Porphyry Cu ± Au deposits predominantly occur at shallow crustal levels (i.e. <5 km below the paleosurface) in Phanerozoic arcrelated settings. In active orogens (e.g. arc-related settings), the susceptibility of shallow porphyry Cu ± Au deposits to erosion during rapid uplift has been considered as a relevant factor for the rarity of these deposits in Precambrian terranes (Groves et al., 2005;Wilkinson & Kesler, 2007). However, the thick (∼250-350 km), refractory, and buoyant subcontinental lithospheric mantle roots underlying Archean cratons make the cratons resistant to reworking by younger orogenic cycles, thus leading to the preservation of many Archean volcanoplutonic terranes (Kerrich et al., 2005). The fact that these volcanoplutonic terranes have been preserved suggests that the rarity of porphyry Cu ± Au deposits in the Archean may not reflect erosion (Richards & Kerrich, 2007), but may instead be attributed to unfavorable tectonomagmatic settings and related geochemical features (Evans & Tomkins, 2011;Richards & Mumin, 2013).
Most known porphyry Cu ± Au deposits are of Phanerozoic age and are exclusively associated with moderately oxidized ( FMQ +1 to +2; where FMQ is the log unit of f O 2 relative to the fayalite-magnetite-quartz mineral redox buffer), sulfur-rich, and hydrous magmas of calc-alkaline to mildly alkalic affinity derived from partial melting of the mantle wedge previously metasomatized by slab-derived fluids (Richards, 2011). In contrast, intrusive rocks in many Archean terranes are dominated by sodium-rich tonalite-trondhjemite-granodiorite-diorite suites (abbreviated as TTG hereafter; Fig. 1). The origin of the TTG magmas remains highly contentious, but they are generally attributed to processes including melting of thickened mafic crust or subducted oceanic crust, crystal fractionation of hydrous arc basalts, or upper crustal differentiation of magmas from the melting of basalt (Kleinhanns et al., 2003;Nagel et al., 2012;Jagoutz et al., 2013;Martin et al., 2014;Hastie et al., 2016;Johnson et al., 2017;Laurent et al., 2020). The rarity of porphyry Cu ± Au deposits in the Archean could therefore be attributed to the absence of modern-style plate tectonics (Fig. 1), particularly prior to 3•0 Ga (Moyen & van Hunen, 2012;Bédard, 2018;Johnson et al., 2019), and/or predominance of relatively reduced hydrothermally altered seafloor basalts and sulfate-poor (<200μM) seawater (Fig. 1; Habicht et al., 2002;Jamieson et al., 2013;Crowe et al., 2014;Stolper & Keller, 2018). Neither of these factors allows for the recycling of oxidized materials from Earth's surface to the mantle, thus rendering the sodium-rich TTG magmas in the Archean either reduced or sulfur-poor (Prouteau & Scaillet, 2012;Jagoutz et al., 2013). Such reduced magmas would, however, favor the formation of magmatic sulfides, retained either in the source region or during magma ascent by assimilation of the overriding reduced lithosphere, which would therefore deplete the silicate melts of their chalcophile element contents (Candela & Holland, 1984;Evans & Tomkins, 2011;Audétat & Simon, 2012) and limit the Cu ± Au potential of the evolved magmas to form porphyry Cu deposits upon emplacement in the upper crust (Richards, 2011;Simon and Ripley, 2011).
However, there are currently few constraints on the magmatic f O 2 and volatile contents of sodic TTG magmas (based mainly on zircon geochemistry; e.g. Yang et al., 2014;Madon et al., 2020;Mole et al., 2021), particularly for those associated with porphyrytype Cu ± Au deposits in the Neoarchean. Given the general lack of constraints of the intensive parameters (e.g. f O 2 ) and presence of some, albeit rare, significant porphyry-type Cu ± Au mineralization (e.g. Côté Gold Au ± Cu deposit in the Abitibi greenstone belt, with ∼10 Moz Au and geochemically significant but undefined Cu resource; Kontak et al., 2013;Katz et al., 2017Katz et al., , 2021)), it remains unclear whether the metallogenic processes of Phanerozoic porphyry Cu systems are necessary to form the same type of deposits in the Archean. The issue is also relevant for identifying the actual cause of the apparent rarity of porphyry-type Cu ± Au deposits in the Archean.
To address the issue, three porphyry-type Cu ± Au deposit settings (i.e. Côté Gold, St-Jude, and Clifford) previously reported in the well-mapped ∼2•7 Ga southern Abitibi subprovince, Canada, were selected for study. These settings were chosen for the following reasons: (1) they are host to disseminated-and veinlet-type mineralization associated with potassic alteration, thus resembling 2008) is illustrated in the pink histogram. The interpretation of tectonic regimes is from Cawood (2020). The lithological transition from sodic tonalite-trondhjemitegranodiorite-to potassic granite-granodiorite-dominated upper continental crust is based on statements from Jagoutz (2013) and references therein. The timing of the rise of the atmospheric O 2 during the Great Oxygenation Event (GOE; in grey) is from Holland (2002). The Fe 3+ / Fe ratios (in pale brown) of the submarine basalts are from Stolper & Keller (2018). Marine sulfate contents on average (in blue; uncertainty not marked for simplification) were compiled in previous studies (Lyons & Gill, 2010;Planavsky et al., 2012); where data are not reported, we extrapollate marine sulfate contents along the blue dashed line.
aspects of Phanerozoic porphyry Cu systems (Sillitoe, 2010); (2) they are well characterized and their absolute timing of emplacement is constrained by high-precision U-Pb geochronology (Galley & Van Breemen, 2002;Piercey et al., 2008;Kontak et al., 2013;Katz et al., 2017Katz et al., , 2021)); (3) they are among the earliest known cluster of TTGrelated porphyry-type Cu ± Au deposits worldwide (Singer et al., 2008).
Specifically, the goal of this study is to estimate the magmatic f O 2 conditions of the causative magmas for these deposits using the chemistry of primary mineral phases, which can be hindered by variable, post-emplacement modification owing to alteration, deformation, and metamorphism (e.g. Madon et al., 2020;Meng et al., 2021). Where analysis of apatite composition was possible, the magma S contents were evaluated. Whole-rock geochemistry and zircon U-Pb-Hf-O isotopes were also analyzed, and used in combination with previously published results, to constrain the crystallization ages and geochemical processes, as well as sources of these magmas.
The three porphyry-type Cu ± Au deposits studied are located in the southern part of the Neoarchean Abitibi greenstone belt (abbreviated as AGB hereafter) of the southeastern Superior Province, Canada. The AGB is bounded to the north and south by sedimentary rocks of the Opatica and Pontiac subprovinces, respectively, and is separated from the Wawa subprovince to the west by the NEtrending, west-dipping Kapuskasing thrust zone (Fig. 2). The AGB is composed primarily of ∼2•70-2•75 Ga east-trending volcanic assemblages and unconformably overlying fluviatile siliciclastic sedimentary sequences with rare alkaline volcanic rocks, as well as ∼2•70-2•75 Ga TTG rocks and younger (<2•68 Ga) granitemonzonite-syenite-dominated intrusions (Figs 2 and 3;Feng & Kerrich, 1992;Beakhouse, 2011;Mathieu et al., 2020).
Six semi-continuous submarine volcanic assemblages have been defined based on rock types and high-precision zircon U-Pb geochronology, including four dominantly tholeiitic ± komatiitic assemblages (i.e. Pacaud, Stoughton-Roquemaure, Kidd-Munro, and Tisdale) and two calc-alkaline-dominated assemblages (i.e. Delore and Blake River groups; Figs 2 and 3; Ayer et al., 2002;Thurston et al., 2008). Thurston et al. (2008) showed that most of the six volcanic assemblages are locally separated by ∼200 m thick chemical sedimentary layers such as iron formation (Fig. 3). The Porcupine and Timiskaming sedimentary assemblages unconformably overlie these six volcanic assemblages (Fig. 3).
The ages and distribution of the volcanic and sedimentary assemblages, which are summarized in Figures 2 and3 Two sedimentary-dominated sequences (i.e. Porcupine and Timiskaming) overlie the volcanic-dominated successions. The older Porcupine assemblage (≤2690-≤2682 Ma) mainly includes finegrained, deep-water deposited, flysch-like clastic sedimentary rocks that unconformably overlie the submarine volcanic assemblages noted above (Figs 2 and3). These clastic rocks include greywacke and mudstone, which are locally intercalated with conglomerate, banded iron formation, and calc-alkaline rhyodacitic volcanoclastic deposits (Bleeker, 2015). Detrital zircons from greywackes yielded ages of 2750-2690 Ma, with some having Mesoarchean ages, and thus consistent with the age spectrum of the putative sources such as pre-Porcupine igneous rocks of the AGB (Ayer et al., 2005;Frieman et al., 2017). The younger sedimentary sequence is the molasse-like Timiskaming assemblage (≤2679-≤2670 Ma), which accumulated in extensional basins, unconformably overlies folded Porcupine and older volcanic assemblages, and consists mainly of subaerial alluvialfluvial conglomerate and sandstone with lesser subaerial alkaline volcanic rocks (Bleeker, 2015).
Five stages or suites of intrusive rocks are present in the southern AGB based on rock type as well as relative timing with respect to the volcanic assemblages and regional tectonic deformation (e.g. Feng & Kerrich, 1992;Beakhouse, 2011;Mathieu et al., 2020).
(1) ∼2750-2695 Ma: this earliest group includes syn-volcanic, high-Al and minor low-Al tonalite (see definition given by Barker and Arth, 1976), trondhjemite, and diorite of the well-known TTG clan.
(2) 2695-2685 Ma: syn-deformational granodioritic and tonalitic intrusion and high-level porphyritic stocks formed synchronously with deposition of the Porcupine assemblage. The extent of the extrusive equivalents remains unclear (e.g. MacDonald & Piercey 2019).
(3) 2685-2679 Ma: a younger, late syn-deformational suite of granodioritic and quartz monzonitic intrusions emplaced between the termination of the Porcupine sedimentation and the start of the Timiskaming sedimentation.
(4) 2679-2660 Ma: an alkalic intrusive suite comprising diorite, monzonite, quartz syenite, syenite and granite emplaced during and immediately after the Timiskaming sedimentation.
(5) <2660 Ma: peraluminous granitic intrusions formed during crustal exhumation of the southern AGB that intrude the metamorphosed volcanic and sedimentary assemblages as mentioned above.
