedral chlorites (~100-200°C).

Evolution of temperature of mineralizing fluids
The circulation of such basinal brines is considered to be related to the formation of U1 and/or U2. Calculating the temperature of formation of sudoite is not possible, precluding any comparison with temperatures obtained by illite thermometry. Pre-U1/f5 white mica that crystallized at low temperatures (~220°C in average) could represent either early localized alteration along fault zones (f1) during development of the earlier Baker Lake and Wharton basins, or circulation of diagenetic fluids from the Thelon Basin. White mica synchronous with U1 and U2 returned temperatures of ca. 280-290°C, but slightly higher in the case of U1. These temperatures are higher than those typically obtained for illite in the Athabasca Basin (220-280°C, Ng et al. 2013;Chu and Chi 2016), which could indicate crystallization of illite and URU oxides at greater depth, and/or under a higher geothermal gradient for the Thelon Basin compared to the Athabasca Basin. A higher geothermal gradient, linked to magmatism associated with emplacement of the Kuungmi lavas at 1540 ± 30 Ma (Chamberlain et al. 2010), i.e., the age proposed for the emplacement of U1 (Sharpe et al., 2015), is our preferred hypothesis, as no geological process could explain an increase in depth of crystallization. A similar explanation was proposed by Chi et al. (2018) in order to reconcile elevated temperature recorded by FIs and shallow URU mineralization in the Athabasca Basin: they proposed the potential occurrence of a magma chamber underneath the basin.
H o m o g e n i z a t i o n t e m p e r a t u r e s o b t a i n e d b y microthermometry on type 2 FIPs (syn-/post-Thelon event) range from 62 to > 220°C. These temperatures are comparable to those obtained in the Athabasca Basin for diagenetic-hydrothermal fluids (Derome et al. 2005;Richard et al. 2016), and at End (100-300°C, Chi et al. 2017) from fluid inclusions in quartz spatially associated with post-Thelon uraninite and calcite in the Kiggavik area.
The shift of~100°C between temperatures given by fluid inclusions and those given by illite geothermometry has already been reported in the Athabasca Basin by Chu and Chi (2016). Such a difference could be related to the fact that the T h measured for the fluid inclusions represent the minimum value for the trapping temperature. However, pressure correction of the fluid inclusion data does not completely explain the difference in temperature (Richard et al. 2016;Chi et al. 2018).
Post-ore illite alteration, observed in f7 fractures, returned higher temperatures of 290-350°C. This increase in temperature could be linked to the emplacement of large volumes of magma associated with the giant dyke swarm of the Mackenzie event (Ernst et al. 1995;Hou et al. 2010). This magmatic event could have triggered the circulation of "hot" hydrothermal fluids shortly before emplacement of the diabase dykes and sealing of the faults, which would also be consistent with illite Ar-Ar ages obtained at ca. 1300 Ma at End, Andrew Lake, Kiggavik Main Zone, and Bong (Ashcroft et al. 2017;Shabaga et al. 2017; Fig. 10).

REE signatures
The chondrite-normalized REE pattern of U oxides is usually considered to be specific to each type of U prospect and to directly reflect its conditions of formation (temperature, redox conditions, fluid composition, REE source(s); (Mercadier et al., 2011a, b;Frimmel et al. 2014;Alexandre et al. 2015)). The U oxides from End (U1) and Andrew Lake (U2) display an overall bell-shaped REE pattern centered on Tb which is typical of URU systems (Bonhoure et al. 2007;Mercadier et al., 2011a, b;Eglinger et al. 2013;Alexandre et al. 2015). Such patterns were also obtained by Fayek et al. (2017) at Kiggavik Main Zone and unambiguously indicate that U1 and U2 at Kiggavik area crystallized under physicochemical conditions similar to U oxides from the Athabasca Basin URU prospects (Pagel 1975;Derome et al. 2005;Richard et al. 2010;Martz et al. 2018). Some U oxides however present modified bell-shaped REE patterns compared to unaltered samples from URU prospects of the Athabasca Basin, with enrichment in LREE, while concentrations of HREE remain the same (Fig. 6a). This modified bell-shaped REE pattern, observed for U oxides in remobilization fronts in the Kiggavik area (Fig. 8a), was first described by Mercadier et al. (2011a) for U oxides in a remobilization front at the Eagle Point U prospect in the Athabasca Basin, and is interpreted to be linked to the interaction between primary UO 2 and low-T (< 50°C) meteoric water. Such modified spectrum is observed for U1 and U2 from the Kiggavik area (Fig.  6a) and indicates that these initial U generations were chemically affected by meteoric water. This is consistent with their reset U/Pb isotopic ages (younger than 600 Ma; Fig. 6b) and with macroscopic observations of redox fronts containing goethite that formed in low-T and low-salinity conditions (Mercadier et al. 2011a). Such alteration and/or remobilization of brine-related U oxides by late low-T meteoric waters appears to be a common and widespread feature in the Kiggavik area, based on published and present data (Sharpe et al., 2015;Shabaga et al. 2017).
