Porphyry deposits supply the bulk of the world’s Cu and Mo and significant amounts of Au, as well as other minor and trace metal(loid)s, including Ag, Re, Te, Pd, Se, Bi, Zn, and Pb. Porphyry deposits are gaining in importance as a source of critical raw materials with the increasing global demand for these commodities. To date, minor and trace metal(loid)s are still commonly recovered as by-products from porphyry ores without prior characterization of their host mineralogy that could inform more efficient processing and improved recoveries. We report a comprehensive metal(loid) deportment study on a complete vein paragenetic series in samples from the northwestern high-grade zone of the Bingham Canyon Cu-Mo-Au porphyry deposit, Utah. The polyphase Bingham stock comprises an early premineralization equigranular monzonite phase that was intruded by a series of five successive, ore-related porphyry intrusions. Veins with hypogene Cu-(Fe) sulfide assemblages from all five porphyry intrusions were characterized for their trace metal(loid) contents by laser ablation-inductively coupled-mass spectrometry (LA-ICP-MS).

It was found that bornite and digenite contain elevated Bi, Ag, Te, and Se relative to chalcopyrite, whereas the latter contains elevated concentrations of Co, Ga, and In. A stepwise decline in sulfide abundance occurs over the porphyry intrusion sequence and is more pronounced in digenite and bornite than in chalcopyrite. The related diminishing concentration per rock volume (inventory) of Bi, Ag, Te, and Se in the youngest porphyry dikes could have been caused in part by a geochemical change in the mineralizing fluid supply across successive intrusive-hydrothermal cycles.

Element mapping of exsolved digenite within bornite revealed characterstic partitioning of metal(loid)s between bornite and digenite; most notably Ag, but also Te and Au are enriched in digenite relative to enclosing bornite. Bornite domains within these composite grains reveal complex zonation of Sn, In, and Bi, which are attributed to stress-induced diffusion within bornite, resulting from the digenite exsolution process. The selective partitioning of metal(loid)s between bornite and digenite is likely a common feature in many porphyry Cu deposits, given the fundamental mineralogical characteristics of these two sulfides. Our results contribute to an improved understanding of the distribution (from mineral to deposit scale) of critical trace metal(loid)s in porphyry deposits, particularly those containing exsolved digenite. This knowledge can be applied to determine more accurately the value of ore resources, to improve geometallurgical models and by-product recoveries, and to help limit the environmental effects of metal(loid) dispersion.

Bulk concentrations of major commodities such as Cu, Mo, and Au are well constrained for most large porphyry deposits. For some of these deposits, minor and trace metal(loid) s, including Ag, Re, Te, Pd, Se, Bi, Zn, and Pb are routinely analyzed in geochemical assay programs and are recovered as by-products (Sillitoe, 2010). Several of these elements are key components in energy production, conversion, and storage technologies and have been classified as energy-critical elements (Jaffe et al., 2011). Selenium and Te, for example, are important constituents of thin film photovoltaics and are almost exclusively extracted by refining Cu concentrates (U.S. Geological Survey, 2019). Porphyry Cu deposits also contain other energy-critical elements, such as In, Co, Ga, and Ge and platinum group elements (see Yano, 2012; Cioacaˇ et al., 2014; Crespo et al., 2018). If recovered, these elements could potentially increase the value of the ore but despite the growing importance of energy-critical elements, their deportment and distribution within porphyry ores have only recently begun to receive focused research attention; this is partly due to a lack of routine analyses. In addition to a scarcity of data on energy-critical elements, information on the deportment and distribution of penalty elements in porphyry ores (Fountain, 2013) such as As, Cd, Sb, Hg, Pb, and Bi are also scarce.

Advancements in microanalytical techniques and the growing demand for critical raw materials have led to an increasing number of trace element deportment studies of porphyry Cu ores (e.g., Tarkian et al., 2003; Reich et al., 2010, 2013a; Cook et al., 2011; Yano, 2012; Cioacaˇ et al., 2014; George et al., 2016; McFall, 2016; Zarasvandi et al., 2018; Keith et al., 2018; Crespo et al., 2020; Aird et al., 2021; Rivas-Romero et al., 2021). This research has shown that the major hypogene Cu-(Fe) sulfides in porphyry ores, chalcopyrite (e.g., George et al., 2018), and bornite (e.g., Cook et al., 2011) are important hosts of a range of trace metal(loid)s, including Bi, Se, Ag, Te, Au, and Zn. Supergene Cu sulfides (digenite) have been found to incorporate Ag, Au, As, Sb, Se, and Te in their crystal structure (e.g., Reich et al., 2010). Apart from the major Cu-(Fe) ore sulfides, electrum and tellurides can be important host minerals for Au, Ag, and Te in porphyry Cu ores (Kesler et al., 2002; Crespo et al., 2018; Kojima et al., 2021). Although pyrite is not an ore mineral in porphyry Cu ore, it is an important host for Co, Ni, As, Au, Ag, and Cu, among other elements in many porphyry deposits (Reich et al., 2013b; Deditius et al., 2014; Crespo et al., 2020; Rivas-Romero et al., 2021). If present in the porphyry Cu ore, sulfosalts such as tetrahedrite, tennantite, and enargite can host a wide range of metal(loid)s, in particular As, Sb, Fe, Ag, Cu, Zn, and Pb, but also Cd, Hg, Bi, Te, and Se (Deyell and Hedenquist, 2011; George et al., 2017; Crespo et al., 2020). Despite this growing knowledge of energy-critical element abundance and association in porphyry deposits, the trace metal(loid) distributions at the mineralogical and the deposit scales remain undocumented for most deposits. Most recently, Rivas-Romero et al. (2021) and Crespo et al. (2020) have examined deposit-scale metal(loid) deportment in the world-class Chuquicamata and Río Blanco porphyry deposits.

Here we aimed to investigate the spatial and temporal trace metal(loid) deportment within the world-class Bingham Canyon porphyry Cu-Mo-Au deposit, Utah. We systematically examined the hypogene Cu minerals, comprising chalcopyrite, bornite, and digenite from the high-grade ore zone on the northwest side of the deposit, referred to as the QMP-LP zone by Redmond (2002).

The Bingham Canyon deposit has been mined and studied for more than a century and the sequence of porphyry intrusions (Stringham, 1953; Bray, 1969; Moore, 1973; Moore and Czamanske, 1973; Lanier et al., 1978; Babcock et al., 1995; Phillips et al., 1997; Redmond, 2002) and the paragenetic sequence of ore formation (Redmond, 2002; Redmond et al., 2004; Landtwing et al., 2005, 2010) are well documented. Using vein truncation relationships, coupled with abrupt changes in Cu-Au grades, sulfide ratios, and potassic alteration intensity at porphyry intrusive contacts, Redmond and Einaudi (2010) concluded that the mass of introduced Cu and Au decreased notably during successive porphyry intrusive-hydrothermal cycles.

In this study, we analyze trace element concentrations of Cu-(Fe) sulfides from paragenetically well-constrained samples from each successive porphyry intrusion, using samples from Redmond (2002) and referred to in Redmond et al. (2004) and Redmond and Einaudi (2010). The objectives were to document and identify the host phases of energy-critical and precious metals and to investigate whether these rare metal(loid)s are contained within the crystal structure of Cu-(Fe) sulfides, in microinclusions therein, or as discrete minerals. These questions were addressed using time-resolved laser ablation profiles (e.g., George et al., 2015) and backscattered electron imaging. In addition, the trace element distribution between bornite and digenite was examined to qualitatively determine the partitioning behavior of trace elements during digenite exsolution, since digenite exsolution textures in bornite are common in samples from the high-grade ore zone at Bingham Canyon (Redmond, 2002).

