The study area, the Millungera Basin, is located in north-west Queensland, Australia (Fig. 1a). The Millungera Basin is surrounded by the Palaeoproterozoic–Mesoproterozoic Mount Isa Province to the west and Neoproterozic–Ordovician Georgina Basin to the south-west, which developed along the eastern margin of the Proterozoic North Australian Craton. Proterozoic Australia comprises three main tectonic units including the North, West, and South Australian cratons. These units were independently accreted from older crustal fragments by ∼1830 Ma (Myers et al., 1996). The North Australian Craton has Archean and/or early Palaeoproterozoic cores that have been superimposed by later Palaeoproterozoic orogenic belts and basins (1800–1575 Ma) such as the Mount Isa and Etheridge provinces (Scott et al., 2000; Withnall et al., 2013) (Fig. 1a). The Mount Isa Province is a world-class mineralized terrain with large deposits of copper, lead, and zinc, recording polyphase deformation and a multi-staged metamorphism that affected the terrane during the Isan orogeny between 1600 and 1500 Ma (O'Dea et al., 1997).

Australian continental fragments amalgamated as an early component of the Rodinian supercontinent between ∼1300 and 1100 Ma (de Vries et al., 2008; Li et al., 2008). The North Australian Craton was first joined to the north-western margin of the West Australian Craton. The combined West and North Australian cratons were joined to the South Australia Craton along the Albany–Fraser Orogen. The Musgravian Orogen in central Australia was responsible for substantial crustal thickening and high-grade metamorphism at ∼ 1200–1150 Ma associated with granite intrusion (Evins et al., 2010; Kirkland et al., 2013). Thereafter, uplift and erosion were followed at ∼1080 Ma by the deposition and of post-tectonic volcanism, accompanied by the intrusion of widespread plutons (Giles Complex) during the extension along the former collision zones (Schmidt et al., 2006; Evins et al., 2010; Aitken et al, 2013). At the same time, major swarms of dolerite intrusion (e.g. the Lakeview Dyke) were emplaced into the North Australian Craton in the Mount Isa Province (Tanaka and Idnurm, 1994).

During the Neoproterozoic, an extensive intracratonic basin (Centralian Superbasin) developed over the junction between the North, South, and West Australian cratons. This was followed by the Rodinia break-up with a mantle plume that initiated continental rifting (Walter et al., 1995; Li et al., 1999). Rodinia's break-up resulted in the generation of a number of fault-bounded sedimentary basins. It has been proposed that the initial period of extension occurred at about 900 Ma and was associated with igneous activities (e.g. Stuart Dyke swarm; Black et al., 1980) and intracratonic basin formation in north-central Australia (e.g. Amadeus Basin; Korsch and Lindsay, 1989; Shaw et al., 1991). The Georgina Basin, located in close proximity to the Millungera Basin (Fig. 1a), represents another intracratonic basin, which consists predominantly of late Neoproterozoic, Cambrian–Ordovician, and Devonian strata unconformably overlying a Proterozoic crystalline basement (Shaw et al., 1991; Greene, 2010). The oldest sedimentary unit of the basin is considered to be ∼ 825 Ma (Greene, 2010, and references therein). The southern Georgina Basin was deformed during the mid-Palaeozoic Alice Springs Orogeny whereby the Neoproterozoic normal faults of the rift basin were reactivated, which are now expressed as high-angle reverse faults (Greene, 2010).

The Millungera Basin is a recently discovered sedimentary basin in northern Queensland, Australia (Korsch et al., 2011, Fig. 1a). It occurs to the east of the Palaeoproterozoic Mount Isa Province and is covered by the thin Jurassic–Cretaceous Eromanga–Carpentaria Basin (Fig. 1b). An angular unconformity between the Eromanga and Millungera basins indicates that the upper part of the Millungera Basin was eroded prior to the deposition of the Eromanga–Carpentaria Basin (Korsch et al., 2011) (Fig. 1b), allowing sampling of the deeper part of the basin consisting of flat-lying to gently dipping sedimentary strata (Fig. 1b), which are strongly deformed and faulted (see below). A marked angular unconformity is also interpreted between the basement of the granite and metasedimentary basement rocks and the base of the Millungera Basin (Fig. 1b) The interpretation of gravity profiles indicates that the basin deepens to the south with a possible maximum thickness of 4000 m subsurface. The interpretation of aeromagnetic data suggests that the basin might have dimensions of up to 280 km by 95 km (Korsch et al., 2011). Apart from geophysical data, almost no geological information exists on the basin. Prominent thrust fault systems truncate both the western and eastern margins of the basin. Particularly the eastern part of the basin has been cut by several deep-penetrating, north-east-dipping thrust faults with the associated development of hanging wall anticlines. Based on SHRIMP U–Pb geochronology of detrital zircons from the Millungera Basin sandstones, the maximum depositional age of the Millungera Basin is constrained to 1574±14 Ma (Neumann and Kositcin, 2011); however, it is a minimum age and not well constrained between the Cretaceous (overlying sediments) and Mesoproterozoic.