A REE pattern similar to the one obtained for the 85W sample has only been previously described for a U oxide from Zambia (Eglinger et al. 2013), which was considered an URU prospect. Such a pattern is however not characteristic of URU oxides, as observed for all prospects in the Athabasca basin (Mercadier et al. 2011b;Alexandre et al. 2015). The REE composition and the age (post-Mackenzie dyke) of this U oxide indicate that the conditions for the crystallization of these U oxides were different than U1 and U2 U oxides, but are still unknown.

Fluid inclusion compositions
Dense arrays of dark FIs (type 1 FIPs) cross-cut by syn/post Thelon high-salinity inclusions have been described previously in the Athabasca Basin (Mercadier et al. 2010;Wang et al. 2015;Pascal et al. 2016;Martz et al. 2017). However, those inclusions were dominated by CO 2 , CH 4 , and N 2 and interpreted as retro-metamorphic fluids equilibrated with graphite-rich lithologies. Here, the vapor-dominated aqueous inclusions could be explained by boiling of an aqueous fluid that could be linked to a pre-Thelon epithermal event, which has been characterized in/close to the Kiggavik area (Turner et al. 2001;Grare et al. 2018a, b). A similar event was also proposed for the Athabasca Basin . Therefore, considering that type 1 FIPs are cross-cut by type 2 FIPs, it is much more likely that these trends of vapor-rich FIs characterize the QB event. In addition, the absence of CO 2 , CH 4 , and N 2 could be linked to lack of graphite-rich lithologies in the Kiggavik area. H 2 and O 2 observed in FIs from type 1 and type 2 FIPs indicate radiolysis of H 2 O in presence of U (Dubessy et al. 1988;Richard 2017).
In type 2 FIPs, the occurrence of similar FI types as described in Derome et al. (2005) and Richard et al. (2016) attests to the presence of NaCl-rich and CaCl 2 -rich brines and a low-salinity-fluid at 85W. Such brines have been observed throughout the Athabasca Basin for the URU deposits. The characteristics of FIs described here show many similarities with those described at End by Chi et al. (2017). The major difference here is that no fluid unmixing was observed. Therefore, fluid unmixing observed by Chi et al. (2017) is probably related to a different fluid episode than that associated with type 2 FIPs. Temperatures and salinity of brine inclusions are higher than those obtained by Renac et al. (2002) in the Thelon sandstone away from mineralized zones: 100-160°C, ca. 17 wt% NaCl. The brines are thought to be primarily derived from evaporated seawater (Richard et al.  Fig. 10 Summary of ages within the Thelon/Baker-Lake area, through U-Pb isotopes on U oxides (Farkas 1984;Miller et al. 1986;J. Bridge et al. 2013;Lach et al. 2013;Sharpe et al. 2015;Chi et al. 2017;Shabaga et al. 2017; this study), Ar-Ar ages on illite (Friedrich et al. 1989;Riegler 2013;Ashcroft et al. 2017;Shabaga et al. 2017), Ar-Ar on muscovite (Ashcroft et al. 2017;Shabaga et al. 2017), K-Ar on illite (Friedrich et al. 1989) and whole rock analysis Roddick 1988, 1992a, b) 2011, Richard et al., 2013, Richard et al., 2014. The CaCl 2rich brines have been proposed to reflect interaction of NaClrich brine with the basement rocks (Derome et al. 2005;Mercadier et al. 2010;Richard et al. 2010Richard et al. , 2016Martz et al. 2018). In this study, we observed dominantly CaCl 2 -rich brine FIs, which would reflect important interaction of the basinal brines with basement rocks. The slightly higher temperatures measured for the CaCl 2 -rich