Regional geology

The Bingham Canyon porphyry Cu-Mo-Au deposit is located in the Oquirrh Mountains of northern Utah, about 30 km southwest of Salt Lake City. The deposit is associated with a late Eocene composite pluton, the Bingham stock (Butler et al., 1920) that belongs to the Stockton-Park City intrusive belt (Waite et al., 1997) emplaced along the Uinta arch, an E-W-trending lineament. The Bingham stock intruded into Paleozoic siliciclastic and carbonate rocks after a change from regional compression to minor extension (Presnell and Parry, 1996). Basin and Range extension-induced block faulting during the Oligocene to Pliocene tilted the deposit eastward by 10° to 25° (Atkinson and Einaudi, 1978; Lanier et al., 1978).

The polyphase Bingham stock comprises an early premineralization equigranular monzonite phase that was intruded by a series of ore-related porphyry dikes (Fig. 1). At least five distinct crosscutting porphyry phases have been recognized at Bingham Canyon in a series of investigations (Stringham, 1953; Bray, 1969; Moore, 1973; Moore and Czamanske, 1973; Lanier et al., 1978; Babcock et al., 1995; Phillips et al., 1997), culminating with Redmond and Einaudi (2010) formally defining the intrusive sequence as follows: (1) quartz monzonite porphyry (QMP), (2) latite porphyry (LP), (3) biotite porphyry (BP), (4) quartz latite porphyry breccia (QLPbx), and (5) quartz latite porphyry (QLP). Relative ages were defined by vein truncation and crosscutting relationships. Each porphyry intrusion is characterized by a similar sequence of vein types, potassic alteration, and sulfide mineralization (Redmond and Einaudi, 2010). Vein abundance and the intensity of potassic alteration decreased from early to late intrusive phases. Mineralization-related hydrothermal alteration began about 0.75 m.y. after emplacement of the equigranular monzonite, inferred from zircon U/Pb ages of the monzonite phase (38.55 ± 0.19 Ma; Parry et al., 2001) and hydrothermal biotite 40Ar/39Ar ages in the porphyry intrusions, ranging from 37.74 ± 0.11 to 37.07 ± 0.21 Ma (Deino et al., 1997; Parry et al., 2001).

Porphyry intrusions

The QMP is an elongate NE-striking and NW-dipping (55°–60°) porphyry dike that intruded along the northwest contact between the equigranular monzonite and Paleozoic sedimentary rocks (Boutwell et al., 1905; Stringham, 1953; Bray, 1969; Fig. 1A, C). The QMP is characterized by 50 to 60 vol % phenocrysts (plagioclase, orthoclase, hornblende, book biotite, rare quartz eyes) in an aplitic groundmass (Redmond, 2002). The QMP is the volumetrically largest porphyry intrusion with a strike length of 1,500 m and near-uniform width of 350 m, extending to at least 2-km depth below the premining surface (Redmond and Einaudi, 2010).

The LP cuts the QMP and comprises a suite of narrow (5–25 m thick) N-NE-striking, steeply dipping (65°–70°) dikes and sills (Bray, 1969; Moore, 1973; Lanier et al., 1978) that extend over a length of more than 3 km along the northwest margin of the intrusive stock. Mineralogically, the composition of the LP and QMP are similar (Redmond, 2002) but the LP groundmass tends to be finer grained and darker, and phenocrysts are smaller in the LP than the QMP.

The BP is a narrow (2–4 m), dark brown-colored dike in the QMP-LP zone that strikes in a northwest direction (Fig. 1B) with a length of about 500 m (Redmond and Einaudi, 2010). It has a similar strike as the other porphyry intrusions and dips northwest (60°–65°). The BP is characterized by a high abundance of biotite phenocrysts (12–15%). Other phenocrysts include plagioclase and lesser K-feldspar and hornblende, with feldspar antecrysts exhibiting embayed edges consistent with partial resorption in a mafic magma (Redmond, 2002).

The QLPbx forms irregular shaped bodies; the largest mapped example had a variable width of 10 to 20 m and extended for ~150 m along the NNE-striking QMP-LP contact (Redmond and Einaudi, 2010). The QLPbx comprises abundant angular to subrounded wall-rock xenoliths, including QMP, LP, and BP fragments up to 10 cm across. Phenocryst composition is dominated by plagioclase over K-feldspar, with lesser book biotite, biotitized hornblende, and minor quartz eyes.

The youngest intrusive phase, the QLP, is typically less than 10 m in width, dipping steeply (75°–80°) to the NW, and occurs as dikes and irregular-shaped bodies as first described by Stringham (1953). The main NE-trending QLP dike has a strike length of at least 2 km, with a number of NW-trending apophyses, the location of which was apparently controlled by NW-striking faults (Redmond, 2002). The QLP is commonly gray to brown colored; feldspars dominate phenocryst composition over quartz and biotite, with the groundmass comprising the same minerals (Bray, 1969). Nonbiotitized QLP is greenish in color, owing to the presence of unaltered amphibole in the groundmass and as phenocrysts (Wilson, 1978).

Vein types and alteration

The paragenetic sequence of vein formation in each porphyry intrusion (Redmond and Einaudi, 2010) begins with early barren biotite veinlets (Phillips et al., 1997), followed by weakly mineralized, early dark micaceous (EDM) veins that predate multiple generations of quartz stockwork veins (A-quartz veins, after Gustafson and Hunt, 1975). A-quartz veins have K-feldspar alteration halos and host the majority of the Cu-Au mineralization at Bingham Canyon. Although A-quartz veins occur in all porphyry intrusions, the highest A-quartz vein abundance occurs within the QMP. Similar veins are also found in the premineralization equigranular monzonite phase and in the surrounding sedimentary rocks (Gruen et al., 2010). Five distinct A-quartz vein types (A1–A5) were defined by Redmond and Einaudi (2010); the youngest vein type, A5, can only be resolved using cathodoluminescence (CL) microscopy. The A5 veinlets cut both A1 to A4 veins and the porphyry wall rock but tend to be more abundant in earlier A1 to A4 veins. Quartz-molybdenite veins postdate all porphyry intrusions and are in turn crosscut by late barren quartz-sericite-pyrite veins (Lanier et al., 1978), or D veins after Gustafson and Hunt (1975).

In the high-grade core of the Bingham Canyon deposit, the intensity of potassic alteration can be directly correlated with A-quartz vein abundance. In this zone, Redmond and Einaudi (2010) identified distinct changes across porphyry contacts from older to younger porphyry intrusions, including: (1) sharp decreases in A-quartz vein abundance, (2) intensity of potassic alteration, (3) bornite/chalcopyrite ratios, and (4) Cu-Au grades. However, with the exception of the BP, all porphyry intrusions locally contain zones of high quartz vein abundance with associated intense potassic alteration and elevated Cu-Au grades. Texture-destructive sericitic alteration is rare in the high-grade core of the deposit and is limited to areas where late D veins are most abundant and alteration selvages overlap (Lanier et al., 1978; Babcock et al., 1995). Late argillic alteration, recorded as smectite, illite ± kaolinite replacement of plagioclase phenocrysts, is most commonly associated with N-NE striking structures (Atkinson and Einaudi, 1978; Lanier et al., 1978) and seems unrelated to vein formation or sulfide mineralization (Redmond and Einaudi, 2010).

Cu-(Fe) sulfide mineralization

Sulfide mineralization in A-quartz veins comprises chalcopyrite, bornite, digenite, and rare molybdenite. The bulk of the Cu-(Fe) sulfides in A1 to A4 veins is interpreted as having been introduced by later A5 veinlets (Redmond, 2002; Landtwing et al., 2005; Redmond and Einaudi, 2010). Coeval precipitation of bornite and chalcopyrite into secondary pore space, formed by retrograde dissolution of earlier quartz, is texturally indicated by cathodoluminescence microscopy. Fluid inclusion microthermometry of the dark-luminescing A5 quartz showed that precipitation of Cu-(Fe) sulfides occurred between 350° and 425°C (Redmond, 2002; Redmond et al., 2004; Landtwing et al., 2005, 2010).