These types of intrusive rocks reflect a broadly evolutionary pattern with time comparable with those in many other Archean cratons globally, which may be attributed to evolving geodynamic processes in the early Earth (e.g. cooling; Laurent et al., 2014).
The autochthonous process for the formation of the southern AGB is supported by the broadly correlated volcanic stratigraphy and intrusive rocks across the belt (Ayer et al., 2002). The recurring tholeiitic and calc-alkaline volcanic assemblages over ∼50 Myr prior to ∼2•70 Ga are interpreted to indicate either episodic short-term plate tectonics (Moyen & van Hunen, 2012) or 'plume-arc' interaction (Ayer et al., 2002;Sproule et al., 2002;Wyman et al., 2002) under a broadly extensional environment (Fig. 3). Based on an interpretation of the LITHOPROBE seismic reflection profiles suggesting that subduction-zone scars are present within the sub-Opatica lithosphere (Calvert & Ludden, 1999), the model of the plume-arc interaction in the AGB was extended by Benn & Moyen (2008) to suggest that an ocean basin was subducted beneath an active and partially differentiated oceanic plateau. However, Bédard et al. (2013) instead argued that the calc-alkaline volcanic rocks formed as a result of the intermittent melting at the base of the thickened basaltic crust (e.g. in vertical tectonics or a sagduction model).
Tectonic shortening of the southern AGB postdating ∼2•70 Ga was interpreted to be a consequence of subduction-like processes (Daigneault et al., 2002). The syn-deformational granodiorite and tonalite plutonic suites are thought to be more probably derived from melting of subducted flat slab versus accreted oceanic crust or 6) Blake River succeeded by minor volcanic rocks hosted in (7) Porcupine and (8) Timiskaming sedimentary assemblages (Ayer et al., 2002;Thurston et al., 2008). The timing of evolving plutonic rock phases from syn-volcanic tonalite-dioritetrondhjemite, early syn-tectonic granodiorite, syn-tectonic sanukitoid (mainly quartz syenite-quartz monzonite-granite), and late-tectonic alkalic intrusion, to crust-derived granite, is based on data compiled by Ayer et al. (2005) and Beakhouse (2011). Deformation events are adapted from Bateman et al. (2005). The timing of the Côté Gold, Don-Rouyn, St-Jude, and Clifford deposits are marked (see text for the new and published geochronological results).
thickened plateau crust (Feng & Kerrich, 1992;Sutcliffe et al., 1993;Wyman et al., 2002). Crustal thickening in the southern AGB starting at ≤2690 Ma was followed by the Porcupine sedimentation and deformation of the previously formed volcanic and intrusive rocks (Monecke et al., 2017). The intrusion of the late syn-deformational quartz syenite-quartz monzonite-granite suites (sanukitoid sensu lato) is interpreted to have been derived from a combination of metasomatized hydrous mantle sources and adakite-type sources (Wyman et al., 2002).
Large-scale folding and thrusting, commensurate with a second stage of tectonic deformation, occurred prior to the Timiskaming sedimentation (i.e. before ∼2679 Ma). This event was succeeded by emplacement of post-deformational shoshonitic trachytic and alkali-feldspar syenitic magmas localized to trans-lithospheric strikeslip structures. These magmas are related to the partial melting of refractory 'sub-arc' mantle fertilized by alkali-rich fluids or melts (Feng & Kerrich, 1992;Wyman et al., 2002). Lastly, crustal melting and exhumation after ∼2660 Ma were associated with formation of peraluminous granite (Daigneault et al., 2002).
The southern AGB mainly experienced prehnite-pumpellyite-to greenschist-facies metamorphism during 2677-2643 Ma, but locally up to amphibolite-facies metamorphism in the contact aureoles of plutons (Powell et al., 1995;Ayer et al., 2002). In contrast to the greenschist-to amphibolite-facies metamorphism of rocks in the northern AGB, the generally lower metamorphic grade now observed in outcropping rocks of the southern AGB (Fig. 2) may reflect the shallower levels of exposure of the supracrustal rocks because of differential uplift (Benn & Moyen, 2008).
The main mineral deposit types and their exploited resources in the AGB include volcanic-hosted massive sulfide (Cu, Zn, Au), komatiite-hosted magmatic Ni-Cu-PGE, banded iron formation, and mesothermal and intrusion-related Au deposits, as well as the unusual Chibougamau-type Cu-Au veins (see Monecke et al., 2017, for a review). The intrusion-related Au deposits are reported to be associated with quartz-carbonate-albite alteration hosted in the late-to post-deformational quartz-monzonitic to syenitic intrusions (Robert, 2001;Bigot & Jébrak, 2015), and thus are not comparable with either the Phanerozoic porphyry Cu-Au deposits (Sillitoe, 2010) or the reduced intrusion-related gold deposits (e.g. Thompson et al., 1999;Hart, 2007). As the associated rock types with the intrusionrelated Au deposits are not sodic TTG, they are therefore not the focus of this study.
Only a few porphyry-type Cu ± Au deposits are reported for the southern AGB. This includes the Côté Gold and St-Jude deposits, both associated with low-pressure syn-volcanic TTG suites formed at ∼2•74 Ga and ∼ 2•70 Ga, respectively, and the Croxall deposit associated with ∼2•69 Ga syn-deformational medium-pressure tonalitegranodiorite at Clifford (Goldie et al., 1979;Galley & Van Breemen, 2002;Piercey et al., 2008;Katz et al., 2017Katz et al., , 2021)). Each of these deposits consists dominantly of breccia-hosted mineralization that is mainly associated with potassic (biotite ± K-feldspar) alteration.
To avoid confusion, we use Côté Gold, St-Jude, and Clifford to refer to the three porphyry-type deposits and related intrusions hereafter.
The salient features of four mineralized magmatic-hydrothermal settings (i.e. Côté Gold, St-Jude, Don-Rouyn, and Clifford) in the AGB are summarized below. Of particular relevance to this study is that these deposits are constrained using high-precision geochronology to have formed at three different times: the Côté Gold Au ± Cu deposit of ∼2•74 Ga, the St-Jude and Don-Rouyn Cu-Mo deposits related to the ∼2•70 Ga Flavrian-Powell intrusive complex, and the Clifford Cu ± Mo ± Au deposit setting of ∼2•69 Ga.
Côté Gold Au ± Cu deposit (∼2740 Ma)
The Côté Gold Cu ± Au deposit is located in the Swayze greenstone belt, an interpreted SW extension of the AGB (Fig. 2). The deposit is hosted by the ∼2•74 Ga Chester Intrusive Complex, a composite tonalite, diorite, and quartz diorite suite that is cut by magmatic and hydrothermal breccia bodies (Figs 4a and 5a,b). The tectonic regime is generally thought to have been a back-arc-like setting based on the geochemistry of synchronous massive-and pillowed mafic volcanic rocks of the Arbutus Formation, which is part of the Pacaud assemblage noted above (Fig. 3; Katz et al., 2017).
The deposit contains an indicated ore resource of 355•4 Mt at 0•87 g t -1 Au or 9•97 Moz Au (Katz et al., 2021) with an associated but undefined amount of Cu (Kontak et al., 2013;Katz, 2016;Katz et al., 2021). The intrusive bodies, in particular the hydrothermal breccia, host sheeted-, stockwork-, disseminated-, and veinlet-types of mineralization associated mainly with biotite alteration overprinted by albite and sericite alteration. Four molybdenite separates from mineralized quartz veins yielded a weighted mean Re-Os age of 2740•2 ± 5•6 Ma [2σ, n= 4, mean square of weighted deviates (MSWD) = 0•73], which is consistent with the isotope dilutionthermal ionization mass spectrometry (ID-TIMS) zircon U-Pb ages of the intrusive phases in the Chester Intrusive Complex (Kontak et al., 2013;Katz et al., 2017Katz et al., , 2021)). The diorite is interpreted to be the causative magma for the Cu ± Au mineralization (Kontak et al., 2013;Katz, 2016;Katz et al. 2021). The area was overprinted by lower greenschist-facies metamorphism related to later (i.e. <2•7 Ga) deformation (Katz et al., 2017) and was cut by a series of later dike rocks of variable type, with the most dominant being the ∼2•45-2•5 Ga Matachewan swarm (Fig. 4a).
The St-Jude and Don-Rouyn porphyry-type deposits are reported to be associated with the ∼2•70 Ga Flavrian-Powell intrusive complex (Fig. 4b) estimated to have been emplaced at a pressure of 90-150 MPa (Feng & Kerrich, 1990). The complex intruded intermediate to felsic volcanic rocks of the synchronous Blake River Group, which hosts many massive sulfide deposits (Monecke et al., 2017) and formed during cauldron subsidence or early rifting in an arc-to back-arc basin setting (Goldie et al., 1979;Laflèche et al., 1992;Galley & Van Breemen, 2002). The Flavrian pluton comprises diorite (Fig. 5c), tonalite, two phases of low-Al trondhjemite (phases I and II; Fig. 5d ande), and a late-stage quartz diorite (Galley, 2003). The high-level pluton has a wide contact aureole (<1 km) developed in the enveloping synchronous volcanic rocks that reflects extensive convection of circulated seawater (Hannington et al., 2003). The Downloaded from https://academic.oup.com/petrology/article/62/11/egab079/6377495 by Tsinghua University user on 16 November 2021 volcanic and plutonic rocks record the effects of later overprinting greenschist-facies metamorphism.
The St-Jude deposit consists of porphyry-type Cu ± Mo ± Au mineralization hosted in a trondhjemite stock (Figs 4b and 5f), which is interpreted to be genetically related to the Flavrian amphibolebearing trondhjemite phase II (Fig. 5e; Goldie et al., 1979;Galley & Van Breemen, 2002;Galley, 2003). The ∼600 m diameter magmatic-hydrothermal breccia pipe (inset in Fig. 4b) hosts disseminated-and veinlet-type Cu ± Mo ± Au mineralization associated with a core of biotite alteration and outer biotite-sericite alteration where the altered rocks are crosscut by aplite dikes (Kennedy, 1985;Galley & Van Breemen, 2002;Galley, 2003). The ore resources have not been reported, but the previously published drill logs suggest that many drilling intervals yielded significantly variable grades of Cu (0•1-2•0 wt%, up to 5•0 wt%) and Au (0•2-8 g t -1 , locally up to 12 g t -1 ; data from reports GM51641 and GM62642 compiled by Québec Ministry of Energy and Natural Resources; gq.mines.gouv. qc.ca/documents/examine). The lower limit of the age for the brecciahosted mineralization is constrained by an ID-TIMS zircon U-Pb age of 2697 ± 2 Ma for the aplite dike (Galley & Van Breemen, 2002), which is temporally consistent with the Blake River Group volcanic rocks (Péloquin et al., 2008, and references therein) and with the laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) U-Pb zircon ages for the trondhjemite (see the section 'Zircon U-Pb geochronology' below). We interpret the trondhjemite in the core of the breccia body to be the causative magma for the mineralization. The aplite dikes may indicate volatile saturation of the Flavrian-St-Jude trondhjemite magma chamber during late-stage crystallization (Galley, 2003).