brines in the Thelon Basin, compared to the Athabasca Basin, in line with the temperatures calculated for illite, could indicate a higher thermal gradient or a deeper percolation of brines within the basement. This could have well been the case in the Kiggavik area where the mineralization formed at an unconstrained depth in the basement rocks, below a now eroded cover of unknown thickness. Brine inclusions display similar major and trace element concentrations compared to NaCl-rich and CaCl 2 -rich brines from the URU deposits in the Athabasca Basin , Richard et al., 2013, Richard et al., 2016Martz et al. 2018), being extremely rich in U, Zn, Rb, Sr, Cs, Ba, and Pb. It is noteworthy that Mg is particularly low compared to the Athabasca basin and that boron was not detected, which is in line with the lack of Mg-tourmaline noted in the Kiggavik area compared to the Athabasca Basin. The data therefore indicate that metal-rich (in particular U-rich) and highly saline basinal brines circulated in the basement rocks of the Kiggavik area and formed U1 and U2 at ca. 1500-1275 Ma, in agreement with the REE pattern for unaltered U oxides of these two mineralizing stages.

Metallogenic model of the Kiggavik area
The combination of all structural, mineralogical, fluid inclusion, dating, and geochemical characteristics of the different U prospects from the Kiggavik area allows proposing a structurally controlled metallogenic model of U mineralization. The main faulting/fracturing events, the related tectonic stress, and the fracture network evolution are presented in synthetic block diagrams (Figs. 11,12,13,14,and 15). These block diagrams were built for Bong, Kiggavik Main Zone (both located oñ ENE-WSW trending faults), Andrew Lake, and Contact (both located on~NE-SW trending faults).
Pre-Thelon Basin faulting/fracturing events and the first stage of U mineralization (magmatic type U0) (f0, U0) The first U mineralization event was recognized in all major prospects in the Kiggavik area. U0 is characterized macroscopically by microbrecciation (f0) of the host rock and is better preserved at greater depth. Hydrothermal rutile cementing microfractures and coated with pitchblende at Contact (Grare et al. 2018a) are interpreted to be relicts of this first mineralization stage. Solution seams characteristics of this stage are traditionally observed in sedimentary environments (e.g., Rutter 1983; Benedicto and Schultz 2010) but have also been described in igneous rocks, for example in rhyolite and welded tuffs (Donald Bloss 1954;Burma and Riley 1955;Golding and Conolly 1962). Mineralized microfractures and the characteristic solution seams were also described by Miller (1980) and Lecheminant et al. (1979) in their study of the Kazan Fall U mineralization in the Baker Lake basin. Kinematic indicators associated with f0/U0 are represented by solution seams oriented NNW-SSE (Fig. 11). These solution seams likely indicate a roughly WSW-ENE trending σ1 consistent with what was proposed by Hadlari and Rainbird (2011) for the formation of the Baker Lake basin. Major faults (Andrew Lake Fault, Judge Sissons Fault, Main Zone Fault, probably Thelon Fault) were active at that time (Fig. 11a), probably under WSW-ENE σ1 and NNW-SSE σ3 (Fig. 11b), and mineralized. The prevalence of this mineralizing stage in prospects controlled by major ENE trending faults highlights the role played by major fault zones in focusing both fracturing events and fluid circulations through time for the Kiggavik area.
