Excess ⁴⁰Ar in Alkali Feldspar and ^206,207Pb in Apatite Caused by Fluid-Induced Recrystallisation in a Semi-Closed Environment in Proterozoic (Meta)Granites of the Mt Isa Inlier, NE Australia

Abstract

Interpretation of ⁴⁰Ar/³⁹Ar dates of alkali feldspar and U-Pb dates of apatite depends on the dominant mechanism of isotopic transport in these minerals, which can be either diffusion or fluid-assisted dissolution-reprecipitation. To clarify the contributions of these processes, we have conducted a holistic study of alkali feldspar, apatite and other minerals from the Mt. Isa Inlier in NE Australia. Mineral characterisation by electron microscopy, optical cathodoluminescence imaging and element mapping reveal a complex interplay of textures resulting from magmatic crystallisation, deuteric recrystallisation, local deformation with subsequent higher-temperature alteration, and finally ubiquitous low-temperature alteration. U-Pb and Pb isotopic data for zircon, apatite, fluorite and alkali feldspar suggest that the latter event occurred at ~300 Ma and was associated with fluid-assisted exchange of Pb isotopes between minerals in the same rock, causing some apatite grains to have ²⁰⁷Pb-corrected U-Pb dates that exceed their crystallisation age. However, this event had no unequivocal effect on the ⁴⁰Ar/³⁹Ar or Rb-Sr systematics of the alkali feldspar, which were disturbed by higher-temperature alteration at ~1450 Ma. The age of the latter event is derived from Rb-Sr data. ⁴⁰Ar/³⁹Ar dates are very scattered and suggest that ⁴⁰Ar redistribution proceeded by diffusion in the presence of traps in some places and by dissolution-reprecipitation with variable amounts of recycling in other places. Our results demonstrate the complex effects that interaction with limited amounts of fluids can have on ⁴⁰Ar/³⁹Ar dates of alkali feldspar and U-Pb dates of apatite and thereby reinforce previous critique of their suitability for thermochronological reconstructions. We further identify and discuss potential implications for noble gas geochronology of groundwaters and fission track dating of apatite.

1. Introduction

Isotopic dating of minerals is one of the most important tools in modern geology because it provides critical information about the timing and duration of various events and processes. The accuracy of geochronological interpretations hinges on our ability to identify and account for the processes that modified the isotopic compositions of the dated minerals after their initial crystallisation, which is often not trivial. This contribution seeks to improve our understanding of such processes using the examples of alkali feldspar ⁴⁰Ar/³⁹Ar and apatite U-Pb geochronometers.

Both ⁴⁰Ar/³⁹Ar dating of alkali feldspar [1,2,3] and U-Pb dating of apatite [4,5,6] have been used as thermochronological tools for reconstructing time-temperatures paths of rocks between ~150–350 °C and ~350–550 °C, respectively. Thermochronology for a given mineral—isotope system requires that the redistribution of parent and daughter isotopes within it was dominated by volume diffusion. As shown by several studies, neither alkali feldspar [7,8,9,10,11] nor apatite [12,13,14,15] always meet this prerequisite because they often undergo significant fluid-induced recrystallisation at temperatures near or below their closure temperature window for efficient diffusive transport of Ar (alkali feldspar) or Pb (apatite). In such cases, the isotopic dates are typically suggested to lie somewhere between the age of crystallisation and the time of interaction with fluids, depending on the proportion of domains that were not altered by the fluids [7,8,9,10,11,12,13,14,15]. Following this reasoning, the youngest date provides the maximum estimate for the age of fluid interaction, and thus an opportunity emerges to use both ⁴⁰Ar/³⁹Ar dating of alkali feldspar and U-Pb dating of apatite to track fluid flow in the Earth’s crust. However, is this logic always correct? The data presented below suggest not.

We report the results of our inquiry into the mechanisms that governed the redistribution of radiogenic ⁴⁰Ar in alkali feldspar and ^206,207Pb in apatite from Proterozoic granitic batholiths of the Mt. Isa Inlier (NE Australia). The initial goal of this study was to obtain and compare time-temperature paths from alkali feldspar ⁴⁰Ar/³⁹Ar and apatite U-Pb dates to test their utility for thermochronology. However, as the data accrued, we realised that our samples are unsuitable for thermochronology due to ubiquitous fluid-induced recrystallisation, and that the changes that those fluids imparted on the isotopic compositions of minerals cannot always be adequately described by the simple removal of daughter isotopes. Our data are better explained by a more complex process of fluid-mediated redistribution of daughter and, sometimes, parent isotopes within and between minerals during their recrystallisation in a semi-closed system. In extreme cases, this process may cause isotopic dates from recrystallised domains within altered minerals to exceed the time of their initial crystallisation. Our results challenge the prevailing expectation that fluid-induced recrystallisation removes radiogenic isotopes from minerals and prompt more rigorous interpretations of geochronological data from altered samples.

2. Sample Provenance

The Mt. Isa Inlier in NE Australia exposes a complex assembly of Paleo- to Mesoproterozoic rocks surrounded by subhorizontal Phanerozoic sedimentary cover (Figure 1). The oldest sequences are amphibolite-facies metamorphic rocks that formed during the ~1900–1870 Ma Barramundi Orogeny, which was followed by felsic magmatism at ~1875–1850 Ma that produced the Leichhardt Volcanics along with the Kalkadoon Batholith and associated intrusions [16,17,18,19]. Subsequently, volcano-sedimentary sequences were deposited during ~1790–1590 Ma [16,17]. Although unconformities exist in these strata, leading to their current subdivision into the ~1790–1730 Ma Leichhardt, ~1730–1670 Ma Calvert and ~1670–1590 Ma Isa Superbasins [16], they only suffered minor deformation during this stage caused by the emplacement of contemporaneous felsic plutons, such as the Wonga and Sybella Batholiths [16,17,20,21]. The entire region was affected by extensive deformation and metamorphism during the ~1600–1500 Ma Isan Orogeny, which terminated with yet another episode of felsic magmatism, forming the ~1550–1490 Ma Naraku and Williams Batholiths and associated intrusions [16,17,21,22]. Except for the emplacement of some mafic dykes at ~1115 Ma, the following history of the inlier was generally quiescent with alternating episodes of accelerated erosion and sedimentation [16,17].

Peak metamorphic grades during the Isan Orogeny varied between the lower greenschist and upper amphibolite facies (Figure 1; the highest estimates for the pressure and temperature are 6 kb and 714 °C) [23]. Recent studies suggest that these conditions were reached at ~1595–1580 Ma [22,23,24], although younger estimates have been published, e.g., ~1532 Ma in [20]. The timing of post-peak cooling has been estimated by interpreting ⁴⁰Ar/³⁹Ar dates of amphibole (1843–1398 Ma), muscovite (1490–1363 Ma), biotite (1534–1122 Ma) and alkali feldspar (1259–767 Ma) [25,26,27] and fission track dates of apatite (390–225 Ma) [25,26,28,29] within the framework of thermochronology. It was suggested that cooling through ~530–300 °C spanned the period from ~1530 to ~1370 Ma [25,26,27], while cooling to <110 °C was inferred to have mainly occurred between ~350 and ~250 Ma [28,29]. These cooling intervals were bridged by time-temperature paths retrieved from multi-diffusion domain modelling of alkali feldspar ⁴⁰Ar/³⁹Ar data, which featured periods of accelerated cooling at ~1280–1050, ~740–640 and ~600–500 Ma following possible reheating during the first of these intervals [25,26].

Our present work focuses on 4 samples (Figure 1). The youngest was acquired from the part of the I-type Williams Batholith that was emplaced towards the end of the Isan Orogeny [21,22,30]. Two more samples were obtained from the southern and northern parts of the I-type Sybella Batholith, while the remaining sample was taken from the I-type Kalkadoon Batholith [19,21]. The latter three samples were acquired from batholiths that predate the Isan Orogeny [19,20,21] in regions that were subjected to amphibolite facies metamorphism during the Isan Orogeny [23]. Hereafter, we will refer to our samples by the name of their source batholith as shown in Figure 1.

3. Methods

The samples were studies using a wide range of techniques, which are briefly summarised in

Table 1. Summary of characterisation and analytical techniques used here.

Type of Work	Technique(s) and Key Details
Petrological characterisation	Classical microscopy, optical cathodoluminescence (CL) and backscattered electron (BSE) imaging, qualitative analysis and phase identification by energy-dispersive X-ray spectroscopy (EDS), and semi-quantitative element mapping by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) with analytical data processing in Iolite 2.5 [31].
In situ U-Pb for zircon	LA-ICP-MS. GJ-1 [32] and Plešovice [33] zircon standards were used. Analytical data were processed in Excel following [34,35] and using 235,238U decay constants from [36].
In situ U-Pb for apatite	Laser ablation multiple collector inductively coupled plasma mass spectrometry (LA-MC-ICP-MS). McClure Mountain [37,38] Emerald Lake and Durango [39] apatite standards were used. Analytical data were reduced in Excel following [34,35,38]. 235,238U decay constants were taken from [36], while common Pb composition was estimated following [40].
Bulk grain U-Pb for apatite	Isotope dilution thermal ionisation mass spectrometry (ID-TIMS). 205Pb-233U-235U tracer solution ET535 [41,42] was used. Procedures for chemical separation and analysis followed [43,44]. Analytical data were processed in Tripoli and YourLab software [45,46,47] and in Excel using 235,238U decay constants from [36] and model common Pb compositions from [40].
Pb isotopes for alkali feldspar	Multiple collector inductively coupled plasma mass spectrometry (MC-ICP-MS). Chemical work followed [48], and SRM981 [49] was used for corrections.
In situ Rb-Sr for alkali feldspar	Reaction cell laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS/MS). MicaMg-NP [50,51], NIST 610 [52], SagaB biotite and alkali feldspar [53,54] and Phalaborwa biotite [53,55] were used as standards. Background and drift correction followed [56], while elemental fractionation correction followed [57], and all calculations were performed in Excel with 87Rb decay constant from [58]. Some chemical data were simultaneously acquired and processed using HDIP software. See Supplementary Archive for a discussion of our analytical approach in the context of prior studies [53,59,60].
In situ, bulk grain 40Ar/39Ar for alkali feldspar	Noble gas mass spectrometry. Fish Canyon Tuff sanidine was used as the neutron flux monitor with age as reported in [61]. Analytical data were reduced in ArArCALC [62] using 40K decay constants and 40Ar/36Ar ratio in air from [63].

. The detailed description of the methodology is provided in the Supplementary Archive. This archive further contains all of the isotopic data discussed below, all of the isotopic data from other granite samples from the region that we have acquired and processed, and representative images demonstrating mineral textures.

4. Results

4.1. Rock and Mineral Textures

4.1.1. Hand Specimens

Based on their mineral composition, all four samples can be classified as biotite granites [64]. This is consistent with how their source bodies are referred to in geological maps and previous literature. However, the degree to which these samples were affected by various combinations of alteration and metamorphism calls into question whether it is valid to refer to them as to magmatic rocks.

The closest to representing a magmatic rock is the Williams sample, which is almost as fresh as granites can be (Figure 2a). It has a porphyritic texture with cm-scale phenocrysts of alkali feldspar embedded in an equigranular matrix of up to mm-scale crystals of biotite, plagioclase, alkali feldspar and quartz. Prominent accessories are magnetite, titanite, apatite and zircon. In places, plagioclase is replaced by albite with muscovite and/or epidote, while biotite is replaced by chamosite with occasional epidote. Other minerals are likewise altered, which we detail below.

At first glance, the Kalkadoon sample appears as a similarly fresh magmatic rock (Figure 2b). With the exception of one cm-scale alkali feldspar phenocryst, it is aphyric and consists of mm-scale equigranular crystals of the same primary rock-forming minerals and main accessories as the Williams sample. However, a closer inspection reveals that a sequence of initial alteration, metamorphism and post-metamorphic alteration has completely reworked almost every part of this rock. For example, virtually all biotite is replaced by aggregates that mainly consist of chamosite and epidote, while what used to be plagioclase is typically represented by aggregates of albite, muscovite and/or epidote, where the latter two minerals are frequently zoned and have Fe-poor cores and Fe-rich rims. More examples of alteration are given below. Hence, the term metagranite is probably more appropriate for this sample [64,65].