In contrast to the St-Jude deposit, the Don Rouyn Cu-Mo deposit (36 Mt @ 0•15 wt% Cu) is hosted by the Powell intrusion, which is interpreted to be structurally offset from its Flavrian counterpart (Fig. 4b). The deposit area is extensively fractured and overprinted by retrograde metamorphic chlorite that renders identifying the original mineralogy difficult. It is reported to be a porphyry-type deposit with disseminated-and veinlet-type mineralization and zoned in Cu with a chalcopyrite-bornite core to chalcopyrite outer zone having grades from <0•1 wt% to >0•25 wt%, respectively (Goldie et al., 1979). According to Galley (2003), the associated alteration includes a silicic core, a biotite-rich halo, and an outer zone of pervasively chlorite-altered host rock. Relevant in this context is the similar occurrence of chlorite replacing hydrothermal biotite at the Côté Gold deposit (Katz et al., 2017), which may together reflect a common feature of later metamorphic overprinting in the AGB. Limited access to the formerly mined area prevented verification of the causative magma, leading us not to consider the deposit as part of this study.
The Clifford Cu ± Mo ± Au deposit setting (∼2689 Ma)
The porphyry-type Cu ± Mo ± Au mineralized Croxall breccia pipe (∼40 m in diameter) is hosted by the Ben Nevis volcanic complex of the ∼2701-2694 Ma Blake River Group (Péloquin et al., 2008) and is one of several mineralized breccia bodies in this area. The Croxall pipe, well described by Chaloux (2005) and Piercey et al. (2008), has a core of andesitic breccia cemented by quartz and sulfide that grades outwards to a marginal 'milled' breccia enriched in quartz, pyrite, and chalcopyrite with elevated Au, Ag, and Mo abundances. The mineralization produced 3-5 % sulfides (locally up to 10 %) with surface assay samples yielding anomalous metal contents up to 0•853 g t -1 Au and >1 wt% Cu (Chaloux, 2005), but the ore resources have not been defined. The pipe contains silicification and disseminated pyrite, whereas the wall rocks contain a network of millimeter-to centimeter-size veins of quartz, epidote, K-feldspar, hematite, and calcite (± sericite). The sulfide mineralization of the dark gray quartz veins with pyrite, chalcopyrite and molybdenite (± gold) is cut by hematite-calcite ± K-feldspar veins, which are similar to 'J veins' in porphyry systems because of their jasperoidal appearance, as noted by Piercey et al. (2008).
A molybdenite separate from a quartz-pyrite-molybdenite vein yielded a Re-Os model age of 2682 ± 8 Ma, which overlaps with the ID-TIMS weighted mean zircon 207 Pb/ 206 Pb ages of 2689 ± 2 Ma and 2687 ± 1 Ma (2σ; MSWD = 0•6 and 0•47, respectively; uranium decay constant uncertainties not considered) for the plagioclasephyric porphyry and tonalite from the Croxall breccia, respectively (Piercey et al., 2008). The porphyry-type mineralization is therefore interpreted to be genetically related to the subvolcanic Clifford stock composed mainly of tonalite, with minor granodiorite (Fig. 5g; Piercey et al., 2008), and quartz diorite. The stock is surrounded by a number of quartz-feldspar porphyritic (Fig. 5h) and mafic dikes. The limited extent of an albite-epidote-magnetite alteration zone defines a pseudo-contact aureole in the enveloping basaltic to andesitic volcanic rocks (Fig. 4c) and has been interpreted to be related to syn-intrusion high-temperature fluid circulation (Hannington et al., 2003;Piercey et al., 2008). The area experienced prehnitepumpellyite-facies metamorphism.
The petrographic features of the major igneous phases from Côté Gold, Flavrian-St-Jude, and Clifford have been described in detail by Katz et al. (2017), Galley (2003), and Piercey et al. (2008), respectively. Representative samples from the different phases of these intrusions were collected for petrological characterization. Locations and petrographic descriptions of the representative samples are provided in Table 1. Heavy minerals from nine representative samples were separated using a Wilfley table after electric-pulse disaggregation at Overburden Drilling Management Inc., Ottawa, Canada. Zircon grains were hand-picked using a binocular microscope and were mounted in epoxy pucks, which were then polished to their mid-section using diamond grits (final grit size was ∼0•25 μm).
Despite most samples having loss on ignition (LOI) values <2 wt% (<2•5 wt% for all), variable metamorphic effects are indicated by petrographic features (Fig. 6). To constrain magmatic f O 2 of the causative magmas for the mineralization, analyses were limited to the primary igneous minerals (i.e. zircon, amphibole, apatite, and magnetite-ilmenite pairs) that are demonstrated to have preserved magmatic features such as twinning or zoning, or to be enclosed as inclusions in fresh, coarser phases (i.e. amphibole and plagioclase) and in robust zircon hosts.
We focused initially on identifying primary mineral inclusions hosted in zircon, preparing ∼300-500 grains for each sample, except for the samples FLV-22 and CS-22, which had poor zircon yields. Zircon grains were mounted and polished to identify mineral inclusions, in particular for inclusions remote from fractures because they are known to have most probably preserved primary compositions (e.g. Bell, 2016). However, the exposed number of mineral inclusions varied considerably among zircon samples, which we consider to reflect the intrinsic abundances of inclusions and/or an unintentional bias towards selective cutting (i.e. along a single plane) of the zircon grains. Following on this, we also attempted to identify primary minerals in some of the polished thin sections of the selected samples.
The abundance of apatite inclusions in zircon grains most probably reflects the relative timing of apatite versus zircon crystallization (Bell et al., 2018). If zircon crystallized earlier than apatite, which is manifested as elevated or nearly identical Ti-in-zircon temperature (T Ti-Zr ) relative to the model apatite saturation temperature (AST), the chance of finding apatite inclusions in zircon is much reduced, and vice versa. For samples yielding elevated or close T Ti-Zr compared to AST (e.g. diorite sample Z10896 from Côté Gold with T Ti-Zr of 815 ± 30 • C and lower AST of 668 that zircon-hosted apatite inclusions are rarely identified, efforts were made to seek apatite inclusions protected by other mineral phases or in the matrices. We scanned the thin sections of these samples using cathodoluminescence (CL) and backscattered electron (BSE) modes of the scanning electron microscope (SEM) to identify the apatite grains and examine their chemical zoning and textures.
Occurrence and features of primary zircon, amphibole, apatite, and Fe-Ti oxides Zircon grains have crystal sizes that vary from 50 to 250 μm and typically exhibit oscillatory or sector zoning, as revealed by CL imaging (Supplementary Data Fig. A1; supplementary data are available for downloading at http://www.petrology.oxfordjournals.o rg), which is consistent with a magmatic origin (e.g. Corfu et al., 2003). The presence of mineral inclusions, as summarized in Table 1, varied and included apatite, K-feldspar, plagioclase, quartz, and minor biotite, rutile, titanite, and Fe-Ti oxides. Amphibole inclusions were not found in any zircon grain, and zircon grains from the Côté Gold diorite sample, of which several hundred polished grains were examined, were completely devoid of mineral inclusions.
The nature of amphibole in the different intrusions varies considerably. Most of the amphiboles in the studied Côté Gold diorite samples record actinolitic alteration (Fig. 6b), although rarely less altered samples were identified and analyzed (Katz et al., 2017). In contrast, the trondhjemite-II phase from the Flavrian intrusion contains amphibole (edenitic variety; Figs 6e and7a), which is not present in the trondhjemite from the St-Jude breccia (Fig. 6f). For the Clifford stock, the calcic amphiboles are common in the equigranular tonalite (Figs 6g and7b) and are also rarely found as phenocrysts in the plagioclase-phyric porphyry dike (Figs 6h and7c) proximal to the Clifford stock (Fig. 4c). Of particular note is a plagioclasephyric porphyry sample that was found to contain an amphibole glomerocryst with two optically different zones of similar magnesiohornblende; the core and overgrowth areas are noted as Amp-I and Amp-II, respectively (Figs 6h and7c).
The number of exposed apatite inclusions in the studied zircons varies for Côté Gold, Flavrian-St-Jude, and Clifford (Fig. 7d andf-i). For diorite sample Z10896 from Côté Gold, one large interstitial equant apatite grain in the matrix having sector and oscillatory zoning was identified (Fig. 7e), whereas other smaller apatite grains with fractures and patchy zonings, probably reflecting the effect of metamorphism, were noted. A few apatite grains have been found in zircons from sample CS-20 (tonalite) collected from the Clifford stock (Fig. 7h andi), and apatite grains hosted by amphibole in the matrix were identified, with most having a patchy or mosaic CL response indicative of metamorphism and alteration (e.g. Fig. 7k). Apatite grains of sufficiently large size (>4 μm) for analysis have not been identified in zircons from sample CS-22 (plagioclase-phyric porphyry), but one large grain with faint magmatic zoning hosted in an amphibole phenocryst (Amp-I, Fig. 7j) was identified.
Primary magnetite and ilmenite are rare, but a single magnetite inclusion was found in a zircon grain from the Côté Gold tonalite (Fig. 8a) and an ilmenite inclusion was found in a zircon grain from a trondhjemite sample from the St-Jude breccia (sample SJ-04; Fig. 8b). Subhedral inclusions forming a primary magnetiteilmenite pair showing textural equilibrium were identified in primary plagioclase (andesine with polysynthetic twinning) from the Clifford tonalite (samples CS-03 and CS-07; Fig. 8c-f), but some of the magnetite grains are pitted and may reflect alteration (Fig. 8g-j).