The f0/U0 is cross-cut by quartz veins of the QB (inferred to have formed at ca. 1750 Ma) and ages of ca. 1830 Ma were obtained on pitchblende within the Baker Lake Basin by Miller and LeCheminant (1985) and Bridge et al. (2009, Fig. 10). Such mineralization event is plausible in the Kiggavik area, since a breccia pipe of the Christopher Island formation (ca. 1833 Ma; Rainbird et al. 2006) was observed in the eastern part of the Kiggavik area, along with ultrapotassic minette and bostonite dykes (Anand and Jefferson 2017) of the Dubawnt igneous suite. These lithologies are enriched in U (LeCheminant et al. 1987;Miller and Blackwell 1992;Peterson et al. 2011) and represent a potential source of U. It is inferred that the U0 mineralization event occurred in the Thelon-Baker Lake area at the end of the Trans-Hudsonian orogeny (ca. 1830 Ma), in response to retro-arc extension with deposition of the Baker Lake formation and emplacement of U-rich peralkaline magmas of the Dubawnt igneous suite (Cuney 2014;Cuney et al. 2015). This U mineralization event is the first one in the Kiggavik area and is likely of magmatic origin. U0 constitutes the first significant "stock" of U oxides in the Kiggavik area that will be later remobilized. U0 is observed in significantly mineralized brecciated rocks, and would have been an easily available source of U for basinal brines during their circulations in basement structures along with possible other U sources, to form the younger U mineralization. The presence of a pre-Thelon-basin source of U is comparable to what is described near the Athabasca basin: volcanic and metasomatic U in the western margin (Dieng et al. 2013(Dieng et al. , 2015 or magmatic and metamorphic U described in the eastern margin . Secondary U sources in the Kiggavik area could have been the metamorphosed epiclastic rocks of the Puqik Lake formation  (Johnstone et al. 2017), rhyolitic flows of the Wharton group (Blake 1980;Peterson et al. 2015), bostonite dykes of the Dubawnt igneous suite (ca. 1830 Ma;LeCheminant et al. 1987;Miller and Blackwell 1992;Peterson et al. 2011), Urich pegmatite of the KIS (Scott and Peterson 2012), and Urich minerals such as monazite (Hecht and Cuney 2000;Richard et al. 2015).
(QB, f2) Intense quartz brecciation related to the QB event (linked to magmatism of the KIS dated at ca. 1750 Ma; Peterson et al. 2015;Grare et al. 2018b) occurred along major fault zones (i.e., Main Zone fault, Judge Sissons Fault, Andrew Lake Fault) in the Kiggavik area and also south of this area (Baudemont and Reilly 1997;Turner et al. 2001). Circulation of high-temperature, low-salinity fluids was focused along major faults and relay zones (Fig. 12a), silicifying previously formed fault rocks. The formation of the QB occurred under a likely stress regime with WNW-ESE trending σ1 and NNE-SSW trending σ3 (Fig. 12b). The silicification event controlled, at least partly, later fracturing and fluid circulation, depending on the thickness of the quartz breccia. The most fractured and silicified part of the QB (i.e., the breccia core) behaved as a transverse barrier for fluids but enhanced along-strike fluid migration and entrapment through reopening of veins along its outer boundaries. The QB played therefore a major role in partitioning fracturing and fluid flow during the later U mineralization stages (Grare et al. 2018a, b).
(Syn?) Post-Thelon faulting/fracturing events: second and third stages of U mineralization (URU-type U1 and U2) (f5, U1) This stage is characterized by monometallic to polymetallic mineralization within reduced narrow fault zones (Fig. 13a). With the exception of Contact, the f5 stage reflects the orientation of the main fault zones in which they are hosted (NE-SW trend of the Andrew Lake Fault for the Andrew Lake prospect, WNW-ESE trend of the Main Zone fault for the 85W prospect). Illite is the main alteration product but sudoite was observed in altered mineralized fault rocks at Jane (Miller 1997), End (Lida 1997), Bong, (Riegler et al. 2014;Sharpe et al. 2015), and Kiggavik Main Zone (Pacquet 1993a). Aluminum phosphate sulfate minerals, synchronous with U mineralization, were observed at 85W (this study), and by Riegler et al. (2016) at Bong. U1 postdates quartz veins and dolomite veins precipitated from highly saline, basin-derived