Introduction

Geochemical studies of gold deposits in metamorphic terranes aim to constrain the nature of mineralizing fluids, metal reservoirs, and ore-forming processes (Goldfarb and Groves 2015). However, using geochemical data to achieve this is not always straightforward. For example, mass-dependent sulfur isotope fractionation is influenced by intensive fluid properties (T, fO2, and pH; Ohmoto and Rye 1979; Seal 2006), which can produce δ34S values in ore-stage sulfides that are non-characteristic of their source. A recent example of this is Schlegel et al.’s (2017) study of the Prominent Hill IOCG deposit where δ34Spyrite varies from − 34 to + 30‰ despite the original fluid having an inferred δ34S of about 4‰. This effect in part can account for some of the significant inter- (e.g., McCuaig and Kerrich 1998) and intra- (e.g., Godefroy-Rodríguez et al. 2020) deposit variance in δ34S sulfide values not uncommon in orogenic gold deposits. Thus, geochemical tracers sensitive to changes in the chemical and physical properties of a fluid are limited in uniquely defining a link between a fluid’s source and the site of ore formation. Trace metal associations are another parameter commonly considered, but are often characterized by a similar Au-Ag ± As ± B ± Bi ± Sb ± Te ± W association for geologically dissimilar syenite-associated, orogenic, and intrusion-related gold deposits (Robert 2001; Groves et al. 2003; Goldfarb et al. 2005; Hart 2007). Such apparently similar metal associations in different deposit types largely obscure their genetic significance. Laser ablation inductively coupled plasma-mass spectrometry (LA-ICP-MS) elemental mapping of sulfides has also introduced cautionary aspects, such as the realization of a protracted element paragenesis and likely remobilization and decoupling of Au from other elements (Neyedley et al. 2017; Gourcerol et al. 2018a, b; Kerr et al. 2018), which appears to be the norm rather than exception in orogenic gold systems (Kontak et al. 2018).

Vein quartz and carbonate δ18O values are considered to be relatively consistent among orogenic gold deposits (e.g., Kerrich 1989; McCuaig and Kerrich 1998; Goldfarb and Groves 2015). However, compilations suggest fluid types and/or vein-forming conditions (e.g., temperature, fluid mixing) are in fact variable, as it has been shown that significant δ18Oquartz variation occurs in both gold districts (e.g., ~ 5‰ in Val-d’Or; Beaudoin and Pitre 2005) and within the orogenic deposit class in general (e.g., ~ 20‰; Beaudoin 2011). Furthermore, the calculated δ18OH2O values generally overlap fields for magmatic and metamorphic waters (Kerrich 1989; McCuaig and Kerrich 1998), which causes uncertainty in fluid source identification. Overprinting hydrothermal events may present further difficulties, as the traditional use of bulk isotopic analysis values may reflect multiple fluid types.

The above discussion highlights how geochemical data for gold deposits in metamorphic terranes can be equivocal (Groves et al. 2003; Goldfarb et al. 2005; Goldfarb and Groves 2015). Here, we provide further insight into this phenomenon by using multiple geochemical methods to characterize three geologically well-constrained hydrothermal events in the ca. 1.1 Moz Au Missanabie-Renabie gold district (Wawa, Ontario, Canada; Fig. 1). These hydrothermal events are well-documented by recent studies (McDivitt et al. 2017, 2018): (1) pre-regional metamorphic, pre-D1, gold-bearing laminated quartz veins with phyllic alteration halos that account for the majority of gold endowment (the Au1 event); (2) post-peak metamorphic, syn-D3, gold-barren quartz ± epidote ± chlorite ± K-feldspar ± hematite veins related to extensive zones of hematite-bearing alteration (the Retrograde event); and (3) late-stage, post-peak metamorphic, syn- to post-D4 gold-bearing pyrite veins (the Au2 event). Whereas the study of McDivitt et al. (2017) constrains the different hydrothermal events within the structural framework of the Missanabie-Renabie gold district, the study of McDivitt et al. (2018) assesses the hydrothermal events from the perspective of fluid chemistry, alteration characteristics, and metamorphic timing. Results of these earlier studies illustrate that the Au1 event records the ingress of a H2O-CO2, Au-Ag-Bi-Cu-Mo-Pb-Te-W-Zn-S and large ion lithophile element bearing fluid into early (i.e., pre-D1), brittle, pre-metamorphic structures related to potassium-poor granitoid magmatism. In contrast, the Retrograde event reflects a post-peak metamorphic, Na-S-H2O fluid event that reactivated earlier brittle structures during later D3 deformation. Lastly, the Au2 event is represented by gold-bearing pyrite veins that overprint phases of the Retrograde event during or after a D4 reactivation of D3 structures. Because the different hydrothermal events contrast in their metal signatures and fluid chemistries, as well as their structural and metamorphic timing, they are interpreted to record different genetic processes. For example, because the Au1 event is localized to pre-metamorphic and pre-orogenic brittle structures related to potassium-poor granitoids, an intrusion-related origin is inferred (McDivitt et al. 2018). In contrast, D3 and D4 are both post-peak metamorphism and late-stage brittle-ductile deformation events thereby favoring an orogenic affiliation for the Retrograde and Au2 events (McDivitt et al. 2018) Here, we geochemically characterize these three different and contrasting hydrothermal systems using: (1) δ34S and Δ33S values of pyrite; (2) trace-metal signatures using whole-rock analysis and LA-ICP-MS elemental mapping of pyrite; and (3) in situ secondary ion mass spectrometry (SIMS) for δ18Oquartz. We also report in situ LA ICP-MS U-Pb geochronology results for titanite related to the Retrograde event to constrain its age.

Fig. 1
figure 1

Simplified regional geology map of the Wawa gold camp (modified after Williams et al. (1991) and Stott et al. (2010)). Locations of gold mines with current or past production are from the Ontario Mineral Deposit Inventory hosted on the web site of the Ontario Geological Survey. Inset displays the location of the main figure relative to the Superior craton

A corollary but related intent of this study is to use our δ18O data to further assess the similarities among deposits of the Wawa gold camp highlighted in McDivitt et al. (2017, 2018). Particularly relevant based on earlier studies (Callan 1991; Samson et al. 1997; Haroldson 2014) is that vein δ18Ocarbonate values across the camp tend to be low when compared to gold deposits of the Abitibi greenstone belt (Kerrich 1990a). To provide insight into these δ18O values, we evaluate the results in the context of our in situ SIMS quartz analysis and fluid inclusion δD data from quartz to test potential origins of this 18O-depleted hydrothermal fluid.