Petrographic features of the mounted zircons, including both the zircons and their inclusions, were characterized using a microscope and a Tescan Vega 3 SEM equipped with a Bruker energy-dispersive spectrometer (EDS) and CL detector at Laurentian University. Details of whole-rock geochemical analysis, zircon U-Pb-Lu-Hf isotope and trace element analysis, zircon O isotope analysis, electron microprobe analysis (EMPA), and sulfur K-edge micro X-ray absorption near edge structure (μ-XANES) analysis of apatite are provided in Supplementary Data Electronic Appendix I.
Four unique and complementary oxybarometers were used to estimate the f O 2 of the causative magmas. We used the zircon oxybarometer of Loucks et al. (2020), the amphibole oxybarometer of Ridolfi et al. (2010), the S-in-apatite oxybarometer of Konecke et al. (2017Konecke et al. ( , 2019)), and coexisting magnetite-ilmenite pairs (see Ghiorso & Evans, 2008). Details are presented in Supplementary Data Electronic Appendix II.
The emplacement pressure of the Flavrian intrusion has been estimated by Feng & Kerrich (1990) using an amphibole geobarometer to be 90-150 MPa (∼120 MPa on average). The emplacement pressure of the Clifford stock is estimated from amphibole composition using the method of Mutch et al. (2016) to correct sulfur-in-apatite oxybarometry data (such estimation is not required for samples from Côté Gold because μ-XANES spectra were not obtained). The Alin-hornblende geobarometer is suitable for tonalitic to granodioritic rocks containing amphibole, biotite, plagioclase (An 15-80 ), quartz, alkali-feldspar, apatite, titanite, and zircon, which are broadly consistent with the mineral assemblage identified in the Clifford tonalite and plagioclase-phyric porphyry (biotite has locally been found).
Crystallization temperatures of zircon, amphibole, apatite, and Fe-Ti oxides were calculated using the following thermometers: Tiin-zircon (Ferry & Watson, 2007), amphibole (Ridolfi et al., 2010), model apatite saturation (Piccoli & Candela, 1994), and magnetiteilmenite (Lepage, 2003). For the Ti-in-zircon thermometer, based on the presence of quartz and Ti-bearing minerals (e.g. titanite and ilmenite) in the studied samples, the activities of SiO 2 and TiO 2 are assumed to be one and 0•5, respectively (Schiller & Finger, 2019). The model apatite saturation temperature (AST) is calculated to represent the temperature at which apatite began to crystallize from the metaluminous magmas [molar Al 2 O 3 /(CaO & Candela, 1994). We assume that the whole-rock SiO 2 and P 2 O 5 abundances approximate the concentration of these oxides in the melt at the time of apatite crystallization.
In addition to the analyses presented here for the selected 17 samples, which include diorite and tonalite at Côté Gold, diorite and trondhjemite at Flavrian-St-Jude, and tonalite and plagioclase-phyric porphyry at Clifford (Supplementary Data Table A1), previously published whole-rock major and trace element data (filtered by LOI < 2•5 wt%) for 47 samples of similar phases are compiled in Supplementary Data Table A1 from Galley (2003), Piercey et al. (2008), and Katz (2016). The major elements and Cu contents are reported for those samples lacking noted alteration with LOI < 2•0 wt%. The major elements were recalculated on a normalized anhydrous basis.
The Côté Gold diorite samples have K 2 O/Na 2 O = 0•04-0•57, total Fe 2 O 3 = 5•1-13•7 wt%, MgO=1•5-6•1 wt%, and Mg# (d,f). (g-j) Partially altered magnetite-ilmenite mineral pairs hosted in plagioclase in sample CS-07; cross-polarized light in (g, i) and BSE in (h, j). The ilmenite in (h) is probably exsolution from an originally homogeneous Ti-rich magnetite. The magnetite in (h) has partially been altered with rim retaining primary composition, whereas the magnetite in (j) has been altered. Bt, biotite; Cpx, clinopyroxene; Ilm, ilmenite; Mt, magnetite; Pl, plagioclase; Ser, sericite; Zrn, zircon. (See Table 1 for sample locations and descriptions.) (Katz, 2016). One sample of tonalite is a white-filled green triangle. (c, d) Flavrian-St-Jude diorite and trondhjemite compared with data (in grey) from literature (Galley, 2003). (e, f) Clifford tonalite and plagioclase-phyric porphyry (in red) compared with data (in grey) from literature (Piercey et al., 2008). Normalization values are from Sun & McDonough (1989).
or K) and depleted in Nb, Ta, and Ti (Fig. 9a). On a chondritenormalized rare earth element (REE) diagram, the samples display a concave pattern (Fig. 9b) with low La/Yb ratios of 0•7-9•0 and moderate to significant negative Eu anomalies (0•25-1•10, 0•70 ± 0•19 on average; 1σ, n = 32). Moderate depletion of Sr, P, and Ti, as well as the negative Eu anomalies, reflects minor fractionation of apatite and Fe-Ti oxides and moderate fractionation of plagioclase.
One diorite sample from the Flavrian intrusion yielded flat trace and rare earth element patterns similar to Côté Gold diorite and has comparable K 2 O/Na 2 O = 0•08, total Fe 2 O 3 = 10•1 wt%, MgO = 4•3 wt%, and Mg# = 41. In contrast, the trondhjemite samples (Richards et al., 2001(Richards et al., , 2012(Richards et al., , 2017;;Wainwright et al., 2011;Hou et al., 2013;D'Angelo et al., 2017;Grondahl & Zajacz, 2017;Zhu et al., 2018). (b) Dy/Yb ratios versus SiO 2 (wt%) diagram. The 'adakite-like rocks' and 'normal andesite-dacite-rhyolite' zones in (a) are from Defant & Drummond (1990). The fractionation trends of garnet and amphibole during magma evolution in (b) are from Davidson et al. (2007). The two least altered samples contain an average Cu content of 36 ± 18 ppm (1σ, n = 2). However, these samples yielded noticeably different trace and rare earth element patterns and lower REE contents in comparison with the Côté Gold and Flavrian samples. In the primitive mantle normalized trace element diagram, they show enrichment in fluid-mobile LILE (Rb, Ba, Th, U, K, and Sr) and LREE, and are depleted in Nb, Ta, and minor Ti (Fig. 9e). Their REE CN (where CN indicates chondrite-normalized) profile shows a strongly fractionated pattern with high La/Yb ratios (9•7-15•3, 12•4 ± 1•6; 1σ, n = 13), and weak negative to positive Eu anomalies (1•09 ± 0•27 on average, 1σ, n = 13; Fig. 9f). These samples also have high Sr/Y (20-45, 31 ± 7 on average, n = 13; Fig. 10a) and V/Sc (13•1 ± 0•8; 1σ, n = 3) ratios. These features suggest fractionation of amphibole and Fe-Ti oxides, but the lack of plagioclase fractionation. The amphibole-dominated fractionation trend is also supported by broadly decreasing Dy/Yb ratios with SiO 2 in Figure 10b (Davidson et al., 2007).
The LA-ICP-MS zircon U-Pb ages for the samples are summarized in Figure 11 and Table 2, with the complete database provided in Supplementary Data Table A2. The major intrusive phases in each of the studied areas have previously been dated with ID-TIMS (Galley & Van Breemen, 2002;Piercey et al., 2008;Kontak et al., 2013;Katz et al., 2017). The new data for Côté Gold provide further age constraints on previously dated diorite and tonalite, and this study also supplements the published data for the St-Jude and Clifford areas with five new intrusive samples including three samples of the two trondhjemite phases from the Flavrian intrusion, the trondhjemite from the St-Jude breccia, and one each of the tonalite and plagioclasephyric porphyry from the Clifford stock. Representative zircon grains from the previously dated diorite and tonalite were also incorporated into the analytical session to monitor zircon inheritance and postcrystallization possible Pb loss.
The diorite and tonalite samples from Côté Gold previously dated using chemical abrasion (CA-)ID-TIMS method were analyzed. Thirty-five analyses of zircon grains from sample Z10896 (diorite) yielded an upper concordia intercept age of 2736 ± 6 Ma (2σ, n = 35, MSWD = 0•01; Fig. 11a). Thirty-one analyses of zircon grains from sample Z10993 (tonalite) yielded an upper concordia intercept age of 2736 ± 4 Ma (2σ, n = 31, MSWD = 0•13; Fig. 11b). These results agree within uncertainty with the published CA-ID-TIMS ages of 2741•5 ± 0•7 Ma and 2741•4 ± 0•9 Ma for the samples Z10896 and Z10993, respectively, reported by Katz et al. (2017), as well as an age of 2741•1 ± 0•9 Ma for a tonalite sample reported in Kontak et al., 2013. These new and published results constrain the emplacement age of the Côté Gold intrusions to ∼2•74 Ga.
For the Flavrian intrusive complex, four trondhjemite samples were dated. Thirty-five analyses of zircon grains from sample FLV-22 (trondhjemite phase I) yielded an upper concordia intercept age of 2691 ± 9 Ma (2σ, n = 35, MSWD = 0•31; Fig. 11c). If four analyses are excluded, two each with Pb loss or reverse discordance (grey ellipses in Fig. 11c), a concordia intercept age of 2700 ± 9 Ma (2σ, n = 31, MSWD = 0•7; Fig. 11c) is defined. Thirty-two analyses of zircon grains from sample FLV-28 (trondhjemite phase I) yielded an upper concordia intercept age of 2689 ± 6 Ma (2σ, n = 32, MSWD = 0•23; Fig. 11d). If the single analysis with Pb loss is excluded, this yields a concordia age of 2692 ± 6 Ma (2σ, n = 31, MSWD = 1•18). Forty-three analyses of zircon grains from sample FLV-04 (trondhjemite phase II) yielded an upper concordia intercept age of 2694 ± 8 Ma (2σ, n = 43, MSWD = 0•26; Fig. 11e), which agrees within uncertainty with the previously published CA-ID-TIMS age of 2700 + 3/-2 Ma (2σ, MSWD = 0•62; Galley & Van Breemen, 2002). Twenty-two analyses of zircon grains from sample SJ-04 (St-Jude trondhjemite) yielded a similar upper concordia intercept age of 2697 ± 13 Ma (2σ, n = 22, MSWD = 2•6; Fig. 11f), with two other analyses not shown in the plot yielding similar discordant older ages ( 207 Pb/ 206 Pb ages of 2882 ± 38 Ma and 2847 ± 7 Ma; Supplementary
The modelled apatite saturation temperature represents the temperature at which apatite started to crystallize. Amp, amphibole; Ap, apatite; Ilm, ilmenite; Mt, magnetite; Zrn, zircon. (2011). Oxygen isotope ratios reported in the standard per mil ( ) relative to standard mean ocean water (VSMOW). The δ 18 O isotope value range (5•3 ± 0•6 , 2σ) for zircon in equilibrium with mantle is from Valley et al. (1998). (See Table 1 for sample locations and descriptions.) Cathodoluminescence images of representative zircon grains are shown in Supplementary Data Figure A1.