The overall abundance of Cu-(Fe) sulfides generally decreases from the oldest to youngest porphyry intrusions, with some minor local exceptions (Fig. 2); the QMP hosts the highest abundance of mineralized veins among all porphyry intrusions. Also the ratios of the main Cu-(Fe) sulfides: chalcopyrite, bornite, and digenite vary within individual porphyry intrusions and change abruptly across porphyry contacts (Redmond and Einaudi, 2010). Chalcopyrite is ubiquitous and the dominant Cu-(Fe) sulfide in the LP, QLPbx, and QLP. By contrast, bornite is the most abundant Cu-(Fe) sulfide phase (with lesser chalcopyrite) in the QMP and BP; veins in the QMP exhibit the highest bornite/chalcopyrite ratios (greater than 1/1 and as high as 10/1), particularly in a high-grade zone intersected by drill hole D211 (Fig. 1D). With the exception of a single high-grade sheeted A-quartz vein sample (Redmond, 2002), bornite is generally absent in the QLP. Digenite is common in veins from the QMP but is rare in the LP, BP, and QLPbx and absent in the QLP.

The samples examined in this study were collected by Redmond (2002) from the QMP- and LP-hosted high-grade Cu-Au zone (>1% Cu and >1 g/t Au), on the northwest side of the Bingham Canyon deposit. Samples selected for this study are from three diamond drill holes (D109, D195, and D211), as well as benchface samples from the open pit (bench elev 4,890, 4,990, and 5,090 ft). They represent mineralization from all five porphyry dikes. In advance of LA-ICP-MS analysis, polished thin sections were petrographically examined and characterized using reflected light microscopy and scanning electron microscopy (SEM). This initial petrography guided the selection of targets for chemical analysis, represented by Cu-(Fe) sulfides and inclusion-poor areas within individual sulfide grains.

Major element compositions of bornite, chalcopyrite, and digenite, hosted in A-type veinlets, were determined using a Tescan Mira field emission gun-scanning electron microscope (FEG-SEM, TESCAN), equipped with an Oxford Instruments X-MaxN 80 mm2 energy dispersive spectrometer (EDS). Data were obtained with the AZTEC X-ray microanalysis software (Oxford Instruments); rigorous control over instrument conditions and quantitative calibration of major element concentrations was achieved with standard materials from the MAC standard 217 block (Micro-Analysis Consultants Ltd.). Samples were analyzed with an acceleration voltage of 20 KV and a beam current of 250 pA. For each analysis, at least 1 M counts were accumulated. Analyses were screened using element totals (99–101 wt %). In three of the samples, trace minerals with a minimum grain size of 5 µm2 were detected with the Oxford Instruments “Feature Analysis” tool of AZTEC software. For this, brightness thresholds were defined to select and analyze only minerals by SEM-EDS that appear brighter than all Cu-(Fe)-sulfides in backscattered electron images due to the presence of elements of higher mass (e.g., Se, Ag, Te, Au, Pb, and Bi). Trace minerals were further classified and separated from other detected high-mass phases (e.g., monazite) by defining threshold concentrations.

The element chemistry of Cu-(Fe) sulfides was analyzed in 14 thin sections by laser ablation-inductively coupled-mass spectrometry (LA-ICP-MS) at Trinity College, Dublin. The setup couples a Teledyne Photon Machines G2 (Teledyne Technologies, Inc.) 193-nm excimer Ar-F laser to a Thermo-Fisher Scientific iCAP-Qc quadrupole mass spectrometer. The laser ablation system is equipped with a HelEx ll active 2-volume ablation cell (Teledyne Cetac Technologies). After the sample is ablated in an He atmosphere, the generated aerosol passes a pulse homogenizing device before entering an inductively coupled plasma and eventually the mass spectrometer. The laser ablation system was operated with Chromium 2.1 software (Teledyne Photon Machines Inc.), and data were acquired with Qtegra (version 2.2, Thermo-Fisher) in time-resolved analysis mode. The dynamic aperture was used to project square-shaped laser spots of 30 µm in size. A shot count of 180, combined with a repetition rate of 6 Hz yielded an ablation of 30 s for each spot analysis. For an adequate cell washout between ablations, a delay of 20 s was used. A total of 24 elements were analyzed (Table 1).

Dwell times were chosen according to expected relative elemental abundances, determined by preceding test spots on the samples and standard reference materials. The 115Sn interference on 115In was corrected, using ratios of the natural unfractionated isotope abundances, (115Intrue = 115Inraw – 0.0148637 × 118Sn). At the beginning and end of each experiment, five MASS1 (external standard, Wilson et al., 2002) and five UQAC-FeS-1 (quality control standard, Savard et al., 2018) spot analyses were performed. For every 20 sample analyses, both standards were analyzed in triplicate, allowing correction for instrumental drift over the course of an experiment. Detection limits are presented in Table 1 while monitored accuracy and precision are shown in Table A1 in the Appendix.

Data reduction was performed with the software package Iolite 3.65 (University of Melbourne), implemented in Igor Pro 6.37 (Wavemetrics Inc.). For spot analyses, concentration data were obtained by applying the data reduction scheme “Trace-Elements” (Woodhead et al., 2007) in “Internal Element Standard” mode. Fully quantitative concentrations were calculated applying the internal standardization method (Ulrich et al., 2009, 2011). For Cu-(Fe) sulfides, Fe was used as an internal standard element, and similarly Cu for digenite. Average abundances of Cu and Fe were derived from SEM-EDS analyses (Fe concentration in chalcopyrite: 30.70 ± 0.2 wt % σ, n = 119; Fe concentration in bornite: 11.4 ± 0.3 wt % σ, n = 110; Cu concentration in digenite: 76.1 ± 1.5 wt % σ, n = 24; uncertainties are presented as one standard deviation (σ).

Principal component analysis (PCA) was applied to the chalcopyrite and bornite data sets using IOGAS software; the digenite data set was not considered large enough to conduct meaningful statistical analysis. A centered log ratio transformation was performed prior to PCA.

Trace element maps of selected bornite-digenite composite grains were also acquired by LA-ICP-MS. The chemical maps were generated by ablating adjacent lines of equal length with a 1.0-µm overlap. The number of analytes was restricted to allow maximum dwell time (duty cycle of 300 ms) for all elements and maximize x-resolution calculated as follows: x-resolution [µm] = scanning speed [µm/s] × total dwell time [s]. Along with 57Fe (20 ms) and 65Cu (20 ms) as internal standard elements, the distributions of the following isotopes were analyzed (dwell times in brackets): 71Ga (40 ms) 77Se (25 ms), 109Ag (25 ms), 115In (40 ms), 118Sn (20 ms), 125Te (40 ms), 197Au (50 ms), 209Bi (20 ms). Individual dwell times of the analytes were defined according to relative abundances, recorded by running test lines on the samples and standard materials. Each map was bracketed by 2 × 5 MASS1 and 2 × 5 UQAC-FeS-1 (quality control standard) line scans with identical laser parameters as applied to the map. Typical mapping parameters were as follows: beam size: 12 µm, laser repetition rate: 12–20 Hz, scanning speed: 8–12 µm/s; fluence: 1.1 J/cm2. Data reduction followed the same general workflow as for spot analysis but trace element distribution maps were generated in semiquantitative mode, using the “Image From Selections” module and the commonly used “ColdWarm” color scheme (Ulrich et al., 2009; Ubide et al., 2015; Cook et al., 2016). Concentration limits were selected to highlight internal inhomogeneity.