brines (Chi et al. 2017 andRiegler, 2013;respectively), and is synchronous with the first generation of stepped calcite veins. Fluid inclusions in calcite were studied by Chi et al. (2017) at End. Although not characterized using cathodoluminescence, it is likely that this calcite material was made of Ca1 (as defined in our study) because veins of the second generation are usually too thin for a reliable study of fluid inclusions. This indicates that Ca1 precipitated from Thelon-derived highly saline basinal brines. The synchronicity between Ca1 and U1 suggests that faulting/ fracturing stage f5 associated with U1 corresponds to a stress regime with WNW-ESE σ1 and NNE-SSW σ3 (Fig. 13b). Such a stress regime is consistent with type 2 FIPs that display dominant WNW-ESE trends (Fig. 8f) indicating a NNE-SSW to NE-SW trending σ3. The WNW-ESE trending σ1 is further consistent with the dextral slip component along E-W to ENE-WSW trending fault relay zones in the QB and with WNW-ESE trending type 1 FIPs. This is also in agreement with the observations made by Anand and Jefferson (2017) who concluded that reactivation of ENE-WSW, north-dipping extensional faults (such as the Thelon Fault or the Judge Sissons Fault) occurred under a~N110-140°trending σ1. The oldest ages for U have been obtained by Farkas et al. (1984) and Sharpe et al. (2015) at 1403 ± 10 Ma and 1520 ± 79 Ma, respectively. Absolute ages obtained in this study for U1 are between 1293 ± 08 and 1187 ± 20 Ma for End prospect and 354 ± 47 Ma at Andrew Lake. Local heating and circulations of hydrothermal fluid associated with emplacement of the Kuungmi lavas at ca. 1540 ± 30 Ma (Chamberlain et al. 2010) are described as one possible first event of URU-type mineralization in the Kiggavik area (Sharpe et al. 2015). This age was observed throughout the Thelon-Baker Lake area (Turner et al. 2003;Bridge et al. 2013). Ages at 1500-1400 Ma are comparable to ages obtained on oldest URU mineralization in Athabasca (1514 ± 18 Ma, Cumming and Krstic 1992;1519± 22 Ma, Fayek et al. 20021540± 19 Ma, Alexandre et al. 2009). The time frame defined by age-dating studies and the reconstructed stress regime are consistent with the Racklan orogeny that occurred to the west of the Kiggavik at ca. 1600-1380 Ma (Cook 1992;Cook and MacLean 1995;Thorkelson 2000). This orogeny would be at the origin of the WNW-ESE shortening, as proposed by Anand and Jefferson (2017).
The ENE-WSW, north-dipping faults formed earlier during the deposition of the Baker Lake and Wharton Groups were reactivated at that stage with a dextral-normal kinematics, favoring the circulation of the NaCl-rich basinal brines from the Thelon Basin down to the basement rocks and the probable formation of CaCl 2 -and U-rich mineralizing brines (U1). Minerals fingerprinting highly saline basinal brines (sudoite, APS), fluid inclusion characteristics (metal-bearing, high salinity, Ca-rich, T h between 100 and 220°C), and REE bellshaped patterns of U oxides, along with the timing of the mineralizaton event (post-Thelon Formation), support the conclusion that U1 precipitated from Thelon-derived basinal brines, and is thus of unconformity-related type, like the U deposits in the Athabasca Basin.
(f6, U2) This stage is characterized by monometallic mineralization within reddish oxidized fault zones (Fig. 14a, b). As already pointed out by Grare et al. (2018a) at Contact, these oxidized faults are not always mineralized and fault cores that display strong clay alteration are usually not mineralized. Uranium mineralization is stronger in well-developed, moderately clay-altered damage zones. These observations are in line with the classical view of fault zones (Chester and Logan 1986;Kim et al. 2004;Faulkner et al. 2010) where fracturing and fluid flow are more important in damage zones than in core zones of faults (Caine et al. 1996). The orientations of U2-bearing faults do not necessarily reflect the orientations of the main faults that host the mineralization. Mineralized U2 faults display a complex range of orientations, sometimes with antithetic dip to the main fault. These variable orientations are typical within damage zones linking main faults and has been described in a variety of settings (Kim et al. 2004;Rotevatn and Bastesen 2014;Fossen and Rotevatn 2016).