Geological setting and gold mineralization

From 1941 to 1991, ~ 1.1 million ounces of gold were produced from shear zones containing laminated quartz veins in the Renabie mine area (Fig. 2a, b; Turek et al. 1996; Callan and Spooner 1998). This historic production was sourced from the Renabie underground mine and the C-zone, Nudulama, and Braminco open pits (Fig. 2b), which are hosted in 2720.8 ± 1.4 Ma biotite tonalite of the Missinaibi Lake batholith (MLB; Kamo 2015; Robichaud et al. 2015). The batholith is a plutonic component of the Wawa gneiss domain, which is a predominantly amphibolite-facies region of felsic gneisses, granitoid plutons, and minor supracrustal rocks. The granitoid plutons and supracrustal rocks range in age from ca. 2.92 to 2.66 Ga, with deformational/metamorphic events having occurred between ca. 2.70 to 2.64 Ga (Moser 1989, 1993, 1994; Percival and West 1994). The Wawa gneiss domain terminates to the east at the subprovince-bounding Kapuskasing structural zone, which is a linear, amphibolite- to granulite-grade metamorphic belt exposed in a Neoarchean to Paleoproterozoic intracratonic uplift (Percival and West 1994). The Kapuskasing structural zone records a protracted deformational and metamorphic history from ca. 2.70 to 2.50 Ga, with metamorphic U-Pb zircon ages representing granulitization from ca. 2.60 to 2.58 Ga, and U-Pb titanite ages recording progressive cooling from ca. 2.60 to 2.50 Ga (Moser 1989, 1993, 1994; Percival and West 1994). The transition from the Kapuskasing structural zone across the Wawa gneiss domain into the Michipicoten greenstone belt (MGB; Fig. 1) has been considered an oblique crustal cross section (Percival and Card 1985), with the Kapuskasing structural zone representing the deep-crustal equivalent of the MGB.

Fig. 2
figure 2

a Location and simplified geology map of the Missanabie-Renabie gold district (modified after Bruce and Horwood (1942), Riley (1971), and Bennett (1972)). Locations of gold occurrences are from the Ontario Mineral Deposit Inventory hosted on the web site of the Ontario Geological Survey. b Geology map of the Renabie mine area (modified after Ferguson (1968) and McDivitt et al. (2015)). Coordinates in NAD83 UTM zone 17N

The MGB is dominated by greenschist-facies metavolcanic rocks formed during three principal bimodal volcanic cycles (ca. 2.90, 2.75, and 2.70 Ga; Sage 1986; Turek et al. 1982, 1984; Sage and Heather 1991; Turek et al. 1992; Heather and Arias 1992). Granitoid plutons of various ages intruded the MGB: (1) the 2745 ± 3 Ma dioritic- to granodioritic Jubilee stock (Sullivan et al. 1985); (2) the 2720.8 ± 1.4 Ma biotite tonalite of the MLB; and (3) the compositionally variable granitoids of a ca. 2.68 to 2.66 Ga intrusive suite (Turek et al. 1996; Robichaud et al. 2015).

Other granitoid-hosted deposits of the Wawa gold camp are in plutons which intrude the MGB. These include the Surluga, Jubilee, and Minto deposits hosted in the Jubilee dioritic stock (Samson et al. 1997), and the Magino deposit hosted by the undated Webb Lake trondhjemitic stock (Fig. 1; Haroldson 2014). Features of these mineralized sites are considered to be similar to those in the Missanabie-Renabie district (McDivitt et al. 2017, 2018): (1) gold-bearing, pre-orogenic quartz which host low- to moderate-salinity H2O-CO2 fluid inclusions and their unmixed counterparts; (2) these are overprinted by syn-orogenic, gold-barren breccia veins and hematite-bearing alteration assemblages characterized by CO2-poor aqueous fluids of variable salinities; and (3) fracture-hosted, auriferous pyrite cross-cuts the earlier hydrothermal events in the Missanabie-Renabie gold district and at Magino. In the Missanabie-Renabie gold district, the pre-orogenic, syn-orogenic, and late-stage, pyrite-associated mineralization styles correspond to the Au1, Retrograde, and Au2 events, respectively. Correlative hydrothermal events at the Jubilee stock and Magino deposits are considered to represent regional equivalents.

Analytical methods

Polished thin sections and epoxy resin mounts containing quartz and/or pyrite from the different vein types were examined in reflected and transmitted light. These samples were further characterized on a JEOL 6400 scanning electron microscope (SEM) coupled to an Oxford Sight energy dispersive X-ray spectrometer (EDS) at the Micro-Analytical Centre (MAC), Laurentian University. Analytical conditions employed an accelerating voltage of 20 kV, a 1.005 nA beam current, a working distance of 15 mm, and an acquisition time of 5 s. Cathodoluminescence (CL) imaging on a single quartz vein sample from the Retrograde event was completed using a Gatan ChromaCL™ detector attached to it with a linear array of photomultiplier tubes with 16 separate photocathodes.

Representative gold mineralized samples were analyzed for trace metals at the Geoscience Laboratories of the Ontario Geological Survey, Sudbury, Ontario. Sample location information and analytical data is provided in McDivitt (2016). Samples comprised ~ 100 cm3 of rock material and consisted of Au1 laminated veins and phyllic alteration (analyzed separately) and Au2 pyrite veins. Both vein and alteration material was separated from adjoining wall rock. Samples were pulverized and dissolved in an open-vessel agua regia digestion and concentrations were determined by ICP-MS (IML-100 technique; McDivitt 2016). For the purpose of plotting, element concentrations reported below analytical are shown as one-half the detection limit value, which are as follows (in ppm): Au (0.002), Ag (0.01), As (0.8), Bi (0.01), Cu (0.7), Mo (0.03), Pb (0.2), Sb (0.06) Te (0.01), U (0.01), W (0.02), and Zn (3).

One representative pyrite grain from each of the Au1 and Au2 events was analyzed using LA-ICP-MS in traverse mode to generate element maps and assess gold-trace metal associations. Elements analyzed were Ag, As, Au, Bi, Cd, Co, Fe, In, Mo, Ni, Pb, Sb, Sc, Te, U, W, and Zn. This work was done at the Geochemical Fingerprinting Lab of Laurentian University following the details outlined in the Electronic Supplementary Materials (ESM Table 1) and elsewhere (Neyedley et al. 2017; Gourcerol 2018a, b; Kerr et al. 2018). In addition to generating element maps, the LA-ICP-MS traverse data was broken down into time-slice domains or TSD which allow for data representing element concentrations over specific time intervals (i.e., 0.23 s for one complete analysis) to be displayed graphically on scatter plots (Gourcerol et al. 2018a, b). As discussed elsewhere (Gourcerol et al. 2018a, b), the error associated with the TSD approach is estimated to be < 20%. Regions of stepped data occur at low elemental concentrations when the TSD data is plotted on scatter plots (e.g., Gourcerol et al. 2018a, b). These regions of stepped data define domains in which the data is close to or below the lower limits of detection. The stepped nature of low-concentration data is due to the pulse counting detector and low backgrounds combined with variable intensities from the pyrite. The TSDs that returned negative concentration values were assigned a value of 0.1 ppm for the purpose of plotting and excluded from the statistical analysis described below.