Cu-(Fe) sulfide mineralogy

In samples from the high-grade core of the Bingham Canyon deposit, the Cu-(Fe) sulfide assemblage consists of chalcopyrite, bornite, and digenite. Mutual grain boundaries between bornite and chalcopyrite are constantly sharp, whereas digenite exclusively forms a mesh-like network of fine µm-scale lamellae within bornite (Fig. 3); this has been referred to as “grating-textured digenite” (Redmond, 2002). The irregular anhedral shapes of the Cu-(Fe) sulfide grains are ubiquitous in A-type quartz veins from all porphyry intrusions and occur in pore space created by quartz dissolution. Aside from chalcopyrite and bornite forming single-phase grains, three types of composite grains are common: (1) grating-textured digenite enclosed in bornite, along with chalcopyrite (Fig. 3A, B); (2) bornite enclosing grating-textured digenite without chalcopyrite (Fig. 3C); and (3) digenite-free bornite and chalcopyrite with mutual boundaries (Fig. 3D-F). These observations confirm the general distinction of two types of bornite occurrences at Bingham Canyon, one free of digenite, the other hosting grating-textured digenite. The latter is widely described as an exsolution texture (e.g., Ramdohr, 1969), reflecting digenite exsolution from a Cu-rich bornite solid solution.

Chemical composition of Cu-(Fe) sulfides

Chalcopyrite compositions plot close to stoichiometric values, with an average Fe content of 30.7 ± 0.2 wt % σ and average Cu contents of 34.4 ± 0.2 wt % σ (Fig. 4, Table 2). Digenite compositions are more variable with an average Cu content of 76.1 ± 1.5 wt % σ and an average Fe content of 1.8 ± 0.9 wt % σ. Bornite in composite grains with digenite overlaps with stoichiometric bornite (avg Cu and Fe contents of 63.2 ± 0.4 wt % σ and 11.4 ± 0.2 wt % σ, respectively), whereas digenite-free bornite grains are slightly Cu depleted relative to stoichiometric bornite (avg Cu contents of 62.6 ± 0.6 wt % σ).

Comprehensive trace element concentration data for Cu-(Fe) sulfides are provided in Tables A2 to A9 and a summary is listed in Table 3. Box and whisker plots of trace element mineral deportment are shown in Figure 5. Bismuth exhibits the highest concentration contrast between the Cu-(Fe) sulfide phases; medians are on the order of 1,000 ppm in bornite and digenite and only 1 ppm in chalcopyrite. Median Se contents reach levels of a few hundreds of ppm in all Cu-(Fe) sulfides, ranging from 240 ppm (chalcopyrite) to 730 ppm (bornite), with intermediate contents of 580 ppm in digenite. Silver concentrations are on the hundred parts per million order for digenite-free bornite and digenite (280 and 350 ppm, respectively), but distinctly lower for bornite in digenite-bornite composite grains (71 ppm); median Ag contents are lowest in chalcopyrite (2 ppm). Overall, Sn contents (10–100-ppm range) exhibit the smallest relative variations between the different sulfide phases. Cobalt, As, Cd, Sb, and Te contents are commonly in the 1- to 10-ppm range, while Ga (except for chalcopyrite), Ge, In, and Au only occur at sub-parts per million levels. Chalcopyrite is the primary host for Co, Ga, and In (median contents of 2.6, 2.9, and 2.8 ppm, respectively). Digenite yields the highest median concentrations in Ag, Te, and Au (325, 7.0, and 0.28 ppm, respectively), whereas Se and Bi contents are highest in bornite (median contents of 605 and 1,606 ppm).

Energy-critical elements and precious metal-bearing trace minerals

Aside from Cu-(Fe) sulfides, the presence of additional trace minerals that may contain weight percent concentrations of energy-critical elements and precious metals was also investigated. Selenium-rich galena, Ag- and Ag-Au-tellurides (hessite, petzite), electrum, and wittichenite (Cu3BiS3) were identified with SEM-EDS (Fig. 6). The largest grains of these trace minerals commonly occur along bornite-chalcopyrite grain boundaries and less commonly along molybdenite-bornite grain boundaries. Determination of the abundance of trace minerals is challenging and submicrometer-scale minerals cannot be quantified reliably with SEM-EDS. Trace minerals with a minimum grain size of 5 µm2 in thin section were detected with the Oxford Instruments “AZTEC” software in-built “Feature Analysis” tool. Due to the small grain size of the target grains and the large size of scanned areas (approx 30–40% of a thin section), a full quantification of the compositions of trace minerals could not be achieved with reasonable counting times. Therefore, qualitative results of mineral abundance and grain size were recorded rather than fully quantitative mineralogy.

Trace mineral abundances are highly variable between the three screened samples (Table 4). Galena inclusions are most common (up to 53 occurrences), followed by wittichenite (up to 14 grains), which is preferentially associated with digenite. The majority of galena inclusions was found in a sample (109–217) associated with chalcopyrite replacing bornite along fractures. Silver tellurides, Ag-Au tellurides, and electrum grains are even rarer. Where present, these are typically less than 5 µm2 in size; only five of these trace phases in excess of 5 µm2 were found within the screened areas. However, they are larger in size than galena and wittichenite, illustrated by relatively high-percentage mineral proportions by area, despite low abundance in Table 4; the largest gold grain identified was ~50 µm in diameter.

LA-ICP-MS trace element distribution maps

Five bornite-digenite composite grains were mapped using LA-ICP-MS to examine the trace element distribution within and between coexisting sulfide phases and to investigate element partitioning between phases. Figure 7 illustrates representative element distribution patterns that were consistently observed within individual thin sections. The Fe distribution map clearly shows the outline of the digenite crystal centered within bornite. Selenium is homogeneously distributed between the two phases, whereas Ag is strongly enriched in digenite relative to bornite. Silver-rich inclusions are abundant in both phases. Interestingly, the In and Sn distribution maps show a distinct zonation feature; a 30- to 60-µm-wide zone (halo) of very low In and Sn concentrations within the bornite surrounding the inner digenite phase. Disregarding this inner low concentration halo in bornite, Sn is elevated in bornite relative to digenite; indium contents are similar in both phases with local zones of elevated In within bornite. Indium-rich inclusions occur at the boundary between digenite and bornite and also within bornite. Tellurium concentrations are enriched in digenite over bornite with the highest concentrations along the boundary between the sulfide phases and low Te concentrations (near detection limit) within bornite. In some composite grains, Te concentrations were below the detection limit in both phases. Gold is slightly enriched in digenite over bornite. Rare Au-rich inclusions are present in bornite, and corresponding Te indicates the presence of Au telluride inclusions. Bismuth concentrations are distinctly lower in digenite relative to bornite, and within bornite the Bi concentration decreases gradationally from the outer rim to a sharp inner grain boundary with digenite. Overall, Ag, Te, and Au (the latter less pronounced) are enriched in digenite relative to bornite, whereas Bi, Sn, and less so, In are depleted in digenite, and Se is equally distributed between both phases.

