Nanoscale Characteristics of Carlin-Type Auriferous Pyrite from the Nadaleen Trend, Yukon

The Au deposits of the Nadaleen trend occur at the eastern end of the Rackla gold belt in central Yukon, Canada (Fig. 1). The company ATAC Resources Ltd. discovered the trend in 2010 while following up on regional stream sediment arsenic anomalies identified by the Geological Survey of Canada (Goodfellow and Lynch, 1978). Drilling in the Conrad, Osiris, Ibis, and Sunrise deposits has delineated a combined total of about 1.7 Moz Au, including an inferred open pit resource of 8 million tonnes (Mt) at 4.1 g/t Au and an inferred underground resource of 4.3 Mt at 4.52 g/t Au (Ristorcelli et al., unpub. report, 2018). Gold is hosted in Neoproterozoic limestones, siltstones, and minor mafic dikes associated with decalcification and lesser silicification and argillization. The Au occurs in hydrothermal arsenian pyrite, which commonly overgrows earlier pyrite phases. Although the origins of gold on the Nadaleen trend remain uncertain, previous studies have characterized the deposit geochemistry and lithological and structural setting of the deposits (Moynihan, 2013; Colpron et al., 2013; Palmer, 2014; Beaton, 2015; Tucker, 2015; Tucker et al., 2018; Steiner et al., 2018; Pinet et al., 2018, 2022 a, b; Ristorcelli et al., unpub. report, 2018; Pinet and Sack, 2019; Moynihan et al., 2019; Pinet et al., 2020a, b; Pinet, 2021).

The deposits in the Nadaleen trend share many characteristics with Carlin-type gold deposits in Nevada. These similarities provide the opportunity to test genetic models for sedimentary rock-hosted Au mineralization that have been hitherto defined largely based on a single region. Both the Nadaleen trend deposits and Nevada’s Carlin-type gold deposits are located in the footwall of a major thrust (Palmer et al., 2017; Pinet et al., 2020a) in miogeoclinal clastic and carbonate rocks on the margin of North America (Tucker, 2015; Tucker et al., 2018). Unlike Nevada, where mineralization occurred during regional extension, the Nadaleen trend is interpreted to have been under transtension (strike-slip deformation) during mineralization (Pinet et al., 2022a, b). As in Carlin-type gold deposits, Nadaleen trend ore fluids traveled vertically along faults, into fold hinges, and laterally into permeable zones (Beaton, 2015; Steiner et al., 2018; Pinet et al., 2020a; Steiner and Hickey, 2022). Like Carlin-type gold deposits, ore fluids on the Nadaleen trend are governed by primary and secondary porosity and permeability (Pinet and Sack, 2019; Pinet et al., 2022a), resulting in replacement-style ore in sedimentary, tectonic, and hydrothermal breccias (Pinet et al., 2018). On the Nadaleen trend, the associations between Au and S, As, Sb, Hg, and Tl are similar to Carlin-type gold deposits (Palmer, 2014), as are the alteration styles of decarbonatization, minor silicification, realgar, orpiment, and calcite (Coulter et al., 2018; Pinet et al., 2020b, 2022a). Clumped isotope data from calcite indicate that Nadaleen trend ore was deposited from low-temperature fluids similar to Carlin-type gold deposits (Pinet et al., 2023). As in Carlin-type deposits, Nadaleen trend gold was deposited by sulfidation (Beaton, 2015) in hydrothermal arsenian pyrite overgrowths on earlier pyrite (Tucker, 2015; Pinet et al., 2020b).

Carlin-type deposits occur on five main trends in northern Nevada and collectively constitute one of the world’s largest accumulations of gold. Many hundreds of studies have examined the characteristics of Carlin-type deposits and the potentially causative processes. Carlin-type gold has been alternately attributed to circulation of meteoric fluids that scavenged Au from a sedimentary host rock source (Hofstra, 1994; Hofstra et al., 1999; Ilchik and Barton, 1997; Emsbo et al., 1999; 2003; Hofstra and Cline, 2000; Large et al., 2011) or from Eocene calc-alkaline magmas (Sillitoe and Bonham, 1990; Henry and Boden, 1998; Johnston and Ressel, 2004; Seedorf and Barton, 2004; Ressel and Henry, 2006; Muntean et al., 2011). In the magmatic model for Carlin-type gold formation, fluids are assumed to have exsolved from a magma at an oxygen fugacity of FMQ+1 (where FMQ = fayalite-magnetite-quartz), resulting in very high H₂S/SO₂ ratios (Muntean et al., 2011). In either the magmatic or the meteoric model, Au is thought to be transported as a bisulfide complex and then incorporated into fine-grained hydrothermal pyrite overgrowths on earlier pyrite. All geologic evidence suggests that the fluids remained reduced from source to precipitation of the hydrothermal pyrite, and thus sulfur isotopes have been used to constrain the source of Carlin-type auriferous fluids (e.g., Hofstra and Cline, 2000; Kesler et al., 2003; 2005; Henkelman, 2004; Cline et al., 2005; Barker et al., 2009; Gopon et al., 2019; Holley et al., 2022, 2024). However, Carlin-type hydrothermal pyrite is finely zoned, and high-spatial-resolution analyses are needed to examine the trace elements and isotopic signatures of individual growth zones (e.g., Barker et al., 2009; Gopon et al., 2019). In prior studies of Carlin-type pyrite, nanoscale secondary ion mass spectrometry (NanoSIMS) analyses show that the δ³⁴S of Au-rich growth zones corresponds to the sulfur isotope composition of Eocene magmatic fluids in the region, whereas meteoric fluids contributed during growth of low-Au zones (Holley et al., 2022, 2024; Holley and Phillips, 2022).

It remains uncertain whether Carlin-type gold deposits and their giant gold endowment are unique to Nevada or if geologically similar areas elsewhere might be equally prospective. Deposits with Carlin-like characteristics have been identified in southern China, Indonesia, Iran, and Malaysia, but the Nadaleen trend deposits in the Yukon appear to be the closest analogues to those in Nevada (Berger et al., 2014; Pinet et al., 2018; Cline, 2018; Xie et al., unpub. data, 2023). This study aims to determine whether the causative processes that led to mineralization on the Nadaleen trend were the same as those operating in northern Nevada during Carlin-type mineralization.

In the present contribution, we examine the characteristics and geochemical signatures of the pyrite-hosting gold on the Nadaleen trend. Of relevance to the broader study of gold deposits, the contribution provides an approach to quantify the source of sulfur in auriferous fluids even where a wide range of sulfur isotope data has been obtained. We use NanoSIMS mapping and NanoSIMS depth profile analyses to quantify the nanoscale distributions and concentrations of Au, As, Ag, Cu, Sb, and δ³⁴S in zoned pyrites from Conrad and Sunrise. We complement these results with Bayesian tracer modeling to examine the potential source contributions to the hydrothermal pyrite. Stable isotope mixing models can be used to estimate the proportional source contributions to a mineral phase that formed due to input from multiple geologic sources (Holley and Phillips, 2022). We use sulfur isotopes as tracers of the sulfur source, since the available data from the Nadaleen trend suggest that the fluids have remained consistently reduced from source to final precipitation of the ore pyrite. We use the Bayesian model MixSIAR (Stock et al., 2018) to examine how the Au concentration of the hydrothermal pyrite covaries with proportional contribution of two potential sources of gold and sulfur: sedimentary pyrite in the local host rocks and igneous rocks in the deposit area.

Since the discovery of the Nadaleen trend in 2010, five main areas have been drilled: Conrad, Osiris, Ibis, Sunrise, and Anubis (Ristorcelli et al., unpub. report, 2018). The deposits occur in the Rackla fold and fault belt at the transition between the Selwyn basin and the Ogilvie platform (Fig. 1; Gabrielse, 1967; Gordey and Anderson, 1993; Moynihan et al., 2019). Neoproterozoic to mid-Paleozoic sedimentation was affected by syndepositional tectonism along the 250-km-long, E-W–trending Dawson fault, which marks the boundary between the shelf and basin (Pinet et al., 2020a). The Rackla belt is bounded on the south side by the Dawson fault and on the north side by the subparallel Kathleen Lakes fault; both faults splay at the eastern end of the belt in the area of the Nadaleen trend gold deposits. Neoproterozoic stratigraphy in the Nadaleen trend consists of Hay Creek Group siliciclastic and dirty limestone units, conformably overlain by the Rackla Group (Moynihan et al., 2019). The Rackla Group comprises clastic rocks of the Sheepbed Formation; mixed siliciclastic and carbonate rocks of the Nadaleen Formation; limestone and dolostone of the Gametrail Formation; fine-grained siliciclastic-dominated rocks of the Blueflower Formation; and limestone of the Algae Formation (Fig. 1). The sedimentary rocks were heavily folded and faulted during formation of the Jurassic to Cretaceous Selwyn fold belt (Mair et al., 2006). The eastern Nadaleen trend is characterized by steep, oppositely dipping thrusts and steeply plunging folds cut by faults (Steiner and Hickey, 2022). The Osiris-Sunrise area is characterized by steeply plunging z-folds, and geometry at Conrad has been interpreted as a doubly plunging anticline (Palmer and Kuiper, 2017; Pinet et al., 2022a) or steepened soft-sediment folds (Steiner and Hickey, 2022).