Core samples were taken from the lower parts of boreholes Julia Creek 1 (JC) and Dobbyn 2 (Dob), which were drilled as part of a (Queensland) state-wide geothermal investigation. The wells are 150 km apart (Fig. 1a and Table 1) and intersect the Mesozoic Eromanga–Carpentaria Basin in the upper part. Julia Creek 1 intersected 320.05 m of the Eromanga Basin sequence and 179.97 m of the Millungera Basin sequence; Dobbyn 2 intersected 332.40 m of the Carpentaria Basin sequence and 155.64 m of the Millungera Basin sequence. It should be noted that the succession within the Millungera Basin has not been formally defined. According to deep seismic reflection surveys, a number of large-scale structures are interpreted as occurring as basin-bonding and intra-basin fault systems (Fig. 1a–b) (see Korsch et al., 2011). Small-scale faults and fractures have also been described from the logging of the cores extracted from JC and Dob (Faulkner et al., 2012; Fitzell et al., 2012). We collected a total of nine Julia Creek 1 and six Dobbyn 2 fault gouge samples which were all analysed for the <2 µm clay mineral content (Table 1; Fig. S1 in the Supplement) and some of which have been selected for K–Ar, 40Ar–39Ar, and Rb–Sr dating and trace element studies. We also sampled representative host rock samples adjacent to the fault gouge zones (Table 1).

Samples were prepared for clay-fraction separation by gently hand crushing the rocks to sand size to avoid artificially reducing the grain size of detrital/primary mineral components and then washed thoroughly by deionized water. Samples were then disaggregated in distilled water using an ultrasonic bath. Clay fractions were separated by the sedimentation method (for <2 µm) and centrifugation (for <1 µm subfractions to <0.1 µm). Oriented slides were prepared by pipetting the suspension onto a 30 mm × 30 mm glass slide to give a concentration of about 3 mg cm-2 or more (Warr and Rice, 1994). X-ray diffraction (XRD) on whole rock samples and clay separates of different size fractions was carried out (Table 1). The XRD analyses were conducted on a Bruker D4 ENDEAVOR and D8 Advance (CoKα and CuKα radiation, respectively), operated at 40 kV and 30 mA at a scanning rate of 1∘ 2θ per minute and 0.05∘ per step. Following XRD analysis of air-dried samples, the oriented clay-aggregate mounts were placed in an ethylene–glycol atmosphere at 30–40 ∘C overnight prior to additional XRD analyses. For polytype analyses, clay fractions of random powder from fault gouge samples (if a sufficient amount of material was available) were scanned from 16 to 44∘ 2θ in the step-scanning mode with a step size of 0.05∘ and a counting time of 30 s per step.

Illite polytypes for randomly oriented pure illite samples have been distinguished with the diagnostic peaks suggested by Grathoff and Moore (1996). To determine the 2M1, 1M, and 1Md % contents of illite and/or muscovites, the ratios of (2.80 Å - 3.0 Å)/(2.58 Å) and (3.07 Å)/(2.58 Å) peak areas for 2M1 and 1M, respectively, were used, as proposed by Grathoff and Moore (1996). The presence of 1Md illite was detected by the presence of the illite hump around the illite 003 diffraction peak (Grathoff and Moore, 1996). WINFIT decomposition by profile fitting was used for the determination of areas of the specific peaks of polytypes. Polytype absolute quantification errors are estimated at about ±5 %.

The Kübler index (KI) determinations are defined as the width of the first order illite basal reflection (10 Å peak) at half height and expressed in Δ2θ values. The Kübler index decreases with increasing illite crystallinity (a measure of the ordering/thickness of illite crystallites), with temperature being the most important controlling factor (Ji and Browne, 2000, and references therein). The KI is a well-accepted mineralogical indicator of anchizone, hydrothermal, and low-temperature regional metamorphism and thermal conditions during fault activity (Merriman and Frey, 1999; Ji and Browne, 2000; Bense et al., 2014).

However, the Árkai index (AI) is becoming an additional or alternative technique (particularly in mafic rocks) to evaluate palaeo-temperature conditions (Árkai, 1991; Warr and Cox, 2016). The AI is determined through the measurement of chlorite 002 peak width (Arkai, 1991). The KI and AI results of this study were calibrated against the Crystallinity Index Standard (CIS) scale using the procedure and interlaboratory standards of Warr and Mählmann (2015).

The thin sections were first examined under plane-polarized light and cross-polarized light conditions using a Nikon Eclipse LV100N POL and a Zeiss Axio Imager.A2m polarizing microscope. Further examination of the thin sections was undertaken using a Philips XL 40 scanning electron microscope (SEM) equipped with an X-ray energy-dispersive spectrometry (EDS) system for chemical spot analyses. The sections were analysed using 30 kV accelerating voltage and a working distance of 12 mm. Images were collected in back-scattered electron mode. Additionally, clay separates were carbon coated and examined using an EDS equipped Zeiss Ultra Plus SEM qualitative phase identification. The samples were analysed under high vacuum with a 15 kV accelerating voltage and a working distance of 6 mm. Images of clay separate were collected in secondary electron acquisition mode.