Mode of trace element residence in Cu-(Fe) sulfides

For identification of submicrometer-scale inclusions below the resolution of the SEM, time-resolved LA-ICP-MS spot profiles were examined. Smooth, slightly decreasing downhole trace element ablation profiles that mirror those of the major elements (Cu, Fe, S) indicate homogeneous substitution into the host grain crystal structure (Fig. 8A); although, theoretically these profiles could also result from homogeneously distributed nanometer-scale inclusions (<100 nm) (Cook et al., 2011; George et al., 2015). By contrast, distinct spikes or peaks in the signal indicate the presence of inclusions (e.g., Ulrich et al., 2009) or encroaching grain boundaries from adjacent phases (Fig. 8B). The ablation profiles of V, Cr, Mn, Ni, Mo, As, Cd, Sb, and W did not follow major element patterns in any of the ablated Cu-(Fe) sulfides. Instead, where concentrations were above detection limit, they were exclusively represented by spikes in the signal, inferring submicroscopic inclusions (Fig. 8C, D). Selenium, and for the most part, Sn show slightly decreasing downhole profiles and hence are most likely structurally bound in all three Cu-(Fe) sulfides. Structural incorporation is also indicated for Co, Ga, Se, In, and Sn in chalcopyrite, due to flat ablation profiles (Fig. 8A); Zn, Ge, Ag, and Pb show similar patterns in some grains. In bornite, flat profiles of Se, Ag, and Bi suggest structural residence (Fig. 9A). Tin and In exhibit similar downhole ablation profiles, although superimposed by many inclusion-related spikes; some grains show similar profiles in Te and Pb. In digenite, flat profiles of Se, Ag, Sn, In, Au, Pb, and Bi suggest their homogeneous distribution in the crystal structure (Fig. 9C, D); Ga and Te exhibit a similar distribution in several grains. Consequently, digenite is the only Cu-(Fe) sulfide that appears to host structurally bound Au in the highgrade core at Bingham Canyon (Fig. 9D). Inclusions with high trace metal(loid) contents are most abundant in bornite, moderately so in chalcopyrite, and rare in digenite.

Qualitative observations on the mode of trace element residence were examined using principal component analysis (PCA) to test the associations without inferring a genetic origin. The PCA results support the qualitative findings on the mode of trace element residence from the discrimination of ablation signals. For both the chalcopyrite and bornite data sets, PC1, the component with the highest variance, separates elements that are dominantly inclusion hosted from the structurally bound trace elements. This is expressed in the PC1 factor loading diagrams in Figure 10.

Elemental substitution in Cu-(Fe) sulfides

Investigation of substitution mechanisms in Cu-(Fe) sulfides on the sole basis of LA-ICP-MS data is challenging, because it relies on the (anti-)correlation of elements. We observed variable abundances of trace micrometer- to submicrometer-scale phases in backscattered electron images and time-resolved ablation profiles. If undetected when using large laser spots sizes (30 µm), these trace minerals can lead to bias in spot laser ablation trace element data for Cu-(Fe) sulfides. This is particularly the case for elements that were found to be both structurally bound and inclusion-hosted (e.g., Se, Ag, Te, and Bi). Hence, element correlations that were detected statistically were subsequently screened in time-resolved laser ablation profiles.

Elemental covariance was evaluated with Spearman rank correlation matrices and log-log bivariate plots. For bornite, positive correlations of Ge with Ga (r2 = 0.55), Cd with Co (r2 = 0.54), Sn and In (r2 = 0.45), as well as As with Co (r2 = 0.31) were detected (Fig. 11). In bornite ablation profiles, Ge spikes very rarely overlap with more common Ga peaks, indicating that a correlation between these two elements is not driven by variable abundance of both Ge- and Ga-rich inclusions. Both Ga and Ge contents are highest in bornite where ablation profiles display a homogeneous distribution, indicating a coupled affinity for residence in the bornite structure. In contrast, As-Co, Cd-Co, and As-Cd peak overlaps are common, suggesting the presence of minerals that are carrying these element pairs (i.e., sulfosalts and cobaltite). However, the observed Cd-Co and As-Co correlations are not necessarily always the result of mineral inclusions that contain both elements. The correlation of two elements can likewise be generated when different inclusion types have a similar occurrence in the ablated Cu-(Fe) sulfide; this was indicated in some bornite grains where Cd, As, and Co spikes appeared at a similar abundance, but without overlap in backscattered electron image profiles. The Cd-As-Co-host mineralogy could not be resolved further, because profile patterns were too inconsistent. Indium appears to most commonly reside within Sn-bearing minerals (Fig. 9A, e.g., in stannite or mawsonite) and locally with Zn-bearing phases (e.g., sphalerite), but in rare cases there is no overlap of In with Sn, or Zn, suggesting the presence of roquesite. In digenite, positive correlation trends between Pb and Se (r2 = 0.75), and between Te and Ag (r2 = 0.89) can be attributed to variable abundances in submicrometer-scale inclusions of Se-rich galena and silver tellurides (e.g., hessite). Coarser grains of these phases were identified by SEM. Chalcopyrite data show no significant correlation patterns, but the heterogeneity of the entire chalcopyrite data set potentially obscures correlations in individual samples. This is indicated in Figure 8D, where Sb spikes seem to commonly overlap with the As signal in chalcopyrite, potentially due to sulfosalt inclusions. This example highlights the importance of LA-ICP-MS ablation profiles in revealing elemental correlations that might be undetectable by statistical treatment of large spot analysis data sets.

Our data set does not provide clear evidence for coupled substitution in Cu-(Fe) sulfides, unlike that detected by Reich et al. (2010) for Ag or Au with As in supergene digenite. Similarly, Cook et al. (2011) studied a variety of hypogene bornite and chalcocite samples from different deposits and found no correlation between As and Ag or between As and Au; they attributed this distinction to the differing crystal-chemical characteristics of supergene and hypogene Cu-(Fe) sulfides.

Digenite exsolution from bornite solid solution

The main purpose of this study was to document mineral-element associations, particularly for energy-critical elements. Inferring the detailed genetic processes that led to the observed associations is beyond the scope of this work. The most striking mineralogical association is the grating-textured digenite within bornite, which requires further examination to better understand the partitioning of minor metal(loids). In line with many previous investigations of this texture, we interpret it to be the product of exsolution from a Cu-rich bornite solid solution (Kullerud et al., 1969; Craig and Vaughan, 1994; Redmond, 2002). Based on major element compositions of bornite in both grain types, it is clear that digenite-free bornite is not the unexsolved equivalent of composite bornite-digenite grains, because digenite-free bornite compositions are systematically more Cu deficient than bornite in bornite-digenite composite grains and do not plot on the tie line between stoichiometric bornite and digenite (Fig. 4). The relative age relationships between the two bornite grain types are unknown, however, both types are present in A5 quartz veinlets (Redmond et al., 2004) and likely precipitated at roughly the same time.

The exsolution process interpreted to be responsible for the occurrence of digenite within bornite grains was first illustrated in the experiments of Kullerud (1960; Fig. 12), in which high bornite-digenite forms a complete solid solution above 330°C. The unit cell of the solid solution consists of a cubic close-packed framework of sulfur atoms (Kanazawa et al., 1978), with eight tetrahedral interstices within the sulfur framework. In these interstices, Cu and Fe atoms as well as vacancies are randomly distributed (metal vacancy population varying from 10% in digenite to 25% in bornite; Grguric and Putnis, 1999). Through the ordering of vacancies, a series of superstructures of variable sizes are produced (Pierce and Buseck, 1978). When digenite exsolves, it occurs as high digenite (Kullerud, 1960) with a closely packed cubic structure (Morimoto and Kullerud, 1963; Will et al., 2002). The bornite solid solution undergoes multiple transitions while cooling (Kanazawa et al., 1978) from about 270°C (Grguric et al., 1998) to 150°C (Grguric and Putnis, 1999) until low bornite, the only stable natural orthorhombic polymorph of bornite, begins to form (Koto and Morimoto, 1975). Between 90° and 73°C, high digenite transforms into low digenite. Low digenite requires substitution of 0.4 to 1.6 atom % Fe for Cu to stabilize its cubic superstructure (Morimoto and Gyobu, 1971). This may explain why digenite from Bingham Canyon contains several wt % Fe, although this could also be the result of partially analyzing bornite within the fine digenite meshlike network.