With the exception of small outcrops of Paleozoic gabbros, magmatic rocks near the Nadaleen trend are notably absent. Despite the steep and rugged topography in the area with plenty of outcrop exposure and recent mapping at 1:20,000 to 1:50:000 (Moynihan, 2016; Steiner and Hickey, 2022), the only magmatic rocks known are two gabbroic dikes. These 0.25- to 4-m-wide dikes have a pooled U-Pb zircon age of 74.4 ± 1.0 Ma (Tucker, 2015). Farther afield, the Selwyn basin area has abundant mid-Cretaceous (ca. 115–90 Ma) granitoid intrusions associated with reduced intrusion-related gold systems (e.g., Hart, 2007). The closest mapped mid-Cretaceous intrusion is 40 km south of the Nadaleen trend. The Selwyn basin area also hosts a string of latest Cretaceous to Paleocene (ca. 75–63 Ma) granitoids of the McQuesten suite (Colpron et al., 2016). The McQuesten suite is associated with reduced styles of mineralization, specifically tin and tungsten skarn deposits (Murphy, 1997).

In the Conrad deposit, gold is mostly hosted in the Conrad limestone of the Nadaleen Formation. The mineralized zone has a domal shape at Conrad, extending vertically over 500 m and over 800 m along strike (Ristorcelli et al., unpub. report, 2018). In the Osiris, Sunrise, and Ibis deposits, gold is hosted in the Osiris lime mudstone of the Gametrail Formation, generally concordant with bedding in an anticlinal structure, mineralized over a 300-m vertical interval. At the Anubis deposit, gold is hosted in the overlying Paleozoic sequence, in Middle Ordovician to Early Permian black shales and siltstones alternating with carbonates (Moynihan, 2016; Pinet et al., 2020a).

Gold mineralization in the Nadaleen trend was characterized by precipitation of Au in disseminated hydrothermal arsenian pyrite. The hydrothermal pyrite is visible in thin section as micron-scale overgrowths on precursor pyrite (Sack et al., 2019). The host rocks were decarbonatized, brecciated, and affected by minor silicification and argillization, similar to Carlin-type deposits in Nevada. The main stage of mineralization included gold in arsenian pyrite, minor very fine grained quartz (“jasperoid”), and minor illite, followed by a late-stage assemblage of realgar, orpiment, blocky calcite, and fluorite (Pinet et al., 2022a). Orpiment and realgar alteration is ubiquitous in mineralized zones and is generally more visually apparent than in Carlin-type ores in Nevada.

Mineralization was structurally and lithologically controlled. In the Osiris, Isis, and Conrad deposits, reverse faults are thought to have served as aquitards; gold-bearing fluids were channeled up-dip and focused into hinges of plunging anticlines (Palmer, 2014; Palmer et al., 2017; Steiner and Hickey, 2022). Hydrothermal fluid flow was also localized along faults, at fault intersections, at dip closure fold/fault intersections, and in areas of high fracture permeability (Pinet et al., 2020b, 2022a). The mineralizing fluids appear to have been channeled through debris flows and focused into calcareous mudstones and silty carbonates bounded by low-permeability siliciclastic units (Pinet and Sack, 2019). Hydrothermal calcite in realgar-altered zones containing >6 ppm Au (ATAC data) from Sunrise and Conrad yielded a U-Pb calcite age of ca. 74 Ma—which Pinet et al. (2022b) interpret to represent the timing of Au mineralization—as well as a younger event at ca. 48 Ma, which was attributed to U mobility in calcite grain fractures. Apatite fission tracks were analyzed in two host rock samples 0.9 and 4.3 km from the deposits, respectively (Pinet et al., 2023). Inverse modeling of the apatite fission tracks allows for T-t paths associated with mineralization around 74 Ma followed by cooling and then exhumation around 35 Ma (Pinet et al., 2023).

All the available evidence suggests that the Nadaleen trend deposits formed by processes similar to Carlin-type deposits in Nevada, from similar fluids. Carlin-type fluids were acidic, low temperature (~180°–240°C), CO₂ bearing (<4 mol %), low salinity (≤ 6 wt % NaCl equiv), and nonboiling (Muntean et al., 2011). Carlin-type Au was transported as a bisulfide complex, and much of the pyrite precipitated by sulfidation of earlier pyrite (Cline et al., 2005). The fluids that formed the deposits on the Nadaleen trend were initially acidic, dissolving carbonate and leaving decalcified textures and illite alteration (e.g., Pinet et al., 2022a). In the Nadaleen trend deposits, the main-stage fluid that transported the gold was reduced, as demonstrated by the dominance of pyrite, and Au was likely carried as a bisulfide complex. Textural evidence from the Nadaleen trend suggests that most styles of the hydrothermal pyrite grew by sulfidation of preexisting pyrite or other Fe-rich phases. Temperatures of the ore-forming system on the Nadaleen trend have been constrained by calcite and fluorite fluid inclusions (T_h of 123°–173°C) and calcite clumped isotope thermometry (91°–162°C), whereas organic matter records regional temperatures of >150°C and maximum temperatures of 220°C (Pinet et al., 2023). Late-stage fluids on the Nadaleen trend were low temperature (91°–173°C) with a salinity of 4.8 wt % NaCl equiv (Pinet et al., 2023). To predict the redox state of the ore-forming fluids, we draw a parallel between the crustal setting of the Nadaleen trend and Carlin-type gold deposits in Nevada. On the Carlin Trend, intrusions emplaced into reduced crust at FMQ+1 are thought to have resulted in exsolution of fluids with H₂S/SO₂ ratios of ~300 (Muntean et al., 2011). Independent thermochemical calculations corroborate these assumptions, yielding H₂S/(H₂S + SO₂) ratios of 0.95 (Muntean et al., 2011). Given the reduced nature of the rocks in the Nadaleen trend, we assume that aqueous fluids exsolving from a magma would have similarly high H₂S/SO₂ ratios and would have remained consistently reduced while traveling through the host rock package.

This study presents NanoSIMS analyses and modeling results from auriferous pyrite in two high-grade drill core intervals from Conrad and Sunrise. Sample selection was based on detailed logging, sampling, and petrographic study of drill core by Pinet and Sack (2019), Pinet et al. (2020a, b, 2022a, b), Sack (unpub. data, 2019), and ATAC Resources. We selected intervals that were well mineralized and display lithological and alteration characteristics typical for the Sunrise and Conrad deposits. Our specific aim was to compare mineralized sedimentary-hosted and dike-hosted pyrites, at grain sizes coarse enough for NanoSIMS mapping and spot analyses (see below). The interval from Sunrise was from drill hole OS-10-001 at a depth of 83.15 m, grading 46 g/t Au over 0.5 m, based on fire assay from the ATAC database. The hole was drilled in the hinge of the Osiris anticline (Figs. 2, 3, 4). The interval comprised heavily decalcified polymict sedimentary breccia debris flows and lime mudstone between dolomitized and massive crystalline limestone. Pyrite occurs in this sample as early and late diagenetic pyrite and early and late hydrothermal pyrite. Each of these generations is distinguishable by morphology. Early diagenetic pyrite is the most morphologically diverse pyrite. This generation is typically finer grained (surface area <1,000 µm²) and varies in morphology from framboidal with radii ≥10 µm to microeuhedral framboidal clusters to porous anhedral masses. Late diagenetic pyrite typically overgrows early diagenetic pyrite as coarser-grained, solid, subhedral to euhedral crystals with surface areas up to 2,500 µm². Hydrothermal pyrite is small and generally occurs as ≤2-µm rims on preexisting pyrite or as single-stage grains. Early hydrothermal pyrite is blocky and either grows directly on diagenetic pyrite (Fig. 5) or as single-stage crystals. In some cases, a very thin (≤1-µm) rim of delicate, fuzzy late hydrothermal pyrite overgrows either early hydrothermal pyrite or diagenetic pyrite. Veinlets of realgar crosscut the bedding and occur along fracture surfaces (Fig. 4). The Ca content of the interval is low, likely due to decarbonatization (Fig. 3a). The drill core interval from Conrad was from drill hole OS-17-259 at a depth of 181.5 m, grading 37 g/t Au over 2 m based on fire assay from the ATAC database. The hole was mostly drilled in Conrad siliciclastics and limestones but intersected the north Conradgabbroic dike (Figs. 3, 4). We selected an interval of Conrad dike gabbro, which was altered to a light gray color and crosscut by calcite veinlets, orpiment, and realgar (Fig. 4). The pyrite in this sample occurs as euhedral magmatic pyrite overgrown by early and late hydrothermal pyrite. The euhedral magmatic pyrite is relatively coarse grained with surface areas typically >500 µm² and a distinct “soccer ball” shape. The early and late hydrothermal pyrite are the same as that described for sample OS-10-001, with the significant difference that the early hydrothermal rims are the thickest (5–10 µm) seen in the study. The dike occurs at the contact between orpiment-altered, decarbonatized siliciclastic rocks, with low Ca content and high As content, and limestone with calcite veining.