Both Sybella samples are significantly foliated, have a granoblastic texture and appear as purely metamorphic rocks in thin sections (Figure 2c,d). However, a porphyritic texture is readily recognised in hand specimens, so the rocks can be referred to as metagranites [64,65]. These rocks are composed of biotite, plagioclase, alkali feldspar and quartz, and prominent accessory phases include magnetite, zircon, titanite, apatite, thorite and an altered mineral that probably was allanite. Subsequent to the development of the foliation, biotite was partially replaced by chamosite with rare epidote, while portions of plagioclase were replaced by aggregates of albite with muscovite and/or epidote. Further examples of alterations are provided below.

4.1.2. Zircon

Zircon from the studied samples has a variably intense brown colour and generally forms crystals with well-defined faces and habits that are typical of a magmatic origin [66], although crystals from both Sybella samples sometimes have rounded tips that probably result from post-magmatic deformation. BSE (Figure 3) and CL (Supplementary Archive) imaging reveals that zircon from all samples has regions with oscillatory zoning that is typical of a magmatic origin [66]. These regions are cut by irregular patches and veins that are usually ascribed to fluid-induced dissolution-reprecipitation of zircon and have been reproduced in experiments simulating hydrothermal conditions [66,67,68].

Altered crystals are particularly common in the Williams, Kalkadoon and Sybella S samples, which reflects their enrichment in U (and Th). Although we did not analyse U quantitatively, a simple comparison of U signal intensities obtained for our samples and the GJ-1 standard during U-Pb analysis indicates that U contents in zircon from the Williams sample are in the order of 10² ppm, zircon from the Kalkadoon and Sybella S samples has 10¹–10³ ppm of U, while zircon from the Sybella N sample generally has 10¹ ppm of U with only occasional analyses reaching 10² ppm. These variations in U contents correlate with the number of cracks observed in the samples (more U—more cracks) and the type of contrast in BSE images (structural contrast with high-U zones being dark in the Williams and Kalkadoon samples, compositional contrast with high-U zones being bright in the Sybella N sample, and mixed in the Sybella S sample). Both microfracturing [69] and the development of structural contrast [70] can be interpreted as a result of radiation damage to the zircon structure, although previous work showed that the BSE signal intensity is positively correlated with the dose [70], which is opposite to our observation. The cause of the latter disparity is unclear.

4.1.3. Other U-Th-Bearing Silicates

Titanite is present in all four samples and typically shows patchy textures in BSE images, such as those in Figure 4a,b. Such textures in titanite are usually suggested to have formed by fluid-induced dissolution-reprecipitation [71,72], and we have no reason to offer an alternative interpretation for our samples. While some patchiness in titanite from the Sybella samples may have developed during the foliation-forming metamorphic event (as in apatite, see below), much of it in these and other samples was clearly simultaneous with the alteration of biotite to chamosite and plagioclase to albite with muscovite and/or epidote (Figure 4a). Titanite from the Sybella S sample is sometimes decomposed to an aggregate of minerals that we did not attempt to identify (Figure 4b). This type of alteration is spatially distributed in a manner that correlates poorly with the occurrence of the biotite and plagioclase alteration, but it can often be linked to visible cracks and grain boundaries, suggesting that it probably occurred under supergene conditions.

Spectacular polyphase aggregates that form pseudomorphs, presumably after allanite (Figure 4c) have been encountered in the Sybella S and most of our other samples from the Sybella batholith. The constituent minerals of these aggregates have not been systematically identified. Among them are REE carbonates that occur within replaced allanite and are also spread over mm-scale distances along cracks that radiate from allanite (Figure 4c). The fact that the carbonate-filled cracks are readily identifiable, unhealed and generally lack spatial correlation with zones of biotite and plagioclase alteration suggests that allanite decomposition occurred after the rocks became exposed to supergene conditions. No such conspicuous allanite pseudomorphs were recorded within the Kalkadoon and Williams samples, although both have individual veins or patches of REE-rich minerals that could have formed at the expense of rare allanite.

The Sybella S sample further contains thorite that forms rounded elongated grains (Figure 4d). The rounded shape of these grains could be related to the deformation of the rock. Each grain is in fact a polycrystalline aggregate composed of tiny platy crystals of thorite and other phases. One of those other phases is an Fe and Pb-rich mineral or mixture of minerals, which also accumulates at the boundaries of thorite grains and sometimes propagates into cracks that extend from them. Again, a clear correlation of thorite alteration textures with visible cracks suggests that they developed late under supergene conditions.

4.1.4. Apatite

Apatite is abundant in all four samples and exhibits complex CL textures and chemical zoning reflective of its protracted history of magmatic crystallisation followed by fluid-mediated alteration (Figure 5, Figure 6 and Figure 7 and Supplementary Archive). Overall, the freshest crystals (e.g., Figure 5a–d, Figure 6a and Figure 7a) have (more) euhedral habits and regular oscillatory zoning, which are typical of apatite from volcanic [73] and fresher plutonic [74,75,76] rocks (as well as hydrothermal veins [77,78,79,80]). These are characterised by relatively high Th/U and La/Y ratios of ~10 and ~2.5, ~10 and ~2, and ~2 and ~0.1 in the Williams, Kalkadoon and Sybella N+S samples, respectively. Many crystals display overgrowth rims (e.g., Figure 5g and Figure 6c), irregular replacement zones along the edges (e.g., Figure 5h,l,m, Figure 6d and Figure 7c) or veins and patches cutting the interior parts (e.g., Figure 5f,p, Figure 6a and Figure 7b,c,g), and the advanced development of these textures is clearly associated with the loss of euhedral crystal morphology (e.g., Figure 6e and Figure 7f). Such textures are often related to reactions with fluids in altered and metamorphosed rocks [13,14,15,78,79,81,82,83,84,85], although some authors ascribed them to various process occurring in crystallising magma bodies [76,80,86,87]. In our samples, the former interpretation is favoured because the development of these textures is clearly associated with the decrease in Th/U ratios and, in most cases, La/Y ratios and Th, U and REE+Y contents. This chemical trend is typical for apatite that grew during lower-grade metamorphism in the presence of epidote, which scavenged Th, U and REE+Y from fluids and also reduced Th/U and La/Y ratios in apatite [13,14,83,88]. In detail, however, the styles and effects of alteration vary from sample to sample, which we describe below.

The freshest apatite from the Williams sample has a magmatic appearance both in element maps and CL images, where it displays regular oscillatory zoning with CL colours ranging from deep purple in U, Th and REE-rich zones to light purplish grey in U, Th and REE-poor zones (Figure 5a–d). Such crystals are characterised by Th/U ratios of ~10 (Figure 5a–c). They are outnumbered by crystals with highly variable appearance that we interpret as altered. In the most obvious cases, alteration formed overgrowth rims (Figure 5g) or replacement zones, either along the edges (Figure 5f–h,l–n,p) or as patches or veins (Figure 5f,g,k,o,p), where apatite has much lower Th/U ratios, as low as ~0.5. While in most cases this late apatite is depleted in U, Th and REE relative to the pre-existing apatite and has light blueish grey, light greenish grey and green CL colours (Figure 5h), some veins of late apatite have U contents that are higher than and Th and REE contents that are similar to those in the pre-existing apatite, and their CL colour is deep purple (Figure 5f,g). In other cases, alteration is pervasive and cryptic: affected crystals have a magmatic appearance in that they display oscillatory zoning in CL images and element maps, however their Th/U ratios are significantly lower than ~10 either throughout or in certain zones. These include crystals with Th/U ratios of ~4 and CL colours that range from deep violet in U, Th and REE-rich zones to dark grey in U, Th- and REE-poor zones (Figure 5e), and crystals with Th/U ratios of ~8 and CL colours that range from deep purple in U, Th and REE-rich zones to light greenish grey in U, Th and REE-poor zones (Figure 5g,i–k). La/Y ratios vary sympathetically with Th/U ratios and range from ~2.5 in the freshest zones to ~1 in the most altered zones and overgrowths.

Figure 5q summarises textural evidence for apatite from the Williams sample. We identify two major stages in its history. The first stage is represented by its crystallisation in a magma that produced euhedral oscillatory-zoned crystals with Th/U ratios of ~10 (and La/Y ratios of ~2.5). The second stage is represented by alteration textures, and the observed overlap between different alteration textures (e.g., Figure 5h) can be interpreted as evidence for multiple discrete alteration events. However, these had a unidirectional effect on apatite chemistry and could have occurred over a short period of time, so we do not have substantial reasons to introduce any temporal division between these textures. The development of alteration textures in apatite can be correlated with the replacement of plagioclase by albite with muscovite and/or epidote.

The freshest apatite in the Kalkadoon sample resembles that from the Williams sample: it displays oscillatory zoning in CL images and element maps, while its Th/U and La/Y ratios reach ~10 and ~2, respectively (Figure 6a). However, some notable irregularities are observed in the element maps, and the CL textures and intra-crystal distributions of different elements are decoupled. Although in general zones with brighter CL have lower contents of Th, U and REE+Y, some U-rich veins visible in U and Th/U maps are nearly invisible in CL images and other chemical maps. There is also a striking contrast between irregular and diffuse zoning in La/Y ratios, generally uniform distribution of Th/U ratios and regular oscillations in CL brightness that form well-defined zones with euhedral outline (Figure 6a). Similar decoupling between CL textures and intra-crystal zoning in Th/U and La/Y ratios is also observed in crystals that we interpret as substantially more altered (Figure 6b–e). Most crystals have been affected by pervasive alteration that lowered their Th/U and La/Y ratios down to ~2 and ~0.2, respectively (Figure 6a–d), some also feature rim overgrowths and replacement zones (Figure 6c,d). Some crystals were either completely recrystallised or maybe even newly grown, so that they have a completely irregular morphology and zoning (Figure 6e). We further note that some crystals, including that in Figure 6a, have variations in Pb contents that are not correlated with other chemical parameters (e.g., U contents) and CL textures (Supplementary Archive).

The range of textures of apatite from the Kalkadoon sample is summarised in Figure 6f,g. We identify at least three major stages in its history, which are magmatic crystallisation of oscillatory zoned crystals with Th/U and La/Y ratios of ~10 and ~2, respectively, and 2 subsequent generations of fluid-mediated alteration. The existence of at least 2 generations of alteration is suggested by the disparities in zoning patterns revealed by different chemical maps. The earlier generation primarily removed La from the exterior parts of crystals by pervasive alteration to produce diffuse zones with irregular outline and low La/Y ratios. This event probably also formed some of the replacement zones/rim overgrowths with lower La/Y and higher Th/U ratios. The later generation had little if any effect on La contents and La/Y ratios but decreased Th/U ratios through the preferential addition of U and/or removal of Th. Crystals were altered pervasively or along some veins and replacement zones in their outer parts. Some rim overgrowths (where both La/Y and Th/U ratios are low) probably also formed at this stage. The irregular variation of Pb contents in some crystals may have been caused by yet another generation of alteration, although evidence for this is scarce. The post-magmatic origin of all features that we relate to alteration is attested by their poor correlation with oscillatory zoning revealed by CL imaging.

Apatite from both Sybella samples is significantly more altered, such that it is unclear whether any purely magmatic textures and compositions are preserved. CL colours vary from deep blue in U, Th and REE+Y-rich zones to blueish or yellowish green in U, Th and REE+Y-poor zones. The crystals which are freshest in appearance have predominantly deep blue CL and oscillatory zoning where individual zones have a euhedral outline, and their Th/U and La/Y ratios are 2 and 0.1, respectively (Figure 7a). Although some veins are enriched in U (with no considerable variation in Th and REE+Y contents) and have deep blue CL (Figure 7b), altered regions are generally depleted in U (as well as Th and REE+Y) and have a green CL response (Figure 7a–i). As in the Kalkadoon sample, there is a clear decoupling between different chemical parameters, such as La/Y and Th/U ratios, and their correlation with CL colours is not always obvious (Figure 7a–c,h). Altered regions have relatively low Th/U ratios that are as low as ~0.1, while their La/Y ratios vary from ~0.03 to ~0.2. Alteration produced various combinations of rim overgrowths (most evident in the uppermost and lowermost crystals in Figure 7f), replacement zones along crystal edges (essentially every crystal in Figure 7) and veins (most evident in Figure 7f,g). Individual thin veins often interlace with each other to form pervasively altered regions that show the entire range of CL colours from blue to green (compare Figure 7a–e,g,h, as well as individual grains in Figure 7f), so that the original blue cores are sometimes barely visible (Figure 7d,e, top crystal in Figure 7f). Some crystals display oscillatory zoning where the outline of individual zones is highly irregular, which presumably resulted from alteration of a more regular magmatic texture (Figure 7b). In other crystals, textures with a blueish green CL response appear to be overprinted by those with a yellowish green CL response (bottom crystal in Figure 7f, Supplementary Archive). Additionally, some grains have late cracks that are enriched in La and Fe (Figure 7d,e), as well as Pb, V, Mn, Mg, Ti, Zr and Ti (Supplementary Archive). These late cracks were not always visible under a binocular or an optical microscope before or even after polishing, however their compositional difference with cracks that certainly formed during polishing, which are invisible in CL images and element maps (Figure 7b, Supplementary Archive), indicates that they existed before sample preparation. Notably, the late cracks are usually completely invisible on U, Th, Y and Th/U maps (Figure 7d,e, Supplementary Archive).