At a high level of crystallographic similarity, the exsolving phase will grow along the crystallographic orientation in the host phase, known as coherent exsolution (Craig and Vaughan, 1994). In this process, elastic strain absorbs spatial differences between the lattice structures (Baker et al., 1959). As exsolution proceeds, lattice distortions will produce dislocations and vacancies in the periphery, leading to noncoherent exsolution (Brett, 1964). In this process, exsolution will not follow the previously preferred orientations. At some point, the matrix will recrystallize and the exsolving phase will form along grain boundaries (Shockley et al., 1952). If supersaturation still prevails after recrystallization, oriented lamellae might form again (Brett, 1964).

In the LA-ICP-MS element maps of bornite-digenite grains at Bingham Canyon, digenite lamellae are seen to follow the crystallographic orientation of bornite in the grain center, as observed by Ramdohr (1969), indicating the coherent exsolution process. However, grain boundaries with bornite are characterized by embayed margins and irregular digenite protrusions, indicating peripheral recrystallization of bornite with protrusions potentially marking a secondary coherent exsolution phase (Fig. 3). “Sea and island structures” are common, which are another characteristic feature of noncoherent exsolution (Brett, 1964).

The exsolution process is driven by temperature-controlled supersaturation in a solid solution (Brett, 1964), whereby nucleation is initiated at imperfections in the lattice (grain boundaries, dislocations, slip planes). The exsolved crystals grow while metal ions diffuse through the crystal lattice (Craig and Vaughan, 1994) and deplete the matrix in their vicinity. Grguric and Putnis (1999) postulated that the presence of vacancy populations in both digenite and bornite allows for rapid diffusion of Cu and Fe atoms at temperatures below the solvus. Rapid diffusion was advocated by Berger and Bucur (1996), who determined diffusion rates for Cu in both phases, and by Grace and Putnis (1976), who described the strong mobility of Fe in bornite. In the transformation from bornite solid solution to digenite, excess Fe is released. As bornite in the examined bornite-digenite composite grains from the Bingham ore is not enriched or zoned in Fe (Fig. 7), the excess Fe must have been removed, which suggests that at least to some degree, a fluid was involved during exsolution.

Hence, it is likely that Bingham Canyon digenite exsolved by fluid-induced solid-state diffusion (Hidalgo et al., 2020; Adegoke et al., 2021); the latter authors have experimentally shown that only if fluids surrounded Cu-(Fe) sulfide grains, a chemical gradient would be established that enhanced metal diffusion and triggered exsolution.

Trace element distribution between bornite and digenite

Bornite-digenite composite grains have spatial trace element distribution patterns reflecting the exsolution processes, controlled in part by variable solubility limits and diffusion rates for elements in both Cu-(Fe) sulfides. However, complex phase transformations upon cooling may also have played an important role in the redistribution of at least some trace elements. Figure 7 shows that Ag, Te, and Au preferentially partition into digenite, whereas Bi is retained in bornite and Se is uniformly distributed between both phases. These distribution patterns appear to be the result of exsolution-induced partitioning of Ag, Te, and Au from bornite into growing digenite, and Bi diffusion away from digenite nuclei into surrounding bornite. This internal redistribution (without major fluidderived input or output of trace elements) is at least supported for Ag and Bi by marked trace element concentration differences between digenite-free bornite and bornite in composite grains with digenite (Table 3); median Ag contents are highest in digenite (325 ppm), lower in digenite-free bornite (279 ppm), and lowest in bornite that engulfs digenite (71 ppm). Exsolution-induced partitioning is also supported by the highest Bi concentrations associated with bornite in bornite-digenite grains (2,760 ppm), intermediate contents in digenite-free bornite (1,566 ppm), and lowest concentrations in digenite (954 ppm). The Ag and Bi distribution patterns between bornite and digenite are in agreement with results from Cook et al. (2011) and interpretations by Rivas-Romero et al. (2021), who deduced that Ag and Bi concentrations in bornite are likely controlled by the presence of coexisting Cu-S sulfides, such as chalcocite, digenite, covellite, and wittichenite. Equivalent Se concentrations in bornite and digenite suggest that both phases can accommodate Se, which is most likely the result of Se substitution for S in the sulfur framework of their crystal structures (McNeal and Balistrieri, 1989).

The more complex distribution of Sn, however, would be difficult to produce through a single-stage process. The elemental map in Figure 13B demonstrates that Sn distribution is not always concentrically zoned; the highest Sn concentrations appear most commonly in areas within bornite that are most distal to digenite crystals. This nonconcentric distribution argues against episodic Sn input from fluids as being responsible for the observed Sn variability. Moreover, the Sn-depleted envelope in element maps of a composite grain in Figure 7 is found to overlap with the periphery of the digenite core of the sulfide grain, where a stress field would have developed due to accumulating lattice distortions during coherent exsolution, as described by Brett (1964). Hence, we propose that initially, Sn was preferentially retained in bornite during coherent exsolution of digenite (although less so than Bi). As digenite growth led to an increasing stress field in the surrounding bornite, Sn diffused out of these zones and into areas within bornite that were peripheral to the digenite core of the crystal.

Similar distribution patterns are also observed for other elements with lower concentrations in digenite relative to bornite, including In and Bi. For In, the spatial extent of the lower concentration envelope outboard of digenite mirrors the low Sn envelope; Bi concentrations are likewise lower in a more diffuse inner bornite zone around the digenite core relative to the outermost areas in bornite. Therefore, we propose that In and Bi were also affected by stress-induced diffusion. The absence of a sharp inner Bi zone in bornite can potentially be attributed to the more sluggish diffusion kinetics of the relatively large Bi ions, compared to Sn and In. Although, there are no studies on Sn, In, and Bi diffusion in sulfides, stress-induced diffusion of Sn has been observed in other materials. Hektor et al. (2019) have suggested Sn diffusion from areas of high compressive stress to adjacent zones of lower compressive stress in Sn coatings on Cu wires. Additionally, Baliga and Ghandhi (1974) identified interfacial stress as a driving force for the larger lateral diffusion of Sn over Zn in gallium arsenide.

Elements that preferentially partition into digenite, such as Ag, Te, and Au, show very low concentrations in enclosing bornite (median Ag: 71 ppm, Au: 0.13 ppm, and Te: 2.2 ppm), compared to the bornite data presented by Cook et al. (2011) from a suite of hydrothermal deposits, reporting mean Ag concentrations in the range of 4,000 to 8,000, up to 3 ppm Au (with smooth LA-ICP-MS downhole profiles) and 10s to 100s of ppm Te in bornite. Thus, Ag, Te, and Au appear to be markedly below their solubility limits in the Bingham Canyon bornite within bornite-digenite composite grains and are therefore less likely to be affected by stress-induced diffusion. Tellurium-rich particles appear to have formed at the immediate contact between digenite and bornite; these particles do not consist of Ag-(Au) or Bi tellurides, as Ag, Au, and Bi are not similarly elevated at this contact zone.

Trace metal(loid) patterns of the porphyry intrusion sequence

Digenite-free bornite and chalcopyrite grains were chosen for a systematic examination of trace metal deportment in Cu-(Fe) sulfides across the sequence of porphyry intrusions due to their ubiquity in all of the porphyry dikes (Fig. 14). Emphasis was placed on potential element-mineral association patterns in structurally bound trace metal(loid)s in bornite (Se, Ag, In, Sn, and Bi) and chalcopyrite (Co, Ga, Se, In, and Sn). The clearest trend for the oldest to youngest porphyry phases is a decrease in Sn in both digenite-free bornite and chalcopyrite. Unique to bornite, Ag also shows a generally declining trend, while there are no obvious trends for other dominantly structurally bound elements (Se, In, Bi) across the porphyry sequence. A trend of increasing As, Cd, and Au is recognized for digenite-free bornite but only if data from the QLP is excluded. In the QLP, bornite yields the lowest contents of inclusion-hosted elements (Co, Ga, As, Cd, In, and Sb), whereas this is not the case for chalcopyrite; Te is distinctly elevated in the LP. Among the remaining structurally bound elements in chalcopyrite, Ga shows a decreasing trend while In increases, from older to younger intrusions; Co content is relatively elevated in the LP (up to 10 ppm). Among the inclusion-hosted elements, As, Ag, and Sb display increasing trends. The QMP stands out with elevated Ga and Se, but the lowest median In and Au contents for chalcopyrite. The decreasing Sn trend from older to younger porphyry generations appears to be the most significant pattern, as this decline exceeds an order of magnitude and is recorded in both bornite and chalcopyrite.