We prepared thin sections from the drill cores and used petrographic screening to identify areas for NanoSIMS analysis. We targeted areas of each sample where abundant pyrite hosted hydrothermal rims visible in thin section. We selected grains for follow-up analysis based on size (>5 µm), and we documented the presence and characteristics of finer-grained pyrite present in the samples. The petrography was conducted on polished 30-µm thin sections in the Colorado School of Mines Mining Geology Research Laboratory on a Carl Zeiss Axio Scope A1 polarizing microscope. Scanning electron microscopy (SEM) was used to further characterize the grains at the Colorado School of Mines Mineral and Materials Characterization Facility using a TESCAN MIRA3 LMH Schottky FE-SEM. Backscattered electron (BSE) imaging was conducted with a single-crystal YAG BSE detector at an accelerating voltage of 20 kV, a 5-µm spot size, and a working distance of 10 mm. Energy-dispersive X-ray (EDX) spectrometry was accomplished using a Bruker XFlash® 6/30 silicon drift detector.

We acquired NanoSIMS element maps and spot analyses for ³²S^–, ³⁴S^–, ⁶³Cu^–, ⁷⁵As^–, ¹⁰⁷Ag^–, and ¹⁹⁷Au^–. In some analyses an attempt was made to collect ¹²¹Sb^–, but its peak position could not be confirmed (see below). The data were collected using the CAMECA NanoSIMS 50L at the Stanford Nano Shared Facility. The thin sections were polished to 0.25 µm, cut to 7- × 7-mm squares, and coated with 15 to 20 nm of carbon prior to analyses.

Qualitative NanoSIMS maps were collected over square regions of the sample ranging from 20 × 20 to 30 × 30 µm. A slightly larger region was presputtered at 500 pA to remove carbon prior to mapping. Mapping was achieved using electron multiplier detectors, a Cs⁺ ion beam accelerated at 8 keV, a 3- to 4-pA beam current, and a 100-nm beam diameter. Mass peak positions were confirmed with chalcopyrite, arsenopyrite, and Ag-bearing tennantite, identified by SEM-BSE, and gold coating working references. The ¹²¹Sb^– mass peak cannot be resolved from a ¹⁰⁹Ag¹²C interference, as it would require a mass resolution of >130,000. For this reason, the Sb data is used here strictly for qualitative purposes. For each map, between 15 and 45 frames were collected in sequence. The OpenMIMS plugin package for ImageJ was used to reduce the image files, including correction for 44 ns deadtime, correction for quasisimultaneous arrivals using a factor of 0.75 (Traxlmayr et al., 1984; Hillion et al., 2008), and summing of the frames. NanoSIMS mapping measures ion intensity in counts per second, so the data are presented qualitatively rather than as absolute concentrations.

Quantitative NanoSIMS spot analyses were collected at locations of interest within individual grains, determined based on the mapping and petrographic screening. Prior to analysis, the spot locations were presputtered for up to 2 minutes to remove the carbon coat and any surface contamination, as well as to implant Cs⁺ into the sample surface. For each spot analysis, the primary beam was rastered over a 2- × 2-µm area divided into 64 × 64 pixels. The per-pixel dwell time was 245 µsec collected in 2,400 frames. The data were recorded as counts per second, which allows for comparable time-resolved signals across all masses. Spot analyses were achieved using a ~350-pA Cs⁺ primary ion beam with an accelerating energy of 8 keV. Faraday cups with 10¹¹ Ohm resistors were used to collect the ³²S^– and ³⁴S^– data in order to avoid the effects of quasisimultaneous arrival and fast electron multiplier aging (Zhang et al., 2014). The other masses were collected on electron multipliers, and the mass peak positions were confirmed as above. Mass resolution ranged from ~6,300 to 7,400 and was set to >6,300 for ³²S^–, ³⁴S^–, and ³⁶S^– to resolve isobaric interferences, such as ¹⁶O₂ and ³¹P¹H^– on ³²S^–; ³³S¹H^– on ³⁴S^–; and ³⁵Cl¹H^– and ³⁶C₃ on ³⁶S^–. Entrance and aperture slits were adjusted to achieve flat-topped, steep-sided peaks. The corrections for background and Faraday Cup were applied based on Hauri et al. (2016). Craters from several representative analyses were measured as <2 µm on the Colorado School of Mines Helios NanoLab 6001 FIB-SEM, indicating that each individual data point in a 2,400-frame spot analysis represents a depth interval of less than 1 nm.

The NanoSIMS spot analysis ³⁴S/³²S isotope ratios are reported in parts per thousand (per mil or ‰). The data are presented relative to Vienna-Canyon Diablo Troilite (V-CDT) in delta notation (δ) (Krouse and Coplen, 1997) using equation (1):

The standard ratio for V-CDT was ³⁴S/³²S_V-CDT = 0.04416259 (Ding et al., 2001).

Instrumental mass fractionation (IMF) corrections were accomplished with standard-sample-standard bracketing. Between one and five sample analyses were bracketed on each side by one, two, or three analyses of Elba pyrite (δ³⁴S_V-CDT 9.37 ± 0.36; Holley et al., 2022) or Balmat pyrite (δ³⁴S_V-CDT 16.39 ± 0.20; Kozdon et al., 2010). The reference pyrite was also checked by bracketing with Canyon Diablo Troilite (CDT) as a secondary standard. The day-to-day standard reproducibility (1σ) of Balmat pyrite was 0.4 to 0.7‰; the Elba pyrite was 0.7 to 1.2‰; and the Canyon Diablo Troilite was 0.5 to 1.1‰. The normalization method was used for the time-dependent IMF correction, where each individual frame of the analysis is normalized to the frame of the standard collected at the corresponding time. Samples in the same analytical section were reanalyzed by laser ablation-multicollector-inductively coupled plasma-mass spectrometry (LA-MC-ICP-MS) to ensure that the NanoSIMS results were not affected by spurious fractionation “matrix effects” as described in Holley et al. (2022).

Each NanoSIMS spot analysis penetrated through multiple compositional zones of the pyrite. Instead of averaging the data from an entire analysis, we present the time-resolved data showing the variation in element concentration or δ³⁴S according to depth within in the sample. We refer to these x-y plots as depth profiles, where time is on the x-axis and the δ³⁴S or element concentration is on the y-axis. We also calculated the values for peaks and plateaus in the time-resolved data based on a limited number of frames with relatively consistent composition. Plateaus were defined as relatively flat zones of a depth profile with consistent chemical composition, bounded by peaks, troughs, or inflection points. For plateau calculations, the isotope ratios were calculated based on the sum of total counts to avoid statistical bias, rather than using the mean of the ratios (Ogliore et al., 2011). The plateau values were used in the isotopic modeling as described in the section below. The depth profiles were also used to calculate Pearson correlation coefficients, showing the extent to which the concentrations of two masses covary within an analysis.