For the Rb–Sr dating (conducted at the Radiogenic Isotope Facility laboratory, RIF, and the University of Queensland, UQ), illitic clay separates were leached for 15 min at room temperature in 1 N distilled HCl (Clauer et al., 1993). Leachate and residue were separated by centrifuging. The residue was rinsed repeatedly with Milli-Q^® water, dried, and reweighed. Acid leached residues and untreated samples were measured directly by Thermo Xseries 1 quadrupole ICP–MS with a precision better than 0.5 % (1σ). The Sr-enriched fraction was separated using cation-exchange resins. Sr isotopic ratios were measured on a VG Sector 54 thermal ionization mass spectrometer (TIMS). Sr was loaded in TaF5 and 0.1 N H3PO4 on a tantalum or tungsten single filament. Sr isotopic ratios were corrected for mass discrimination using 86Sr/88Sr=0.1194. Long-term (6 years) reproducibility of statically measured NBS SRM 987 (2σ; n=442) is 87Sr/86Sr=0.710249±0.000028. More recent dynamically measured SRM 987 had 86Sr/88Sr ratios of 0.710222±0.000020 (2σ; n=140). Rb–Sr isochron ages were calculated using the Isoplot programme (Ludwig, 2012) and decay constant recommended by Villa et al. (2015). For isochron age calculation, standard errors of ±0.01 % for 87Sr/86Sr and of ±1 % for 87Rb/86Sr ratios were assigned to the results. Individual analytical uncertainties were generally smaller than these values.

The K–Ar dating was performed at the CSIRO Argon facility in Perth, Australia, according to standard methods given in detail by Dalrymple and Lanphere (1969). Potassium content was determined by atomic absorption. The error of K determination of standards is better than 1.2 % (1σ). The K blank was measured at 0.50 ppm. Argon was extracted from the separated mineral fraction by fusing the sample within a vacuum line serviced by an online 38Ar spike pipette. The isotopic composition of the spiked Ar was measured with a high sensitivity, online, VG3600 mass spectrometer. The 38Ar was calibrated against standard biotite GA1550 (McDougall and Roksandic, 1974). Blanks for the extraction line and mass spectrometer were systematically determined, and the mass discrimination factor was determined periodically by airshots (small amounts of air for 40Ar/36Ar ratio measurements). During the course of the study, 16 international standards (8 HD-B1 and 8 LP-6) and 16 airshots were analysed. The results are summarized in Table 2. The error for the 40Ar/36Ar value of the airshot yielded 296.08±1.23, (0.41 %; 1σ). The general error for argon analyses is below 1.3 % (1σ) based on the long-term precision of 330 measurements of international Argon standards. The K–Ar age was calculated using 40K abundance and decay constants recommended by Steiger and Jäger (1977). The age uncertainties take into account the errors during sample weighing, 38Ar/36Ar and 40Ar/38Ar measurements, and K analysis.

Four fault gouge illites were dated by the 40Ar–39Ar method at the University of Michigan, United States. Illitic clay samples were re-suspended in 1 mL of deionized water, spun-down at 10 000 rpm in a microcentrifuge, and carved into a ∼1 mm pellet following decanting. To avoid loss of 39Ar due to recoil, clay pellets were placed in 1 mm ID-fused silica vials prior to being sent for neutron irradiation for 90 MWh in medium flux locations of the McMaster Nuclear Reactor (hole 8C for irradiation 1, 8A for irradiation 2). Following irradiation, samples were attached to a laser fusion system, the vials were broken under a 1×10–8 Torr vacuum, and the samples were step heated in situ using a defocused beam from a 5 W Coherent INNOVA continuous Ar-ion laser operated in multi-line mode. Argon isotopes were then analysed using a VG1200S mass spectrometer equipped with a Daly detector operated in analogue mode using methods by Hall (2014). Ages in this study are calculated relative to an age of 520.4±1.7 Ma for standard hornblende MMhb-1 (Samson and Alexander, 1987). The total gas age obtained from the vacuum encapsulated sample is equivalent to a conventional K–Ar age and quoted at 1σ.

For the trace element analysis conducted in the Radiogenic Isotope Laboratory at the University of Queensland (RIF, UQ), clay samples were dissolved with a mixture of HF and nitric acids on a hotplate, then evaporated to dryness, refluxed twice with nitric acid, and dissolved in 2 N nitric acid. Aliquots of the solutions were spiked with internal standards, diluted, and analysed on a Thermo X-series 1 quadrupole inductively coupled plasma mass spectrometer (ICP-MS). Sample preparation and analytical procedures used were similar to those of Eggins et al. (1997) except that Tm was not used as an internal standard and duplicate low-pressure digestions of United States Geological Survey W-2 diabase standard and a known concentration profile (pre-analysed by laboratory) were used for calibration (Li et al., 2005). The 156CeO/140Ce ratio for the run was 0.016. Long-term precision (RSD, relative standard deviation) was based on duplicate analyses of the duplicate digestions of AGV1, whilst precision for the run was based on five duplicate analyses of W-2, which were better than 3 % for most elements except for Li, Zn, Mo, Cd, and Cs, which ranged between 5 % (Li, Cd, and Cs) and 15 % (Zn).