Minerals specific to URU mineralization were observed associated with U2: APS minerals were observed in oxidized fault zones at Contact (Grare et al. 2018a) and 85W (this study), and sudoite associated with illite was described, at Contact (Grare et al. 2018a), Jane (Miller 1997), Kiggavik Main Zone (Hasegawa et al. 1990), End (Lida 1997), and Andrew Lake (Hasegawa et al. 1990;Pacquet 1994). Along with hematization of the host rock, the mineralogical differences (e.g., absence of Ni-Co Arsenides) between U1 and U2 highlight that U2-bearing fluids had a different composition (although still being basinal brines) than U1-related fluids, with slightly higher f O2 and pH (see Pourbaix diagram in Romberger 1984). This difference may be explained by the evolution of the fracture conduits through time, since this new faulting/ fracturing event formed new fractures while likely reactivating the complex, pre-existing fracture network of the Kiggavik area.
The response of the fracture network to the newly applied tectonic stress was obviously different depending on its orientation: at Andrew Lake and Contact, hematitized and mineralized damage zones of faults trending NE-SW are well developed in contrast to those associated within ENE-WSW fault zones of the Kiggavik Main Zone prospect (Fig. 14a). The f6 fault zones underwent sinistral and/or reverse movements (Fig. 14). Sinistral reactivation of the ENE-WSW Judge Sissons Fault overprinted, in some places, the dextral motion on this fault that offsets the Schultz Lake Intrusive Complex. These observations, together with the NE-SW crack-seal calcite (Ca2) veins (f6) coated with U2 spherulitic pitchblende support a local stress regime with NE-SW to ENE-WSW trending σ1 and NW-SE to NNW-SSE trending σ3 for this tectonic event. This stress regime may account for the secondary NE-SW to ENE-WSW trends of type 2 FIPs (Fig. 8f) associated to a second episode of circulation of basinal brines. The regional meaning of this stress regime however still remains poorly understood.
In terms of timing, the f6 faults cross-cut the sandstones of the Thelon Formation (Grare et al. 2018a). In addition, crosscutting relationships indicate that f6/U2 stage occurred prior to the emplacement of the Mackenzie diabase dykes at 1267 ± 2 Ma (LeCheminant and Heaman 1989) as well as the f5/U1 event. These observations bracket the f6 stage between 1500 and 1267 Ma. Note however that no available absolute age dating supports this timing. The dated U2 samples display ages at 565 ± 38 Ma and 345 ± 19 Ma at Andrew Lake. Shabaga et al. (2017) obtained ages at ca. 1000 Ma and 550-300 Ma. One likely explanation is related to alteration and reset/reprecipitation of U oxides that accounts for the modified REE bell-shaped pattern for U2 oxides (Fig. 6a).
Like U1, U2 formed in relation to the circulations of basinal brines derived from the Thelon Basin, and is of unconformity-related type. The differences (fracture orientations, intensity of clay alteration, reduced vs. oxidized fault zones, polymetallic vs. monometallic character) between mineralization events U1 and U2 are here interpreted as the distinctive response of pre-existing reactivated faults to a change in the regional tectonic stress field. This reactivation induced maturation of fault zones and likely a change in fluids pathway, hence contrasting fluid-rock interactions. U1 and U2 mineralization probably occurred between 1530 and 1267 Ma, an interval previously proposed for the crystallization of hydrothermal UO 2 in the area (Sharpe et al. 2015;Chi et al. 2017). URU-type mineralization in the Kiggavik area present two main differences with the Athabasca Basin are noted; the absence of dravite and Mg-foitite (Mg-tourmaline, Mercadier et al. 2012) and the presence of a Mg-rich sudoite rather than a Al-Mg sudoite.