Minor and trace element concentrations were calculated using relative sensitivity factors, according to the method of Zhang et al. (2017). Since sulfide standards for minor and trace elements have not been extensively studied at the nanoscale (Gopon et al., 2021), the bracketing standards used for sulfur analysis could not be used as standards for the other masses collected. Calibration curves were not used because most NanoSIMS depth profiles collected in this study had varying concentrations of minor and trace elements. Furthermore, electron probe microanalysis (EPMA) data for the Sunrise and Conrad samples could not be used to standardize the NanoSIMS data due to the large sampling volume of the EPMA instrument relative to the scale of the elemental zoning in the pyrites. Instead, a single-point standard was used to calculate relative sensitivity factors (RSFs) from a uniform trace-element–rich spot analysis of pyrite from the Red Dot deposit, Nevada, where the zoning was relatively coarse. The NanoSIMS data for the Red Dot standard were collected in the same analytical sessions, and the corresponding concentrations collected during electron probe microanalysis are presented in Holley et al. (2024). Calculated RSFs and the associated uncertainties with each RSF data point were 0.57 ± 0.09 for Cu, 0.034 ± 0.002 for As, 4.2 ± 1.7 for Ag, and 0.94 ± 0.09 for Au. The optimal detection limits for SIMS are in the sub-ppb (parts per billion) range. Specific detection limits depend on composition and instrumental methods and are more difficult to quantify than for other methods (e.g., relative to EPMA), and quantification of detection limits for the NanoSIMS data set would also require rigorous study of SIMS reference materials at the nanoscale. Therefore, approximate concentrations of trace and minor elements were calibrated using the equation:

where C_A is the concentration of A (a given element) and C_32S is the concentration of S measured by EPMA. The A/³²S is the ratio of secondary ion intensity of element A to ³²S measured using NanoSIMS. In some analyses, As and Cu showed some variation in RSF according to depth in the sample due to crater effects. This was corrected by normalizing to the As/³²S curve of the bracketing standards, although the same approach could not be used to correct Cu since the bracketing standards contained little Cu.

Holley and Phillips (2022) reviewed the application of stable isotope modeling to solve mixing problems in geology and demonstrated the utility of the method in examining source contributions to Carlin-type pyrite. Following the same approach, we modeled the relationship between Au concentration and source contributions to the hydrothermal pyrite from Sunrise and Conrad using the model Mixing Stable Isotope Analysis in R (MixSIAR; Stock et al., 2018). The model combines MixSIR (Sampling-Importance-Resampling; Moore and Semmens, 2008) and SIAR (Stable Isotope Analysis in R; Parnell et al., 2010) in the R software for statistical computing. The model allows the user to examine the range of possible proportional source contributions to a mixture, based on isotopic or other geochemical tracers. The MixSIAR model is useful for determining probability distribution while accounting for analytical uncertainty in isotopic signatures of the source or mixture, in systems where the number of potential sources exceeds the number of isotopic tracers +1, or in systems where other factors covary with source contribution. In this contribution we use the modeling to examine source contributions to the hydrothermal pyrite. The following equations are the basis of the tracer mixing models used in this study:

The mean tracer value of the mixture, Y_j, is equal to the sum of k source tracer means, $μjks$⁠, multiplied by p_k, their proportional contributions to the mixture. Bayesian tracer models randomly generate q vectors of the possible proportional contributions from the sources, f_q. The modeling calculates the probability of each f_q, where L(data | f_q) describes the likelihood of the data, and P(f_q) is the probability (Ellison, 2004):

Markov Chain Monte Carlo (MCMC) simulation is used to fit the model. The model produces true probability density distributions as the source contribution posterior distributions, accounting for the data and any prior information on the expected contributions (Parnell et al., 2010).

Following the approach of Holley and Phillips (2022), we set the hydrothermal pyrite as the “mixtures” and several used locally present host rock types as the potential sources. We use the NanoSIMS δ³⁴S plateau values and corresponding Au concentrations as tracers to examine the potential source contributions to the hydrothermal pyrite at Sunrise and Conrad. We examine the possible contributions of sulfur and Au from the Conrad gabbro as well as the local host stratigraphy in the host rocks at Sunrise. Since δ³⁴S fractionation occurs between pyrite and mineralizing fluids, we based our modeling on the δ³⁴S of fluids in equilibrium with the hydrothermal pyrites at 200°C, as an estimate for the temperature of mineralization. As described previously, there is no evidence for redox changes in the fluid from source to the final ore pyrite. With a constant H₂S/SO₂ ratio with very high H₂S, redox-associated fluid-pyrite fractionation is assumed not to have occurred. The input data and parameters are described in detail in the Stable Isotope Modeling section, and example codes are provided in Holley and Phillips (2022).

The lime-mudstone samples from Sunrise contain 1 to 3% modal abundance of fine-grained disseminated pyrite, ranging from less than a micron up to about 100 µm in diameter. We observed three styles of pyrite grains in these samples. The first pyrite type is coarsest and most visually apparent. This pyrite type consists of grains that are subhedral to euhedral, 20 to 100 µm in diameter, and yellow to bright white in reflected light (Fig. 5). These coarse grains are competent with sharp grain boundaries and minor to moderate internal fractures. Most grains of this pyrite type are overgrown by thin rims of hydrothermal pyrite, which occurs as tendrils irregularly distributed along fractures and around the edges of the earlier pyrite. The hydrothermal rim thickness ranges from <1 to 5 µm but is most commonly <1 µm. Detailed NanoSIMS maps show that the coarse pyrite contains low concentrations of Cu, As, Ag, Sb, and Au. Spot analyses show that the pyrite grain cores contain up to 4.2 ppm Cu, 25 ppm As, 7.1 ppm Ag, and 3.7 ppm Au (Table 1). The correlations among these elements are weakly positive in most analyses (Table 2). Two analyses show strong positive Ag-As correlations, and one analysis shows strongly positive Au-Ag and Au-As analyses (>0.6). The As, Ag, and Au are more concentrated in the hydrothermal pyrite than in the coarse-grained pyrite, although they are irregularly distributed within that phase (Fig. 5). Spot analyses of this hydrothermal pyrite are up to 29 ppm Cu, 1,053 ppm As, 15.1 ppm Ag, and 44.8 ppm Au. The concentrations of Au, Cu, and As show strong positive correlations with one another in most of the Sunrise hydrothermal rims, whereas the correlations between Ag and these elements can be strongly positive, moderately positive, or weakly negative (Table 2). The mapping of this grain shows that Au is elevated in the matrix and is not limited to the hydrothermal rim but also occurs within the matrix surrounding the pyrite grain (Fig. 5Ll). This pyrite type is discussed in the penultimate paragraph below.

The second most apparent pyrite type in the Sunrise samples is spongy or porous framboidal pyrite, appearing yellow in reflected light (Fig. 6A-L). Overgrowths of sphalerite occur along some of the margins. The qualitative NanoSIMS mapping shows that this pyrite type comprises aggregates of fine-grained pyrite ranging from <1 to 5 µm, clumped together in clusters that have a total diameter of 20 to 80 µm. This pyrite type is too small for NanoSIMS spot analyses. The fine-grained pyrites making up the framboid contain low concentrations of Cu, As, Ag, Sb, and Au, as indicated by the qualitative NanoSIMS mapping. They are overgrown and cemented together by <1- to 3-µm rims of hydrothermal pyrite that contain higher concentrations of these elements. The distributions of Cu, As, Ag, Sb, and Au are irregular in the pyrite overgrowths, which are characterized by both oscillatory zoning and patchy areas of enrichment. The distribution of Au is particularly irregular, corresponding to areas of high As, Ag, Cu, and Sb, but not equally distributed in all of the areas where the other elements are enriched (Fig. 6A-L).

The third pyrite type in the Sunrise samples consists of very fine grained <1-µm grains disseminated in the sedimentary rock matrix (Fig. 6M-P). These grains occur at 1 to 3% modal abundance and can be concentrated in groups adjacent to the coarse-grained pyrite (Fig. 5I, L) and the spongy pyrite framboids (Fig. 6B-L). The very fine grains were imaged and identified as a consequence of mapping the larger grains that could be identified using the NanoSIMS optics during setup. We do not know whether the very fine grains also occur in locations where the two previously described pyrite types are absent. We do note that, despite their proximity to the other pyrite types and their occurrence in groups, the very fine grained pyrite grains are disseminated in the host rock and do not appear to touch one another (Fig. 6M-P). Although the very fine grained pyrites are visible as bright specks in high-magnification reflected light and BSE images (Fig. 6B, C), they are too small to examine in detail using those methods. The mapping shows that the very fine grained pyrites contain Cu, As, Ag, Sb, and Au concentrations that are elevated relative to the host rock (Fig. 6M-P). The relative enrichments of these elements vary depending on the individual grain, and not all of the grains contain Au. The distribution of trace elements within these grains suggests that they may be single stage, although they could contain nanoscale cores of earlier pyrite or other Fe-rich phases that would be undetectable at the resolution of the NanoSIMS maps.