Pyrite samples were powdered using a Dremel™ drill and analyzed at the Manitoba Isotope Research Facility, University of Manitoba, Winnipeg, Canada. Samples were analyzed by a Costech™ 4010 Element Analyzer coupled to a Thermo Finnigan™ Delta V Plus isotope-ratio mass-spectrometer via an open-split interface. Each sample was analyzed three times non-consecutively as a test of reproducibility. Isotopic ratios are reported as δ34S and δ33S (‰) relative Vienna Cañon Diablo Troilite (VCDT). The international standards IAEA-S1, IAEA-S2, and IAEA-S3 were used as calibration standards and analyzed at the beginning, middle, and end of each run. Least-squares linear regression of the known and measured standard values was used to produce a calibration line. Anderson pyrrhotite was used as an internal standard and repeated analysis of this standard yielded uncertainties (1σ) of 0.2‰ for δ34S and 0.3‰ for δ33S. The uncertainty for Δ33S is estimated to be 0.1‰ by the standard error of the mean Δ33S value calculated for the internal standard.

Vein quartz was crushed into mm-size fragments and hand sorted under a binocular microscope to avoid impurities. Select grains were mounted in epoxy resin with ~25 grains per mount. Spot locations selected for SIMS O-isotope analysis were chosen based on areas that lacked fractures and mineral inclusions. In addition, a euhedral crystal of quartz from the Retrograde event with primary fluid inclusions was cut perpendicular to its C-crystallographic axis and investigated along a detailed SIMS O-isotope traverse following its CL imaging. The SIMS δ18O analysis was completed at the University of Manitoba Isotope Research Facility using a CAMECA 7f ion microprobe with a 3 nA cesium (Cs+) primary beam accelerated to 10 kV and focused to a 20 μm spot size. An offset voltage of 300 V was used to eliminate molecular ion interferences. Ions were detected with a Balzers SEV 1217 electron multiplier coupled with an ion-counting system using a dead time of 52 ns. Both 18O and 16O were detected by switching the magnetic field. The results are presented using the δ-notation (‰) relative to Vienna Standard Mean Ocean Water (VSMOW). Internal standards used for calibration were analyzed at multiple times during the analysis; the spot reproducibility on the standards ranged from 0.4 to 0.5‰. The 1σ error value associated with unknown measurements is estimated to be 1.2‰.

Two samples of quartz from the Retrograde event at Nudulama East were analyzed at the Facility for Isotope Research at Queen’s University for H isotopic composition. Fluid inclusions from the same samples were studied by McDivitt et al. (2018). Fluid inclusions within the quartz comprise two-phase (liquid-vapor) H2O inclusions with ~ 10–20 vol% vapor and occur as primary, pseudosecondary, secondary, and indeterminate fluid inclusion assemblages. Samples were weighed into silver capsules, degassed at 100 °C for 1 h, and then loaded into a zero-blank auto sampler. The hydrogen isotopic composition was measured using a Thermo-Finnigan thermo-combustion element analyzer coupled to a Thermo-Finnigan Deltaplus XP Continuous-Flow Isotope-Ratio Mass Spectrometer. Values are reported relative to VSMOW with a precision of 3‰.

Measurements for in situ U-Pb geochronology of titanite (n = 86) within hematite-bearing Retrograde event alteration (Sample NUE-28C) was conducted using LA-ICP-MS. Prior to analysis, the grains were imaged in BSE mode on the SEM. The LA-ICP-MS analysis was completed at the Geochemical Fingerprinting Lab of Laurentian University using the operating conditions and data processing procedures given in ESM Table 2. Titanite OLT-1 (1014.8 ± 2 Ma; Kennedy et al. 2010) was used as a calibration standard and titanites BLR-1 (1047.1 ± 0.4 Ma; Aleinikoff et al. 2007) and FCT (28.53 ± 0.05 Ma; Schmitz and Bowring 2001) were used as secondary standards. The secondary standards were analyzed 6 time each throughout the run and returned ages of 1033 ± 18 Ma (BLR-1) and 29.6 ± 1.2 Ma (FCT).

Vein-hosted quartz, pyrite, and alteration-related titanite: defining features and textural characteristics

The analyzed materials were collected from mineralized outcrops in the MGB (Pileggi No. 1; Fig. 2a) and the MLB (Baltic D, C-zone, Nudulama, and Nudulama East; Fig. 2b). The different quartz and pyrite generations are displayed in the context of hydrothermal events and the camp’s deformational framework in Fig. 3 (McDivitt et al. 2017, 2018).

Fig. 3
figure 3

a An integrated paragenesis illustrating how different vein types and key mineral phases in this study relate to the Au1, Retrograde, and Au2 hydrothermal events. Relationships to deformation are after McDivitt et al. (2017). b A schematic summary diagram emphasizing the hybridized nature of the ore zones. The diagram illustrates how the Au1, Retrograde, and Au2 vein and alteration types relate to structural features (McDivitt et al. 2017). Also shown is a summary of isotopic characteristics of the Au1, Retrograde, and Au2 events. The diagram is largely based on the Nudulama East outcrop

Au1 laminated quartz veins

Quartz (Qtz1) crystals range from equidimensional, coarser crystals (~ 1–2 mm) with well-defined triple junction boundaries to smaller (~ 0.1–0.2 mm), irregular, amoeboid-shaped crystals with lobate to cuspate boundaries. The smaller crystals occur in association with sericite domains or lithons, which give Au1 veins their characteristic laminated texture (Fig. 4a, b). Pyrite (Py1) within these veins is fine- to medium-grained, euhedral to subhedral, and commonly hematite altered on its margins (Fig. 4c). Magnetite occurs in textural equilibrium with the late-stage hematite. Quartz is the dominant mineral in Au1 veins (~ 80%), with the lesser sericite, carbonate, and accessory titanite, allanite, epidote, tellurides, and sulfides (pyrite, chalcopyrite, molybdenite, galena, sphalerite). These veins are surrounded by phyllic alteration halos (quartz (~ 30%)-sericite (~ 65%)-pyrite (~ 2–5%)) containing the same accessory mineralogy as the laminated veins.