Rare earth element (REE) data and Th, U, and Sc contents of illites (<2 µm clay-size fractions) from the fault gouge samples are given in Table 4. Chondrite-normalized REE patterns of illites from the fault gouges are shown in Fig. 9a. In addition, the REE pattern of post-Archean average shale (PAAS; Taylor and McLennan, 1985) is included in the REE diagram. The fault gouge illites are substantially enriched in light REEs (LREEs) relative to PAAS with La contents as high as 10 times PAAS. The illites are, however, somewhat depleted in heavy REEs (HREEs) relative to LREEs (Fig. 9a). The chondrite-normalized (La / Lu)c ratios of the illites are significantly higher (up to 76) than the (La / Lu)c ratio of PAAS (10) (Table 4). Fault gouge illites are also enriched in Th and U (up to 10 times) in comparison to PAAS (Table 4).

Brittle deformation and faulting are evident from cores in the sampled intervals in Julia Creek 1 and Dobbyn 2, which are probably associated with the large-scale faults inferred from interpretations of seismic surveys (Fig. 1). Under upper crustal conditions, fault zones accommodate intense shear strain often localized in bands of cataclastic deformation formed by friction-dominated faulting within the seismogenic regime (Sibson, 1977; Schmid and Handy, 1991). While cataclastic fault rocks are generally considered to display random fabric, foliated fault rocks such as fault gouge and foliated cataclasites have also been reported in different lithologies ranging from crystalline rocks to siliciclastic- and carbonate-dominated sediments at different burial or deformation depths (Chester et al., 1985; Rutter et al., 1986; Lin, 1999; Ujiie et al., 2007; Laurich et al., 2004; Delle Piane et al., 2017; Nicchio et al., 2018). Frictional sliding and abrasion are common processes during repeated fault movements and result in a strong grain size reduction in the fault rocks with respect to the constituting minerals in the undeformed portion of the host rocks. The abundant presence of micro-sized and nanosized particles in cataclasites and gouges may result from combined effects of cataclasis and pressure solution precipitation during deformation in the presence of fluids (e.g. Vrolijk and van der Pluijm, 1999; Solum et al., 2005). Foliated cataclasites have also been observed at very shallow depths in siliciclastic sediments (<500 m; e.g. Balsamo et al., 2014) and carbonate rocks (<2 km; e.g. Smeraglia et al., 2016) as the result of cataclasis, clay smearing, and/or pressure solution precipitation in the presence of fluids during deformation. The corroded grain boundaries of quartz grains in faulted samples from Julia Creek 1 (Fig. 4) and the presence of the injection veins and hydrothermal hematite in the cataclasites from Dobbyn 2 (Fig. 5) indicate that deformation occurred in a fluid-rich environment that promoted detrital muscovite dissolution and the new growth of illite. The small injection veins that are observed to cut through the foliated cataclasites and the detrital quartz grains (Fig. 5) may represent the effect of hydraulic fracturing due to a fast increment of fluid pressure in the fault zone during a seismic slip (e.g. Sibson, 1989; Cowan et al., 2003; Ujiie et al., 2007; Rowe et al., 2012). At the core scale, some samples show no shearing-related fabrics in the sandstone cores (fresh and hard) adjacent to clay-filled cracks (Fig. 2a–d). This may be a result of the precipitation of clay-rich material and the injection of granular material from seismically mobilized circulating fluids (cf., Smeraglia et al., 2016).

K–Ar results show that illites from fault gouges and matrix illites in undeformed adjacent sandstones precipitated contemporaneously (Fig. 7). In some tectonically active regions, mineral assemblages from the fault rocks and their parent rocks are significantly different whereby parent rocks do not contain any alteration minerals, with new mineral growth being restricted to the fault rocks. This indicates that the heat and fluid flows associated with mineral authigenesis were not controlled by regional tectonic events in these regions but were rather confined to the areas within the fault zone (e.g. Uysal et al., 2006; Işik et al., 2014; Babaahmadi et al., 2019). However, the relation between large-scale fluid flow and seismic events has long been reported (e.g. Bruhn et al., 1994; Eichhubl et al., 2010; Faulkner et al., 2010; Lupi et al., 2010; references therein). Brittle faulting in the upper crust involves episodic changes in the stress level that can expel large volumes of fluids, leading to the generation of hydrothermal/geothermal systems (e.g. Maffucci et al., 2016). Faults and veins and their immediate surroundings represent zones of fluid passage and the transfer of mass through those fluids (e.g. Sibson, 1987). Mineral alteration in slip zone gouges extends outward from the fault zone into the undeformed wall rock (e.g. Parry et al., 1991; Craw et al., 2009). The wall rock alteration is attributed to the diffusion and advection of fluids and hence chemical mass and heat transfer associated with deformation. For example, metasomatic alteration zones develop around fluid pathways by advection with mineral dissolution and precipitation increasing towards the conduit and dictated by infiltrating fluids (Ferry and Dipple, 1991; Rossetti et al., 2011; Maffucci et al., 2016). Metasomatic mineral alteration is common in sedimentary basins contemporaneous with regional extensional tectonics. Alteration is driven by the reactivity of sandstone host rocks, with illitic clay minerals, K-feldspar (adularia), hematite, calcite, and quartz being some common minerals precipitating from tectonically mobilized K-bearing basin brine (Fedo et al., 1995; Michalski et al., 2007; Eichhubl et al., 2010; Rossetti et al., 2011). Similarly, in this study, sandstone immediately above or below the fracture and shear (fault gouge) zones represents zones of the alteration of detrital minerals to illite by tectonically moved fluids.