Depth profiles through the coarsest Sunrise pyrites (the first pyrite type described above) enabled us to compare the geochemical signatures of the pyrite grain cores and the hydrothermal overgrowths (Fig. 7A-C). Each analysis was initiated in the grain core that is exposed at the thin section surface, and each analysis penetrated through the contact into the hydrothermal rim underneath the grain core. The pyrite grain cores typically have relatively low δ³⁴S values. Individual data points are as low as –15‰ (Fig. 7A-C), with plateau averages from –3.9 to 0.8‰ (Table 3), similar to diagenetic pyrite in marine shales (e.g., −29.4‰ to −9.4‰; Grema et al., 2022). The contact between the grain core and the hydrothermal rim is marked by a sharp increase in the concentrations of As and Cu, whereas the innermost zones of hydrothermal pyrite at the contact do not contain much Au (Fig. 7A-C). Instead, the Au concentrations peak in the middle or outer zones of the hydrothermal pyrite. In some analyses, the highest concentrations of Au, Cu, Ag, and As occur in the same zone (Fig. 7A), and in other analyses there are minor variations in the relative enrichments of these elements in each zone (Fig. 7B, C). The δ³⁴S compositions of the hydrothermal pyrite vary somewhat between the samples but are between 1 and 8‰ higher than the earlier grain cores in the coarse-grained Sunrise pyrite (Fig. 7B, C; Table 3). Of all the Sunrise pyrite types, only the coarse-grained pyrite could be analyzed using depth profiles. The second (spongy framboids) and third (fine-grained disseminated) pyrite types were too small for depth profile analysis.

The dike-hosted pyrite from Conrad contains 1 to 3% modal abundance of pyrite disseminated in the altered gabbro (Fig. 8A-D). Individual pyrites are 5 to 100 µm in diameter. The pyrites are subhedral to rounded, and many have pits and minor fractures. They occur as individual grains dissociated from any other pyrite grain, as well as in 500- to 1,000-µm-diameter stringers or clusters of pyrite grains. In each cluster, there are typically between 5 and 50 grains that touch one another, as well as an additional 5 to 100 grains that are detached from other grains but disseminated near the aggregated grains. Most of the dike-hosted pyrite have 1- to 5-µm rims of hydrothermal pyrite visible in reflected light and BSE imaging (Figs. 8A-G, 9A-B). The hydrothermal overgrowths have relatively consistent thickness around the earlier pyrite in many grains, and they also occur as a cement infilling the spaces between grains in the aggregated clusters. The contact between the earlier dike pyrite and the hydrothermal pyrite is smooth and sharp, but the hydrothermal pyrite has fuzzy exterior margins at the contact with the host rock. The NanoSIMS mapping shows that the early pyrite contains little Cu, As, Ag, Sb, or Au, with maxima of 9.4 ppm Cu, 383 ppm As, 14.4 ppm Ag, and 4.9 ppm Au (Table 2). All of these elements are enriched in the hydrothermal pyrite, with values up to 116 ppm Cu, 23,400 ppm As, 386 ppm Ag, and 672 ppm Au (Table 3; Figs. 8H-L, 9C-L). The spatial distribution of these elements varies in the hydrothermal pyrite, with evidence for both oscillatory zoning and patchy or irregular zoning. The Au appears to be most enriched in the latest phases of the hydrothermal rims, occurring at the very outer margins of the pyrite at the contact with the host rock (Figs. 8L, 9G, L).

The coarser size of the Conrad dike-hosted pyrites enabled us to collect a larger number of NanoSIMS spot analyses than for the Sunrise sedimentary-hosted pyrites. However, the grain cores and hydrothermal rims were too thick to intersect all zones in a single analysis. To survey the most material possible, three types of depth profiles were collected from the Conrad dike-hosted pyrites:

1. Four analyses initiated in the center of the dike pyrite grain cores, where the underlying dike pyrite was presumed to be thickest, and the resulting analyses only encountered dike pyrite (e.g., representative analyses in Fig. 7D, E).

2. Four analyses were initiated at the midpoint of the hydrothermal rim, showing zoning in the rim itself (Fig. 7F-I).

3. Four analyses were initiated roughly equidistant between the center of the grain and the inner edge of the hydrothermal rim, enabling us to characterize zoning in the outer part of the dike pyrite as well as the inner part of the hydrothermal rim (Fig. 7J-L).

The δ³⁴S signatures of the early dike pyrite grain cores range from about 0 to just over 11‰ in individual data points in the NanoSIMS depth profiles (Fig. 7D-F, J-L), with δ³⁴S plateau values from 3.7 to 11.2‰ (Table 3). The δ³⁴S of the dike pyrite grain cores is distinctly higher than the sedimentary pyrite analyzed from Sunrise. The Cu, As, Ag, and Au are distributed in submicron-scale zones. The zoning patterns of these elements varied by analysis, with Ag and As showing nearly identical zoning (Fig. 7D), similar but not identical zoning (Fig. 7E), and dissimilar zoning (Fig. 7J). The Au enrichment occurs only in some zones of As enrichment (Fig. 7E). The Cu and Sb distributions are also zoned in the dike pyrite grain cores, but the zoning does not appear to correspond to the zoning patterns of the other analyzed elements. The correlations among the measured elements are all weakly positive or weakly negative in the dike pyrite grain cores, except for a strongly positive Cu-As correlation and moderately positive Au-Cu and Au-As correlations in one analysis (Table 2).

Depth profiles that intersect the hydrothermal pyrite rims show that there are differences between the inner and outer parts of the rim. Although both rim types contain higher trace and minor element concentrations than the dike pyrite grain cores, the enrichments are far greater in the outer rims. The inner parts of the rims closest to the dike pyrite contain up to 51.1 ppm Cu, 935 ppm As, 58 ppm Ag, and 19.1 ppm Au (Table 1; Fig. 7F, J-L). Elemental correlations in the inner part of the hydrothermal rim vary by analysis, with the only consistent trend in the Cu-As correlation, which is strongly positive in all analyses (Table 2). The inner rims show numerous narrow zones, with inconsistent correlations among the minor and trace elements. For example, As is enriched wherever there are Au peaks, but the Au peaks are relatively narrow and are less common than the As peaks (Fig. 7F, J-L).

The outer parts of the hydrothermal rims contain higher trace and minor element concentrations than the inner rims, up to 116 ppm Cu, 23,400 ppm As, 386 ppm Ag, and 672 ppm Au (Table 1; Fig. 7G-I). The analyses commence with As plateaus that are several hundred ppm As compared to plateau values near 0 ppm As in the cores and inner rims. Elemental correlations in the outer rims also vary by analysis for most element pairs. However, the Cu-As correlations are all strongly positive, and the Cu-Ag and Ag-As correlations are all weakly positive (Table 2). In the outer rims, zones of enrichment are typically correlated across all the elements, with the exception of a dissociation between Au and Ag in one analysis (Fig. 7G, H vs. I). The outer rims also possess several zones of Au enrichment, and some of the Au peaks are relatively broad (Fig. 7G, I), indicating thicker zones of Au enrichment than in the inner rims.

The δ³⁴S values of individual data points in the hydrothermal rims range from –1 to 17‰ (Fig. 7F-L). Depth profile δ³⁴S plateau values range from 1.2 to 11.0‰ (Table 3). There are differences in δ³⁴S between the dike pyrite grain cores and the hydrothermal rims, but the magnitude and direction vary. In three analyses, the lowest δ³⁴S values are associated with the most Au-rich and As-rich zones (Fig. 7E, F, I). The lowest δ³⁴S values in the entire Conrad data set come from inner rim hydrothermal pyrite moderately enriched in Au, with δ³⁴S around –1‰ and 3 to 15 ppm Au (Fig. 7F). In contrast, two analyses of the Conrad hydrothermal pyrite have higher δ³⁴S values in the Au- and As-rich zones and lower δ³⁴S in the Au- and As-poor zones (Fig. 7J, L). The highest δ³⁴S values in the data set are associated with relatively high Au (Table 3; Fig. 7I). The sulfur detectors malfunctioned in the two outer rim analyses with the highest Au values in the entire Conrad data set (Fig. 7G, H). In one analysis there is little difference between the low-As, low-Au pyrite grain core and the high–As-Au hydrothermal inner rim (Fig. 7K). None of the Conrad dike pyrite depth profiles cross the contact from the moderately Au-enriched inner hydrothermal rim to the highly Au-enriched outer rim, likely due to the coarse grain size.