The range of textures of apatite from the Sybella samples is summarised in Figure 7j. We envisage that its history started with the formation of oscillatory zoned crystals with Th/U and La/Y ratios of at least ~2 and ~0.1, respectively, but potentially even higher. This stage was followed by protracted alteration. The occurrence of altered crystals in fresh plagioclase (Figure 7i) indicates that some recrystallisation preceded or was coeval with the deformation event responsible for the development of metamorphic foliation. Such early alteration textures are typically blueish green in CL images. Similar alteration textures with CL colours shifted towards yellowish green probably developed while biotite and plagioclase were being decomposed. The overall appearance of the late cracks and their enrichment in La, Fe and Pb makes them very similar to the cracks and grain boundaries filled with the products of the allanite and thorite decomposition, so we suggest that all these are coeval.

4.1.5. Alkali Feldspar

Alkali feldspar textures in the Williams sample match those reported from relatively fresh plutonic rocks such as the Shap granite and the Klokken syenite [89,90,91,92,93], suggesting that they formed in a similar way. First, magmatic crystallisation produced crystals with multiple resorption zones, which are revealed by contrast variations in CL and BSE images (Figure 8a,b). Such zoning patterns are typical of sanidine [94,95] and plagioclase [96,97] from volcanic rocks, where their formation is related to changes in physicochemical conditions in a magma chamber, for example due to injections of new batches of melt. As the host rock cooled to sub-solidus temperatures, Na-K interdiffusion produced Na-rich lamellae in a K-rich matrix, while recrystallisation caused by interaction with deuteric fluids produced veins of replacive perthite that consist of discrete Na and K-rich subgrains (Figure 8). These processes and resulting textures have been investigated in great detail in alkali feldspar crystals from the Shap and Klokken plutons [90,91,92,93]. Finally, even later fluid-induced recrystallisation produced very thin micropore-rich veins, which are usually composed of feldspar of similar composition to that which they cut, although in some cases they exhibit increased Ba contents (Figure 8c). Similar veins were extensively studied in alkali feldspar from the Klokken intrusion, where their formation was associated with low-temperature fluid-mediated alteration hundreds of Ma after the emplacement [92]. In our Williams sample, their occurrence is clearly associated with the replacement of plagioclase by albite with muscovite and/or epidote. The cross-cutting relationships between the observed alkali feldspar textures are summarised in Figure 8d.

Alkali feldspar from the Kalkadoon sample displays a considerably different range of textures, full analogues of which to our knowledge have not been previously described. It has two sets of Na-rich lamellae in a K-rich matrix (Figure 9c). The first set is represented by regularly spaced lamellae that have thicknesses of up to 10⁰ μm and are oriented along the Murchison plane, the usual orientation of exsolution lamellae formed by Na-K interdiffusion. Hereafter, we refer to these as normal lamellae. The second set comprises unevenly spaced lamellae with thicknesses of up to 10¹ μm, which have either a straight or bended shape, the latter resembling flames. We refer to these as flame lamellae in the following text. Such lamellae are predominantly oriented along the (001) cleavage in the crystal shown in Figure 9, although they can have different orientations in other crystals. Although normal lamellae can occur between flame lamellae, most of the regions surrounding the latter only consist of K-rich feldspar with deep blue CL, while most normal lamellae form swarms away from flame lamellae in regions that have a light blue CL response (Figure 9a,b). Furthermore, CL images reveal many intersecting irregular veins of replacive K-feldspar, the latest of which have deep purple colours, while the earlier ones exhibit lighter, more blueish colours (Figure 9a,b). These veins are usually devoid of normal exsolution lamellae, while flame lamellae were seemingly unaffected by these veins and sometimes even presented barriers to their propagation. Occasionally, K-feldspar in such veins alternates with albite, quartz, calcite and/or chamosite. Some zones that appear as those with blue CL in BSE images display variably strong purple hue in CL images, indicating that they have been pervasively altered (Supplementary Archive). Finally, numerous strands of micropore-rich feldspar are revealed by BSE images (Figure 9c). These either occur as isolated features or interlace with each other to form extensive altered regions with a high quantity of micropores.

The relationships between the observed feldspar textures in the Kalkadoon sample are summarised in Figure 9d. It is not entirely clear how they have developed. Obviously, the starting point was crystallisation from a magma, although any evidence of this other than the outline of the crystals and the distribution of inclusions in them is completely eradicated. Next came the development of two sets of Na-rich lamellae, and how this happened is most perplexing. Previous studies of flame lamellae argued that they form by fluid-induced dissolution-reprecipitation in deforming rocks [98,99,100]. However, the Kalkadoon sample lacks any evidence of deformation despite going through the Isan Orogeny (see Section 2), so an alternative explanation is needed. We speculate that its flame lamellae along with the surrounding K-feldspar with deep blue CL are a result of annealing of replacement perthite veins, such as those in the Williams sample, which occurred due to protracted residence at elevated temperatures and the infiltration of fluids prior to and during the Isan Orogeny. Thus, we speculate that flame lamellae are what irregular groups of Na-rich crystals in replacement perthite veins evolve to if chemical and textural relaxation continues for sufficient time (cf. alkali feldspar from the Shap Granite and our Williams and Kalkadoon samples). This suggestion is supported by the lack of discernible variations in composition and BSE contrast that mark resorption zones, which may have been smoothed out over the same time period. Regions with normal lamellae and lighter blue CL probably represent annealed magmatic feldspar. Subsequently, the numerous replacive veins of (mostly) K-feldspar formed, which, considering the presence of chamosite in them, probably happened simultaneously with (some stages of) the biotite to chamosite replacement. At the same time, some parts of annealed magmatic feldspar were altered pervasively, perhaps due to the infiltration of fluids along nanotunnels that might have formed at the boundaries of normal lamellae [91]. Finally, highly porous veins developed, which was probably simultaneous with the formation of Fe-rich overgrowths in muscovite and epidote that occur within altered plagioclase.

Alkali feldspar from both Sybella samples is significantly deformed and often has a granoblastic texture, such that former phenocrysts are now composed of a mosaic of equidimensional crystals (Figure 10a,b). Tartan twinning is frequently observed. Regular Na-rich exsolution lamellae are rare and occur in less deformed areas, and when they are present, they often have a wavy outline and co-exist with Na-rich blebs resembling those interpreted as another generation of exsolution in [101]. These textures presumably predate or formed simultaneously with deformation. Also present are chains of albite crystals that are reminiscent of those commonly observed in replacement perthite veins, some of which probably formed prior to deformation, while others postdate it (Figure 10a,b). K-rich feldspar has a brighter blue CL response in regions that seem to have been relatively unaffected by post-metamorphic alteration and a darker blue CL response along grain boundaries, some chains of Na-rich crystals and in irregular regions away from both, all of which are interpreted as post-deformation fluid-mediated alteration (Figure 10a,b). Finally, like the other two samples, both these samples contain thin veins of highly-porous K-feldspar that cut all the other textures. The relationships between the observed textures are summarised in Figure 10c.

4.1.6. Fluorite

All four samples contain fluorite, whether confirmed or inferred from its blue CL, in regions where plagioclase was replaced with albite and muscovite and/or epidote. Such fluorite typically forms small crystals with sizes of 10⁰ μm. A few larger scale crystals with sizes of up to 10² μm were observed in the Sybella S sample, where they occurred in zones of early post-metamorphic replacement of alkali feldspar along the crystal boundary.

4.1.7. Relationships Between Alteration Textures in Different Minerals

Establishing which alteration processes were coeval is challenging. One contemporary association of textures that can be deduced with good certainty in the Williams sample and seemingly holds in other samples, although not always as clearly, includes the late highly porous veins in alkali feldspar, aggregates of albite with muscovite and/or epidote replacing plagioclase, chamosite and occasional epidote formed in place of biotite, patchiness in titanite and (later stages of) alteration of apatite. It is unclear from the petrological evidence alone whether zircon was altered at the same time, although the isotopic evidence that we discuss below suggests that it was. The development of the replacement perthite veins in alkali feldspar from the Williams sample clearly predated these textures and had no obvious effect on other minerals. In the Kalkadoon and both Sybella samples these textures are predated by the development of the early post-metamorphic veins of albite and K-feldspar, which may have occurred simultaneously with the early stages of the replacement of biotite with chamosite and epidote and, especially in the Kalkadoon sample, the development of larger muscovite crystals within altered plagioclase. Apatite (and also titanite) may have been affected by this process. In the Sybella samples, some apatite alteration clearly predates their deformation. The breakdown of titanite and allanite and the late alteration of thorite in the Sybella S sample occurred during the waning stages of or after the development of the association of textures that includes the late highly porous veins in alkali feldspar. All of the textures mentioned in this subsection must have formed during or after the Isan Orogeny (see Section 2). It is unclear from the evidence discussed whether this happened during punctuated short-lived events or gradually as the rocks were exhumed.

4.2. Isotopic Data

4.2.1. Zircon

In situ LA-ICP-MS U-Pb dating of zircon yielded data sets with many discordant analyses for each of the four samples discussed here in detail (Figure 11) and also for our other samples from the Mt. Isa Inlier (Figure 12). Analyses obtained for individual samples typically define linear trends that intercept the concordia at Proterozoic and Phanerozoic ages, although the scatter about these trends usually exceeds that expected from analytical uncertainties. To extract statistically plausible linear trends that account for most of the data, we have iteratively rejected analyses with the highest contribution to MSWD values calculated for discordias fitted by the method of York [102,103] until meeting the criterion of Wendt and Carl [104]. Final calculations of the intercept dates were performed in Isoplot [105]. Upper intercept dates for the Williams (1508 ± 8 Ma), Sybella S (1658 ± 7 Ma) and Sybella N (1667 ± 15 Ma) samples (and also for the other samples in Figure 12) are in good agreement with the published crystallisation ages of the respective intrusions, while that for the Kalkadoon sample (1832 ± 8 Ma) is 18 Ma younger than the previously reported lower age limit for its source batholith [16,17,18,19,20,22]. Lower intercept dates of 320–315 Ma have been obtained for the Williams, Kalkadoon and Sybella S samples, while that of the Sybella N sample is not well defined.

For the purpose of our study, the upper intercept dates are taken as the crystallisation ages, although the validity of doing so in the case of the Sybella Batholith requires further investigation. The U-Pb analyses from different samples of the Sybella Batholith spread out near the concordia resembling a discordia vectored to a Proterozoic Pb loss event. Of greater relevance to our aims are the lower intercept dates. The immediate interpretation for these would be a Pb loss event at ~300 Ma, which can be readily attributed to fluid-mediated recrystallisation that produced the observed zircon textures (see Section 4.1.2). However, it has been previously suggested that lower intercept dates can be spurious [106], and we examine this possibility more closely below.

Pb loss from zircon is generally considered to be a fluid-mediated process affecting crystals that were significantly damaged by radioactive decay and leading to either partial or complete resetting of their U-Pb systematics [67,68,106,107,108]. The analyses that we obtained from oscillatory zoned regions typically fit well into this scheme, and greater discordance is recorded for domains with higher U contents and apparent degree of metamictisation (Figure 3b,d, Section 4.1.2). However, some of the least discordant analyses originate from patches of replacive zircon with a homogenous bright BSE signal, which is most apparent in the Williams sample (Figure 3a). While perplexing, this is not entirely inconsistent with the experimental results on hydrothermal recrystallisation of zircon. Thus, despite the high fluid/zircon ratio in the system, one of the in situ analyses from altered zones in the experiments of Geisler et al. [68] was concordant and matched the crystallisation age, while Sinha et al. [108] obtained concordant crystallisation ages by bulk-grain analyses of reacted zircon. An alternative explanation is that given by Pidgeon [107] for a very similar case from an Archean granitic intrusion elsewhere in Australia, who concluded that analogous homogenous replacive patches in zircon yielded concordant dates representing the emplacement age because of having formed shortly after magma solidification. This explanation seems less viable to us because the development of such recrystallisation textures seems to be associated with old rocks with much younger lower intercept dates and thus high degrees of radiation damage of zircon. Evidence for fluid-induced zircon replacement close to the emplacement age is not found in many younger rocks, such as the Devonian Shap granite [109] or the youngest exposed granites of the Quaternary Takidani pluton [110], which otherwise display similar textures ascribable to interactions with deuteric fluids (e.g., cf. our characterisation results for the Williams sample and those reported in [89,111]).