Fault gouges from the Millungera Basin in Australia (Fig. 1) contain a mixture of coevally formed 1M/1Md and 2M1 illite from which the crystallization age was determined by a combined application of Rb–Sr, 40Ar–39Ar, and K–Ar techniques. It is demonstrated that applying both the Rb–Sr and K–Ar (40Ar–39Ar) techniques for dating the same fault gouge minerals provides more robust and complementary age constraints on faulting episodes and minimizes the inherent disadvantages for each isotopic system. A common drawback of Ar geochronology when dating white mica is that Ar apparent ages are either significantly older or younger than the Rb–Sr isochron ages of the same samples (Kelley, 2002; Di Vincenzo et al., 2006). Rb–Sr isotopic systematics may remain unaffected because Rb–Sr resetting requires higher closure temperatures and sufficient fluids in the system to facilitate recrystallization (e.g. Di Vincenzo et al., 2006). However, a potential pitfall of the Rb–Sr dating technique could result from heterogeneous initial 87Sr/86Sr ratios on a mineral scale (cf., Davidson et al., 2005). In this study, as discussed in Sect. 4.2.3, we minimized the effect of mineral-scale initial isotopic heterogeneity by analysing different aliquots (untreated, leachates, and residues) and different sub-size fractions of one clay sample. Indeed, different sub-size fractions and aliquots of some samples plot on different isochron lines indicating different 87Sr/86Sr ratios (Fig. 8; see the discussion below). Illites from the Millungera Basin fault gouges display well-developed linear data arrays on Rb–Sr isochron diagrams which we interpret as reflecting statistically valid late Mesoproterozoic ages (Fig. 8).

Numerous case studies, as discussed in detail below, indicate that such linear relations can result either from a mixing between different mineral populations with different initial 87Sr/86Sr ratios or from a complete isotopic equilibration of the entire mineral assemblage at a given time. In the former case, the linear relationship between 87Rb/86Sr and 87Sr/86Sr could have developed from a mixing line of two end members with potentially no genetic relationship and therefore without meaningful age information, whereas the latter relation provides a valid isochron whose slope yields the age of illitic clay generation during a fault reactivation event. However, valid and geologically significant isochrons and mixing lines can also be obtained simultaneously from samples with different mineral populations, comprising minerals with different Rb/Sr ratios but identical initial 87Sr/86Sr ratios. In this case of identical initial Sr-isotopic compositions of two components of a mixture at time t=0, the two components and mixtures thereof define a horizontal line both in a classic isochron diagram (which is the key condition for validity of calculated Rb–Sr isochron ages) and in the 87Sr/86Sr vs. 1/86Sr diagram commonly used for the evaluation of binary isotopic and compositional mixing (cf. Wendt, 1993; Schneider et al., 2003, for theoretical background). In sedimentary basins or hydrothermal systems, samples with various Rb/Sr ratios can precipitate from a chemically homogeneous basinal fluid that can yield the same isotopic composition across the entire sedimentary basin (e.g. Uysal et al., 2001; Golding et al., 2013). Fault gouge clay separates from the Millungera Basin contain a mixture of illites and other authigenic clay minerals such as chlorite and kaolinite (see the Sect. 4.1.2) and minor carbonates, which include considerable amounts of Sr but no or very little Rb (in contrast to illite). Since samples used for Rb–Sr analysis contain these different minerals in various amounts, the obtained linear relations can be considered to have evolved from (initially horizontal) mixing lines, but they simultaneously represent geologically meaningful isochron correlations. The isochron ages are consistent with K–Ar ages of the same clay-size fractions of the corresponding samples, which provides further strong support in favour of isochrons with meaningful age information (see below). Similarly, valid Rb–Sr ages based on linear relations representing both isochrons and mixing lines were commonly obtained from leachate, untreated, and residue aliquots of fault gouge and matrix illites (Clauer and Chaudhuri, 1995; Mutlu et al., 2010; Uysal et al., 2011; Golding et al., 2013; Işik et al., 2014; Middleton et al., 2014; Rosenbaum et al., 2015; Babaahmedi et al., 2019).