Source contributions to the hydrothermal pyrite were modeled in MixSIAR using δ³⁴S as a tracer and Au as a covariate. The hydrothermal pyrites from Sunrise and Conrad were modeled as the “mixtures,” represented by 13 plateau values from the NanoSIMS data obtained in this study (App. 1). A description of the approach for calculating plateau values is included in the “Methods” section on presentation of the NanoSIMS data. The mixture data set (App. 1) includes plateau values from the Sunrise hydrothermal pyrite rims as well as the inner and outer hydrothermal rims from the Conrad dike pyrites. We used these values to calculate the corresponding δ³⁴S compositions of fluids in equilibrium with the hydrothermal pyrite at 200°C, using the pyrite-H₂S fractionation equation of Ohmoto and Rye (1979) to account for cooling from the temperatures of magma emplacement to the anticipated temperature of hydrothermal pyrite precipitation. This equation is relatively insensitive to temperature, and the effects of cooling are inconsequential relative to the variance in the mixture data set, since pyrite-H₂S fractionation only varies by 1.8‰ over the entire expected temperature range. In addition to the δ³⁴S model inputs, we used the depth profile Au concentration as a continuous covariate in the model, enabling us to determine how Au enrichment of the hydrothermal pyrite covaries with the contributions from the two sources. Sulfur and metals in the hydrothermal pyrite mixtures theoretically could have been sourced from any number of rock packages, including the sedimentary rocks that locally host ore or other sedimentary rocks beneath or adjacent to the orebodies; the Conrad gabbroic dike; other unknown intrusions in the district; or the crystalline Precambrian basement. At the deposit scale, pyrite values for Osiris and Sunrise range from 4 to 28‰ (Beaton, 2015); these data cannot be used to represent the source rocks because they are conventional analyses of microdrilled grains that mix hydrothermal, diagenetic, and base metal-stage pyrites, and therefore the data are already mixtures, obfuscating the particular compositions of the early-stage (nonhydrothermal) pyrite. At the regional scale, published δ³⁴S data are limited to broad surveys with wide isotopic ranges, or locations potentially too distant from the deposit area to be informative. For example, pyritic shales commonly have negative δ³⁴S values (Rollinson, 1993), but the δ³⁴S of pyrite in Neoproterozoic to Paleozoic siliciclastic rocks of the Selwyn basin can range from –30 to +35‰ due to bacterial reduction of sulfate (Goodfellow, 1987; Ross et al., 1995). Other available δ³⁴S data from host rocks and sulfide minerals are from distant locations, including the Cassiar terrane several hundred km south of the Dawson fault (Cooke and Goodwin, 1984; Turner, 2009; Rasmussen, 2013) and several sites in the Tombstone belt more than 150 km west of the Nadaleen trend. The latter include the Clear Creek area (Marsh et al., 2003), Brewery Creek (Lindsay, 2006), Keno Hill (Boyle, 1979), and the Paleocene Rackla pluton at the Tiger deposit about 80 km west of the Nadaleen trend (Thiessen et al., 2016).

Our existing NanoSIMS data provide an elegant solution to this lack of an exhaustive δ³⁴S data set: we can use the NanoSIMS data to test whether the observed compositions of the hydrothermal pyrite could be achieved by mixing from locally present units, or whether additional source contributions of sulfur and gold were required. Although the modeling cannot “rule out” the contributions of other source rocks, it can effectively “rule in” additional sources if the model cannot be solved with only two sources. This approach enables us to test whether the two-modeled sources of sulfur (the lime-mudstone and magmatic rocks with sulfur compositions matching the gabbroic dike) are sufficient to explain the sulfur in the hydrothermal pyrite mixtures. We used the NanoSIMS data for the pyrites in these two units (App. 1), calculating the depth profile plateau values for δ³⁴S as described in the “Methods” section. Although fractionation is expected between the hydrothermal pyrite “mixture” and the hydrothermal fluid from which it precipitates (accounted for as described previously for the mixture inputs to the model), fractionation is not expected between the sulfur sources and the hydrothermal fluid. First, given the geologic setting as previously described, sulfur would be transported as H₂S with negligible SO₂ and would remain consistently reduced from source to final ore pyrite, so thermochemical sulfate reduction and redox effects on sulfur fractionation would not be expected. No sulfur fractionation is expected between a magmatic source and an exsolving hydrothermal fluid, given the high temperatures. Finally, fractionation would not occur during stripping, for example during hydrothermal fluid circulation and leaching of host rock sulfur from a preexisting magmatic unit.

We used a two-source model configuration and a generalist or “uninformative” prior in MixSIAR, since no prior information was available on the proportional contributions. The model parameters are shown in Appendix 1. The model converged using the standard Markov Chain Monte Carlo chain length of 100,000. The diagnostics and summary statistics were reasonable given the limited size of the data set (App. 1). Histograms of the probability distributions for each of the two sources and the correlation coefficient of –1.00 (Fig. 10A) show a perfect inverse correlation between proportions of the two sources, as defined for a two-source model.

Most importantly, the two-source model can be solved, which demonstrates that the δ³⁴S compositions of the hydrothermal pyrite can be achieved using only the magmatic fluid that formed the Conrad gabbroic dike and the sedimentary pyrite. The model calculates the scaled relative probability of proportional source contribution of δ³⁴S to the hydrothermal pyrite, producing credible intervals (Stock et al., 2018) that are wide for both Conrad gabbroic dike pyrite and sedimentary pyrite over all samples (Fig. 10B). All proportional contributions are possible, although there is high probability that the contribution of sulfur from sedimentary pyrite is greater than the contribution from the Conrad gabbroic dike. At low Au concentrations there is high probability that there is little contribution from the Conrad pyrite and major contribution from the sedimentary pyrite (Fig. 10C), whereas at high Au concentrations there is high probability that most of the sulfur in the hydrothermal pyrite comes from a source isotopically equivalent to the Conrad gabbroic dike (Fig. 10D). At median Au concentration, most of the sulfur likely comes from the Conrad gabbroic dike pyrite, but there is still some contribution from the sedimentary pyrite (Fig. 10E). The source contribution varies with Au, with an increasing proportion of sulfur coming from the dike pyrite (or an isotopically equivalent source) as the Au concentration of the hydrothermal pyrite increases (Fig. 10F).

The pyrite data are consistent with the characteristics of Carlin-type gold deposits, including the pyrite types, morphologies, trace element concentrations, and zonation. At both Conrad and Sunrise, the hydrothermal pyrites contain Au, Cu, As, Ag, and Sb (Table 1; Fig. 11). The trace and minor elements are distributed in submicron-scale oscillatory and patchy zones, with elemental correlations that vary by zone and grain. At both Sunrise and Conrad, the hydrothermal rims overgrow earlier pyrite grain cores that contain minor Au and Cu alongside moderate As and Ag. Sunrise hosts more variety in hydrothermal pyrite styles than Conrad, potentially due to the presence of disseminated individual grains as well as the framboid clusters that served as substrates for the hydrothermal pyrite. Previous EPMA mapping of As by Beaton (2015) and Tucker (2015) clearly shows hydrothermal pyrite overgrowing coarse pyrite grains in sedimentary rocks from Osiris and Conrad, respectively. Both studies also identified spongy pyrite framboids at Osiris and Conrad, and the published reflected light and BSE images suggest that hydrothermal overgrowths are present on these grains, although they were not mapped compositionally in those studies. Some BSE images from both studies include abundant <1-µm bright spots that may be the very fine grained pyrite we observed in NanoSIMS maps. These previously published images corroborate with our observations and indicate the representativity of the sedimentary-hosted pyrite grains that we analyzed. One unresolved question is whether the very fine grained disseminated pyrite has cores of precursor sedimentary pyrite. Backscatter images from other studies suggest that this is arsenian pyrite that may have a core with lower As (Penny, 2020; Gopon et al., 2023; Gopon et al., unpub. data, 2023). The spatial resolution of our depth profile technique would be ideal for these grains if the optics of the NanoSIMS instrument could be improved.

An important difference between our dike-hosted and sedimentary-hosted samples is the relative grain size, which we attribute to the different host rocks. Even though the Conrad dike pyrite grain cores are approximately the same size as the coarser sedimentary pyrite grains, the dike-hosted hydrothermal rims are ubiquitously coarser than the sedimentary-hosted hydrothermal rims. The dike-hosted hydrothermal rims also contain thicker and more numerous zones of elemental enrichment. The concentrations of Au, Cu, As, and Ag are higher in the Conrad hydrothermal rims than at Sunrise based on our NanoSIMS spot analyses (Table 1). However, similar drill core assay grades (typically over 0.5 to 2 m) and similar pyrite modal abundances between the Sunrise and Conrad samples suggest that there is more Au in the samples than detected in our data set. Our highest-Au spot analysis from Sunrise ended in a high-Au zone and did not reach the outermost margin of the hydrothermal pyrite, where the NanoSIMS mapping indicates the highest Au concentrations occur. Furthermore, we were not able to obtain quantitative spot data from the fine-grained clusters or very fine grained disseminated pyrite styles. Since the measured Au concentrations of the coarser hydrothermal pyrites at Sunrise (this study) are not much higher than the Au grades of the samples (cf. Sack et al., 2019), we suspect that the very fine grained pyrite contributes significantly to the gold budget, and future studies should quantify the Au concentrations in this pyrite type.