Data Table A2) and thus probably reflecting xenocrysts. All four of the LA-ICP-MS U-Pb dates are consistent within uncertainty and yield a calculated weighted mean age of 2695 ± 5 Ma (2σ, n = 4, MSWD = 1•11). Thus these dates constrain the emplacement age of the Flavrian-St-Jude trondhjemite phases to ∼2•70 Ga.
For the Clifford stock, two samples with tonalitic composition were dated. For sample CS-20 (tonalite), 36 analyses of zircon grains yielded an upper concordia intercept age of 2690 ± 7 Ma (2σ, n = 36, MSWD = 0•78; Fig. 11g). Twenty-nine analyses of zircon grains from sample CS-22 (plagioclase-phyric porphyry) yielded an upper concordia intercept age of 2694 ± 12 Ma (2σ, n = 29, MSWD = 0•61; Fig. 11h). These two dates overlap within uncertainty, and a weighted mean age of 2692 ± 8 Ma (2σ, n = 2, MSWD = 0•05) was calculated. These dates overlap within uncertainty with the previously published ID-TIMS zircon weighted mean 207 Pb/ 206 Pb ages of 2687 ± 1 Ma and 2689 ± 2 Ma (Piercey et al., 2008), constraining the formation of tonalite and plagioclase-phyric porphyry to ∼2•69 Ga.
The age results are consistent within the uncertainty of previously published ID-TIMS zircon U-Pb age results, although the new ages reported here have relatively poorer precision. Of particular interest and relevance to this study is that most of the zircon analyses for each sample yielded concordant data that are consistent with closedsystem behavior despite variable degrees of post-crystallization tectonic disturbance in the study areas.
The zircon Hf-O isotopic data are summarized in Figure 12 and Table 2, with the full dataset reported in Supplementary Data Tables A3 andA4 (3•91 ± 0•13 on average; 1σ, n = 8), with a xenocrystic zircon grain yielding a similar δ 18 O value of 4•16 ± 0•29 (2SE). The low-δ 18 O zircons from the Clifford and Flavrian-St-Jude intrusions have low OH/O ratios that are comparable with those of the measured reference zircons in the analytical session; we therefore suggest that the apparent δ 18 O depletions are intrinsic features of the protolith rather than as a result of post-crystallization modification owing to fluid-mediated alteration.
We interpret these results to indicate that the Côté Gold magmas were derived from a mantle source, whereas the protoliths (also with juvenile mantle-derived sources) for the Flavrian, St-Jude, and Clifford magmas experienced variable degrees of high-temperature hydrothermal alteration or assimilated with hydrothermal altered wall rocks comparable with low-δ 18 O magmas (e.g. Valley et al., 2005;Bindeman & Simakin, 2014;Hammerli et al., 2018, and references therein). It should be noted that the scattered and broadly lower δ 18 O zircon values for the plagioclase-phyric porphyry versus the tonalite in the Clifford stock may indicate preferential assimilation of altered volcanic supracrustal rocks of the Blake River Group (reported to locally have a relatively depleted δ 18 O signature; Cathles, 1993) during the magma ascent.
The compositions of igneous apatite from the Côté Gold, Flavrian-St-Jude, and Clifford igneous suites are listed in Supplementary Data Table A5 and illustrated in Fig. 13a. The average chlorine concentration of zircon-hosted apatite inclusions from the Côté Gold tonalite is 0•07 ± 0•05 wt% (1σ, n = 9; Supplementary Data Table A5), whereas the S contents of these inclusions are below the EPMA detection limit (i.e. 0•01 wt%). The S concentrations of groundmass apatite grains from Côté Gold diorite are below the EPMA detection limit, whereas the average Cl concentration is 0•04 ± 0•01 wt% (1σ, n = 43; Supplementary Data Table A5), significantly lower than the Cl contents reported for apatite micro-phenocrysts in Phanerozoic arcrelated plutonic and volcanic rocks (e.g. Streck & Dilles, 1998;Imai, 2004;Scott et al., 2015).
The S contents of all of zircon-hosted apatite inclusions (n = 17) from the Flavrian-St-Jude trondhjemite samples are below the EPMA detection limit. For the apatite from ore-unrelated trondhjemite samples (FLV-22 and FLV-28), the average Cl concentration is 0•07 ± 0•02 wt% (1σ, n = 5) excluding analyses below the detection limit and a single value of 1•10 wt%. In contrast, the ore-related trondhjemite samples (SJ-04 and FLV-04) have much higher apatite Cl contents of 0•75 ± 0•44 wt% (1σ, n = 4) with four analyses below detection limit excluded. The altered apatite grains hosted in amphibole from sample FLV-32 (diorite) yielded S contents below the detection limit (n = 6; except one analysis of 0•01 wt% S) and Cl contents of 0•37 ± 0•11 wt% (1σ, n = 7).
The primary and altered apatite grains from the Clifford tonalite and plagioclase-phyric porphyry yielded distinctly different S and Cl contents. Two zircon-hosted primary apatite grains from the Clifford tonalite (sample CS-20) yielded S and Cl contents of 0•05 ± 0•03 wt% and 1•86 ± 0•48 wt%, respectively (n = 5), whereas four analyses of apatite hosted in the Amp-I phenocryst in the plagioclase-phyric porphyry (sample CS-22) yielded S and Cl contents of 0•09 ± 0•02 wt% and 0•90 ± 0•03 wt% (1σ, n = 4), respectively. In contrast, the altered apatite grains (samples CS-20 and CS-22) yielded similar compositions with S contents of 0•017 ± 0•003 wt% (1σ, n = 5, but excluding 17 analyses below the detection limit) and Cl contents of 0•16 ± 0•08 wt% (1σ, n = 22), respectively. A altered zircon-hosted apatite inclusion (cut by a fracture; Fig. 7i) was also analyzed and yielded similarly low S (0•018 wt% for one analysis and the second analysis yielding S content below the detection limit) and Cl (0•16 ± 0•06 wt%; 1σ, n = 2) contents. The noticeable decreases of S and Cl contents from primary to altered apatite in these samples are interpreted to reflect the effect of metamorphism and alteration.
To evaluate whether the studied apatite grains crystallized in a volatile-undersaturated or volatile-saturated environment, we followed the thermodynamic models of Stock et al. (2018) and accordingly plotted the X Cl /X OH versus X F /X Cl and X Cl /X OH versus X F /X OH ratios for analyses yielding halogen contents above the detection limits (Fig. 13b-f). Based on experimental results that demonstrate higher fluid-melt partition coefficients for Cl than for F (Webster et al., 2009), the models predict that the apatite X F /X Cl ratio decreases or remains constant with variable X Cl /X OH ratios under volatile-undersaturated conditions. In contrast, for volatile-saturated conditions the apatite X F /X Cl ratio increases and is commensurate with a decrease in its X Cl /X OH ratio and, in addition, the apatite X Cl /X OH ratio rapidly decreases with increasing X F /X OH ratio.
Following the thermodynamic models, we plotted EMPA only results obtained using a 5 μm beam or for a 2 μm beam where the apatite was oriented perpendicular to its c-axis so as to limit the effects of beam damage (Webster et al., 2009;Meng et al., 2021). Thus, having taken the data screening criteria into account, the observed correlations between ratios of X Cl /X OH and X F /X Cl in the plots (Fig. 13b-f) are interpreted to represent primary signatures. These results suggest most of the apatite analyses for the Côté Gold tonalite and the Flavrian-St-Jude trondhjemite evolved along volatile-saturated and volatile-undersaturated trajectories, respectively. The low S contents of apatite grains from the Flavrian-St-Jude trondhjemite samples are consistent with primary compositions unaffected by magmatic degassing and hydrothermal alteration. In comparison, the altered apatite grains hosted in amphibole grains in sample FLV-32 indicate evolution along a volatile-saturated trajectory (Fig. 13b, Supplementary Data Fig. A2), consistent with hydrothermal alteration of the apatite grains.
Most of the analyses for the Cl-rich primary and Cl-poor altered apatite grains from the Clifford tonalite and plagioclase-phyric porphyry evolve along volatile-undersaturated and volatile-saturated trajectories, respectively, of which the volatile-saturated trajectory for the Cl-poor altered apatite analyses supports hydrothermal alteration associated with metamorphism. Five analyses of the Cl-poor apatite grains hosted in amphibole from the tonalite evolve along a volatile-undersaturated trajectory (Fig. 13e), possibly reflecting selective preservation of pre-degassed apatite in the matrix. Nevertheless, the Cl-rich primary apatite grains are interpreted to have formed prior to magma degassing and also record their primary composition. (b) X Cl /X OH versus X F /X Cl ratios. (c-f) X Cl /X OH versus X F /X OH ratios. X F , X Cl , and X OH represent mole fractions of F, Cl, and OH in apatite, respectively. The volatile-undersaturated and volatile-saturated trajectories in (b-f) are adapted from Stock et al. (2018). (See Table 1 for sample locations and descriptions.)
The S content of the silicate melt from which apatite crystallized in each of the three mineral systems was estimated based on the S concentration of apatite in conjunction with published apatite/melt S partition coefficients with the knowledge that the S content of apatite is controlled by the S content of the melt, temperature, and oxidation state (Peng et al., 1997;Parat et al., 2011;Webster & Piccoli, 2015;Konecke et al., 2017Konecke et al., , 2019)). The apatite S concentrations for the Côté Gold diorite and tonalite and the Flavrian-St-Jude trondhjemites are below the detection limit, which precludes such evaluation. However, we note that their S contents (below detection limit) significantly lower than that for apatite micro-phenocrysts from Phanerozoic arcrelated magmas (Streck & Dilles, 1998;Imai, 2004;Scott et al., 2015), which are interpreted to reflect either low magmatic S content or a relatively reduced environment based on f O 2 estimation.