(f7) The last, post-ore main fracturing event is characterized by fault zones displaying strongly desilicified, illitized, and bleached fault rocks with numerous narrow fault cores. This event is temporally dissociated of U mineralization and reworks/offsets previously formed prospects along E-W and NW-SE faults (Fig. 15a). This stage had a significant impact on the 3D architecture of some prospects (e.g., Bong, End) because the related faults offset U orebodies with a significant dip-slip and a minor strike-slip component. This is especially obvious at Bong where the orebody is offset by W-E to WNW-ESE trending, nearly dip-slip normal faults, but also at End where the orebody is offset by NW-SE to NNW-SSE trending faults. Moreover, observed U1-and U2-bearing fault zones are also less mineralized when they are cross-cut by white clay-altered faults, indicating remobilization of the U oxides at this stage. Only trace relicts of pitchblende were observed in f7 fault zones (Fig. 4j). As this event corresponds to one of the latest fracturing stages in a complex and polyphase fault system, fracture orientations of this stage represent more reactivation than neoformation. Thus, it is difficult to process oriented data in order to retrieve the "true" fault orientations. A WNW-ESE to NNW-SSE trend and a steep dip to the northwest however appear to be the dominant orientation of these faults in the Kiggavik area. Reactivation of faults during this stage is illustrated by NE-SW white clay-altered faults at Contact, initially belonging to the f6 stage (Grare et al. 2018a). This observation that most of the faults related to this post-ore fracturing stage are inherited is consistent with models in which reactivation of faults is often easier compared to neoformation (e.g., Pinheiro and Holdsworth 1997). f7 faults with kinematic data display two main trends: WNW-ESE with dextral to dextral-normal slip component and NNW-SSE with sinistral slip component. Combined with the dominant dip-slip component observed on some WNW-ESE faults observed at Bong, we infer a transtensional stress regime with σ1 trending NW-SE and σ3 trending NE-SW. This faulting event would be responsible for the down-drop offset of the U orebodies observed at Bong, End, and Kiggavik Main Zones (example in Fig. 4e). The NW-SE trending σ1 and NE-SW trending σ3 would also account for the latest sinistral offset of the Judge Sissons Fault and Thelon Fault by NNW-SSE faults, observed in the field and on aeromagnetic maps (Tschirhart et al. 2017). Such NNW-SSE faults were likely reopened during the emplacement of the Mackenzie dykes. The inferred NW-SE trending σ1 and NE-SW trending σ3 (Fig. 15b) are consistent with a regional compressional stress active on the southeast margin of the Canadian Shield (Hou et al. 2010) at ca. 1270 Ma. It is inferred that the emplacement of a mantle plume triggered the emplacement of Mackenzie mafic dykes swarm at that time. This magmatic event is likely at the origin of the circulation of hot (~300°C), probably acidic, fluids during the f7 fracturing stage. This is supported by age dating on illite (K-Ar and Ar-Ar) from f7 fractures showing an alteration event at ca. 1300 Ma (Fig. 10) at End, Andrew Lake, Bong, and Kiggavik Main Zone. This tectonic/magmatic event ended by the emplacement of the MacKenzie diabase dykes which cross-cut orebodies and fault rocks. Along these dykes, hot fluids would have caused local remobilization and reprecipitation of U oxides, hence resetting of the U/Pb isotopic system. Such strong clay alteration has previously been described as synchronous with U alteration (Hasegawa et al. 1990;Riegler et al. 2014;Shabaga et al. 2017), but this study unambiguously shows that white clay illitization postdates U0, U1, and U2.
Post Mackenzie dykes minor fracturing events and U mineralization/remobilization (U3) Several ages at ca. 1000 Ma and ca. 800 Ma were previously obtained from U oxides in the Kiggavik area (Fig. 10). These ages, not reproduced in the present study, could reflect a farfield tectonic activity associated with the Grenville orogeny (Gordon and Hempton 1986) and the initial rifting event of Rodinia (Badger et al. 2010;McClellan and Gazel 2014). It is currently unclear whether these two events have led to the formation of new mineralization or simply caused alteration and reset of existing mineralization. Younger ages at ca. 550 Ma and ca. 350 Ma correspond to a reset of the U-Pb isotopic systems of U1 and U2 but also to their dissolution linked to circulation of meteoric fluids and precipitation of U3. The specific modified bell-shaped pattern marks the alteration and remobilization of the URU-type U oxides by low-T meteoric fluids, which is consistent with the observation of goethite in these mineralized samples. U3 occurred at a post-Mackenzie stage and does not belong to the unconformityrelated type. It likely happened when a significant thickness of the Thelon Basin, Wharton Basin, and/or basement rocks was definitely eroded, letting supergene, low-temperature oxidizing fluids circulating through the fracture network and remobilizing U. Such (weak) reactivation of the fracture network is probably linked to a far-field stress associated with the breakup of West Rodinia supercontinent (ca. 500 Ma, Bond et al. 1984) and the Appalachian Orogeny (ca. 350 Ma, Hatcher, 2002). This event has also been recognized in the Athabasca Basin (Dieng et al. 2013).