The significance of the lower intercept dates obtained for the Williams, Kalkadoon and Sybella S samples is demonstrated by their presence in multiple samples with different crystallisation ages (Figure 12). Similar lower intercept dates were further obtained for several of our other samples and have also been reported in previous zircon U-Pb surveys of the Mt. Isa Inlier [18,20,22], although more Phanerozoic apparent Pb loss events can be identified in both sets of data. In our samples, most of the lower intercept dates fall in the range of 339–253 Ma, which is very similar to the published range of apatite fission track dates that span 390–225 Ma [25,26,28,29] and corresponds well with the Middle Ordovician to Middle Jurassic hiatus in the Phanerozoic sedimentary cover [17]. Therefore, we conclude that the lower intercept dates of ~300 Ma are not spurious and do reflect the timing of the zircon-fluid interaction event that caused zircon recrystallisation. The excess scatter about the fitted discordias could have multiple causes, such as the protracted nature of zircon-fluid interaction, the presence of ante- or xenocrystic domains in the analysed crystals or the local effects of recrystallisation on isotopic systematics. Note that the ²⁰⁷Pb/²⁰⁶Pb ratio in the Pb that zircon lost during this event was ~0.1 (Figure 11).

4.2.2. Pb Isotopes in Alkali Feldspar

Bulk-grain MC-ICP-MS Pb isotope analyses of alkali feldspar from each sample are more radiogenic than predicted by the model of Stacey and Kramers [40] for the time of their crystallisation (Figure 11). Although the difference is minor in the case of the Sybella N sample (0.2 rel. % difference in ²⁰⁷Pb/²⁰⁶Pb ratios), the ²⁰⁷Pb/²⁰⁶Pb ratios obtained for the Sybella S and Kalkadoon samples are, respectively, 1.6 and 2.5 rel. % lower than the model values, while that for the Williams sample is 22.2 rel. % lower than the model value, placing it below even the present-day value. Clearly, an uptake of Pb from a source with a low ²⁰⁷Pb/²⁰⁶Pb ratio has occurred, the degree of which appears to be correlated with the content of Pb in alkali feldspar. Although we do not have properly calibrated measurements, our semi-quantitative estimates suggest Pb contents of ~1 ppm for the Williams sample, ~20 ppm for the Kalkadoon and Sybella S samples and ~34 ppm for the Sybella N sample. These estimates were obtained by comparing the intensities of ²⁰⁸Pb signal per weight of dissolved material that we observed for our samples and for alkali feldspar from the Shap granite, which contains ~40 ppm of Pb [112].

4.2.3. Apatite

U-Pb dating of apatite via both in situ LA-MC-ICP-MS and bulk-grain ID-TIMS yielded data that show a great deal of scatter for each sample when plotted in Terra-Wasserburg space (Figure 11). Many analyses in the Williams and Kalkadoon samples and a few in the Sybella S sample fall left of the tie-lines connecting the points on the concordia corresponding to the crystallisation ages of the samples with the isotope compositions of common Pb, whether predicted by the model of Stacey and Kramers [40] or taken to be equal to the ²⁰⁷Pb/²⁰⁶Pb ratios measured in alkali feldspar from the same samples (see Section 4.2.2). In other words, ²⁰⁷Pb-corrected U-Pb dates of some of the analysed crystals exceed the crystallisation ages of their source rocks regardless of which of the standard methods for estimating the common Pb composition was used. Other analyses plot near the said tie-lines or right of them, implying ²⁰⁷Pb-corrected U-Pb dates close to or below the crystallisation age. Overall, the obtained ²⁰⁷Pb-corrected U-Pb dates range from 1664 ± 37 to 1347 ± 86 Ma for the Williams sample, from 1965 ± 69 to 1833 ± 42 Ma for the Kalkadoon sample, from 2376 ± 67 to 1397 ± 23 Ma for the Sybella S sample and from 1619 ± 88 to 1433 ± 24 Ma for the Sybella N sample. In the following, we only refer to these even if the correction method is not indicated.

The acquisition of U-Pb dates of apatite that are older than the crystallisation age of its source rock is surprising. Apart from another study from the same region [113] that was presented in parallel with ours [114], we are not aware of any analogous cases. Although some research suggests that apatite xenocrysts can occur in granitic intrusions [115], a xenocrystic origin is not a viable explanation for results. To begin with, there is no petrological evidence for xenocrystic apatite in our samples of I-type (meta)granites (see Section 2 and Section 4.1). However, even if there were apatite xenocrysts, it is unlikely that they would retain much of the radiogenic Pb accumulated prior to being entrained by the magma. Modelling of conductive cooling of large granitic plutons suggests that these reside at near solidus temperatures for 10⁻¹ Ma or more [116,117,118]. In the ‘coldest’ of the cited models, cooling of an H₂O-rich peraluminous leucogranite from the emplacement temperature of 700 °C to 500 °C spanned ~0.1 Ma [117]. Simulations of diffusive Pb loss from apatite within H₂O-poor rhyolite melt, which we performed using the spherincl.m script from [119] and Pb diffusion and partitioning parameters from [120,121,122], indicate that even this thermal disturbance would be sufficient to completely reset the U-Pb dates of xenocrysts predating the emplacement by as much as 1000 Ma and having radii of up to ~70 μm. Therefore, the old U-Pb dates of apatite require a different explanation.

Insights into the origin of the apatite U-Pb compositions are provided by comparing dates with textural and chemical evidence (see Section 4.1.4). In the Williams sample, unaltered oscillatory-zoned crystals or regions within crystals with high Th/U ratios of ~10 yielded U-Pb dates of 1528 ± 13 to 1509 ± 30 Ma (Figure 5b–d,h, Supplementary Archive), which are close to the crystallisation age. In contrast, much older and much younger U-Pb dates ranging from 1664 ± 37 Ma to 1347 ± 88 Ma were obtained from altered crystals or regions within crystals featuring low Th/U ratios of ~4 and/or irregular replacement zones (Figure 5h–p, Supplementary Archive). This demonstrates that alteration is the cause of both positive and negative shifts in U-Pb dates relative to the crystallisation age in this sample. In the Kalkadoon sample, only one U-Pb date of 1833 ± 42 Ma overlaps with the crystallisation age, while the remaining ones are up to 100 Ma older (e.g., 1965 ± 69 and 1935 ± 4 Ma). It is hard to say if any of these were obtained from fresh apatite due to incomplete characterisation data and significant complexity of alteration in this sample. In the Sybella samples, two crystals that are inferred to be relatively fresh from their Th/U ratios of ~2 and La/Y ratios of ~0.1 yielded U-Pb dates of 1493 ± 5 and 1487 ± 1 Ma (Figure 7b,c), while three crystals that are thought to be relatively fresh because of their predominantly blue CL yielded U-Pb dates of 1497 ± 4, 1491 ± 5 and 1466 ± 1 Ma (Supplementary Archive). These are significantly younger than the crystallisation ages found for these samples and possibly result from resetting by the foliation-forming metamorphic event. U-Pb dates of altered grains are much more variable. Bulk-grain analyses obtained for the Sybella S sample range from 1563 ± 1 to 1457 ± 13 (Figure 7d, Supplementary Archive), while in situ analyses of the Sybella S and Sybella N samples vary from 2376 ± 67 to 1397 ± 23 Ma and from 1619 ± 88 to 1433 ± 24 Ma, respectively (Figure 7f–h, Supplementary Archive). Here again, alteration has clearly caused U-Pb dates to scatter to both older and younger values relative to the U-Pb dates from less affected crystals.

In principle, there are two end-member mechanisms by which alteration can modify-U-Pb dates, and these are changes to contents of either U or Pb. An increase of U-Pb dates can be achieved by either the removal of U or the addition of Pb with a low ²⁰⁷Pb/²⁰⁶Pb ratio, while the inverse changes can cause the opposite effect. The most conspicuous example of the addition of Pb is provided by La- and Fe-rich cracks in the Sybella samples (Figure 7d), which are also enriched in Pb (Supplementary Archive) and can be associated with the late breakdown of allanite and thorite. It is more challenging to establish which of these end-member mechanisms dominated during the earlier generations of alteration. Thus, while most replacement zones and rim overgrowths are depleted in U relative to pre-existing apatite (e.g., Figure 5h, Figure 6d and Figure 7d), some veins are enriched in U (e.g., Figure 5f, Figure 6a and Figure 7b), indicating that U was mobile in fluids and could be added in one local environment and removed in another. Alteration-related changes to Th/U, La/Y and Pb/U ratios (Figure 5 and Figure 6; Supplementary Archive) demonstrate that Th, La, Y and Pb were likewise mobile in fluids. Therefore, Pb could also be gained in some reprecipitated regions and lost in others. Arguably, a preponderant U removal is a more viable process to explain excessively old U-Pb dates in regions affected by pervasive alteration (e.g., crystals in Figure 5i,j, relics of cores with blue CL in Figure 7f–h), while an important contribution from the addition of Pb with low ²⁰⁷Pb/²⁰⁶Pb ratios is expected in replacement zones and rim overgrowths (e.g., those in Figure 5h,p and Figure 7f–h). U-Pb dates younger that those obtained for the least altered apatite imply a more thorough chemical and isotopic re-equilibration with the surrounding rock.

4.2.4. Fluorite

Bulk-grain ID-TIMS U-Pb data were acquired for fluorite from the Williams and Sybella S samples because of its clear association with alteration textures in feldspars (see Section 4.1.6). The analysis obtained for the Williams sample fell left of the concordia (Figure 11a), indicating the presence of Pb with an extremely low ²⁰⁷Pb/²⁰⁶Pb ratio that could only be attained by radioactive decay of U, but for the attainment of which there is not enough U. Thus, we suggest that Pb of such composition was carried by fluids, from which this fluorite precipitated. In contrast, fluorite from the Sybella S sample is in line with U-Pb results for apatite, suggesting a more normal composition of Pb in fluids. We note that the analysed grains were relatively large and thus could be associated with the early post-metamorphic replacement of alkali feldspar.

4.2.5. Rb-Sr Results for Alkali Feldspar

In situ LA-ICP-MS/MS Rb-Sr dating of alkali feldspar from the Kalkadoon sample was undertaken to reveal the ages of different feldspar generations described in Section 4.1.5. Isotopic data for each spot were acquired simultaneously with semi-quantitative content determination for several elements such as Ti and Fe. Overall, the sample is characterised by low ⁸⁷Rb/⁸⁶Sr ratios of ≲50 (Figure 13a). K-rich feldspar with a blue CL response (feldspars 1 and 2 in Figure 9d) generally has ⁸⁷Rb/⁸⁶Sr ratios of ≲5, with only a few spots yielding values up to ~13. Most of the latter are proximal to veins and zones of K-rich feldspar with purple CL (feldspars 3, 4 and 5 in Figure 9d), which generally has ⁸⁷Rb/⁸⁶Sr ratios of >5. No further distinction between feldspars 1 and 2 or 3, 4 and 5 can be made based on our Rb-Sr data. Despite having higher ⁸⁷Rb/⁸⁶Sr ratios, feldspars 3, 4 and 5 are only slightly richer in Rb compared to feldspars 1 and 2 (Supplementary Archive), suggesting that the difference mostly results from lower Sr contents. The former group further differs from the latter by having lower contents of Ti and Fe (Supplementary Archive).

A single isochron fit through all of the data gives a date of 1532 ± 32 Ma (Figure 13b), however the MSWD value of 2.0 is above the acceptance threshold of 1.3 as calculated following Wendt and Carl [104]. An acceptable yet imprecise Rb-Sr isochron at 1841 ± 148 Ma is obtained for analyses from feldspars 1 and 2 that have ⁸⁷Rb/⁸⁶Sr ratios of up to 5, with the remaining analyses being excluded because of the likely contamination from feldspars 3, 4 and 5. No statistically acceptable Rb-Sr isochron can be fitted through all the analyses of K-rich feldspar with purple CL or the veins, although the over-dispersed linear arrays that the two groups of analyses define are characterised by similar slopes with dates of 1457 ± 38 and 1456 ± 41 Ma, respectively. In contrast, distributing analyses from proximally located veins into two subgroups yields Rb-Sr isochron dates of 1468 ± 65 and 1429 ± 42 Ma, while their MSWD values of 1.8 and 1.6, respectively, render them statistically acceptable [104]. Finally, a statistically acceptable isochron can be calculated for the 5 analyses with the highest ⁸⁷Rb/⁸⁶Sr, yielding a date of 1343 ± 60 Ma.