In contrast to the Côté Gold and Flavrian-St-Jude igneous suites, the primary apatite grains crystallized from the Clifford magmas yielded a relatively high pre-degassed S content (0•06 ± 0•03 wt%; 1σ, n = 9) at the model AST of 869 ± 5 • C (1σ, n = 2) and a magmatic f O 2 of FMQ ∼ + 1•5. If considering the f O 2 -T parameters are comparable with those for many Phanerozoic arc magmas, the lower apatite S contents for the Clifford rocks compared with that for Phanerozoic arc magmas (i.e. 0•15 ± 0•07 wt%, 1σ, n = 349; see compilation by Meng et al., 2021) suggest that the S content of the Clifford stock magma was slightly lower than that off the Phanerozoic arc magmas. In addition, we used the method proposed by Meng et al. (2021) that combines the apatite/melt partition coefficient for S (D S apatite/melt ) as a function of f O 2 (Konecke et al., 2019) and temperature (Parat & Holtz, 2004) to calculate the S content of the melt. This approach yielded a calculated D S apatite/melt of 2.0 and a model melt S content of 0•032 ± 0•015 wt% (1σ) for the Clifford tonalitic magmas, which is similar to the lower limit of the average S content of 0•09 ± 0•07 wt% (1σ, n = 69) for many Phanerozoic arc magmas (see compilation by Meng et al., 2021).
Zircon trace element data for representative samples collected from the study areas are illustrated in Figure 14 with the full database reported in Supplementary Data Table A2. To minimize the effects of hydrothermal alteration and contamination from xenocrystic cores and subsurface mineral/melt inclusions on the calculations and related interpretations, we filtered our data using geochemical criteria as follows: (1) La content <1 ppm (Lu, 2016); (2) LREE Index (LREE-I) = (Dy/Nd) + (Dy/Sm) > 10 ( Bell et al., 2016); (3) Ti content <50 ppm (Lu, 2016); (4) age discordance <10 %. Based on these criteria, 32 of 294 analyses (i.e. 11 %) were excluded for the following interpretation. The geochemical features of the zircons support a magmatic origin, based on the criteria of Hoskin & Schaltegger (2003), which include (1) a lack of compositional variation for each sample, (2) positive Ce anomalies, (3) moderate to significant negative Eu anomalies (Fig. 14), and (4) Th/U ratios >0•4 (rarely between 0•3 and 0•4 for the St-Jude trondhjemite).
The Hf contents and Zr/Hf ratios, which are generally used to indicate magmatic fractionation (Claiborne et al., 2006), are not observed to correlate with T Ti-Zr for most of the samples (Supplementary Data Fig. A3). This suggests that insignificant fractionation had occurred at the time of zircon crystallization, although the slight negative correlations between Hf and Zr/Hf for the Côté Gold samples are the exception. Furthermore, the T Ti-Zr versus Y contents and Th/U ratios do not define well-developed trends (Supplementary Data Fig. A3). We therefore infer that the analyzed zircons crystallized in a narrow interval of magmatic evolution and therefore presumably record homogeneous compositions.
Following the above interpretation, the methods of Loucks et al. (2020) and Ferry & Watson (2007) are used to quantify the f O 2 and temperatures for the igneous rocks. The standard error for each analysis is propagated based on standard errors of the calibrated equations and the abundances of related elements (i.e. Ce, Ti, and agecorrected U). The results are reported in Supplementary Data Table A2 and illustrated in Figure 15. The single analysis results of f O 2 and temperature for each sample overlap with each other within 2SE and yielded p-values >0•05 (most of them are above 0•80), suggesting statistically significant similarities (Supplementary Data Fig. A4). We report the average values with standard deviations (1σ) below with results summarized in Table 2. The Clifford tonalite and plagioclasephyric porphyry yielded FMQ +1•2 ± 0•4 (1σ, n = 37) and FMQ +0•2 ± 0•2 (1σ, n = 28), respectively. The lower f O 2 value of the Clifford plagioclase-phyric porphyry dike versus the coeval Clifford tonalite is interpreted to reflect assimilation of the tonalitic magmas with the surrounding volcanic rocks of the Blake River Group during magma ascent, which is presumed to be as reduced as the coeval Flavrian-St-Jude intrusion (i.e. f O 2 values of FMQ -0•3 and 0•4 above).
Zircons from each sample yielded narrow variations of T Ti-Zr with 815 ± 30 • C (1σ, n = 35) and 773 ± 34 • C (1σ, n = 31) for the Côté Gold diorite and tonalite, 736 ± 33 • C (1σ, n = 109) for Flavrian trondhjemite samples, and 809 ± 30 • C (1σ, n = 37) and 810 ± 27 • C (1σ, n = 28) for the Clifford tonalite and plagioclase-phyric porphyry, respectively. Compared with the Flavrian trondhjemite samples, the St-Jude trondhjemite yielded the lowest T Ti-Zr of 692 ± 33 • C (1σ, n = 22; Fig. 15b).
The amphibole composition data for the Flavrian trondhjemite, and Clifford tonalite and plagioclase-phyric porphyry, as well as those reported for Côté Gold quartz diorite in a previous study (Katz et al., 2017), are compiled in Supplementary Data Table A5. Data with TiO 2 < 0•95 wt% that may reflect sub-solidus re-equilibration (Houston & Dilles, 2013) were excluded for the following calculations and interpretations. Three analyses of amphibole from the Côté Gold quartz diorite are classified as magnesio-hornblende and yielded an average crystallization temperature and f O 2 of 762 ± 4 • C (1σ, n = 3) and FMQ +1•6 ± 0•3 (1σ, n = 3), respectively (Fig. 16). Ten analyses of amphiboles from the Flavrian trondhjemite are classified as ferro-edenite; the f O 2 -T estimations are therefore not applicable.
The 38 analyses of the amphibole grains in the Clifford tonalite and plagioclase-phyric porphyry are classified as magnesiohornblende. Sixteen amphibole analyses from the tonalite (after excluding six analyses with TiO 2 < 0•95 wt%) yielded average values for crystallization temperature, pressure, and f O 2 of 802 ± 18 • C (1σ, n = 16), 172 ± 22 MPa (1σ, n = 16), and FMQ +2•9 ± 0•2 (1σ, n = 16), respectively (Fig. 16). Another 22 analyses for Amp-I and Amp-II phenocrysts in the Clifford plagioclasephyric porphyry yielded tight clustering of compositional data that are indistinguishable and are treated as a single population. The average values for crystallization temperature, pressure, and f O 2 are 850 ± 14 • C (1σ, n = 22), 259 ± 7 MPa (1σ, n = 22), and FMQ +2•6 ± 0•1 (1σ, n = 22), respectively. In comparison, the broadly elevated crystallization temperatures and pressures for the amphibole phenocrysts compared with those in the tonalite suggest an earlier crystallization history for the amphibole phenocrysts (Fig. 16).
The S contents of apatite grains from the Côté Gold diorite were too low to yield detectable μ-XANES S K-edge spectra. Thus, data are presented only for samples from the St-Jude trondhjemite in the Flavrian intrusion and Clifford tonalite and plagioclase-phyric porphyry.
The μ-XANES spectra for two zircon-hosted apatite inclusions from the St-Jude trondhjemite yielded relatively broad and discernible S 6+ peaks that reflect relatively noisy spectra owing to their low S contents (i.e. below the detection limit). The μ-XANES spectra yielded an integrated S 6+ / S peak area ratio of 0•98 ± 0•02 (1σ, n = 2; Fig. 17), suggesting that the minor amount of S incorporated into the apatite structure is dominated by S 6+ . The dominance of S 6+ in the Katz et al. (2017). The FMQ values and temperatures are calculated using the method of Ridolfi et al. (2010). Pressures are estimated using the method of Mutch et al. (2016). in red). Spectra in red and pink are for primary and altered apatite grains from Clifford samples, respectively. (See Table 1 for sample locations and descriptions.) The asterisks after sample numbers indicate analyses conducted at APS, whereas the other analyses were conducted at SLS. apatite is interpreted to indicate a relatively oxidized redox state of S at the time of apatite crystallization.
The μ-XANES spectra for two zircon-hosted apatite inclusions from the Clifford tonalite (sample CS-20) yielded an integrated S 6+ / S peak area ratio of 0•97 ± 0•03 (1σ, n = 2; Fig. 17). Another apatite inclusion in a zircon grain (cut by a fracture) from the sample CS-20 has an internal texture seen in its CL image that was previously interpreted (see above) to reflect the effect of metamorphism (Fig. 7i), and yielded a slightly lower S 6+ / S peak area ratio of 0•92 (n = 1, Fig. 17). For the plagioclase-phyric porphyry (sample CS-22) from Clifford, four analyses of primary apatite hosted in the Amp-I phenocryst (Fig. 7c) yielded an integrated S 6+ / S peak area ratio of 0•86 ± 0•07 (1σ, n = 4; Fig. 17). In contrast, three analyses of altered apatite in the Amp-II phenocryst of the same sample yielded a lower integrated S 6+ / S peak area ratio of 0•72 ± 0•09 (1σ, n = 3; Fig. 17).
The μ-XANES spectra for zircon-hosted apatite inclusions from sample CS-22 acquired at the Swiss Light Source (SLS) suggest the presence of S 1+ (see Sadove et al., 2019), whereas μ-XANES spectra for apatite inclusions from the same sample acquired at the APS did not reveal the presence of S 1+ (Fig. 17). We interpret the S 1+ peak in the data from the SLS either as an artifact or as reflecting different detection capabilities for the two beamlines. More work is required to address this issue. We highlight that the μ-XANES spectra from both beamlines reveal that S incorporated in the apatite is mainly S 6+ , consistent with a relatively oxidized environment when the apatite crystallized. The primary and altered apatite grains yielded integrated S 6+ / S peak area ratios of 0•83 ± 0•05 (1σ, n = 3) and 0•73 (n = 1) at SLS, and 0•96 (n = 1) and 0•72 ± 0•13 (1σ, n = 2) at Advanced Photon Source (APS), respectively (Fig. 17). The decrease in the integrated S 6+ / S peak area ratios for the altered versus primary apatite crystals strongly indicates reduction of S by metamorphism and/or alteration. In comparison, the altered apatite in the fractured zircon (Fig. 7i) from sample CS-20 yielded an only slightly lower S 6+ / S ratio of 0•92 (Fig. 17) compared with 0•97 ± 0•03 for the texturally primary grains, which indicates limited modification of the S oxidation state owing to enclosure in the zircon.