Fig. 4
figure 4

Polished slab, photomicrograph, and outcrop images of the different vein types and mineral phases analyzed in the study. a Polished slab of a gold-bearing, laminated Au1 vein with saccharoidal domains of quartz (Qtz1), sericite laminations (dark sections), and sulfide-rich zones (brown sections). b Photomicrograph (cross-polarized light (CPL)) of a gold-bearing, laminated Au1 quartz vein comprising Qtz1 and sericite laminations. c Photomicrograph (combined reflected light and transmitted plane-polarized light) of hematite-replaced pyrite (Py1) encased within quartz (Qtz1) on the margin of a sericitic lamina comprising sericite and epidote (Ser + Ep). Inset in the lower right shows a close-up, reflected light photomicrograph of hematite (Hem)-replaced Py1. d Photograph from the Nudulama East outcrop where a hematite-associated alteration (HAA) breccia vein overprints a shear zone-hosted laminated quartz vein (LQV). Note the red hue in the clasts of the HAA breccia vein; this reflects chlorite-albite-titanite-hematite and K feldspar-hematite alteration assemblages formed after biotite tonalite of the MLB (McDivitt et al. 2018). Compass for scale (7 cm wide). e A large clast of Qtz1 from a shear zone-hosted laminated quartz vein at the Nudulama East outcrop included within a hematite-bearing breccia vein that formed during the Retorgade event (i.e., the Retrograde breccia). The clastic component of the breccia comprises chlorite-albite-titanite-hematite and K feldspar-hematite alteration assemblages formed after biotite tonalite of the MLB (McDivitt et al. 2018). The clasts are surrounded by Qtz2 which occurs as the dominant matrix component of the breccia. Pencil for scale (12 cm long). f Photomicrograph (CPL) of euhedral quartz (Qtz2) within the HAA breccia vein shown in d. The zones interstitial to the quartz that are fine-grained and of a higher birefringence correspond to the altered, hematite-bearing clasts also depicted in d. The Qtz2 crystal in the upper-central portion of the image outlined by the white, dashed line was the subject of cathodoluminescence-coupled SIMS δ18O analysis and is further shown in Fig. 10e. h NW-striking quartz (Qtz3) tension gashes hosted by mafic- to intermediate-volcanic rocks of the Pileggi No. 1 outcrop. Note the earlier, isoclinally folded quartz (Qtz1) vein. Pencil for scale (12 cm long). g Massive, gold (Au2)-bearing pyrite (Py2) vein at Nudulama East. Compass for scale (7 cm wide). The inset in the upper right is a photomicrograph (CBL) of Py2 overgrowing hematite-bearing K-feldspar alteration of the Retrograde event in wall rock marginal to the vein

Retrograde hematite-associated alteration veins

The Retrograde event veins are associated with hematite-bearing alteration assemblages and comprise both discrete-planar and breccia veins that overprint and include phases of the Au1 event (Fig. 4d, e; McDivitt et al. 2017). Vein fill consists of quartz (Qtz2), chlorite, epidote, K-feldspar, and earthy-red hematite (Fig. 4f). These phases are commonly euhedral and locally define comb-like textures, with quartz and epidote generally medium- to coarse-grained whereas chlorite is a late-stage, fine-grained infill. The modal proportions of infill minerals vary, with hematite and K-feldspar the least abundant (typically < 20%), and quartz, chlorite, and epidote the most abundant (locally to 100%). These veins are associated with zoned alteration halos comprising proximal K-feldspar-hematite alteration and distal chlorite-titanite-albite-hematite alteration where titanite forms fine-grained aggregates in chloritized biotite of the MLB (Fig. 5a).

Fig. 5
figure 5

a An SEM-BSE image displaying chlorite-titanite-albite-hematite alteration developed in wall rock to Retrograde veins at Nudulama East. The central portion of the BSE image is overlain by a Ti and Mg EDS map displaying intergrown titanite and chlorite. The white, circular spots display the locations of U-Pb LA-ICP-MS analytical points. b Results of U-Pb LA-ICP-MS titanite analysis (n = 86) depicted on a Tera-Wasserburg concordia diagram. c Reduced U-Pb LA-ICP-MS titanite dataset (n = 38) and the age determination resulting from a common-Pb anchored regression (2586 ± 21 Ma; MSWD = 1.7). d Results of a 207Pb-corrected 238U/206Pb weighted mean using the same 38 data points as in D (2580 ± 21 Ma; MSWD = 1.8)

Au2 massive pyrite veins and quartz tension gashes

The Au2 veins consist of fine- to medium-grained, anhedral pyrite (Py2), which commonly displays jagged- to serrated-grain boundaries (Fig. 4g). These veins lack alteration halos, but pyrite is commonly disseminated in the adjacent wall rocks. Quartz (Qtz3) is the majority infill phase (90–95%) of NW-striking quartz-tension gashes at the Pileggi No. 1 outcrop (Fig. 4h), with lesser, brown, granular oxide and fine-grained pyrite accounting for the remainder. Quartz crystals are typically elongate in thin section (~ 1–5 mm), display undulatory extinction, and have irregular- to jagged-grain boundaries. Micro-fracturing is abundant in some of the quartz, in particular where zones of cataclasite are present. The Qtz3 tension gashes do not carry significant gold grade (avg. 0.017 ppm Au; n = 10), but they are similar in orientation and structural timing to the Au2 pyrite veins (McDivitt et al. 2017). Therefore, for the purpose of this study Qtz3 is assigned to the Au2 hydrothermal event (Fig. 3).

Analytical results

Titanite U-Pb geochronology

The results of 86 U-Pb titanite spot analyses are shown on a Tera-Wasserburg concordia diagram in Fig. 5b and given in the ESM (Table 3). Calculations and plots were completed in Isoplot v. 3.76 (Ludwig 2003). The overall dataset displays a high degree of scatter, and discordance due to both Pb loss and common Pb incorporation (Storey et al. 2006; Kirkland et al. 2017). The dataset was filtered on the basis of percentage discordance (< 0% and < 20% excluded) and uranium concentration (< 9 ppm U excluded). Two points were excluded on the basis of large 2σ error values. Five additional points were excluded on the basis of spot overlap with adjacent mineral phases. An unanchored regression of the reduced dataset (n = 38) yields an age of 2577 ± 51 Ma (95% confidence; MSWD = 1.8) with a 207Pb/206Pb intercept of 0.96 ± 0.25 (95% confidence). The 207Pb/206Pb intercept is in agreement with a common 207Pb/206Pb value of 1.0795 inferred from host rock age (2720.8 ± 1.4 Ma biotite tonalite) and the terrestrial Pb evolution model of Stacey and Kramers (1975). The latter is used in an anchored regression (Fig. 5c) and in a 207Pb-corrected 238U/206Pb weighted mean calculation (Fig. 5d), which return respective ages of (95% confidence) of 2586 ± 21 Ma (MSWD = 1.7) and 2580 ± 21 Ma (MSWD = 1.8).

Whole rock and LA-ICP-MS trace metal associations

Samples analyzed for trace metals are plotted in Fig. 6. In general, the data indicate (1) significant range in Au from 0.004 to 27.05 ppm for Au1 samples and 0.04–3.99 ppm for Au2 samples (Fig. 6a); (2) similar range in Ag from 0.01–11.83 ppm for Au1 samples and 0.53–4.34 ppm for Au2 samples (Fig. 6a); (3) Au1 and Au2 samples are typically enriched in Bi (up to ~ 27 ppm; Fig. 6b) relative to Au1 veins from Pileggi No. 1 (max = 0.07 ppm); (4) a large range in Te and Pb ( ~0.01/0.2 to ~ 40 ppm; Fig. 6c, d) and a restricted range for U (0.01 to 2.53 ppm; Fig. 6e), with Au1 veins from Pileggi No. 1 relatively depleted in Te (max = 0.33 ppm; Fig. 6c), Pb (max = 0.9 ppm; Fig. 6d), and U (max = 0.05 ppm; Fig. 6e); (5) select samples of the Au1 event are markedly enriched in Mo (to > 350 ppm upper detection limit; Fig. 6f); and (6) largely overlapping values among different sample populations for Cu, W, and Zn, but with W generally low (most < 1 ppm), and most Zn and Cu values range from 1 to 100 ppm (Fig. 6g, h). The Pileggi No. 1 samples are enriched in Cu (to 233 ppm; Fig. 6h) relative to most other samples. In addition, Sb (not plotted) is below detection (0.06 ppm) in all samples and As (not plotted) is either below detection (0.8 ppm) or present only slightly above detection (max = 4.5 ppm). It is therefore apparent that for these samples the Au-Sb-As association, so common in other gold systems, is lacking.