Our Rb-Sr analyses reveal two major stages in the history of the alkali feldspar. The first is the formation and textural evolution of all the regions characterised by a blue CL response. The lack of variation of trace element contents across these indicates that they have formed in a sequence of related processes, that is by magmatic crystallisation transitioning to deuteric alteration, which continued for sufficiently long to form flame lamellae and to eliminate chemical zonation. The Rb-Sr isochron date obtained for different textural types of K-feldspar with blue CL coincides with the emplacement age that we established for the Kalkadoon sample, although its poor precision cannot rule out partial resetting by up to 152 Ma. It is expected that the Kalkadoon batholith was reheated during the Isan Orogeny, when the surrounding rocks equilibrated at amphibolite-facies conditions (Section 2, Figure 1). While this event might have advanced the textural and chemical changes mentioned above, it had no noticeable effect on the Rb-Sr systematics. The latter is probably unsurprising, considering that Sr diffusion in alkali feldspar is relatively slow [123] and implies a Dodson’s closure temperature [124] in excess of 500 °C even for a small sphere with the radius of 50 μm. The following stage is the early post-metamorphic replacement, which produced zones and veins of K-rich feldspar with purple CL. Their similar trace element composition and Rb-Sr systematics suggests that this was a relatively short-lived event. The excess scatter about the isochrons fitted through all the analyses of zones and veins with purple CL reveals a lack of large-scale equilibration between the reacting mineral and fluid phases, raising the possibility that the isotopic composition of Sr was in part inherited. Considering the slight depletion in Sr within the replacive K-feldspar, it is reasonable to assume that Sr preferentially entered the fluid phase, facilitating its transport and potentially leading to smaller-scale equilibration in veins. Hence, the overlapping Rb-Sr isochron dates of 1468 ± 65 and 1429 ± 42 Ma obtained from different sets of veins can be interpreted as their crystallisation age. However, the fit obtained for the 5 analyses with the highest ⁸⁷Rb/⁸⁶Sr leaves the possibility that their age is younger. The development of the late highly porous veins had no discernible effect on the Rb-Sr systematics.

4.2.6. ⁴⁰Ar/³⁹Ar Results for Alkali Feldspar

Both in situ and bulk-grain ⁴⁰Ar/³⁹Ar analyses were performed on alkali feldspar from the Kalkadoon sample. The in situ ⁴⁰Ar/³⁹Ar dates were acquired from the crystal shown in Figure 9. When combined with the textural evidence outlined in Section 4.1.5, the dates can only be categorised into two groups (Figure 14a,b). The first group ranging from 1736 ± 29 to 1429 ± 21 Ma was derived from K-rich feldspar having a blue colour in CL images. The second group includes ⁴⁰Ar/³⁹Ar dates between 1781 ± 32 and 931 ± 34 Ma that were obtained from K-feldspar displaying purple CL responses. Virtually none of these analyses yielded detectable quantities of Cl-derived ³⁸Ar or atmospheric ³⁶Ar. The bulk-grain ⁴⁰Ar/³⁹Ar data (Figure 14c,d) were acquired by conventional step-heating analysis of an aliquot of 0.3–0.5 mm fragments that were handpicked from crushed rock. The resulting ⁴⁰Ar/³⁹Ar age spectrum starts with a very old date of ~3500 Ma, which then drops to ~800 Ma in the following couple of steps and after gradually increases to ~1300 Ma to form a relatively flat region accounting for the final ~70% of ³⁹Ar release. Although this flat region does not satisfy the classical criteria for a plateau [125] because of the excess scatter, its less scattered portion can be used to obtain a non-rigorous ⁴⁰Ar/³⁹Ar plateau date of 1303 ± 18 Ma, which we will use as a reference value. Again, virtually none of the steps defining this date yielded detectable quantities of Cl-derived ³⁸Ar or atmospheric ³⁶Ar.

Obtaining such old ⁴⁰Ar/³⁹Ar dates was surprising. None of the previously published alkali feldspar ⁴⁰Ar/³⁹Ar dates for the Kalkadoon batholith and wider Mt Isa Inlier (1259–767 Ma) [25,26] are as old as the ⁴⁰Ar/³⁹Ar plateau date obtained here, let alone the majority of the in situ ⁴⁰Ar/³⁹Ar dates. Many in situ ⁴⁰Ar/³⁹Ar dates even exceed the age of the Isan Orogeny, during which the analysed material is believed to have been subjected to amphibolite-facies conditions (see Section 2) with temperatures far above those needed for diffusive removal of ⁴⁰Ar from alkali feldspar according to laboratory outgassing data [1,2,126]. Perplexingly, most of the in situ ⁴⁰Ar/³⁹Ar dates are higher than the ⁴⁰Ar/³⁹Ar plateau date that we obtained, and the total gas released by laser ablation has an ⁴⁰Ar/³⁹Ar date that is a ~14% older than the total fusion gas (1445 ± 7 vs. 1271 ± 15 Ma). Finally, the relationships between the ⁴⁰Ar/³⁹Ar and Rb-Sr results are also complicated. The oldest and youngest in situ ⁴⁰Ar/³⁹Ar dates were obtained from veins of K-rich feldspar with purple CL, and these are far older and far younger than the Rb-Sr isochron dates that were derived for the same veins. At the same time, the latter coincide with the youngest in situ ⁴⁰Ar/³⁹Ar dates from K-feldspar with blue CL, which also yielded in situ ⁴⁰Ar/³⁹Ar dates that are older by up to ~300 Ma.

First, we will examine the discrepancy between our in situ and bulk grain ⁴⁰Ar/³⁹Ar data. We cannot rule out that the only phenocryst of alkali feldspar that we found in hand specimen is different from the handpicked alkali feldspar fragments, although the probability of this being the case is low. The phenocryst and the groundmass crystals have very similar appearances in CL images (cf. Figure 2b and Figure 9). Therefore, it is more tenable that the discrepancy results from differences in sample preparation. The phenocryst did not suffer direct hits during its extraction from the rock and subsequent cutting and polishing. In contrast, step-heated material was repeatedly put through a jaw crusher to obtain fragments that can pass through 0.5 mm sieve holes. It has been known since the early days of the K-Ar method that crushing can cause some Ar to be released from feldspars [127,128,129,130,131]. The most easily removable fractions usually have elevated ⁴⁰Ar/³⁹Ar ratios due to the release of ⁴⁰Ar either from fluid inclusions, in which case it correlates with the release of Cl-derived ³⁸Ar [131], or from some kind of crystal structure defects that act as traps, in which case no such correlation is observed [129,130]. Subsequent gas fractions usually have ⁴⁰Ar/³⁹Ar ratios that resemble those of the crystal structure itself [129,131]. Losses of up to 85% of ⁴⁰Ar have been observed upon crushing of larger fragments to 10–4 μm sized powders [127,129]. Therefore, we speculate that the ~14% mismatch between the ⁴⁰Ar/³⁹Ar dates was caused by sample preparation.

Regarding the relationships between the ⁴⁰Ar/³⁹Ar and Rb-Sr results, the observation that different ranges of in situ ⁴⁰Ar/³⁹Ar dates were obtained for the metamorphosed altered magmatic feldspar and for the regions affected by the early post-metamorphic replacement clearly indicates that the latter played an important role in ⁴⁰Ar redistribution. However, the way it did so is not consistent with the widespread view that fluid-induced dissolution-reprecipitation of alkali feldspar completely removes ⁴⁰Ar from affected zones [7,8,9,10,11], which is supported by the inferred Ar partition coefficient between this mineral and fluid of 10⁻⁵–10⁻⁶ [132]. Rather, replacive K-feldspar yielded the oldest in situ ⁴⁰Ar/³⁹Ar date, suggesting that its formation was accompanied by the incorporation of excess ⁴⁰Ar. This process can explain why some in situ ⁴⁰Ar/³⁹Ar dates of replacive K-feldspar exceed its Rb-Sr isochron dates. However, could the younger in situ ⁴⁰Ar/³⁹Ar dates represent the age of the early post-metamorphic replacement, or do they result from ⁴⁰Ar loss? The striking coincidence between the youngest in situ ⁴⁰Ar/³⁹Ar dates from K-feldspar with blue CL and the Rb-Sr isochron dates for replacive veins (cf. Figure 13b and Figure 14b) would be hard to explain if the dates do not record the time of replacement. Furthermore, even the youngest regression fitted through our Rb-Sr data translates to a much older date (1343 ± 66 Ma) than the youngest in situ ⁴⁰Ar/³⁹Ar date (931 ± 34 Ma). Therefore, we suggest that some ⁴⁰Ar has been lost from K-rich feldspar with purple CL. A different set of processes must have occurred in the alkali feldspar regions that display a blue CL colour. These have evaded replacement at ~1450 Ma but probably experienced diffusive loss of ⁴⁰Ar while it was happening, which was complete in some places and not in others.

The effects of fluid-induced dissolution-reprecipitation on the ⁴⁰Ar/³⁹Ar systematics that we infer for alkali feldspar from the Kalkadoon sample are reminiscent of those proposed for mica from metamorphic rocks. Specifically, some studies argue that such mica should only lose ⁴⁰Ar when the amount of fluid passing through its host rock is sufficient to flush out ⁴⁰Ar from the intergranular space, while a shortage of fluid should lead to the accumulation of ⁴⁰Ar in the intergranular space and either prevent mica from losing ⁴⁰Ar or even cause it to incorporate more ⁴⁰Ar [132,133,134,135]. There is, however, an important difference with our case. The cited studies suggest that mica maintains local equilibrium with the fluid in the intergranular space, so that any particular region in mica can either lose or acquire ⁴⁰Ar, but not both at the same time. This is at odds with our results, where extremely different in situ ⁴⁰Ar/³⁹Ar dates were calculated for adjacent analyses both within a single replacive vein and across a vein boundary (Figure 9a). The latter implies that the incorporation of ⁴⁰Ar into replacive veins was not controlled by its solubility in alkali feldspar and could occur simultaneously with the loss of ⁴⁰Ar from the regions that did not undergo dissolution-reprecipitation.

How could the loss and incorporation of ⁴⁰Ar be simultaneous? Why did some regions in the metamorphosed altered magmatic feldspar lose less ⁴⁰Ar than others? And what controlled the loss of ⁴⁰Ar from the veins and zones produced by the early post-metamorphic replacement? We speculate that the answer to all three questions lies in our explanation for the discrepancy between the in situ and bulk-grain ⁴⁰Ar/³⁹Ar data. The crushing-induced release of ⁴⁰Ar that we inferred for our sample, and that was experimentally demonstrated elsewhere [127,128,129,130,131], indicates that a portion of ⁴⁰Ar found in alkali feldspar can reside outside the crystal structure. It can be located in fluid inclusions [129,131], incoherent grain boundaries [127] or some crystal structure defects acting as traps [129,130]. Importantly, this kind of easily removable ⁴⁰Ar may be present without concomitant Cl-derived ³⁸Ar [130]. As to its origin, two possibilities have been suggested, including diffusive loss of ⁴⁰Ar from the surrounding alkali feldspar [127,129,130] and the incorporation as excess ⁴⁰Ar during (re)crystallisation [129,131]. Considering all this, we argue that the veins and zones that formed in our sample during the early post-metamorphic replacement incorporated excess ⁴⁰Ar within crystal structure defects or inclusions, so that their ⁴⁰Ar/³⁹Ar dates vary as a function of the abundance of stochastically formed defects or inclusions. At the same time, the content of ⁴⁰Ar in the fluid was low enough to cause diffusive outgassing of the earlier generations of K-feldspar, which in places was hampered by trapping of ⁴⁰Ar into crystal structure defects and gran boundaries. We further argue that ⁴⁰Ar diffused towards crystal structure defects, inclusions and grain boundaries across all textural types of alkali feldspar after this event and was partially removed by sample preparation. It is possible that the formation of the late highly porous veins has likewise caused the incorporation of excess ⁴⁰Ar into features that were partially opened by sample preparation. The difference in apparent loss of ⁴⁰Ar between K-feldspars with blue and purple CL responses could result from systematic variations in the type and properties of ⁴⁰Ar traps.