The Rb–Sr isochron ages of JC samples are concordant with K–Ar ages of the same samples (Table 1). However, K–Ar and Rb–Sr ages are not consistent for a number of Dob samples. K–Ar ages of Dob-389.6 and Dob-441 samples are older than the corresponding Rb–Sr isochron ages, while K–Ar ages of all different size fractions for sample Dob-449.1 are lower (925 to 913 Ma) than the Rb–Sr isochron age (Fig. 8b–d; Table 1). The Rb–Sr isochron age of 1000±12 Ma for <2 µm of sample Dob-476.6 (along with the acid-leached residues of Dob-441B and Dob-389.6 for 2–0.5 to 0.5–0.1 µm fractions) is consistent with K–Ar ages of the same sample for coarser size fractions (975.7±22.2 and 983.7±23.0 Ma) within analytical errors. The <1 µm fraction of sample Dob-476.6 yields a consistent K–Ar age (922.2±21.2 Ma) with clays of all size fractions of sample Dob-449.1 (Table 1).

In summary, all JC samples with their various size fractions yield consistent Rb–Sr isochron and individual K–Ar ages. This finding is similar to those reported by some recent studies that present robust and comprehensive Rb–Sr, 40Ar–39Ar, and K–Ar age data for deep diagenetic and anchizonal fault gauge illites (e.g. Middleton et al., 2014; Rosenbaum et al., 2015; Babaahmadi et al, 2019). Discrepancy in Rb–Sr and K–Ar ages for Dob samples, e.g. Dob-441 <2 µm (K–Ar age is older) and Dob-449.1 (K–Ar ages are younger), and the internal inconsistency of K–Ar ages among different size fractions of sample Dob-4449.3 (e.g. 1048±24 Ma for 2–1 µm and 1117±26 Ma for 1–0.5 µm fractions; see Table 1) may result from sample heterogeneity (hand specimen and microscale; see Figs. 4 and 5) as a result of more than one illite generation due to multiple faulting episodes (see Sect. 5.4).

The mixing of different generations is possible on a small scale as a result of variable degrees of isotopic resetting of a single illite generation during subsequent faulting events. Pervasive overprinting and recrystallization can be hindered even on a microscale, which may result from a lack of permeability and/or limited availability of fluids (e.g. Bröcker et al., 2013).

The trace element composition of authigenic clay minerals reflects the mineral–fluid partition coefficients for different elements, as well as the composition of fluids from which the clays precipitated. Trace element contents and concentrations, as well as some element ratios, can be used to trace the origin of basinal and hydrothermal fluids (e.g. Uysal and Golding, 2003; Uysal et al., 2005, 2011). The fault gouge illites analysed in this study are highly enriched in LREE and other incompatible elements such as Th and U relative to PAAS (Fig. 9a and Table 4). This geochemical characteristic indicates precipitation of the illites from fluids that must have interacted with rocks of the upper crust enriched in incompatible and heat-producing elements. This is also consistent with initial 87Sr/86Sr values of fault gouge illites that reflect the Sr isotope composition of fluids from which the illites precipitated. The radiogenic initial 87Sr/86Sr ratios of about 0.72 indicate the involvement of fluids that equilibrated with old Rb-rich crustal rocks. The trace element and Sr isotope data are in agreement with seismic and potential field data by Korsch et al. (2011), which are interpreted as indicating the occurrence of granites with a thickness of up to 5.5 km below the Millungera Basin. The inferred granites may be a part of the granite (Williams Supersuite) exposed just to the west in the Mount Isa Inlier, which is enriched in Th, U, and K (Korsch et al., 2011 and references therein).

Different 87Sr/86Sr initial values of illites of the parallel isochron lines corresponding to the same Rb–Sr age in Fig. 8 and scatter of Rb–Sr data points for some samples indicate separate circulation pathways for seismically mobilized fluids that might have restricted to unconnected fault planes and fracture systems in different areas of the Millungera Basin.

A valid interpretation of illite isotopic ages in relation to deformation history is subject to a solid mineralogical characterization of samples. Particularly, information about the illite crystallinity and illite polytype data are critical in assessing the illite crystallization temperature and a possible contamination of samples by metamorphic detrital muscovite from the undeformed host rock. Illite crystallinity is commonly used to identify the transitional anchimetamorphic zone between the diagenesis and epimetamorphic zone of low-grade metamorphism. The boundary limit from diagenetic to anchimetamorphic conditions has been reported to be at 0.52∘ 2θ, whereas the anchizone–epizone boundary is set at 0.25∘ Δ2θ (Warr and Mählmann, 2015). Accordingly, KI values of <2 and 2–1 for Julia Creek illitic clays indicate diagenetic and anchizone metamorphic conditions. Since these illites occur as a discrete phase (containing no expandable layers) and contain both 1M/M1 and 2M1 polytypes, they indicate formation temperatures of about 200 ∘C and higher (cf., Hoffman and Hower, 1979; Walker and Thompson, 1990). Although the 2M1 polytype has been known to appear usually at temperatures higher than 250 ∘C (Srodon and Eberl, 1984), its occurrence at lower temperatures of about 200–250 ∘C in co-existence with 1M/Md has also been reported (Walker and Thomson, 1990, Chen and Wang, 2007; Hejing et al., 2008). KI values show considerable differences between samples from Julia Creek and Dobbyn areas (Fig. 10), which clearly indicate different palaeo-thermal conditions in different areas. The <2 µm fractions of Julia Creek fault gouge samples show similar KI values with insignificant changes with depth except a spike for the sample at 440.5 m (Fig. 10). Samples from Dobbyn 2 are characterized by lower KI values and the dominance of 2M1 illite (Fig. 10; Table 1).