Minor metal(loid)-bearing phases

Galena crystals (Fig. 6) constitute the most abundant trace sulfide minerals that are larger than 5 µm2 (59 grains). The majority of these galena grains contain several wt % Se, which may be due to solid solution between clausthalite (PbSe) and galena (Liu and Chang, 1994). Submicrometer-scale galena, most common in bornite, shows no evidence for having formed via replacement. Despite not being able to quantify the percentage of submicrometer-scale galena, at most these inclusions carry a small fraction of the ore’s total Se inventory since Se is structurally bound at concentrations of several hundred ppm in all the Cu-(Fe) sulfides. In rare cases, galena grains of micrometer scale accumulate along fractures where bornite is partially replaced by chalcopyrite. According to the LA-ICP-MS element spot data, median Pb concentrations in chalcopyrite (4 ppm) are approximately 20 times lower than measured concentrations in digenite-free bornite (130 ppm) in the same composite chalcopyrite-bornite grains. Thus, galena crystals presumably formed when bornite was replaced by chalcopyrite and excess Pb from bornite could not be incorporated in chalcopyrite.

Wittichenite (Cu3BiS3) is the second most abundant trace sulfide mineral found in this study (Fig. 6). It occurs preferentially between laths of grating-textured digenite and most likely formed due to excess Bi that was not taken up by exsolved digenite.

The least abundant trace minerals larger than 5 µm2 were electrum and Ag-(Au) tellurides. In contrast to galena and wittichenite, we found no evidence for their origin being related to secondary exsolution and instead propose that they coprecipitated along with Cu-(Fe) sulfides. This is supported by cathodoluminescence petrography (Redmond, 2002), which revealed that even silicate-hosted gold grains are part of the same A5 vein network as Cu-(Fe) sulfides. According to the calculations of Redmond (2002), the Au inventory in the QMP-LP zone is primarily controlled by the abundance of trace phases (electrum ± tellurides). As we found a similar number of electrum and Ag-(Au)- telluride inclusions, the latter might also significantly control the Te inventory. By contrast, the Ag inventory is much less influenced by Ag-(Au) tellurides because, unlike Te and Au, Ag is also structurally bound at concentrations of 100s of ppm in bornite and digenite.

Trace metal(loid) inventory of porphyry intrusions at Bingham Canyon

At the Bingham Canyon deposit, the inventory of the majority of potential valuable by-product elements is controlled by total sulfide abundance and relative abundance of the different Cu(-Fe) sulfide phases. This is the case for Co, Ga, Se, In, Sn, Ag, and Bi. As previously noted, the abundance of Cu-(Fe) sulfides decreases from oldest to youngest porphyry intrusion. This decline is stepwise (Fig. 2), with changes in Cu-(Fe) sulfide ratios leading to different metal(loid) inventories in successive porphyry intrusions. Bornite and digenite contain approximately three orders of magnitude more Bi and at least 40 times more Ag than chalcopyrite, demonstrating that the inventory of these two elements is primarily controlled by the abundance of bornite-digenite. Similarly, Se and Te are elevated in bornite-digenite over chalcopyrite by a factor of ~3. Chalcopyrite abundance controls the In and Ga inventory, with contents being 10 times higher relative to bornite-digenite.

Among the porphyry dikes, the QMP has the highest Cu (-Fe) sulfide abundance and is the only porphyry where bornite and digenite are dominant, leading to an elevated Bi, Ag, Se, and Te inventory. The LP ranks second after the QMP in terms of Cu-(Fe) sulfide abundance. However, in contrast to the QMP, the abundance of bornite and digenite is markedly lower, whereas chalcopyrite is only marginally lower than in the QMP. This results in much lower contents in bornite-digenite-hosted trace elements for the LP, especially Bi, Ag, and to a lesser degree, Te and Se; whereas the In and Ga inventory remains more or less constant. The Co inventory is likely highest in the LP, due to concentrations of Co in LP-hosted chalcopyrite, ~3 times higher than in chalcopyrite from all other porphyry dikes (Fig. 14). The abundance of bornite and digenite is similar in the BP to the LP, but chalcopyrite is less abundant, resulting in lesser amounts of chalcopyrite-hosted elements, such as In and Ga. From the BP to QLPbx, there is a sharp drop in Cu-(Fe) sulfide content, accompanied by a large decrease in all trace metal(loid)s. In the QLP, the youngest porphyry, digenite is absent and bornite is limited to a single-sheeted vein sample; chalcopyrite grains are also less common than in the QLPbx. Consequently, the QLP has the lowest critical metal(loid) inventory and is notably depleted in Bi, Ag, and Te relative to older porphyry intrusions.

Whole-rock Ag assay data (Redmond, 2002) indicate that sulfide ratios can be correlated with the metal(loid) inventories of the individual porphyry dikes. Whole-rock Ag/Cu ratios can be used to estimate the bornite-digenite to chalcopyrite ratio of the whole rock, independent of the total sulfide abundance. This is due to Ag concentrations being at least 40 times higher in bornite-digenite relative to chalcopyrite, whereas Cu contents are only about twice as high as in chalcopyrite. Porphyry dikes with chalcopyrite as the dominant Cu-(Fe) sulfide (the LP and the QLP) have the lowest whole-rock Ag/Cu ratios (Fig. 15). QMP, with the highest bornite-digenite to chalcopyrite ratio, stands out with the highest median Ag/Cu ratio. We suggest that whole-rock Ag and In assay data may be useful to estimate Cu-(Fe) sulfide ratios, and that Ag/In ratios can be used as a general proxy for the inventory of almost all energy-critical elements in porphyry Cu ores.

Metal(loid) deportment at Bingham Canyon in the context of other world-class porphyry deposits

Finally, we compared the metal(loid) deportment at Bingham Canyon with two other world-class Cu-Mo porphyry deposits, namely Río Blanco and Chuquicamata, for which data on element deportment data are available.

At Río Blanco, Crespo et al. (2020) investigated the deportment of Ag, due to its relevance as a by-product. In comparison with Bingham Canyon, a common finding is that the Ag concentrations at Río Blanco are also highest in the hypogene potassic core of the deposit. However, Río Blanco features three Ag mineralization events that are related to three distinct alteration types (potassic, gray-green sericite, and quartz-sericite alteration), whereas at Bingham Canyon each individual porphyry intrusion is linked to a separate mineralization event that is associated with similar potassic alteration (e.g., Redmond and Einaudi, 2010). When comparing metal(loid) concentrations in Cu-(Fe) sulfides from potassic alteration-related mineralization between both deposits, chalcopyrite and bornite from Río Blanco exhibit similar Ag concentrations (5–7 and 236–244 ppm) as chalcopyrite and digenite-free bornite from Bingham Canyon (1–6 and 270 ppm); electrum and Ag tellurides are also common. In contrast to Bingham Canyon, at Río Blanco further bornite from the transitional mineralization (430–460 ppm) and Cu sulfosalts from the late mineralization stages (1,000s of ppm up to wt % levels), as well as fine Ag sulfides (Ag2S) contribute to the high Ag inventory. No hypogene digenite was reported for Río Blanco, whereas for Bingham Canyon, this is an important Ag carrier.