To summarise, we propose that the variation in the in situ ⁴⁰Ar/³⁹Ar dates is a result of ⁴⁰Ar redistribution during and following early post-metamorphic replacement at ~1450 Ma. It is unclear if any ⁴⁰Ar was lost from alkali feldspar prior to this event, and it is possible that none was lost, considering the examples of mica from metamorphic terranes [132,133,134,135]. During the event, ⁴⁰Ar diffusively escaped regions that did not react with fluid to become first caught by the fluid and then incorporated as excess ⁴⁰Ar into regions that underwent dissolution-reprecipitation. As fluid interaction ceased, ⁴⁰Ar probably continued to migrate by diffusion in all generations of alkali feldspar. Any diffusive migration of ⁴⁰Ar was associated with its partial trapping into features such as crystal structure defects, where it remained until being partially released during sample preparation. Some of the ⁴⁰Ar lost in this way may have been redistributed by fluids as they formed the late highly porous veins, although direct the available evidence is insufficient to verify this. The discrepancies between our alkali feldspar ⁴⁰Ar/³⁹Ar data and those published before may result from a variety of factors ranging from the differences in geologic histories of specific rocks (e.g., a greater fluid/rock ratio during alteration) to the variations in sample preparation protocol (e.g., using more cycles of crushing).

4.2.7. Relationships Between Isotopic Disturbances in Different Minerals and Systems

A comparison of all our geochronological and isotopic data is provided in

Table 2. Summary of geochronological and isotopic results obtained here along with previously reported dates that have been used for thermochronological reconstructions for the Mt. Isa Inlier.

Sample	Mineral	System	Type of Data	Result
Williams	zircon	U-Pb	upper intercept date	1508 ± 8 Ma
	zircon	U-Pb	lower intercept date	320 ± 9 Ma
	zircon	U-Pb	estimated 207Pb/206Pb in released Pb	0.104
	K-feldspar	Pb isotopes	207Pb/206Pb, % lower than S&K75 *	0.742 ± 0.001, 22.2
	apatite	U-Pb	207Pb-corrected dates using S&K75 *	1664 ± 37 to 1347 ± 86 Ma
	fluorite	U-Pb	location in the Tera-Wasserburg space	below the concordia
Kalkadoon	zircon	U-Pb	upper intercept date	1832 ± 8 Ma
	zircon	U-Pb	lower intercept date	315 ± 7 Ma
	zircon	U-Pb	estimated 207Pb/206Pb in released Pb	0.123
	K-feldspar	Pb isotopes	207Pb/206Pb, % lower than S&K75 *	0.9607 ± 0.0001, 2.5
	apatite	U-Pb	207Pb-corrected dates using S&K75 *	1965 ± 69 to 1833 ± 42 Ma
	K-feldspar	Rb-Sr	isochron date, group with blue CL	1841 ± 148 Ma
	K-feldspar	Rb-Sr	isochron date, 1st group with purple CL	1468 ± 65 Ma
	K-feldspar	Rb-Sr	isochron date, 2nd group with purple CL	1429 ± 42 Ma
	K-feldspar	40Ar/39Ar	in situ dates for regions with blue CL	1736 ± 29 to 1429 ± 21 Ma
	K-feldspar	40Ar/39Ar	in situ dates for regions with purple CL	1781 ± 32 and 931 ± 34 Ma
	K-feldspar	40Ar/39Ar	step-heating plateau date	1303 ± 18 Ma,
Sybella S	zircon	U-Pb	upper intercept date	1658 ± 7 Ma
	zircon	U-Pb	lower intercept date	315 ± 12 Ma
	zircon	U-Pb	estimated 207Pb/206Pb in released Pb	0.112
	K-feldspar	Pb isotopes	207Pb/206Pb, % lower than S&K75 *	0.95267 ± 0.00005, 1.6
	apatite	U-Pb	207Pb-corrected dates using S&K75 *	2376 ± 67 to 1397 ± 23 Ma
	fluorite	U-Pb	location in the Tera-Wasserburg space	in line with apatite results
Sybella N	zircon	U-Pb	upper intercept date	1667 ± 15 Ma
	zircon	U-Pb	lower intercept date	−117 ± 590 Ma
	zircon	U-Pb	estimated 207Pb/206Pb in released Pb	0.099
	K-feldspar	Pb isotopes	207Pb/206Pb, % lower than S&K75 *	0.96662 ± 0.00004, 0.2
	apatite	U-Pb	207Pb-corrected dates using S&K75 *	1619 ± 88 to 1433 ± 24 Ma
Selected data from the literature	amphibole	40Ar/39Ar	date range from [25,26,27]	1843–1398 Ma
	biotite	40Ar/39Ar	date range from [25,26,27]	1490–1363 Ma
	K-feldspar	40Ar/39Ar	date range from [25,26,27]	1534–1122 Ma
	apatite	fission track	date range from [25,26,28,29]	390–225 Ma

* The model of Stacey and Kramers [40].

. The Pb and U-Pb isotopic analyses of alkali feldspar, fluorite and apatite indicate that all three minerals often contain a non-radiogenic Pb component with a very low ²⁰⁷Pb/²⁰⁶Pb ratio. According to our petrological evidence, the occurrence of this Pb component in apatite is linked to dissolution-reprecipitation in fluids, which also precipitated fluorite in zones of plagioclase replacement by albite with muscovite and/or epidote. The coeval process in alkali feldspar was the formation of the late highly porous veins, and it was probably this process that introduced this Pb component into this mineral. Although it was not clear to us from petrological evidence alone if the recrystallisation of zircon was coeval, the fact that this process caused the release of Pb with a very low ²⁰⁷Pb/²⁰⁶Pb suggests that it occurred at the same time, dating the alteration event at ~300 Ma.

Was the fluid-mobile Pb component with a very low ²⁰⁷Pb/²⁰⁶Pb ratio sourced locally or distally derived? A simple mass balance calculation for alkali feldspar, which makes up ~50 vol. % of the studied rocks, suggests that the difference between its measured Pb isotopic composition and that predicted from the model of Stacey and Kramers [40] can be accounted for by introducing a fairly small quantity of ²⁰⁶Pb and ²⁰⁷Pb. More specifically, given the estimated Pb contents, 1 t of alkali feldspar from the Williams sample would need to acquire as little as ~0.05 g of ^206,207Pb, while ~0.11, 0.17 and 0.21 g of ^206,207Pb are needed for the Sybella N, Sybella S and Kalkadoon samples. The present-day contents of U that are needed to produce this amount of ^206,207Pb between the established crystallisation ages and 300 Ma are 0.23, 0.40, 0.62 and 0.68 ppm for the Williams, Sybella N, Sybella S and Kalkadoon samples, respectively. The latter are well below the U contents of ≳2 ppm that were reported for the rocks of the Willams, Sybella and Kalkadoon Batholiths [19,21], suggesting that the Pb component in question could be sourced locally.

Could the same Pb component be entirely derived from zircon? This can be estimated by converting the above figures for the minimum bulk rock contents of U needed to produce the observed effects to the equivalent bulk rock contents of Zr with the assumption that all U is stored in zircon. Taking 100, 10, 500 and 500 ppm as U contents in zircon, we obtain the equivalent bulk rock Zr contents of 1143, 19931, 613 and 675 ppm for the Williams, Sybella N, Sybella S and Kalkadoon samples, respectively. These values are above the bulk rock Zr contents that were previously reported for the Willams, Sybella and Kalkadoon Batholiths [19,21], implying that zircon from a local volume of rock was not the only source of the Pb component in question, and some of it was derived from other U-bearing phases. One caveat to this inference is that our strategy for zircon separation and analysis might have favoured low-U crystals, and taking higher U contents for zircon can shift the balance in favour of zircon in the Williams, Sybella S and Kalkadoon samples. Thus, previous bulk-grain U-Pb analyses of zircon from the Williams batholith suggest U contents of ~500 ppm [21], and with it the above estimate for bulk rock Zr content decreases to 229 ppm, which is near the reported value [21].

Notably, while the isotope composition of Pb in alkali feldspar appears to have been modified by the development of the late highly porous veins, the same process had no unambiguous effect on the Rb-Sr and ⁴⁰Ar/³⁹Ar data obtained from this mineral. Such data were only obtained from the Kalkadoon sample. The Rb-Sr system in this sample appears to have only been disturbed once during the early post-metamorphic replacement at ~1450 Ma, which caused more or less simple resetting of the Rb-Sr isochron date in the reprecipitated regions. The ⁴⁰Ar/³⁹Ar system, however, was disturbed in a more complex way both during and subsequent to the same event. During the event, regions that were not directly reprecipitated have partially lost ⁴⁰Ar by diffusion in the presence of ⁴⁰Ar traps, while reprecipitated regions lost ⁴⁰Ar as they dissolved and then incorporated ⁴⁰Ar as they crystallised. Subsequent disturbance was relatively minor and is inferred to be caused by diffusion and trapping of ⁴⁰Ar with potential contribution from fluid-assisted ⁴⁰Ar redistribution during the formation of the late highly porous veins.

5. Discussion

5.1. Fluid-Assisted Incorporation of Excess ⁴⁰Ar and ^206,207Pb

It is customary among geochronologists to consider each combination of a mineral and an isotopic system in isolation. Thus, interpretations of ⁴⁰Ar/³⁹Ar analyses of alkali feldspar and U-Pb analyses of apatite have been typically made under the assumption that these minerals invariably lose radiogenic isotopes that are produced in respective decay schemes, whether it be by diffusion [1,2,3,4,5,6] or fluid-induced recrystallisation [7,8,9,10,11,12,13,14,15]. Any incomplete loss would be attributed to insufficient time spent at temperatures high enough for diffusion [1,2,3,4,5,6] or to the presence of domains that did not go through dissolution-reprecipitation and thus preserved their initial isotopic signature [7,8,9,10,11,12,13,14,15], while late incorporation of radiogenic isotopes as an excess component would generally not be considered. However, as we demonstrate here, one mineral’s loss may well be another mineral’s gain! That is, radiogenic isotopes that have been released from one mineral or domain in a mineral grain can be introduced into another mineral or domain in the same mineral grain. This concept has already been entertained, for example, in the context of ⁴⁰Ar/³⁹Ar dating micas from high-grade metamorphic terranes [132,133,134,135], and here we show that it also applies to apatite, alkali feldspar and systems that resided at much lower temperatures and pressures. Apatite fission track data from [28,29] suggest cooling to <110 °C at least as late as during the event at ~300 Ma, but as we will discuss below potentially even earlier.

There is a sort of symmetry in how fluid-induced dissolution-reprecipitation disturbs the U-Pb and ⁴⁰Ar/³⁹Ar systems, so we generalise the effects of this process using conceptually similar Tera-Wasserburg and inverse isochron plots in Figure 15. We envision that the dissolution stage causes Ar and Pb that were stored in the consumed material to be completely released into to the fluid, where they then mix with the Ar and Pb that were already there. During the reprecipitation stage, the resultant mixture becomes partially incorporated into the newly grown material, whether as predicted by a partition coefficient or in a more erratic way in inclusions and, echoing previous suggestions of Scaillet [133,134] regarding Ar in metamorphic micas, some kind of atomic-scale defects. Both the chemical and isotopic composition of this mixture will be spatially variable if fluid propagation through the rock is such that no immediate homogenisation of the abundance and isotopic composition of Ar and Pb occurs so as to eliminate any local changes that are imparted by the act of dissolution. In such a case, each portion of the reprecipitated material will have its own initial isotopic composition, without the knowledge of which no correct date can be calculated for the timing of alteration. With the ⁴⁰Ar/³⁹Ar system, a further difficulty arises from the fact that ⁴⁰Ar will be overwhelmingly more abundant than other Ar isotopes, and the ³⁶Ar/⁴⁰Ar ratios in Figure 15b are greatly exaggerated. The entire process can also be complicated by changes in the contents of the parent isotopes, a potential example of which has been previously reported for the U-Pb system in titanite [136], and the U-Pb system can be further complicated by disequilibrium in the U-series decay scheme [137], but we will leave these aspects aside for the rest of our discussion.