Our interpretation of the significance of K–Ar ages for fault gouge and matrix illite is based on K–Ar age versus grain size and KI relationships (Figs. 7b and 11a, b, c). This relationship can represent either an inclined or a parallel age spectrum. The parallel age spectrum results from identical K–Ar ages of different size fractions within error, representing internal consistency, which is regarded as geologically meaningful (e.g. Clauer and Chaudhuri, 1995; Torgersen et al., 2014). The inclined age spectrum can arise from the presence of multiple illite generations, such as either an earlier authigenic illite generation or inherited (detrital) components mixing with younger authigenic illites. The slope of the spectrum is a function of the age difference between the two age end members.

Plotting KI values vs. K–Ar ages, fault gouge samples from Julia Creek 1 indicate almost flat spectra with identical (within error) or slightly decreasing ages between 1049±25 and 1006±23 Ma for most of the JC samples and at ∼1100 Ma for sample JC-408 (Fig. 11a). These concordant K–Ar ages of samples with changing grain size (Fig. 7b) and KI values can thus be considered meaningful in indicating the timing of major deformation events. Similarly, an age clustering around 1060 Ma of samples with changing grain sizes and KI values is evident for matrix illites (Figs. 7c and 11c).

There are two different KI values vs. K–Ar age populations of Dobbyn 2 fault gouges, displayed by the shallow (above 441 m) and deeper (below 441 m) samples, which are distinguished by higher and lower KI values, respectively. They show two parallel trends with reasonably strong negative correlations (Fig. 11a), which may be considered as the mixing of two possible end members. These may be represented by an earlier illite generation at ∼1100 Ma (similar age as various size fractions of sample JC-408; see above) and a later illitization or isotopic resetting at ∼900 Ma. The correlations in Fig. 11b can also be interpreted as indicating the effect of numerous and superimposed slip episodes during discrete faulting events. A similar K–Ar age range but different extent for KI values of Dobbyn 2 fault gouges indicate that different thermal conditions prevailed in the shallower and deeper parts (see the discussion above for the illite crystallinity) and occurred in the same time period. Although decreasing K–Ar ages with increasing KI values of Dob samples could be due to a decreasing amount of detrital illite/muscovite with decreasing grain size, K–Ar ages for different size fractions and KI values of sample Dob-389.6 (2–1, <2, 1–0.5, and 0.5–0.1 µm), Dob-449.1 (2–0.5, <2, 1–0.5, and <0.5 µm), and Dob-476.6 (2–1 and <2 µm) are consistent within analytical error. This, together with authigenic mineral textures of illites (Fig. 3), suggests that the presence of detrital muscovite in <2 fractions is unlikely.

The lowest K–Ar ages associated with the highest KI values indicate the later recrystallization of illites in finer crystals or the isotopic resetting of finer illites at relatively lower temperatures not affecting the coarser size fractions. Thermally activated volume diffusion in clay minerals leading to 40Ar loss can cause decreasing K–Ar ages with decreasing grain sizes of clays (e.g. Torgersen et al., 2014; Lerman et al., 2007). The finest clay size fractions are more susceptive to younger thermal events due to poor radiogenic argon retentivity because of a smaller diffusion radius and less crystallinity. Consequently, the reheating of the finest particles, formed earlier, during a later thermal event to a temperature high enough to enable 40Ar diffusion from the crystal structure could cause the partial or complete resetting of the K–Ar isotopic systematics (Clauer and Chaudhuri, 1999). Alternatively, the finest fraction represents the last clay growth of newly crystallized tiny illite crystallites, which can occur during fluid flow events related to tectonically active regimes (e.g. Zwingmann and Mancktelow, 2004; Uysal et al., 2006). The size of authigenic clay minerals can be a function of both the duration of crystal growth and the crystallization temperature (Frey, 1987; Cashman and Ferry, 1988). The ages of the finest grain size fractions therefore date either to the time of the last, short-lived thermal and/or fluid flow events (cf., Torgersen et al., 2014) or to cooling events after a prolonged burial and mineral growth, which took place in the Neoproterozoic at the latest.