The structural evolution proposed in this study differs from the model of Riedel shear system of Anand and Jefferson (2017). These authors proposed that the entire fracture network evolved mainly under a~N110-N140°trending σ1 from ca. 1800 Ma to 1540 Ma. In their model, the different fault zones (oriented ENE-WSW, NNW-SSE, NS, NE-SW, E-W) are part of a single Riedel system (i.e., analogue to the P, T, X, R′, R shears). Our structural analysis, together with the compilation of isotopic U-Pb ages and trace element concentrations for the different mineralization observed in the Kiggavik area, indicates that the U system was instead polyphase and that U was mobilized at several times during more than 1.5 billion years. Such extremely long duration of U events is similar to what has been observed in the Athabasca Basin. Additionally, the similarities between the Athabasca and Thelon Basins demonstrate that they had rather close evolution through time in terms of U systems, moving from relatively high-T magmatic conditions (U0) before 1750 Ma, to moderate-T hydrothermal conditions with basinal brines (U1 and U2) between 1530 to 1270 Ma to low-T meteoric mineralization (U3) later.

Conclusions
This paper proposes an integrated structural and metallogenic model of the Kiggavik area based on a multi-scale and multidisciplinary approach. The main conclusions of the study are as follows: -The Kiggavik area is characterized by a polyphase fault/ fracture network that evolved in a brittle style from ca. ca. 1270 Ma. The pre-Thelon Basin U0 mineralization and the Quartz Breccia event highlight the importance of magmatic-related fracturing, fluid circulation, and mineralization on controlling subsequent occurrence of URU prospects. U0 constituted a significant stock of U available for reconcentration through later circulation of oxidized basinal brines. This first mineralization is of magmatic-related type.
-After deposition of Thelon and Lookout Point formations, and before the MacKenzie dyke event, i.e., between 1530 and 1270 Ma, a first circulation of highly saline basinal brines derived from the Thelon Basin in basement structures occurred under a strike-slip stress regime with WNW-ESE σ1 and NNE-SSW σ3. This event was associated with precipitation of U1 in fault relay zones and/or in fault zones where the Quartz Breccia was present. Faulting/fracturing and renewed circulations of basinal brines occurred under a stress regime with NE-SW σ1 and NW-SE σ3 which caused transpressional reactivation of the previously formed fault/fracture network and the formation of U2. U mineralization preferentially developed in moderately altered damage zones of faults. U1 and U2 mineralization belong to the unconformity-related (URU) type. They formed in relative similar conditions and timing than URU-type mineralization in the Athabasca Basin. -Circulation of basinal brines occurred through a fault/ fracture network that evolved though time in response to the changing local/far-field tectonic stress. This caused changes in the fluid pathways hence in fluid-rock interaction (changing characteristics of the fluid such as fO 2 and pH) and led to slightly different U mineralization (polymetallic reduced U1 vs monometallic oxidized U2). -The initial NaCl-rich basinal brines reacted with basement rocks to form CaCl 2 -and metal-rich mineralizing brines at the origin of U1 and U2. The chemistry and physico-chemical characteristics of the CaCl 2 -rich brines in the Kiggavik area are similar to those measured for mineralizing CaCl 2 -rich brines in the Athabasca Basin, indicating rather common processes for the two sedimentary basins and related U mineralization. -Late faulting and associated strong clay alteration and bleaching of the host rock occurred at ca. 1300 Ma in response to local NE-SW extension. This fracturing event, barren of U mineralization, caused the offset of previously formed orebodies and accounts for the formation in the Kiggavik area of strongly altered areas disconnected of any U mineralization. -Weak reactivation of the fracture network after emplacement of the MacKenzie dykes induced circulation of low-T meteoric water, remobilizing and reprecipitating U oxides (U3) at two different times, ca. 550 and 350 Ma.
Other fluid circulation events likely happened at ca. 1000 and 700 Ma with similar processes. -U mineralization in the Kiggavik area are of a mixed type, combining magmatic (U0), URU type (U1 and U2), and meteoric-related ("roll-front," U3) mineralization, as proposed by previous studies in the Athabasca Basin. This area documents more than 1.5 billion years of U mobility.