Fig. 6
figure 6

Log10 bivariate scatter plots of whole-rock data (ppm) from mineralized samples

Pearson product-moment correlation coefficients and associated p values were calculated for different sample populations and illustrate the following (Table 1): (1) well-defined (0.58–0.99), significant (p < 0.05) correlations between Au and Ag, Bi, Pb, Te, and Mo in the Au1 vein and phyllic alteration samples hosted in the MLB; (2) lower, non-significant (p > 0.05) correlation coefficients (0.04–0.54) between Au and Ag, Bi, Mo, Te, and Pb in Au2 samples; and (3) a strong Au-Ag (0.97; p < 0.05) in Au1 veins from the Pileggi No. 1 outcrop, but correlation coefficients between Au and Te, Pb, Mo, and Bi are lower (0.07–0.81) and nonsignificant (p > 0.05).

Table 1 Pearson product-moment correlation coefficients determined from log10 trace element concentrations in whole-rock (WR) and LA-ICP-MS pyrite data. Probability values (P) for the correlation coefficients are shown; those < 0.05 are in italics and those > 0.05 shown in bold

Element maps and corresponding trivariate scatter plots of TSD data for representative samples of Py1 and Py2 are shown in Figs. 7 and 8, respectively. The element map for Py1 (Fig. 7a) shows a central area relatively enriched in Ni (Fig. 7b) with a rim enriched in Co (Fig. 7c) but with obvious primary growth zones lacking. The remaining elements shown all lack spatial correlation with Ni and Co and instead define local or spotty enrichment zones (Fig. 7d–j). It is likely that at least some of the Au, Ag, Bi, Te, and Pb enrichment records the presence of micro-inclusions of Au-bearing tellurides with associated Ag, Bi, and Pb, as shown by SEM-EDS analysis (Fig. 7k, l). In addition to Au, other elements (Ag, Bi, Te, Pb, U, W) also define narrow, linear zones of enrichment (Fig. 7d–j). Similar relationships are evident in the trivariate scatter plots of Fig. 7, which show positive covariation trends between Au and Ag (Fig. 7m–o), Bi (Fig. 7p), and Te (Fig. 7q); in contrast, Pb displays a weak but positive covariation with Au (Fig. 7r) and neither U nor W show an association with Au (Fig. 7s, t).

Fig. 7
figure 7

Results of LA-ICP-MS element mapping of pyrite from the Au1 event (Py1; aj). The presence of Au1 tellurides is shown in the SEM-BSE image of the core of the mapped pyrite grain (kl). The Au1 tellurides are also shown in red on the schematic map of the pyrite grain, which summarizes the salient features of the grain. The plots (mt) are trivariate element plots of time-slice data (the third element is denoted in brackets, with its concentration indicated by the size and color of the points shown on the legend). Points that plot along the ordinates and abscissas at 0.1 ppm are below detection data

Fig. 8
figure 8

Results of LA-ICP-MS element mapping of pyrite from the Au2 event (Py2; aj). The plots (kr) are trivariate element plots of time-slice data (the third element is denoted in brackets, with its concentration indicated by the size and color of the points shown on the legend). Points that plot along the ordinates and abscissas at 0.1 ppm are below detection data

The element map for Py2 (Fig. 8a) contrasts with Py1 in that there is a subtle zoning defined by Co and Ni (Fig. 8b, c). In addition, there is noted enrichment in the central part of the pyrite crystal by Au (Fig. 8d) and other elements (Ag, Bi, Te, Pb, U, and W; also Zn but not shown; Fig. 8e–j). The trivariate plots also document Au-Ag (Fig. 8k–m) and Au-Bi-Te (Fig. 8n, o) associations in the form of strong and moderate- to weak-, positive covariations, respectively. In contrast, Pb, U, and W lack covariation with Au (Fig. 8p–r).

Pearson product-moment correlation coefficients were calculated from the TSD data using log10 values in the same manner as for whole-rock data (Table 1); a graphical comparison of the results is shown in Fig. 9. Significant (p < 0.05) correlation coefficients include values > 0.7 defined by Au-Ag, Au-Bi, and Au-Te in Py1, with the Au-Pb correlation lower (0.37; p < 0.05) and no Au-Mo correlation (− 0.03; p < 0.05). In Py2 there is a significant Au-Ag correlation (0.66; p < 0.05), and Au-Bi, Au-Pb, and Au-Te correlations are lower (0.36–0.45; p < 0.05), with no Au-Mo correlation (0.08; p < 0.05).

Fig. 9
figure 9

Comparative diagram displaying log10 Pearson product-moment correlation coefficients (relative to Au) of mineralized sample populations that are representative of the Au1 and Au2 events

Stable isotopes

Sulfur isotope analysis (Table 2) for Py1 and Py2 returned similarly negative δ34S average values (Py1 = − 5.5‰ ± 0.2‰ (1σ) n = 3; Py2 = − 3.5‰ ± 0.3‰ (1σ) n = 12) and δ33S average values (Py1 = − 3.2‰ ± 0.2‰ (1σ) n = 3; Py2 = − 2.1‰ ± 0.3‰ (1σ) n = 12). The δ34S values are consistent with Callan and Spooner (1998) for pyrite from the Renabie mine (δ34SAvg = − 5.7‰ ± 0.9‰ (1σ) n = 8). Average Δ33S values are similar for both Py1 (− 0.4‰ ± 0.1‰ (1σ) n = 3) and Py2 (− 0.3‰ ± 0.2‰ (1σ) n = 12).

Table 2 Sulfur isotope values (δ34S and δ33S) and calculated Δ33S values for Au1 event pyrite (Py1) and Au2 event pyrite (Py2). δD values returned from analysis of Retrograde event Qtz2

The results of SIMS δ18Oquartz analysis for Qtz1, Qtz2, and Qtz3 are summarized in Fig. 10 (a, b, and c, respectively; ESM Table 4). Whereas the δ18O values of Qtz1 range from 3.9 to 13.4‰ and approximate a standard normal distribution (avg. = 8.4‰; Fig. 10a), the data for Qtz2 (avg. = 6.7‰; Fig. 10b) define a negatively skewed distribution with a larger proportion of values lower than 6‰, with minimum and maximum values of 1.2‰ and 9.7‰, respectively. The δ18O values in Qtz3 are similar in value and distribution to Qtz1 and range from 5.8‰ to 10.9‰ (avg. = 8.0‰; Fig. 10c). For the SIMS transect of a Qtz2 crystal (Fig. 10e, f), the data are summarized in both a histogram (Fig. 10d) and superimposed on the grain image (Fig. 10f). Results for the transect indicate a range for δ18O values of 5.4‰ to 12.6‰ (Fig. 10d) with a general trend of lower values in the core and higher values towards the rims (Fig. 10e, f). There is no apparent relationship between the areas of markedly different CL response and δ18O values. Hydrogen isotope analysis from the two samples of Qtz2 (Table 2; Fig. 10e)) yielded identical δD values of − 37‰.