North-east Australia lies on a cratonic margin that has had a complex crustal history involving the successive development of several Proterozoic to Palaeozoic orogenic systems (Fig. 12). Age data from the faults defining the margins of the Millungera Basin are thus important in revealing concealed major Proterozoic tectonic zones in Australia, which contain energy and mineral resources (Korsch et al., 2011). The age data from the fault gouges provide clear evidence for a late Mesoproterozoic minimum age for the Millungera Basin and are in accordance with the early–middle Mesoproterozoic maximum depositional age of the Millungera Basin as constrained from zircon ages for Millungera Basin sandstones (Neumann and Kositcin, 2011).

The fault gouge ages clustering at ∼1115±26, ∼ 1070±25, ∼1040±24, ∼1000±23, and ∼905±21 Ma (Fig. 7) may be related to the regional extension and associated major thermal event that occurred across Australia at 1120–900 Ma (e.g. Musgrave Orogeny and subsequent Giles Event) due to interactions between Australia and other continents during the assembly of the supercontinent Rodinia (De Vries et al., 2008; Li et al., 2008; Evins et al., 2010). This Australia-wide tectonothermal event that largely developed along former (Mesoproterozoic) collision zones led to the emplacement of widespread dyke swarms, sills, and associated granite plutons in the Central and North Australia cratons largely in a time frame between ∼1040 and ∼1090 Ma (Schmidt et al., 2006; Evins et al., 2010; Aitken et al., 2013). The Musgrave Orogeny involving the widespread emplacement of granite and mafic–ultramafic bodies was recorded in central Australia at 1220 and 1120 Ma (Evins et al., 2010; Kirkland et al., 2013). Major swarms of dolerite intrusions in the North Australian Craton are dated at 1116±12 Ma (Lakeview Dolerite; Tanaka and Idnurm, 1994), and associated hydrothermal events were recorded in Mount Isa Province (adjacent to the study area; Fig. 1) (Uysal et al., 2004). The illite ages clustering around ∼1100 Ma coincide with the latest stage of the Musgrave Orogeny (Figs. 7b, 12).

Another cycle of mafic intrusions in central Australia occurred during the extensional Giles Event between ∼1078 and 1068 Ma (Evins et al., 2010; Aitken et al., 2013; references therein), which was followed by granite magmatism and accompanying felsic volcanism between ∼1050 and ∼1040 Ma (Evins et al., 2010, and references therein). The latest phase of the Giles Event is represented by the felsic Smoke Hill Volcanics yielding an age of 1026±26 Ma. The K–Ar ages clustering around ∼1040 Ma of different illite size fractions from the fault gouges and sandstones are consistent with the timing of the later stage of the Giles event.

Orogenic events post-dating the Giles event are represented by mafic dykes and rare pegmatites emplaced at about 1000 Ma (Evins et al., 2010). Further, a Rb–Sr age of 897±9 Ma is reported for dolerite from the Stuart Dyke Swarm in the southern part of the Arunta Block, Northern Territory (Black et al., 1980). The Rodinia supercontinent was assembled through worldwide orogenic events by 900 Ma. Stresses induced by the ca. 900 Ma event probably caused the reactivation of older orogens within Rodinia (Li et al., 2008). The Amadeus Basin in north-central Australia was initiated at ∼900 Ma in the late Proterozoic by crustal extension, probably in association with mafic intrusions being correlated with the Stuart Dyke Swarm (Korsch and Lindsay, 1989). Fault gouge K–Ar ages of ∼ 950–900 Ma for various size fractions from sample Dob-449.1 and the finest fraction (<0.1 mm) from Dob-389.6 coincide with the timing of deformation associated with these early Neoproterozoic igneous and deformation events (Fig. 7b).

The dated faults of the Millungera Basin may be associated regionally with a series of fault systems bounding rift basins in the southern Georgina Basin. Those fault zones (e.g. Burke River Structural Belt, Pilgrim Fault Zone; Greene, 2010) which are in close proximity to and run parallel with the dated faults framing the Millungera Basin occur extensively in the adjacent Mount Isa Inlier (Greene, 2010; Korsch et al., 2011). The Pilgrim Fault Zone was established in a Mesoproterozoic structural boundary within the Mount Isa Inlier (Greene, 2010). The southern Burke River Fault, just to the west of the Millungera Basin (Fig. 1), represents a rift-bounding normal fault, and it was reactivated and inverted to reverse faults during the mid-Palaeozoic Alice Springs Orogeny (∼ 400–350 Ma) (Greene, 2010). Similarly, samples from this study were taken from thrust faults at the margin of the Millungera Basin. However, the K–Ar, 40Ar–39Ar, and Rb–Sr ages of the fault gouge illites have been essentially preserved, and no tectonic event after about 905 Ma has reset the isotopic systematics of these fault gouges (Fig. 12). This can be explained by the lack of significant fluid or heat flow events allowing recrystallization or 40Ar diffusion from illites. In conclusion, our geochronological age data constrain the timing of fault activity associated with the late Mesoproterozoic and early Neoproterozoic emplacement of the intrusions and crustal regional extension in north-central Australia.