At both Sunrise and Conrad, Au is present only where As is enriched, but not all zones of As enrichment contain Au. This results in the varied correlations between As and Au (Table 2). Numerous studies have investigated the role of As during Au incorporation within pyrite, showing that As governs the solubility limit for solid solution of Au in pyrite (Reich et al., 2005; Deditius et al., 2014; Gopon et al., 2019; Kusebauch et al., 2019). Our data suggest that Au occurs both in solid solution as Au⁺¹ and as Au⁽⁰⁾ nanoparticles (Fig. 12). Some analyses cross the solubility limit line, indicating that both types of Au may be found in a single pyrite grain. The presence of As is also thought to control portioning of Au into pyrite (Kusebauch et al., 2019), as well as the presence of Au in solid solution by inhibiting reduction of Au (Kusebauch et al., 2019), creating locally reduced conditions (Pokrovski et al., 2014) and facilitating Au-As redox reactions (Pokrovski et al., 2021). Since the As is consistently present, we infer that it is not the only factor determining the occurrence of Au as Au⁺¹ (cf. Kusebauch et al., 2019). We note that As can only scavenge and adsorb Au (cf. Kusebauch et al., 2019) if Au is present. Our data point to potential temporal or spatial variation in the availability of Au in the hydrothermal fluid during pyrite growth or remobilization postformation. The dominance of oscillatory zoning suggests that elemental distribution in the hydrothermal pyrite was controlled by availability rather than remobilization, crystallographic structure, or surface enrichment (Watson and Liang, 1995; Watson, 1996; Chouinard et al., 2005; Barker et al., 2009).

We hypothesize that the mechanisms of hydrothermal pyrite growth were consistent between the sedimentary and dike-hosted ores, given the observed similarities in the pyrites. Pinet et al. (2020b) determined that mineralized samples at Sunrise are significantly enriched in S but not Fe compared to their unmineralized counterparts. This suggests that the hydrothermal pyrite formed by sulfidation rather than pyritization. Nucleation of Au-rich hydrothermal pyrite occurred on the precursor pyrite in all of the studied pyrite types at Conrad and in the coarser styles at Sunrise; we presume this is also the case for the finest grained pyrite at Sunrise. Sulfidation requires scavenging of Fe from the host rocks. The spatial scale of Fe remobilization on the Nadaleen trend is unknown, although the width of flow paths at Osiris has been proposed to be mostly <3 m, and hydrothermal fluid transport at Conrad appears to be limited to numerous small flow paths and overall limited fluid-rock interaction (Steiner and Hickey, 2023). It is possible that the need for Fe is lessened by the substitution of hydrothermal fluid-derived As for Fe during hydrothermal pyrite growth, although As substitution for S is thought to be more common (Chouinard et al., 2005, Reich et al., 2005; Blanchard et al., 2007). The mechanisms of pyrite growth must also account for the observed differences between sedimentary-hosted and dike-hosted ores, the most salient of which is the coarser textures in the dike-hosted ores. We observed the same trend at Carlin-type deposits in Nevada, where hydrothermal pyrite rims are relatively thin on fine-grained sedimentary pyrite and relatively thick on the coarser dike and skarn pyrite (Holley et al., 2022, 2024). Collectively, the observations in Nevada and the Yukon suggest that the grain size of the hydrothermal pyrite is influenced by the grain size of the precursor pyrite. Two possibilities thus arise, with implications for district-scale exploration and deposit-scale ore control: either the narrower hydrothermal rims concentrate the same amount of Au in a smaller volume of material, or the coarser rims can accumulate more Au by virtue of their larger volume. In the latter case, units with coarser precursor pyrite would be more favorable mining targets. Testing would require systematic comparison of zone thickness and number, Au concentrations, and total Au grade across a suite of hydrothermal pyrites of varying grain sizes, including the micron-scale disseminated pyrite.

The NanoSIMS data and MixSIAR modeling provide insight into the potential sources of metals, metalloids, and sulfur in the pyrite. The δ³⁴S values in the sedimentary pyrite from Sunrise are consistent with marine shales, although regionally the range is much wider, with high values possible due to bacterial sulfate reduction (Goodfellow, 1987; Ross et al., 1995). The pyrites in the Conrad gabbroic dike have δ³⁴S plateau values of 3.7 to 11.2‰, slightly higher than the normal range of magmatic sulfur but typical for magmatic rocks intruding into sedimentary units. Similar δ³⁴S values have been obtained for intrusions in the Cassiar terrane (2.0–10.2‰; Rasmussen, 2013) and the Rackla pluton (6.0–6.6‰; Theissen et al., 2016), whereas intrusions in the Tombstone suite appear to have incorporated biogenic sulfide and are commonly isotopically negative (Marsh et al., 2003). The NanoSIMS analyses of the pyrite grain cores indicate that metals were present in the sedimentary basin during pyrite formation as well as in the Conrad gabbroic dike-forming magmas, although more Ag than Au, As, or Cu was incorporated into both these types of early pyrite.

The sulfur composition of the hydrothermal pyrite from Sunrise and Conrad is relatively similar to that of the pyrite grain cores in our samples. The depth profiles do not show abrupt changes at the contacts between grain cores and hydrothermal rims. This contrasts with observations from Carlin-type deposits in Nevada, where the hydrothermal rims are in the range of 0.0 to 8.8‰, and the sedimentary pyrite is either isotopically heavy (δ³⁴S up to 54.4‰) or light (as low as –33.5‰). Our Yukon samples are more like Eocene dike pyrite and Jurassic skarn pyrite from Carlin-type deposits in Nevada, with little sulfur fractionation between magmatic sulfide grain cores and hydrothermal pyrite rims (Holley et al., 2024). The relatively homogeneous δ³⁴S compositions in the Sunrise and Conrad pyrites makes the source contributions more cryptic. The wide credible intervals in the MixSIAR model result partly from the similar population means, as well as the variability of sulfur signatures within each of the sources. A larger data set of well-defined δ³⁴S data from possible sources would provide valuable information. Although the MixSIAR modeling cannot rule out the contributions of additional undefined sources, it does demonstrate that the compositions of the hydrothermal pyrite can be achieved by mixing sulfur from the local sedimentary pyrite with sulfur from the pyrite in the Conrad gabbroic dike.

At the highest Au concentrations, the modeled proportional source contribution of δ³⁴S is mostly from the Conrad gabbroic dike, with only minor contribution from the sedimentary pyrite. Our modeling suggests that the gabbro or an isotopically similar unit in the expected range for magmatic sulfur contributed both Au and sulfur to the hydrothermal pyrite. Importantly, this also means that a source with sulfur isotopes in the expected magmatic range cannot be ruled out. This raises the question of whether gabbro-forming magmatic activity drove mineralization on the Nadaleen trend.

Our modeling data are permissive of two possible scenarios:

1. The hydrothermal pyrite could have formed through amagmatic processes, which leached sulfur and metals from preexisting magmatic units. These leached units could have included the Conrad gabbroic dike and a larger underlying volume of magmatic rocks. The magmatic rocks would have acted as passive sources of δ³⁴S and Au, which would have been stripped, remobilized, and precipitated in the pyrite rims in a later hydrothermal event unrelated to the magmatism.

2. The hydrothermal pyrite could have formed from magmatic-hydrothermal processes, from fluids that were exsolved from a cooling magma.

Integrating our modeling with the available geochronology, the second scenario (magmatic-hydrothermal mineralization) appears most likely. The age of the Nadaleen trend mineralization is ~74 Ma, as indicated by U-Pb dates on hydrothermal calcite (Pinet et al., 2022b). This is contemporaneous with the age of the Conrad gabbroic dike at 74.4 ± 1.0 Ma. HeFTy inverse modeling of apatite fission tracks is permissive of mineralization around 74 Ma (Pinet et al., 2023), although a younger age of mineralization cannot be ruled out. The apatite fission tracks record a complex history with the youngest ages of resetting in the Eocene interpreted to represent either cooling of hydrothermal fluids or exhumation (Tucker, 2015; Tucker et al., 2018; Pinet et al., 2023).

Pinet et al. (2022b) relate the potential formation of the Nadaleen trend deposits to a regional episode of magmatism and associated metallogenesis at ~74 Ma. Although the 74.4 ± 1.0 Ma age of the Conrad gabbroic dike does not correlate with any other intrusions currently known in the Rackla belt, regional and high-resolution district-scale aeromagnetic data may indicate the presence of plutons at depth. A large positive magnetic anomaly occurs north of the Nadaleen trend, potentially indicative of a magnetite-bearing intrusion, and the Osiris and Sunrise deposits occur on the northeastern flank of a magnetic low that could indicate the presence of a reduced (ilmenite-bearing) intrusion.