Fig. 10
figure 10

Histograms displaying the results of SIMS δ18O analysis for the different quartz generations (Qtz1, Qtz2, Qtz3) for mounted grains (ac) and the cathodoluminescence (CL) imaged Qtz2 crystal (d); n = number of analyses. e A photomicrograph (plane-polarized light) of the CL-imaged Qtz2 crystal cut perpendicular to its C-axis. The crystal displays primary growth zones that parallel assemblages of 2-phase (liquid-vapor) H2O fluid inclusions which yield δD values of − 37‰. The black rectangle indicates the location of CL image in f. f CL image of the Qtz2 crystal shown in e with the white spots and corresponding values denoting the locations of SIMS spots and δ18O values, respectively

Discussion

A geochemical comparison of the hydrothermal events

When the trace-metal characteristics of Au1 samples from the MGB are compared to those from the MLB, geochemical contrasts are evident that may represent local host-rock effects on fluid chemistry (Polito et al. 2001; Evans et al. 2006; Kontak et al. 2011; Gourcerol et al. 2018b). For example, the Au-Bi and Au-Te plots (Fig. 6b, c) show that laminated vein samples hosted in the MGB (Pileggi No. 1 samples) are lower in Bi and Te (below 35th percentile for Au1 samples), and define a trend discordant to the Au1 samples from the MLB. Similarly, the MGB laminated vein samples are enriched in Cu (above 70th percentile for Au1 samples; Fig. 6h). When comparing correlation values between laminated veins in the MLB and the MGB, the Pileggi No. 1 veins only show a significant (p < 0.05) Au-Ag correlation and lack the Au-Bi, Au-Pb, Au-Te, and Au-Mo correlations shown by the laminated veins in the MLB (Table 1). When correlation coefficients are used to compare whole-rock samples of Au1 and Au2 from the MLB (Fig. 9), Au1 samples show significant (p < 0.05) and consistent Au-Ag, Au-Bi, Au-Te, Au-Pb, and Au-Mo correlations that are not defined by Au2 samples. When LA-ICP-MS pyrite data is considered (Fig. 9), Au1 pyrite stands out with higher (> 0.7) Au-Ag, Au-Bi, and Au-Te correlations than Au2 pyrite. Overall, the correlation coefficient patterns defined by both whole-rock data and pyrite from Au2 are similar. The correlation coefficient pattern defined by Au1 pyrite shows similar Au-Ag, Au-Bi, and Au-Te correlations in comparison to the whole-rock data; however, the Au-Pb and Au-Mo correlations are markedly lower in the pyrite data (Fig. 9). This suggests that the Au-Pb and Au-Mo correlations in whole-rock data reflect the presence of other phases such as galena and molybdenite in the whole-rock samples. The correlation coefficient data shows clear differences in the trace-metal signatures of the Au1 and Au2 mineralization styles, and the results emphasize that correlation coefficients in trace metal datasets are useful ways to characterize and discriminate among different hydrothermal gold events. In the context of using these trace-metal associations to provide insight into ore deposit genesis, the Au-Ag-Bi-Te-Pb associations are not indicative of a certain deposit type, with the same associations displayed by syenite-associated, orogenic and intrusion-related deposits (Robert 2001; Groves et al. 2003; Goldfarb et al. 2005; Hart 2007). The Au-Mo correlation and elevated Mo concentrations in Au1 samples (Fig. 6f) are perhaps the most insightful as to deposit affinity, as these are not characteristic of orogenic gold deposits (Groves et al. 2003; Goldfarb et al. 2005; Goldfarb and Groves 2015), and these are features commonly documented in magmatic-hydrothermal systems, including intrusion-related deposits (Thompson et al. 1999), syenite-associated deposits (Robert 2001), gold deposits with anomalous metal associations (Groves et al. 2003), and porphyry deposits (Seedorff et al. 2005). Thus, it is permissible to infer that the Au1-Mo association and Mo enrichment in Au1 samples reflect an intrusion-related origin, as previously suggested based on both geological (McDivitt et al. 2017) and geochemical (McDivitt et al. 2018) evidence.

The sulfur isotopic signatures of Py1 and Py2 are similar (Table 2), although there is a 2‰ difference between the mean δ34Spyrite values of − 5.5‰ and − 3.5‰, respectively. Taking into account analytical precision (0.1‰ for Δ33S), the Δ33S values indicate components of mass-independently-fractionated sulfur in both Py1 and Py2 (Farquhar and Wing 2003; LaFlamme et al. 2018). The slightly-negative Δ33S values differ from the slightly-positive Δ33S common in gold deposits of the Yilgarn craton (Selvaraja et al. 2017; Godefroy-Rodriguez et al. 2020; LaFlamme et al. 2018; Groves et al. 2019), and in examples of Archean intrusion-related gold mineralization (Helt et al. 2014). The Δ33S values are more similar to those reported for Archean magmatic Ni sulfide and VMS mineralization (Xue et al. 2013; Sharman et al. 2015; Selvaraja et al. 2017; Ripley and Li 2017). The similar δ34S and Δ33S values for Au1 and Au2 pyrites suggests that isotopically similar reservoirs have been utilized throughout successive hydrothermal events in the district. Unlike the trace-metal data, the sulfur isotope data does not discriminate between the Au1 and Au2 events.

The δ18O values of Qtz1, Qtz2, and, Qtz3, which show considerable overlap, are also not a good discriminant of the different hydrothermal events. In addition, the δ18O values in all quartz generations tend to be lower than those typical of Archean lode gold deposits (Fig. 11a). The mean δ18O values of Qtz1, Qtz2, and, Qtz3 are similar to the mean δ18O values derived from equivalent carbonate generations from earlier studies (Fig. 11b). Whereas, the mean δ18O values from Au1 and Au2 quartz/carbonate are similar, the mean δ18O values from Retrograde quartz/carbonate are consistently lower.