Using the oxybarometer of Konecke et al. (2019), the S 6+ / S peak area ratios of the zircon-hosted primary apatite inclusions from the St-Jude trondhjemite yield f O 2 values of ≥ FMQ +1•2. Relevant in this regard is that the sulfide-sulfate transition in f O 2 space for Fe-poor or Fe-free felsic melts (i.e. soda-lime, K 2 Si 4 O 9 , haplo-trondhjemite liquid at 1000-850 • C and 200 MPa; Klimm et al., 2012) is reported to be 1•5 log units lower than that for Fe-rich basaltic and andesitic melts (Jugo et al., 2010;Botcharnikov et al., 2011). If we consider that sample SJ-08 has an Fe-poor trondhjemitic composition with a model AST of 859 • C and an estimated emplacement pressure of ∼90-150 MPa (Feng & Kerrich, 1990) that are comparable with values for the melts of Klimm et al. (2012), and assuming the effect of the composition of the silicate glasses is proportional to the effect of sulfur incorporation into apatite, the corrected f O 2 values from the S 6+ / S peak area ratios are ≥ FMQ -0•3.
For the Clifford stock samples, the integrated S 6+ / S peak area ratios for the primary apatite crystals from samples CS-20 and CS-22 (andesitic compositions) yielded f O 2 values of ≥ FMQ +1•2 and FMQ +0•83 ± 0•10 (1σ), respectively. One analysis without the S 1+ peak for sample CS-22 yielded an f O 2 value of FMQ +1•48, similar to that for sample CS-20. It should be noted that the sulfide-sulfate transition in f O 2 space is not sensitive to changing melt composition from basalt to andesite (Jugo et al., 2010), but has a narrower transition from ∼ FMQ +1•5 to ∼ FMQ +2•0 for dacitic melt (Kleinsasser et al., 2021), and is suggested to shift towards more oxidizing or reducing conditions with decreasing temperature or pressure ( FMQ +0•5 and+ 0•2 log units per 100 • C and 300 MPa), respectively (Matjuschkin et al., 2016;Nash et al., 2019). Sample CS-20 yielded a model AST of 866 • C and emplacement pressure of 172 ± 22 MPa (1σ), hence the P-T-corrected f O 2 value is estimated to be above FMQ +1•79 ± 0•02 (1σ; Fig. 18). For sample CS-22, which yielded a model AST of 893 • C (consistent with the calculated 850 ± 14 • C for the host amphibole phenocryst) and pressure condition of ∼259 ± 7 MPa (1σ), the P-T-corrected f O 2 value is estimated at FMQ +1•34 ± 0•10 (1σ; Fig. 18). We interpret these results to constrain the pre-degassed redox state of the Clifford tonalitic magma to ∼ FMQ +1•4-1•8.
Primary magnetite-ilmenite mineral pairs are not present in the studied samples from Côté Gold, Flavrian, and St-Jude. However, 16 analyses of two representative primary magnetite-ilmenite pairs (Fig. 8c-f) hosted in tonalite from the Clifford intrusion yielded equilibrium temperatures of 619-648 • C and f O 2 values of FMQ +2•3-4•2 (3•3 ± 1•3 on average; 1σ, n = 2; Fig. 18). The result is interpreted to record the redox state during the later stage of magma crystallization.
The new and previously published geochronological results constrain the emplacement ages of the Côté Gold, Flavrian-St-Jude, and Clifford igneous rocks to ∼2•74-2•69 Ga (Figs 3 and 11). Formation of the Côté Gold and St-Jude deposits coincides with a period of plume-arc interaction or intermittent melting of the base of the thickened basaltic crust at ∼2•75-2•70 Ga (Benn & Moyen, 2008;Bédard et al., 2013), whereas the Clifford deposit formed during a mountain-building stage in the Neoarchean southern AGB when subduction or equivalent processes are interpreted to have been in operation (Feng & Kerrich, 1992;Daigneault et al., 2002;Wyman et al., 2002). The rocks associated with the three studied deposits differ in lithogeochemical and zircon Hf-O isotopic features, which suggests variable petrogenesis (Galley, 2003;Piercey et al., 2008;Katz et al., 2017).
The Côté Gold diorite/quartz diorite samples are enriched in Cs, Rb, Ba, and LREE contents with concave to listric REE patterns, and have slight negative Eu anomalies and low to moderate Sr/Y ratios (up to 20; Figs 9 and10a) that are consistent with plagioclasedominated fractionation indicating relatively dry magmas. Zircon from the diorite and tonalite yielded comparable positive ε Hf (t) and mantle-like δ 18 O values (Fig. 12), which are consistent with the interpretation of Katz et al. (2017) that the magmas are derived from a differentiated lithospheric mantle or lower-crust mafic source. In addition, our results indicate that the sources for these magmas have not been contaminated by a crustal component that was previously altered by both low-temperature and high-temperature hydrothermal seawater or meteoric fluids.
In the Flavrian-St-Jude area, the trondhjemitic samples yielded relatively flat mantle-normalized trace element patterns except for significant depletion of Sr, P, and Ti relative to the tholeiitic diorite (Fig. 9c). The depletion of Sr, P, and Ti indicates fractionation of plagioclase, apatite, and Fe-Ti oxides during magma evolution. These features combined with the significant Eu anomalies (Fig. 9d) indicate that the magmas were sourced from a low-pressure reservoir where plagioclase was stable. The geochemistry of the suite, together with their juvenile zircon ε Hf (t) signature, supports previous arguments that the diorite-trondhjemite magma originated from a primitive mantle source (Galley, 2003).
The Flavrian intrusion was emplaced under cauldron subsidence or inferred rifted setting and drove heated seawater for hydrothermal circulation to an >8 km depth (Cathles, 1993;Piercey et al., 2000;Hannington et al., 2003;Galley, 2003). The variably low zircon δ 18 O values (3•7-4•9 ) for the Flavrian-St-Jude trondhjemite samples relative to mantle values (Fig. 12) may suggest that the magma chamber has been assimilated with a low-18 O reservoir such as hydrothermally altered wall rocks or seawater or meteoric fluids at high temperatures (Valley, 2003;Valley et al., 2005;Blum et al., 2016). However, the major hydrothermal events [i.e. for volcanogenic massive sulfide (VMS) deposit formation in the region] were driven by the Flavrian intrusion, rendering the pre-Flavrian volcanic rocks of Blake River Group poorly altered and 18 O-undepleted (Cathles, 1993;Hannington et al., 2003); therefore the low zircon δ 18 O values are less likely to have resulted from assimilation of pre-existing wall rocks, but are, rather, more consistent with the previous argument of extensive interaction of the Flavrian magma chamber with heated seawater (Hannington et al., 2003;Galley, 2003). The fact that addition of aqueous fluids can lower the solidus temperature of magmas plausibly explains the apparently lower T Ti-Zr values for these samples (Fig. 15b). The broadly reduced zircon δ 18 O values and T Ti-Zr for the St-Jude trondhjemite relative to Flavrian trondhjemite (Figs 12 and15b) plausibly reflects enhanced interaction of hightemperature seawater with the magma chamber during formation of the St-Jude deposit.
In comparison, whole-rock trace element compositions of the Clifford tonalite and plagioclase-phyric porphyry are consistent with typical subduction-zone magmas (Kelemen et al., 2007;Mohan et al., 2008;Piercey et al., 2008). The presence of amphibole phenocrysts (Fig. 6h), weakly negative to slightly positive Eu anomalies (Fig. 9f), and high Sr/Y ratios (Fig. 10a) together support suppression of plagioclase fractionation in a relatively hydrous magma. Petrological modeling indicates garnet residue in the source region for the Clifford intrusive rocks (Piercey et al., 2008). However, garnet incorporates HREE over LREE and its fractionation can thus decrease the residual melts of Dy/Yb ratios (Macpherson et al., 2006), which is not consistent with the increasing trend of Dy/Yb ratios with SiO 2 (Fig. 10b). Amphibole incorporates the middle REE relative to HREE, so the broadly decreasing Dy/Yb ratios versus SiO 2 instead support amphibole fractionation during magma evolution (Macpherson et al., 2006). We therefore interpret the results to suggest amphiboledominated fractionation, and the previously modeled garnet residue may reflect the uncertainties in the assumption for the petrological modeling.
Assimilation of wall rocks upon emplacement of the tonalitic magma in the Blake River Group volcanic rocks may have lowered the zircon δ 18 O values of the plagioclase-feldspar porphyry (Fig. 12), making the zircon isotopic values not representative of the original feature. In comparison, the slightly lower but homogeneous zircon δ 18 O values for the tonalite relative to the mantle values (Fig. 12) support a minor degree of high-temperature hydrothermal alteration of the source region (e.g. subducted basaltic oceanic crust or basaltic lower crust) or the magma chamber.
Measuring oxidation state of the ore-related magmas for the Côté Gold, Flavrian-St-Jude, and Clifford deposits has to date been hindered by modification of primary minerals by extensive alteration, deformation, and metamorphism. To address this issue, we restricted our analyses to primary minerals that preserved igneous features or to mineral inclusions protected by the primary zircon, amphibole, and plagioclase.
Zircons from the Côté Gold pre-mineralization tonalite and synmineralization diorite yielded f O 2 results of FMQ -0•7 ± 0•4 and FMQ +0•8 ± 0•4, respectively. Estimates from the published amphibole composition for quartz diorite (Katz et al., 2017) yielded an f O 2 value of FMQ +1•6 ± 0•3. Consistent with the previous argument that amphibole may overestimate magmatic f O 2 by about 1 log unit for plutonic rocks (Wang et al., 2014), our result from amphibole is also ∼1 log unit higher than the f O 2 estimated from zircon geochemistry. These results indicate an oxidation state of the ore-related Côté Gold diorite/quartz diorite of ∼ FMQ +0•8 ± 0•4.
The presence of ilmenite in zircon from the St-Jude trondhjemite indicates a relatively reduced redox state of the magma (Bell et al., 2018). The Ce-Ti-U-in-zircon and S-in-apatite oxybarometers are applicable for the Flavrian-St-Jude trondhjemite rocks and yielded internally consistent results. Zircons from the four trondhjemite samples (FLV-22, FLV-28, FLV-04, and SJ-04) yielded f O 2 values of FMQ +0•07 ± 0•6, 0•4 ± 1•0, -0•0 ± 0•6, and -0•3 ± 0•6, respectively. The S-poor apatite grains crystallized in the Fe-poor trondhjemitic melt yielded f O 2 values of ≥ FMQ -0•3 at the time of apatite crystallization. These results overlap within uncertainty and are interpreted to constrain the primary f O 2 of the trondhjemite samples to around FMQ +0.