At the regional scale, there is further evidence for Late Cretaceous magmatism around 74 Ma. The age of the Conrad dikes differs from magmatic episodes documented on the eastern side of the Tintina fault, including those associated with the Tombstone suite (96–90 Ma; Mair et al., 2006) and the emplacement of the McQuesten suite (67–64 Ma; Murphy, 1997) and the Rau pluton at 63 Ma (Thiessen et al., 2012). However, the 74.4 ± 1.0 Ma U-Pb zircon age of the gabbro is contemporaneous with an episode of magmatism on the western side of the Tintina fault, including formation of Cu-Mo-Au porphyry, epithermal, and skarn deposits, and the Casino and Prospector Mountain igneous suites (Colpron et al., 2016). These intrusions were followed shortly by the subaerial Carmacks Group volcanic succession from 72 to 67 Ma (Grond et al., 1984; Joyce et al., 1996). All of these units are compositionally suggestive of a mantle source (Pinet et al., 2022b). Pinet et al. (2022b) observed that prior to the right-lateral displacement on the Tintina fault, the Nadaleen trend deposits would have been closer to the other areas where mineralization occurred at ca. 74 Ma, and prior to faulting, the Nadaleen trend also could have been within the footprint of the Late Cretaceous Carmacks Group volcanics.

We envision that mineralization on the Nadaleen trend occurred during a Late Cretaceous magmatic episode that contributed heat, fluids, and metals to a magmatic-hydrothermal system, shortly after emplacement of the Conrad gabbroic dikes. Fluids exsolving from the causative pluton would have had very high H₂S/SO₂ ratios, remaining reduced during travel through the host rocks. The fluids dominantly migrated along fractures and preexisting veins, with relatively limited intergranular flow (Steiner and Hickey, 2023). Meteoric fluids carried sulfur from the sedimentary rocks, and magmatic-hydrothermal fluids carried sulfur and gold from the cooling magma. Wall-rock sulfidation by the mixing fluids led to the growth of oscillatory zoned, trace element-rich hydrothermal arsenian pyrite containing high Au concentrations in pyrite zones that derived greater contribution from the magmas. If indeed these deposits formed in association with a locally limited and deep-seated Late Cretaceous magmatic-hydrothermal system, Late Cretaceous intrusions should be considered an important exploration vector for Au mineralization in the district as well as more broadly in the region.

Most previous studies on the Nadaleen trend have made comparisons to Carlin-type deposits in Nevada, and several studies explicitly describe the Nadaleen trend deposits as Carlin-type (Beaton, 2015; Tucker, 2015; Tucker et al., 2018; Pinet and Sack, 2019; Pinet et al., 2020a, b). Our data show that the hydrothermal pyrites from the Nadaleen trend are texturally and geochemically similar to Carlin-type pyrite in Nevada. The Sunrise and Conrad pyrites contain both Au⁺¹ and Au⁽⁰⁾ (Fig. 12), as well as similar concentrations of As, Au, and Cu. The hydrothermal rims on coarse sedimentary pyrite are texturally most like Carlin-type hydrothermal overgrowths on sedimentary pyrite, based on NanoSIMS data from Carlin, Getchell, and Turquoise Ridge (Holley et al., 2022, 2024). The framboidal pyrite from Sunrise is similar to the spongy textures in Carlin-type ores described at Meikle (Emsbo et al., 2003; Simmons et al., 2020), Gold Quarry (Large et al., 2009), and Twin Creeks (Simon et al., 1999; Deditius et al., 2014). The scale of published analyses on spongy pyrite from Nevada is lower than our NanoSIMS mapping of the Nadaleen trend pyrite, but we posit that the Carlin-type spongy grains reported in the literature from Nevada may consist of numerous smaller grains cemented by finely zoned hydrothermal pyrite matrix similar to those observed in our high-spatial-resolution analyses of the grains from Sunrise. Spongy pyrite is also noted at numerous Carlin-like deposits of uncertain origin outside the Great Basin (Basori et al., 2018; Li et al., 2020; Wei et al., 2020; Lin et al, 2021). The hydrothermal pyrites from Conrad are texturally similar to hydrothermal overgrowths on earlier magmatic pyrite at the Deep Star Carlin-type deposit and the Red Dot Carlin-like deposit (Holley et al., 2024). There are no published accounts of the very fine grained disseminated pyrite at Carlin-type deposits in Nevada, although the scale of most analytical methods would be insufficient to identify it. We did not encounter this pyrite type during NanoSIMS mapping of Carlin-type pyrite from Deep Star, Carlin, Turquoise Ridge, and Getchell (Holley et al., 2024) and therefore surmise that this pyrite type is not widespread in Nevada if it does occur. Although the pyrite characteristics are largely consistent between Carlin-type gold deposits and the Nadaleen trend, the characteristics of the pyrite alone cannot prove a Carlin-type origin for the Nadaleen trend pyrite, since such textures can form by various processes (e.g., oscillatory As-zoned arsenian pyrite overgrows authigenic pyrite in areas of As-contaminated groundwater such as Bangladesh; Lowers et al., 2007).

The δ³⁴S data and modeling provide evidence that the Nadaleen trend pyrite could have formed by the same processes as Carlin-type gold deposit pyrite in Nevada. In both Nevada and the Nadaleen trend, the Au-rich pyrite formed during mixing of magmatic and meteoric fluids (Fig. 10F-H). Further studies should compare other key processes in Carlin-type gold deposit mineralization to the processes that were active during Au mineralization on the Nadaleen trend. For example, in Nevada, extensional tectonism was a key contributor to mineralization, generating magmatism and the development of structural conduits for hydrothermal fluids. There is no evidence for widespread Cretaceous or younger extension in the Yukon, so we infer that other processes drove fluid circulation and the development of fluid conduits. In Nevada, the Au in Carlin-type deposits was contributed by Eocene mantle-derived melts generated after a long period of quiescence (Muntean et al., 2011; Holley et al., 2022), and the preliminary evidence for synmineralization mantle-derived magmatism on the Nadaleen trend suggests that similar processes were operating in the Yukon during the Late Cretaceous.

Hydrothermal pyrite on the Nadaleen trend occurs in a variety of textures, overgrowing precursor sedimentary and dike-hosted pyrite grains, and as micron-scale disseminated grains. Hydrothermal pyrite was studied from a single sedimentary-hosted interval at Sunrise and a single dike-hosted interval at Conrad. This pyrite is zoned at the nanoscale due to variation in Au, As, Ag, Cu, and Sb, although the correlations among these elements vary between grains. The Au appears to be hosted as both Au⁽⁰⁾ and Au⁺¹, with both types present in the same grain in some cases, suggestive of temporal or spatial variation in Au availability in the hydrothermal fluid during pyrite growth, or potentially the remobilization of Au after mineralization. Coarser hydrothermal rims occur on coarser precursor pyrite, indicating that the substrate places controls on the characteristics of the pyrite during sulfidation, potentially due to the availability of Fe. The coarsest rims occur on dike-hosted pyrite, which may also indicate host rock control on hydrothermal rim growth. The characteristics of the Nadaleen trend pyrite are strikingly similar to Carlin-type pyrite from Nevada, and Bayesian mixing models show that the pyrite in both regions formed by similar processes. Together these lines of evidence support a Carlin-type model for the Nadaleen trend deposits. On the Nadaleen trend, pyrite source contributions were derived from local sedimentary pyrite and magmatic sulfur, likely contemporaneous with a Late Cretaceous episode of magmatic-hydrothermal activity. Continued geologic framework studies are needed to link our evidence for site-specific characteristics and processes to the regional metallogenic setting in which these deposits formed.

We dedicate this contribution to Julia Lane. The work was funded by NSF Career Award EAR-1752756 to EAH. The Stanford Nano Shared Facilities are supported by NSF ECCS-2026822. We thank ATAC Resources for samples and data, and we are grateful to Adam Coulter, Patrick Mercier-Langevin, and Nicolas Pinet for discussion. Anne Fulton prepared the samples and operated the NanoSIMS with CJR. Jae Erickson, Kelsey Livingston, Katharina Pfaff, and Sage Langston-Stewart helped with sample preparation and analyses at Colorado School of Mines. Brian Stock provided helpful suggestions on the MixSIAR modeling. We thank Crystal LaFlamme, Nicholas Pinet, and an anonymous reviewer for providing comments on the manuscript.

Elizabeth Holley is an Associate Professor at Colorado School of Mines, specializing in mineral exploration and mining geology. This work was conducted under a National Science Foundation (NSF) Career Award on the role of magmatism in sedimentary-hosted gold mineralization. Dr. Holley also leads the NSF-funded Responsible Critical Minerals project and is the site director for the NSF Center to Advance the Science of Exploration to Reclamation in Mining. She is a fellow of the Payne Institute of Public Policy and a fellow of the SEG.