Fig. 11
figure 11

a Plot of δ18O values for vein quartz (Y-axis) versus temperature (T (°C); X-axis) with isopleths of corresponding δ18OH2O values (curved lines) calculated using the quartz-H2O fractionation equation of Matsuhisa et al. (1979). Different fluid types (e.g., metamorphic waters, Archean lode Au waters) are classified after Taylor (1974) and McCuaig and Kerrich (1998). Mounted quartz grain SIMS δ18O values are shown as black dots and categorized by hydrothermal event/quartz generation (Au1/Qtz1; Retrograde/Qtz2; Au2/Qtz3) and host rock (MLB = Missinaibi lake batholith; MGB = Michipicoten greenstone belt). Vertical black bars on the right correspond to ranges in δ18Oquartz values from other areas of gold mineralization in the Superior craton: Red Lake (1; Kerrich 1989), Gutcher Lake (2; Kerrich 1989), and Val-d’Or (3; Beaudoin and Pitre 2005). These latter values emphasize the presence of low δ18Oquartz values in this study. Note that the T values at which the different quartz populations are shown at do not necessarily reflect the real T of quartz formation; rather, they are chosen to facilitate a comparison of the δ18Oquartz values in the context of the different hydrothermal events, host rocks, geographic locations, and fluid types. b Comparison of mean SIMS δ18Oquartz values from this study to compiled carbonate data compiled from earlier studies (Callan 1991; Samson et al. 1997; Haroldson 2014). c Evaluation of a fluid mixing/cooling model over the T range 200–400 °C. Red lines are mixing lines between endmember fluids (δ18OH2O = 10‰ and − 15‰). The proportion of the δ18OH2O = 10‰ fluid in the mixed system is indicated by the numbers associated with the red squares distributed along the mixing lines. The dark-blue, isopleth-parallel arrows indicate cooling trends. The lower, dotted black line depicts the δ18O values of the mounted Qtz2 grains (Qtz2 (M)) evenly distributed over the chosen T range. The upper, dotted black line depicts δ18O values of the CL-imaged Qtz2 crystal evenly distributed over the chosen T range. d Suggested model for the origin of δ18Oquartz and δ18Ocarbonate values in the Wawa gold camp: (1) magmatic fluids with δ18OH2O = 8‰ (?) initially formed pre-orogenic, granitoid-associated, gold-bearing veins (Au1 phases); (2) an 18O-light metamorphic fluid with δ18OH2O = 4‰ (?), generated due to the devolatilization of biotite-bearing granitoids in the mid- to lower crust, was channeled into the Wawa gold camp and overprinted Au1 veins and formed Retrograde veins; (3) late metamorphic fluids became 18O-enriched due to a higher degree of country-rock equilibration and formed veins during the Au2 event. The relative δ18OH2O values shown for the Au1, Retrograde, and Au2 events are in part based on the trends documented by the mean values in b

Origin of low δ18O values in vein quartz and carbonate

Herein, we explore different models to account for the low δ18O values in vein quartz and carbonate throughout the camp. Retrograde breccia vein quartz from Nudulama East (Qtz2) is used to constrain the origin of the low δ18O values for the following reasons: (1) it returned the lowest mean δ18O values in both local and regional datasets (Fig. 11b); (2) δD analysis of primary inclusions (Fig. 10e), and the reproducible δD results (Table 2) agree with both petrographic observations and microthermometric data (McDivitt et al. 2018) indicating a single fluid is present in Qtz2; (3) the quartz is generally euhedral, coarse-grained, and characterized by primary growth textures suggesting it is likely to preserve a primary isotopic signature; and (4) U-Pb titanite geochronology infers the timing of the Retrograde event and quartz formation at 2580 ± 21 Ma, thus allowing it to be interpreted in the context of a regional temporal framework. We consider the temperature range of 200–400 °C based on the following (ESM Fig. 1): (1) a minimum T constrained from the Th values of aqueous fluid inclusions in Retrograde breccia vein quartz (McDivitt et al. 2018); and (2) a maximum T constrained by the upper limit of biotite chloritization reactions in granitic rock bodies (Yuguchi et al. 2015), as such reactions define alteration associated with the Retrograde quartz (McDivitt et al. 2018). Additional maximum to intermediary T constraints include the following: (1) the T range of retrograde silica solubility (Fournier 1985), as inferred by dissolution-reprecipitation textures in the cores of quartz crystals (Fig. 10f); (2) the T range for brittle-ductile deformation of quartz (Smith and Bruhn 1984; Sibson 2001), as brittle-ductile deformation of quartz occurred during the Retrograde event (McDivitt et al. 2017); and (3) similar temperatures of Archean orogenic quartz vein formation (Goldfarb et al. 2005).

Low δ18O values in vein quartz from porphyry-epithermal and epizonal Archean lode gold settings are interpreted to fingerprint the involvement of surficial fluids (Taylor 1979; Hagemann et al. 1994). In order to evaluate the presence of surficial fluids, we utilize the δ18O and corresponding δD values of different fluid types (Taylor 1974) in a fluid-mixing/cooling model. A fluid of δ18OH2O = 10‰, typical for Archean lode gold systems (McCuaig and Kerrich 1998), is modeled to mix with a high-latitude meteoric water (δ18OH2O = − 15‰) from 200 to 400 °C (Fig. 11c). A high-latitude meteoric water is used in the modeling since for any given volume it would produce a larger isotopic shift than a low-latitude meteoric water or seawater; thus, it is used as a conservative constraint. The δ18O values of Qtz2 crystals can be explained if the endmember fluid with δ18OH2O = − 15‰ is present in proportions ranging from ~ 25 to 50%. Corresponding δD values in the mixed system should be approx. − 43 to − 65‰ although this assumes the upper limit of δD reported in Taylor (1974) where δ18OH2O = 10‰ for a metamorphic fluid (δD approx. − 20‰). If the lower limit (δD approx. − 60‰) is considered, the δD values in the mixed fluid should range from approx. − 72 to − 85‰. The reproducible δD values returned from Qtz2 of − 37‰, which is typical of metamorphic and magmatic fluids, do not indicate the involvement of a high-latitude meteoric water. While some studies illustrate that δD values from quartz provide δDH2O values concordant with those calculated from hydrothermal muscovite (Ojala et al. 1995), others conclude that δD values from quartz represent both trapped fluid inclusions and structurally bound water and may not reflect the original δDH2O values (Simon 2001). In this latter case, δD quartz values may be lower than the inferred δDH2O values (Simon 2001). In consideration of this, the δD in Qtz2 of − 37‰ still do indicate the involvement of a high-latitude meteoric water, as the δD values in Qtz2 are too high. It is possible that the δD values in Qtz2 reflect a low-latitude meteoric water or seawater, but a low-latitude meteoric water (δ18O = − 5‰) would need to be present in proportions from ~40–90% to explain the Qtz2 δ18O values in the mixing/cooling model. Seawater would need to be present in proportions from ~ 60–100%. As discussed by Pickthorn et al. (1987), (Kerrich 1990b), and Kyser and Kerrich (1990), the transport of these large volumes of surficial fluid to crustal depths where orogenic gold mineralization forms in an isotopically unmodified state presents issues. This is not our favored interpretation for the low δ18O values in veins quartz and carbonate in the camp.