Previous work commonly assumed that isotopic and chemical transport in rocks only proceeds via grain boundaries, regardless of whether it is fluid-mediated or diffusive [132,133,134,135,138,139,140]. However, as shown by our petrological evidence (Section 4.1), rock-altering fluids do not exclusively flow along the networks of grain boundaries that were formed during magmatic crystallisation or peak metamorphism. Very often, they propagate in arbitrary directions, leaving behind numerous veins that clearly cross-cut the original grain boundaries as well as earlier veins. The latter is much more consistent with the suggestion that a “fluid can literally react its way through a rock” made by Putnis in his go-to review of dissolution-reprecipitation processes [141]. By analogy with the experimentally demonstrated intracrystalline migration of melt inclusions [142], it can probably be imagined, at least for a limited range of cases, that fluids percolate through rocks as isolated droplets. Such a mechanism could explain, for example, why flame lamellae in the alkali feldspar from the Kalkadoon sample occasionally presented barriers for the propagation of the early post-metamorphic veins of K-rich feldspar.

Drawing inspiration from the above considerations, we have explored the potential behaviour of ⁴⁰Ar, ^206,207Pb, Th and U during mineral replacement by simplistic modelling of migration of a fluid droplet through a crystal. Modelling was performed using MATLAB/OCTAVE scripts that are provided in the Supplementary Archive. In brief, these scripts simulate stepwise movement of a unidimensional droplet along a straight line by dissolving a fixed amount of solid material in front of it and immediately precipitating the same amount of solid material behind it, such that the mass balance, fluid/mineral ratio and fluid/mineral partition coefficient remain constant at each step (Figure 16a). For simplicity, the scripts ignore any differences in density as well as potential changes in density, which can be justified by the qualitative, order-of-magnitude nature of our inferences and the lack of any quantitative constraints for most of the other parameters that are needed for more accurate calculations.

The first set of simulations explores whether and how dissolution-reprecipitation by a moving droplet of fluid can cause alkali feldspar to incorporate excess ⁴⁰Ar. Kelley [132] used data describing Ar solubility in alkali feldspar and water to infer that the partition coefficient between the two can be as low as ~7 × 10⁻⁶ and hence suggested that only trivial quantities of excess ⁴⁰Ar can be present in alkali feldspar that interacted with fluids. Taking the same partition coefficient and assuming that the starting fluid had no ⁴⁰Ar, we ran two different simulations for alkali feldspar with 1 ppm of ⁴⁰Ar, which is about the value that alkali feldspar from the Kalkadoon sample should have had during alteration at ~1450 Ma. In the first simulation, a droplet of fluid travelled 1000 μm and completely removed ⁴⁰Ar from the reprecipitated alkali feldspar (Figure 16b). However, this led the fluid to acquire an extremely high ⁴⁰Ar content of 1000 ppm (Figure 16c), which exceeds both the empirical (fluid inclusions) and experimental (at ≲300 °C) upper limits of its solubility cited by Kelley [132] by an order of magnitude. Furthermore, this value would only continue to grow with additional propagation of the fluid droplet. Considering that the experimental solubility data that Kelley [132] used to infer the partitioning behaviour of Ar were obtained in the presence of pure Ar phase [143,144], it seems sensible to assume that the fluid droplet becomes saturated with ⁴⁰Ar, exsolving it as a bubble. Therefore, the content of ⁴⁰Ar in the fluid droplet was capped at 100 ppm in our second simulation, while any excess ⁴⁰Ar was assumed to form a bubble that was immediately trapped by the mineral. The results show that after displacing 1000 μm, the fluid droplet stripped alkali feldspar of all of its ⁴⁰Ar, and the lost ⁴⁰Ar started to be trapped in situ after the fluid droplet travelled 100 μm and reached saturation (Figure 16b,c).

Although there is currently no clear evidence to validate the second simulation, we argue that it is consistent with the alkali feldspar ⁴⁰Ar/³⁹Ar data presented here. The proposed mechanism alleviates the necessity to assume enormous and erratic variations of ⁴⁰Ar content in the fluid over μm-scale distances by instead attributing the inhomogeneous entrapment of excess ⁴⁰Ar to the variations in the quantities of captured Ar bubbles. Furthermore, this mechanism is compatible with the simultaneous diffusive loss of ⁴⁰Ar from regions that do not undergo recrystallisation. Finally, it explains how excess ⁴⁰Ar can be trapped without concomitant Cl, which is probably more likely to partition into the liquid phase. Developing the latter point, we suggest that it is possible to test the proposed mechanism and potentially even quantify the amount of excess ⁴⁰Ar by correlating its release with the release of, say, other noble gases, which would probably partition into the Ar bubble more readily than Cl.

One critical point relating to the modelled scenario with Ar bubbles is that the fluid/alkali feldspar ratio was only 1/1000. Indeed, increasing this number by a factor of 10 by letting 9 more droplets of fluid to travel along the same path would remove all the Ar bubbles that were generated in the modelled segment after the passage of the first one. However, this would only occur because of the assumption that the fluid droplets are initially free of ⁴⁰Ar, and it is unclear what could be the source of an ⁴⁰Ar-free fluid deep in the Earth’s crust. Thus, even the final (10th) fluid droplet in this scenario would become saturated with ⁴⁰Ar after travelling 1000 μm. In fact, from the perspective of our modelling, any loss of ⁴⁰Ar can be viewed with a surprise, as it would require percolation of enormous amounts of fluids with considerable transport of ⁴⁰Ar in a form of bubbles. At the same time, the lack of a discernible effect of the late highly porous veins in alkali feldspar from the Kalkadoon sample on its in situ ⁴⁰Ar/³⁹Ar dates can be considered unsurprising. These inferences are at odds with those made by Kelley [132] presumably because his calculations did not allow for the saturation of fluids with Ar.

The following simulation explores the behaviour of ^206,207Pb in a mineral that can be viewed as either alkali feldspar or apatite. We assume that Pb is mildly compatible and preferentially distributes into the mineral with a partition coefficient of 2, which is consistent, for example, with the lack of strong enrichment or depletion of Pb in replacive apatite (see Section 4.1.4 and the associated figures in the Supplementary Archive). The modelled mineral segment of 100 μm length has an inhomogeneous initial distribution of Pb but a constant ²⁰⁷Pb/²⁰⁶Pb ratio of 0.9. It was traversed by 67 droplets of fluid, each initially having 1 ppm of ²⁰⁶Pb and a ²⁰⁷Pb/²⁰⁶Pb ratio of 0.1. The initial and final ²⁰⁶Pb content and ²⁰⁷Pb/²⁰⁶Pb ratio profiles are presented in Figure 16d,e. They show that despite the 67 cycles of dissolution-reprecipitation of the entire segment, the isotopic and chemical signature of the fluid droplets has only been acquired by the first 30 μm. This result has two interesting consequences. First, it suggests that the lack of compositional and isotopic change in a particular domain of a mineral undergoing recrystallisation does not necessarily imply the lack of replacement. Second, it demonstrates the difficulty of distant transport of a Pb component having a ²⁰⁷Pb/²⁰⁶Pb ratio that is very different from that of the common Pb. Thus, a droplet of fluid that leaves a crystal of zircon, where it has acquired a high content of Pb with a very low ²⁰⁷Pb/²⁰⁶Pb ratio, should rapidly lose this chemical and isotopic signature as it travels through minerals tolerant to Pb in their structure such as apatite and alkali feldspar. We should note that the transportation distance does increase if the partition coefficient of Pb is reduced to 0.1, also a reasonable value given our observations for apatite, however it still remains short relative to geological length scales or even the scales of hand specimens (~600 μm for otherwise the same inputs).

Our final simulation models U and Th transport in apatite. As discussed in Section 4.1.4, alteration of some apatite grains caused a decrease of their Th/U ratios without considerable perturbations to the oscillatory zoning visible in CL images and U and Th maps (e.g., Figure 5i,j and Figure 6b). It is not immediately clear how this might have happened, considering that any sharp peaks in element contents should be shifted and eroded in the simplest case of fluid-induced dissolution-reprecipitation in a unidirectionally flowing fluid (e.g., Figure 16d). For this reason, we attempted to find a more complex solution that would reproduce the observation. Without claiming that the outcome here is representative of the natural process recorded by our samples, we show that 67 droplets of fluid with 2 ppm of U and a Th/U ratio of 0.5 can alter the initial U and Th/U profiles shown in Figure 16f,g as depicted in the same figures with the assumption that the partition coefficients of U and Th are linearly dependent on the content of a third element, which remains completely immobile. This example serves to demonstrate that certain alteration patterns that contradict intuitions formed from simple models may be caused by a complex chemical interaction in the fluid responsible for recrystallisation.

In conclusion to our simulations, we would like to highlight that they did not consider advective transport. Clearly, the latter can significantly increase the distance over which a fluid can travel while retaining an exotic chemical and isotopic signature, and even more so if reactions with the percolated rock are restricted to some particularly susceptible domains such as damaged zircon. However, such large-scale advection of a selectively reacting fluid seems improbable in a (meta)igneous rock with virtually no open porosity. For example, our samples contain very few open cracks, which at best formed at supergene conditions, but possibly also during their collection, shipment and preparation. This contrasts with numerous seemingly isolated micropores in zones of plagioclase and alkali feldspar alteration and myriads of tiny fluid inclusions in quartz. Thus, fluid migration by reacting through minerals as opposed to merely flowing in variably narrow open spaces provided by cracks and grain boundaries appears to play an important role.

5.2. Imlications for ⁴⁰Ar/³⁹Ar Dating of Alkali Feldspar

Non-plateau ⁴⁰Ar/³⁹Ar age spectra of alkali feldspar such as the one in Figure 14d have been predominantly interpreted in the framework of thermochronology, which in the past 35 years relied on the use of the multi-diffusion domain theory of Lovera et al. [1]. This includes previous studies of the Mt. Isa Inlier led by one of the co-authors here, which employed the latter theory to recover continuous time-temperature paths from ⁴⁰Ar/³⁹Ar analyses of alkali feldspar that are compatible with other thermochronological constraints [25,26]. The key assumption behind ⁴⁰Ar/³⁹Ar thermochronology is that ⁴⁰Ar is mostly redistributed by diffusion, while the multi-diffusion domain theory in particular also assumes that laboratory release of Ar isotopes for analysis mimics the process of ⁴⁰Ar loss that occurred in the geological past [1]. Although diffusion of ⁴⁰Ar may have caused its partial loss in the alkali feldspar from our Kalkadoon sample, much of it has been redistributed via fluid-induced dissolution-reprecipitation. No one-to-one correlations can be made between the in situ ⁴⁰Ar/³⁹Ar dates and the ⁴⁰Ar/³⁹Ar age spectrum that we obtained, suggesting that the latter represents a very complex mixture of Ar isotopes derived from a variety of sources ranging from feldspar structures of different age and composition to various fluid inclusions, bubbles and/or defects holding ⁴⁰Ar that was lost into them diffusively and/or incorporated during recrystallisation. Therefore, the release of Ar isotopes from this alkali feldspar by step-heating cannot be taken as mimicking natural ⁴⁰Ar loss, and the multi-diffusion domain theory cannot be used to model its ⁴⁰Ar/³⁹Ar age spectrum, which in the absence of all the supplemental evidence presented here erroneously appears to be suitable for such modelling. We see no good reason why the examined alkali feldspar should be a rare exception. The apparent agreement of the thermal history reconstructions that were previously derived for the Mt. Isa Inlier from step-heating ⁴⁰Ar/³⁹Ar analyses of alkali feldspar [25,26] with other constraints may be fortuitous [145] or represent a more systematic “mathematical mirage of microtexture” [146].

Even though we see no prospect of applying the multi-diffusion domain theory for thermochronology, there might be room for an alternative approach. It appears that the alkali feldspar that we analysed here has partially lost ⁴⁰Ar by diffusion towards the replacive veins in regions that evaded alteration and towards some traps that were opened during sample preparation, whether these be fluid inclusions, bubbles or defects. The latter mechanism has already been proposed by some of us for alkali feldspar from elsewhere [9], we have since found earlier similar suggestions by Gentner and Kley [127] and Pushkarev et al. [129]. In theory, it should be possible to recover thermochronological information by quantifying the occurrence of traps, their physical properties, the amount of ⁴⁰Ar in them and the apparent age of the surrounding alkali feldspar structure. Only time can tell whether this is possible in practice.