At Chuquicamata, Rivas-Romero et al. (2021) determined the trace element association of Cu-(Fe) sulfides from different hydrothermal alteration stages. Concentrations of most metal(loid)s (Se, In, Pb and Sn) are highest in chalcopyrite from the potassic alteration zone, relative to chalcopyrite from lower temperature alteration zones (propylitic and phyllic); silver and Bi concentrations are similar across all alteration zones. When comparing chalcopyrite compositions from the potassic zones of Chuquicamata and Bingham Canyon, median concentrations of critical raw materials (Se, Ag, Bi, In, Co, and Ga) are consistently higher at Chuquicamata (Table 5); selenium even reaches levels above 2 wt % in some grains. But it is important to note that less than 15% of the analyses on chalcopyrite from Chuquicamata was above the limit of detection for Se, Co, and Ga, whereas Ag, Bi, and In contents are consistently higher (by a factor of 2–4) compared to Bingham Canyon. Bornite only occurs in the potassic zone and similarly to Bingham Canyon, hosts high concentrations of Ag and Bi. However, median Se, Ag, and markedly Bi concentrations (268 ppm) in bornite are lower than at Bingham Canyon (1,566 ppm; Table 5). Moreover, Se concentrations above the limit of detection were only measured in 10% of the bornite grains from Chuquicamata, while Se was detected systematically in bornite from Bingham Canyon. Pyrite is the dominant host for Co and Ni at Chuquicamata, while pyrite is absent in high-grade Cu ores from Bingham Canyon. In both deposits, Au-Ag tellurides and electrum are common and contribute to their critical raw material inventories.

These findings show that there are some commonalities in metal(loid) deportment for large porphyry deposits, most prominently the strong association of enrichment in several critical raw materials with potassic alteration zone mineralization. Future studies of more Cu-Au and Cu-Mo porphyry deposits will undoubtedly refine this knowledge but even from the limited data available to date, it is clear that the metal(loid)-mineral association and the extraction evaluation of critical raw materials will need to be assessed individually for known and yet-to-be-discovered porphyry deposits.

In ore samples from the high-grade core of the Bingham Canyon porphyry Cu-Mo-Au deposit, the majority of metal(loid)s are hosted within Cu-(Fe) sulfides, comprising chalcopyrite, bornite, and digenite. In chalcopyrite, Co, Ga, Se, In, and Sn appear to be structurally bound, as evidenced by their smooth, relatively flat concentration profiles in time-resolved LA-ICP-MS signals. Both bornite and digenite host Se, Ag, In, Sn, and Bi in their crystal structures, and digenite also hosts Te and Au. Trace element concentrations of Se, Ag, Te, and Bi are generally of similar magnitude in bornite and digenite but significantly lower in chalcopyrite. Bornite is the dominant host for Bi (>1,000 ppm) and also contains the highest Se concentrations (>700 ppm) of the sulfide phases. Digenite contains the highest concentrations of Ag (>300 ppm), Te (>7 ppm), and Au (>0.25 ppm). Chalcopyrite is the dominant host for Co, Ga, and In, although at relatively low concentrations (1–10 ppm range).

Apart from Cu-(Fe) sulfides, Bingham Canyon ore contains rare metal(loid)-bearing minerals that carry weight percent concentrations of metal(loid)s; these include electrum, Ag-(Au) tellurides, Se-bearing galena, and wittichenite (Cu3BiS3). The presence of electrum and Ag-(Au) tellurides controls the Au and likely also the Te inventory of the ore, whereas Au and Te concentrations in primary Cu-(Fe) sulfide ore minerals are relatively low. Selenium-bearing galena and wittichenite exert only a minor influence on the Se and Bi inventory, due to high Se and Bi concentrations in major Cu-(Fe) sulfides. Other trace metal(loid)s such as V, Cr, Mn, Ni, Mo, W, and deleterious elements (As, Cd, Sb) occur exclusively as submicrometer-scale inclusions in Cu-(Fe) sulfides but not in their crystal structures.

The content of potential by-product metal(loid)s present in porphyry ores such as Bingham Canyon can be estimated through the absolute abundance and relative proportions of Cu-(Fe) sulfide minerals. The QMP has the highest metal(loid) inventory of any porphyry phase at Bingham, including Ag, Bi, Se, Te, and Au, as it exhibits the highest overall Cu-(Fe) sulfide (in particular bornite and digenite) and electrum abundance. The metal(loid) inventory decreases stepwise from QMP to successively younger porphyry intrusions, linked to declined abundances of Cu-(Fe) sulfides. In the youngest porphyry QLP, metal(loid) contents are only locally elevated where rare A-quartz veins contain sulfide grains.

Digenite exclusively occurs as grating-textured lamellae within bornite and is interpreted to have formed via fluid-induced solid-state exsolution from a Cu-rich bornite solid solution. For the majority of digenite grains, textures indicate a change from coherent to noncoherent exsolution. Element mapping (LA-ICP-MS) of sulfide intergrowths clearly shows selective partitioning of trace elements between bornite and exsolved digenite. Silver, Te, and Au partition strongly into digenite, whereas Se is homogeneously distributed between the two phases. Bornite exhibits zonation only for elements that are preferentially retained in bornite (Sn and Bi) or found at similar concentration (i.e., In) in digenite. Trace element depletions within the periphery of exsolved digenite indicate diffusion of Sn and In over Bi during exsolution. It is postulated that this was triggered by accumulating lattice distortions in bornite during digenite growth, resulting in preferential stress-induced diffusion of Sn, In, and Bi into the margins of bornite crystals.

With Cu production projected to greatly increase as the global economy transitions to low carbon technologies, ore characterization will play an important role in estimating global critical raw material reserves. This study demonstrates that porphyry orebodies can host significant amounts of critical raw materials. However, high-resolution LA-ICP-MS mapping and analysis of downhole time-resolved signals demonstrates the complexity of trace metal-mineral associations caused by zoning, inclusions, and exsolution processes. The spatial association of critical raw materials and their distribution within Cu ores have not generally been examined in conventional deportment studies but can be used to inform metallurgical processing and potentially improve overall critical raw material recoveries.

This research project was partly supported by the Science Foundation Ireland (SFI; Grant 13/RC/2092) and cofunded under the European Regional Development Fund and by iCRAG industry partners. We thank Kennecott and their local team at the time, consisting of Ed Harrison, Tracy Smith, Kim Schroeder, and Ken Krahulec, for their help during Patrick Redmond’s sampling campaign, which provided the samples for this research. For the purpose of Open Access, the author has applied a CC-BY-NC public copyright license to any Author-Accepted Manuscript version arising from this submission. We are grateful for access to the iCRAG laboratory at Trinity College Dublin, and we thank Cora McKenna and Garry O’Sullivan for their continued support during LA-ICP-MS data acquisition. The manuscript benefitted from discussions with Emma Tomlinson, Thomas Ulrich, and Larry Meinert. We are further grateful for the insightful reviews by Artur Deditius, Manuel Keith, and Jeff Hedenquist that greatly contributed to the improvement of this manuscript, and for the editorial handling by Larry Meinert and Martin Reich.

Maurice Brodbeck is a postdoctoral researcher at the Department of Geology at Trinity College Dublin, Ireland, and at the Irish Centre for Research in Applied Geosciences (iCRAG). He graduated with B.Sc. and M.Sc. degrees from the University of Tübingen, Germany, and obtained a Ph.D. degree (2021) from Trinity College Dublin. Maurice has worked on layered intrusions (Bushveld), but his scope has extended to porphyry copper deposits. His interests include the deportment of critical elements on various scales, ore formation, and secondary redistribution processes. Maurice is currently operating a raw materials LA-ICP-MS laboratory and is examining novel methods for sulfide matrices.

Gold Open Access: This paper is published under the terms of the CC-BY 3.0 license.

Supplementary data