Fluid-rock interaction provides another means to modify δ18OH2O values (Kontak et al. 2011). However, the shear zone-hosted nature of the breccia at Nudulama East (McDivitt et al. 2017) favors a high fluid/rock ratio system typical of orogenic mineralization (McCuaig and Kerrich 1998). Furthermore, the formation of a fluid with low δ18OH2O values via fluid-rock interaction invokes low δ18O host rock values (< 5.5‰), such as those formed by the interaction of magmas with meteoric water, or by assimilation of altered country rock with low δ18O values (Bindeman and Valley 2001; Hammerli et al. 2018). This situation is not supported by the whole rock δ18O values of 6.4 to 9.5‰ for granitoids in the Wawa gneiss domain (Li et al. 1991).

Rayleigh distillation has been considered in explaining the isotopic data from Archean gold deposits (Kerrich 1989, 1990a; Samson et al. 1997; Kontak et al. 2016; Neyedley et al. 2017). When the δ18O Qtz2 range by is assessed by closed-system Rayleigh distillation using an initial δ18OH2O = 10‰ over a temperature range of 200–400 °C, the δ18Oquartz range requires significant fluid distillation with f values to approx. ≥ 0.2 where f is the amount of original fluid remaining. The SIMS δ18Oquartz values from the CL-imaged crystal show increasing values from core to rim (Fig. 10f), and contradict a model of progressive Rayleigh fluid distillation. In contrast, the isopleth-parallel distribution of the δ18Oquartz values over a 200 to 400 °C range (Fig. 11c) suggests a progressive cooling model associated with crystal growth. In the CL image, it is also evident that some of the lowest δ18O values (i.e., 5.4‰) occur juxtaposed against higher values (i.e., 9.4‰). In this case, analytical error is an important consideration (1σ = 1.2‰), but the juxtaposition of contrasting values in a growth zone of quartz with uniform CL indicates that any difference in δ18O may not reflect T since variation of T is commonly reflected in CL zoning (Alan and Yardley 2007; Rusk et al. 2008; Mao et al. 2017). It is possible that the local, small-scale δ18O variance in zones where CL is uniform records Rayleigh distillation effects, but overall the distribution of δ18Oquartz values is best explained by a cooling model with the initial fluid having a low δ18O value.

A low δ18O fluid as the product of a devolatilized Granitoid source?

An initial δ18OH2O value of ~ 4‰ is indicated for the Retrograde fluid if the average δ18O value from the core (Fig. 10f; 7.9‰) is considered to have formed at 400 °C. Geochronological constraints on the timing of the Retrograde fluid infer its circulation at 2580 ± 21 Ma, which post-dates local granitoid magmatism (ca. 2.72–2.66 Ga; Turek et al. 1996; Kamo 2015), but overlaps metamorphism and deformation in the amphibolite- to granulite-grade Kapuskasing structural zone (Fig. 1; Moser 1994; Percival and West 1994). Devolatilization of greenstone belt lithologies during regional metamorphism is often considered a fluid-generating mechanism responsible for veining and alteration in Archean gold districts (Groves et al. 1987; Phillips et al. 1987; Goldfarb and Groves 2015). The presence of Retrograde event phases in the MLB (Fig. 2), which is external to the MGB, renders this mechanism of fluid generation unlikely due to a lack of greenstone fluid sources at depth. If a metamorphic fluid source at depth is to be considered, then it is likely represented by granitoids and gneisses of the Wawa gneiss domain and Kapuskasing structural zone. Mineral δ18O values reported by Li et al. (1991) from granitoids of the Wawa gneiss domain (~ 9 to 11‰ for quartz; ~ 6 to 8‰ for feldspars; and ~ 2 to 5‰ for biotite) illustrate that devolatilization of such granitoids, in particular the biotite component, could generate an 18O-depleted metamorphic fluid, as δ18OBt18OH2O fractionation at temperatures typical of amphibolite grade and higher is low (approx. − 2.5‰; Zheng 1993). Another consideration is the fractionation between aluminosilicate O and hydroxyl O, as the latter may be depleted in 18O relative to the former (δ18OBt18OOH ~ 10–3‰ at 300–800 °C; Zheng 1993); hence, if the dehydroxylation of biotite is the dominant mechanism in metamorphic fluid production, the δ18OH2O of the fluid may be lower than the δ18O of the precursor biotite. In orogenic gold settings where crustal zones are characterized by high permeability, brittle-ductile deformation, and transitions from lithostatic to hydrostatic fluid pressure (Sibson 2001), open-system metamorphic devolatilization (Valley 1986) may preclude fluid-rock equilibrium and preserve the 18O-depleted nature of a fluid generated from the metamorphic devolatilization of biotite. Such a deep-crustal metamorphosed granitiod source potentially explains the lack of a significant carbonic component and gold mineralization associated with the Retrograde event in the Missanabie-Renabie district (McDivitt et al. 2018), as carbonic fluids and gold mineralization are often attributed to the devolatilization of carbonated greenstone belt lithologies (Powell et al. 1991; Elmer et al. 2006), or the involvement of mantle sources (Hronsky et al. 2012).The similarities of mean values in both our SIMS δ18O data and the compiled regional data (Fig. 11b) suggests the low δ18O values for the Retrograde fluid was a regional-scale phenomenon. Because the Retrograde fluid event overprints earlier Au1 mineralization, the low δ18O values for Au1 quartz may represent dissolution of Qtz1 and precipitation of Qtz2 in Au1 veins during the Retrograde event as is inferred by fluid inclusion evidence (McDivitt et al. 2018). In contrast the Au2 equivalent quartz and carbonate returns to a higher mean δ18O value (Fig. 11b), suggesting the event may record metamorphic fluids affected by a greater degree of country-rock equilibration (Fig. 11d).

Conclusions

A geochemical comparison of three hydrothermal events (Au1, Retrograde, and Au2) in the Missanabie-Renabie gold district using whole-rock and LA-ICP-MS pyrite trace metal analysis in addition to sulfur and oxygen isotope data highlights the following:

  1. (1)

    Pearson product-moment correlation coefficients (log10) of whole-rock and LA-ICP-MS pyrite trace-metal datasets show good suitability in discriminating and characterizing different gold systems and events.

  2. (2)

    From a genetic perspective, Au-Mo correlations and elevated Mo concentrations in the Au1 trace-metal dataset are the most significant and support an intrusion-related affinity for the Au1 event.

  3. (3)

    Sulfur (δ34S/Δ33S in pyrite) and oxygen (δ18O in quartz) isotope data shows poor suitability in discriminating the different hydrothermal events and in characterizing them from a genetic perspective; hence, caution is warranted in using such data for the foregoing purposes.

The involvement of a late-stage, 18O-depleted metamorphic fluid is considered to be the best explanation for the low δ18O quartz and carbonate values documented across the Wawa gold camp. Geological, geochemical, and geochronological data suggest that this fluid was generated during the 2580 ± 21 Ma Retrograde event due to devolatilization associated with the prograde metamorphism of biotite-bearing granitoids at crustal levels represented by the Wawa gneiss domain and Kapuskasing structural zone. The association of this fluid to low- to moderate-salinity H2O fluid inclusions suggests that metamorphically-devolatilized granitoid rocks may be under-recognized source components of orogenic gold deposits.