Multiple oxybarometers have been employed for the Clifford tonalitic magmas because primary barometric minerals in the tonalite and plagioclase-phyric porphyry are demonstrated to have locally survived the low-grade prehnite-pumpellyite metamorphism. Zircon grains from the tonalite sample yielded FMQ +1•2 ± 0•4, consistent with the f O 2 result of FMQ +1•4-1•8 from the S-in-apatite oxybarometer for the primary apatite grains hosted in zircon and amphibole. In contrast, zircon grains from Clifford plagioclase-phyric porphyry may have crystallized after assimilation with relatively reduced Blake River Group volcanic rocks (presumably as reduced as the coeval Flavrian intrusion) and yielded relatively low f O 2 value of FMQ +0•2 ± 0•2.
The amphibole phenocrysts in the plagioclase-phyric porphyry are constrained to crystallize at identical P-T conditions to the equigranular tonalite, suggesting that the amphibole phenocrysts and their host inclusion may record the original f O 2 condition. Amphibole grains from the Clifford stock record oxidation states ranging from FMQ +2•5 ± 0•1 (for amphibole phenocrysts in plagioclasephyric porphyry) to FMQ +2•9 ± 0•2 (for amphibole in tonalite). We interpret the f O 2 result of FMQ +2•5 ± 0•1 to reflect the oxidation state of the less evolved Clifford magma chamber, which again exceeds the result from other oxybarometers by about 1 log unit, consistent with Wang et al. (2014). These results collectively indicate that the Clifford magmas were moderately oxidized at ∼ FMQ +1•5, which is similar to many Phanerozoic arc-related magmas (Richards, 2015). Magnetite-ilmenite pairs hosted in plagioclase from the tonalite yielded significantly higher f O 2 value of +3•3 ± 1•3 at an equilibrium temperature of 634 ± 21 • C, reflecting increased f O 2 values during the late stage of magma crystallization (Lee et al., 2005) or subsolidus re-equilibration.
These results reveal that the Clifford tonalitic magmas are more oxidized than those for the Côté Gold diorite and Flavrian-St-Jude trondhjemite, which together span a broad spectrum of FMQ +0 to FMQ +1•5 (Fig. 18) that equates to f O 2 values for mid-ocean ridge basalt (MORB) and Phanerozoic arc magmas (Kadoya et al., 2020). The f O 2 values reported here are also consistent with lower apatite S contents of Côté Gold diorite and Flavrian-St-Jude trondhjemite than the relatively oxidized Clifford tonalitic magmas. These results together lead us to conclude that there were variable f O 2 controls on TTG-related porphyry-type Cu ± Au deposit formation in the Archean.
Phanerozoic porphyry Cu ± Au deposits commonly form from magmatic-hydrothermal fluid(s) evolved from oxidized, sulfur-rich, hydrous, causative arc-related magmas that are interpreted to reflect metasomatism of the mantle wedge by slab-derived fluids (Kelley & Cottrell, 2009;Richards, 2011). Magmatic redox state is a primary control on the solubilities of Cu and Au in silicate melts (Holzheid & Lodders, 2001;Botcharnikov et al., 2011;Zajacz et al., 2012). It is generally thought that under relatively reduced conditions (< FMQ +1), voluminous sulfide saturation will deplete the silicate melt of Cu and Au, both of which may either be sequestrated in deep crustal cumulates or mantle lithosphere (Richards, 2011(Richards, , 2015;;Audétat & Simon, 2012;Lee et al., 2012;Chiaradia, 2013) or be removed during magma ascent (Richards & Şengör, 2017). If that is the case, then the derivative magmas would have limited fertility to form porphyry Cu ± Au deposits. Such conditions are hypothesized to have limited the ore-forming potential of sodic granitoids in the Archean as a consequence of recycling of the sulfate-poor seawater and reduced submarine basalts via slab descent into the mantle (Jagoutz, 2013;Stolper & Keller, 2018;, Stolper & Bucholz, 2019). However, the f O 2 constraints reported here for the causative magmas for the three porphyry-type Cu ± Au deposits are contrary to that hypothesis in two ways: (1) porphyry-type Cu ± Au deposits can form from magmas with f O 2 values covering a broad range from FMQ +0 to FMQ +1•5;
(2) relatively oxidized sulfurous sodic melts did form, at least locally, in the Neoarchean.
Case 1 (relatively reduced and sulfur-poor magmatic conditions). The St-Jude deposit formed from a relatively reduced (∼ FMQ +0) trondhjemitic melt derived from a primitive mantle source at ∼2•70 Ga. However, such reduced and sulfur-poor characteristics probably reduced the ore-forming potential of the trondhjemitic magmas. The least altered trondhjemite samples have lower Cu content (11 ± 9 ppm; n = 14) compared with Cu concentrations of ∼50 ppm commonly reported for intermediate arc rocks associated with Phanerozoic porphyry Cu deposit formation (Richards, 2015), suggesting a relatively low Cu content of the St-Jude magma. Forming a porphyry deposit from a causative magma chamber with relatively low Cu content may require more efficient partitioning of Cu from the magma by hydrothermal fluids. Because the St-Jude trondhjemite originated from a magma chamber that underwent significant modification owing to interaction with heated seawater (i.e. see interpretation based on zircon O isotope data; Fig. 12), one of the hypothesized consequences for the infiltration of the external seawater, of which the salinity in the Archean is constrained to be 1•5-2 times the modern ocean value (Knauth, 2005), is to enhance fluid saturation, volatile exsolution and the Cl concentration of the exsolved volatile phase, which would increase the efficiency of extraction of Cu from the silicate melt to the volatile phase.
Case 2 (mildly oxidized condition). The Côté Gold Cu ± Au deposit formed from dioritic/quartz dioritic magmas with an f O 2 value of ∼ FMQ +0•6-0•7 that were derived from a mildly enriched mantle source at ∼2•74 Ga. Abundant Au in the system is consistent with experimental results that demonstrate a maximum of Au solubility in silicate melt over a narrow range of redox conditions ( FMQ <1) characterized by the sulfide-sulfate transition (Botcharnikov et al., 2011;Zajacz et al., 2012). The moderate Cu content of 39 ± 21 ppm (n = 22) for the least altered diorite/quartz diorite samples indicates that Cu can be transported in the mildly oxidized melts, or voluminous sulfide saturation may have been delayed in the less evolved melts.
Case 3 (moderately oxidized condition). The Clifford Cu ± Au deposit formed from moderately oxidized (∼ FMQ +1•5), arctype tonalitic magmas at ∼2•69 Ga. The oxidized features of the tonalitic rocks are comparable with those of Phanerozoic island arc magmas (Wallace, 2005;Kelley & Cottrell, 2009;Brounce et al., 2014;Richards, 2015), which supports published models that favor subduction processes operating at least locally in the Neoarchean owing to Earth cooling (Laurent et al., 2014). The results lead us to suggest that the Clifford deposit formed from metallogenic processes similar to those that typify Phanerozoic porphyry Cu ± Au deposits. In the oxidized source magma, voluminous sulfide saturation may have been prevented, which is consistent with the Cu content of 48 ppm for the least altered tonalite sample being comparable with that for intermediate arc rocks for Phanerozoic porphyry Cu deposits.
We note, however, that the presence of oxidized, sulfurous tonalitic magmas in the Clifford stock appears to contradict the hypothesis or experimental results that suggest that sodic magmas in the Archean were either reduced or sulfur-poor (Prouteau & Scaillet, 2012;Jagoutz et al., 2013). Formation of oxidized, sulfurous tonalite magmas at Clifford can plausibly be explained by two recently proposed models: (1) local sulfate-rich seawater is recycled into the mantle (Ohmoto, 2020); or (2) aqueous fluid is a sufficiently effective oxidizing agent of the sub-arc mantle (Lacovino et al., 2020). However, our results themselves are not sufficient to resolve this issue.
Nevertheless, these cases taken together reflect variable f O 2 controls and metallogenic processes for TTG-related porphyry-type Cu-Au deposits in the Archean. However, it is worthwhile to note that: (1) less evolved sodic plutonic rocks (e.g. diorite in the Côté Gold deposit) are minor in the upper crust of Archean cratons such as in the AGB (Matthieu et al., 2020); (2) magmas with low zircon δ 18 O values, such as the St-Jude trondhjemite, are sparse in the Archean, which has been interpreted to reflect a preservation bias (Valley et al., 2005;Hammerli et al., 2018); (3) the formation and extent of oxidized and sulfurous sodic magmas remain poorly constrained. These factors together lead us to conclude that the rarity of TTG-related porphyry Cu ± Au deposits may be attributed to either local restriction of the identified favorable metallogenic conditions, exploration bias, or a preservation bias towards destruction facilitated by vertical tectonics, which is a process favored in the Archean.
In contrast to the typical association of porphyry Cu ± Au deposits with moderately oxidized, sulfur-rich calc-alkaline magmas in Phanerozoic arc-related settings, porphyry-type Cu ± Au deposits in the Neoarchean can form from sodic magmas with f O 2 conditions ranging from FMQ +0 to FMQ +1•5 under variable tectonic settings. The Côté Gold and Clifford deposits formed from mildly to moderately oxidized magmas in which early sulfide saturation is interpreted to have been variably limited. The trondhjemitic magmas of Flavrian-St-Jude yielded f O 2 values of ∼ FMQ +0 that would have favored early sulfide saturation that probably depleted the magmas of chalcophile metals (e.g. Cu) and hence reduced its fertility. However, external hydrothermal fluids (i.e. seawater) infiltrated the Flavrian-St-Jude magma chamber, which may have contributed to form the unique St-Jude porphyry-type Cu ± Au deposit owing to the addition of seawater Cl to the system, considering that Cl is the major ligand responsible for scavenging and transporting Cu from silicate melt.
Constraining the magmatic redox state for old rocks (e.g. Precambrian) is challenging because of the almost ubiquitous pervasive deformation, metamorphism, and alteration as part of Earth's protracted tectonic history. Our detailed petrographic study demonstrates, however, that primary barometric minerals protected as inclusions in other mineral phases, such as zircon, plagioclase, and mineral phenocrysts, can in fact survive from prehnite-pumpellyiteto greenschist-facies metamorphism. As such, the scope of this study, as well as methods and derivative information presented, are valuable in the context of guiding future relevant research
Downloaded from https://academic.oup.com/petrology/article/62/11/egab079/6377495 by Tsinghua University user on 16 November 2021