In general, our present results agree with the suggestions that the redistribution of ⁴⁰Ar in alkali feldspar is predominantly a fluid-mediated process [7,8,9,10,11]. However, they challenge the ensuing idea that ⁴⁰Ar/³⁹Ar analysis of this mineral can be used to date fluid flow: one of the fluid interaction events that affected our Kalkadoon sample was associated with trapping of enormous amounts of excess ⁴⁰Ar into replacive K-feldspar, while another did not leave any unequivocal imprint on the ⁴⁰Ar/³⁹Ar systematics, which can also be seen as a result of complete reincorporation of the released ⁴⁰Ar as excess ⁴⁰Ar. Thus, extracting the age of fluid interaction requires quantification not only of ⁴⁰Ar inheritance due to the presence of unaltered regions, but also of excess ⁴⁰Ar. As our data demonstrate, the latter may be very difficult to do because the presence of excess ⁴⁰Ar may not correlate with the presence of any other Ar isotope, whether natural or formed during neutron irradiation from Cl (and Ca). Our simplistic modelling suggests that the lack of Cl-derived ³⁸Ar may be explained by the incorporation of excess ⁴⁰Ar not in a form of fluid inclusions but as bubbles of Ar that exsolved from the recrystallisation-causing fluid upon saturation. While Cl is not expected to enter such bubbles in significant quantities, they might preferentially uptake other noble gases, as well as N and C-based compounds. Most noble gases formed in the Earth’s crust by radioactive decay, fission and nuclear reactions are produced at much lower rates than ⁴⁰Ar (at around the same rate as ^36,38Ar), however ⁴He has a similar production rate and thus may be present in comparable quantities [147]. Perhaps a fluid that selectively interacted with zircon can become sufficiently enriched in heavy noble gases. N₂, CO₂ and CH₄ are sometimes detected in gas bubbles inside fluid inclusions [148], suggesting that they may occur together with ⁴⁰Ar in relatively high quantities. Of course, even if some element or compound gets trapped alongside excess ⁴⁰Ar, their ratio may be spatially variable.

5.3. Imlications for Noble Gas Geochronology of Groundwater

Isotope analysis of Ar and other noble gases has been used to constrain the residence times for deep-seated groundwater, and some estimates reported from Precambrian crustal blocks amount to astonishingly high values of 10³ Ma [149,150]. This method assumes that the radiogenic, nucleogenic and fissiogenic noble gases that are produced throughout the rock migrate into the pore-filling fluid either diffusively or immediately, with the latter option reportedly providing a minimum age [149,150,151]. Although some related literature does discuss the possibility of reactive release of noble gases [147,152], it is generally not accounted for in the calculations [149,150,151]. Furthermore, the similarity of apparent residence times obtained using different isotopes (e.g., ⁴He, ⁴⁰Ar, ⁸⁶Kr) under the assumption of immediate release is considered to validate their accuracy [149,150,151]. However, as pointed out by Tolstikhin et al. [152], different noble gases have vastly different diffusion parameters, and the rates of diffusion at any given temperature are inversely related with the atomic number [153,154]. Hence, different residence times should be obtained from different isotopes with the assumption of immediate release if the release of noble gases was actually controlled by diffusion, with He-related estimates being the oldest. Similar residence time estimates, on the other hand, directly indicate non-diffusive release. This resonates with the results of our work, which show that the redistribution of at least ⁴⁰Ar in alkali felspar is largely a reactive rather than diffusive process requiring a significant influx of fluid to be efficient. Furthermore, the latest omnipresent recrystallisation event that we documented in our samples occurred at temperatures of ~100 °C or less, which is close to the temperatures of ~50 °C reported for the apparently old deeply seated groundwaters [149,150,151]. This implies that isotopic enrichment of the latter and the isotopic mobility documented here may be more or less the same phenomenon viewed from different sides. If indeed ⁴⁰Ar and other noble gases are released into groundwaters through dissolution-reprecipitation, their concentrations can be acquired literally immediately and therefore cannot be used for residence time estimation. Considering this, it is concerning that such groundwater residence time estimates feed into the evaluation of the safety of nuclear waste disposal projects [155,156].

5.4. Imlications for U-Pb Dating of Apatite

U-Pb dating of apatite started to be used for thermochronology relatively recently [4,5,6], although it also generally accepted that the U-Pb systematics of apatite can be disturbed by fluid-induced dissolution-reprecipitation [12,13,14,15]. However, this processes is typically assumed to simply remove radiogenic Pb from affected zones, implying that U-Pb dates of apatite should lie between the crystallisation and alteration ages depending on the extent of Pb inheritance due to the presence of unaffected zones [12,13,14,15]. Our present work shows that such an assumption is not always justified. In extreme cases, fluids that cause apatite recrystallisation can carry radiogenically enriched Pb that was leached from another high U (and Th) mineral such as zircon, thereby increasing the U-Pb dates. The same effect can be achieved by preferential leaching of U (and Th) from apatite. Furthermore, our modelling attempts highlight that fluid-induced dissolution-reprecipitation can perturb intragrain distributions of different isotopes without significant changes to bulk-grain compositions, suggesting that noticing and quantifying its effects may not be as simple as usually presumed [12,13,14,15]. Insufficient mixing in the fluid can cause replacive apatite to have initial Pb with spatially variable isotopic composition, making it virtually impossible to calculate the age of alteration. The latter possibility can be tested with the use of linear fits through the arrays of U-Pb data in Tera-Wasserburg space.

Overall, our results urge extreme caution when using U-Pb dating of apatite for thermochronology. In our samples, alteration of apatite appears to have affected virtually every crystal that we looked at, even if it was entirely encapsulated in quartz or alkali feldspar. In some cases, alteration was cryptic and caused ²⁰⁷Pb-corrected U-Pb dates to increase and Th/U ratios to decrease with only subtle changes to the magmatic oscillatory zoning pattern revealed by CL imaging (e.g., Figure 5j). Even grains with magmatic Th/U ratios sometimes yielded ²⁰⁷Pb-corrected U-Pb dates that exceed the age of crystallisation (Figure 5b), suggesting that they may have also been affected by alteration. Considering this, one potential path forward for apatite U-Pb thermochronology is targeting apatite inclusions either in particularly resilient minerals or, perhaps even better, in relics of minerals that easily interact with fluids and thus are much easier to distinguish from altered domains. For example, it seems that the plagioclase-hosted inclusion in Figure 7i was not affected by the low-temperature event that disturbed the U-Pb systematics of others. The potential for success is further highlighted by studies that employed melt inclusions in relics of plagioclase and olivine in altered volcanic rocks to reconstruct their original chemistry [157,158,159]. Note that diffusion modelling in this case would need to explicitly account for the presence of the mineral that hosts apatite [119].

5.5. Imlications for Fission Track Dating of Apatite

Annealing of fission tracks in apatite is generally considered to be a temperature-controlled process of reconstitution of perturbed structure via diffusion of atoms and defects, so that apatite fission track data are invariably interpreted in the framework of thermochronology [160,161,162]. However, the coincidence between the fission track dates that were previously reported for apatite from the Mt. Isa Inlier [25,26,28,29] and the age of the event that mobilised Pb from zircon into apatite and other minerals across all the samples described here opens the possibility that some fission tracks are eliminated by dissolution-reprecipitation. This possibility is further highlighted by the evidence of fluid-induced disturbance of the (U-Th)/He system in apatite from elsewhere [163]. According to our modelling results, fluid-assisted recrystallisation of a particular domain inside apatite may not even manifest itself through a change of its composition. This suggests that the identification of the process that caused any given fission track to disappear is challenging. If the fission track dates were indeed reset by dissolution-reprecipitation, they would provide a minimum estimate for the timing of cooling and a maximum estimate for the temperature at any time.

5.6. A Note on the Late Highly Porous Veins

Our views on the processes that disturb the ⁴⁰Ar/³⁹Ar systematics of alkali feldspar and U-Pb systematics of apatite in our samples from the Mt Isa Inlier have evolved over time. Our earlier working hypothesis was that the introduction of radiogenically enriched Pb into alkali feldspar in the Kalkadoon sample occurred during the formation of early post-metamorphic veins [114], and it is only while looking deeper into the data to write this paper that we appreciated the importance of the late highly porous veins. Despite being acquainted with the work of Parson et al. [92] we just did not pay attention to them in all but the Williams sample. Our failure is probably understandable: Parson et al. [92] have only recognised the presence of similar late porous feldspar in alkali feldspar from the Klokken syenite after decades of work with this material. They explain this by the tendency of petrologists “to ignore such highly ‘altered’ regions or dismiss them as trivial, and even perhaps related to weathering.” However, it now becomes increasingly evident that this kind of feldspar is not a triviality, and its development can impact isotopic compositions of alkali feldspar in complex ways. Unfortunately, as we were not conscious about the importance of the late highly porous veins through most of the data collection, we did not test their impact on isotopic data systematically. We urge others not to repeat our mistake and pay attention to this kind of feldspar in their samples.

6. Conclusions

We have combined detailed petrological characterisation with a range of isotopic analyses to provide a holistic view of not only the mechanisms but also the directions of radiogenic ⁴⁰Ar and ^206,207Pb redistribution in alkali feldspar and apatite from Proterozoic (meta)granites of the Mt. Isa Inlier. All of the studied samples have been affected by at least two fluid interaction events that caused considerable recrystallisation of rock-forming and accessory minerals. The earliest event was associated with deuteric recrystallisation and naturally had no impact on the geochronological record. Effects of this process on alkali feldspar textures are obvious in the Williams sample, less obvious but recognisable in the Kalkadoon sample and are believed to have been completely obliterated by deformation in both Sybella samples. The latter three samples record a fluid interaction event that occurred towards the end or shortly after the Isan Orogeny and altered alkali feldspar, plagioclase, biotite, apatite and maybe other minerals. In the Kalkadoon sample, this event is dated at ~1450 Ma and caused both fluid-assisted and diffusive redistribution of ⁴⁰Ar. Diffusion in the presence of traps occurred in regions that remained unaltered, while regions that underwent dissolution-reprecipitation incorporated variable and sometimes very high amounts of excess ⁴⁰Ar. The effects of this event on apatite were not identified, nor was its temperature, although evidence for ⁴⁰Ar diffusion in alkali felspar suggest that it reached several hundred °C. Finally, all samples were affected by fluid interaction at ~300 Ma, which altered basically all minerals in the studied rocks and drove Pb redistribution between them, such that alkali feldspar, apatite and fluorite acquired a radiogenically-enriched Pb component from zircon and other local sources. In apatite, U-Pb systematics were probably also affected by in situ recycling of Pb and U loss. This event had no unambiguous effect on the ⁴⁰Ar/³⁹Ar systematics of alkali feldspar, suggesting that all ⁴⁰Ar released by recrystallisation was reincorporated locally as excess ⁴⁰Ar. The temperature of this event could be ~100 °C or even lower.

Overall, our results concur with previous suggestions that fluid-induced dissolution-reprecipitation can play a key role in the redistribution of ⁴⁰Ar in alkali feldspar and ^206,207Pb in apatite at temperatures that are too low for significant diffusion [7,8,9,10,11,12,13,14,15]. However, the way this happens is not always consistent with the notion of fluid-assisted loss. In some cases, ⁴⁰Ar and ^206,207Pb are recycled locally in the same or neighbouring grains, resulting in the incorporation of excess ⁴⁰Ar and ^206,207Pb. The disparate behaviour of ⁴⁰Ar and ^206,207Pb at ~300 Ma indicates that intergranular mobility of one of them does not necessarily imply that the other will be able to leave its host grain. Due to limited mixing in transport and potentially short transport distances, the initial isotopic composition of Pb in replacive apatite and Ar in replacive alkali feldspar can vary significantly from place to place. Although we found evidence for diffusive redistribution of ⁴⁰Ar in alkali feldspar, and some diffusive redistribution of ^206,207Pb in apatite can probably be assumed for the Sybella samples, the overprint of alteration precludes extraction of thermochronological information from these minerals using traditional approaches [1,2,3,4,5,6]. Finally, our results highlight that fluid-mediated recrystallisation may assist with recovery of fission tracks in apatite and aid with noble gas release into groundwaters, prompting deeper consideration of these processes.

There is a notion among geochronologists that a mineral that is stable across a wide range of pressure-temperature conditions is a good target for analysis. The idea is that since it is so stable, it would probably not get altered. The results of this work rather show that if a mineral is stable across a wide range of pressure-temperature conditions, fluid-induced dissolution-reprecipitation may affect it repeatedly, sometimes only leaving subtle traces and greatly complicating interpretation of geochronological data. Conversely, it may be a lot more straightforward to interpret an isotopic date from a well-preserved relic of a mineral that is prone to replacement and thus would presently not be analysed.