Introduction

Epithermal ores have accounted for a significant proportion of historic Au and Ag production worldwide (7 and 16 %, respectively; Simmons et al. 2005). Epithermal deposits are formed under relatively shallow, low-temperature conditions and occur worldwide in subduction-related volcanic arcs, back-arc rift zones, and postsubduction volcanic settings (Simmons et al. 2005; Richards 2009; Saunders and Brueseke 2012). Their ore-forming solutions appear to be largely heated meteoric (ground) water similar to present-day geothermal systems based on H and O isotope studies (O’Neil et al. 1973; John et al. 2003; Simmons et al. 2005; Saunders et al. 2008), although some studies suggest that there could be a significant component of magmatic waters involved in the main epithermal Au-Ag ore-forming stage (Matsuhisa and Aoki 1994; Hayashi et al. 2000; Faure et al. 2002). The source(s) of ore-forming constituents in these shallow ores is controversial, with some models proposing that metals and sulfur are leached from surrounding country rocks, in contrast to others invoking a magmatic source for metals and sulfur, with perhaps some degree of mixing of the two potential sources. More recently, Heinrich et al. (2004) and Williams-Jones and Heinrich (2005) have proposed that low-density magmatic fluids can transport significant amounts of ore metals and sulfur from the deeper magmatic setting to the shallow epithermal systems. This so called vapor-phase transport of metals can be seen in present-day volcanic fumaroles, where micrometer to sub-micrometer-sized native gold particles (locally alloyed with Ag, Cu) and other metallic phases precipitate directly from volcanic emissions (Meeker et al. 1991; Fulignati and Sbrana 1998; Taran et al. 2000; Yudovskaya et al. 2006). Detailed textural studies of bonanza epithermal ores have shown, at least locally, that nanoparticles (or colloids) of Au-Ag-Cu minerals can be physically transported by epithermal ore-forming solutions, and their aggregation and deposition can form spectacularly high-grade ores (Saunders 1990, 1994, 2012; Marinova et al. 2014). Saunders and Schoenly (1995) and Marinova et al. (2014) proposed that boiling of the epithermal ore-forming solutions led to nanoparticle formation at shallow crustal settings. However, Pb and Re-Os isotope data from these bonanza ores support a deep, magmatic source of metals in these shallow ores (Kamenov et al. 2007; Saunders et al. 2008, 2014). Thus, this study was designed primarily to evaluate Cu and S isotopic constraints on the origin of ore-forming constituents in these ores, as well as to look at Pb isotope values in both ore and gangue minerals from the high-grade vein bands.

Results from these previous studies suggest that Au and Ag appear to be derived from mafic magmatic sources and not the adjacent host rocks. Further, the surrounding host rocks vary greatly in composition and age, yet the epithermal ores are remarkably similar in many respects (Saunders et al. 2008). Here, we present Cu isotope data from epithermal ores from western USA and Japan, along with well-constrained new Pb and S isotope data. All of these data strongly support a magmatic source of both metals and sulfur in epithermal ores and are also consistent with, but do not conclusively prove, nanoparticle transport of metals and sulfur. In addition, the new Pb isotope data support a different source for ore and gangue minerals in some epithermal ores.

There have been many attempts to classify these epithermal ores; Lindgren (1933) proposed a ninefold classification of them based mostly on mineralogy and geochemistry. However, today, two principal end-member types are currently recognized (Sillitoe and Hedenquist 2003; Simmons et al. 2005): (1) The most abundant class by far is generally referred to as “low sulfidation” (LS), deposited from near-neutral pH hydrothermal solutions and (2) “high-sulfidation”, deposits formed from acidic ore solutions. Here, we focus on some examples of the more abundant LS ores from two of the best-documented regions, western USA and Japan (Fig. 1). LS bonanza ores from both countries are renowned for their extraordinarily rich Au-Ag grades, and because of the abundance of precious-metal minerals, it is possible to discern ore depositional features such as timing and mutual association of ore and gangue minerals, which allowed for an optimization of isotopic analyses. The northern Great Basin (USA) (NGB) mid-Miocene bonanza ores are thought to be related to the volcanism attending the initial emergence of the Yellowstone mantle plume (Saunders et al. 2008), whereas the Japanese ores are associated with volcanism in a subduction-back-arc rifting setting (Izawa et al. 1990; Shimizu et al. 1998).

Fig. 1
figure 1

Location map showing deposits or districts in the western USA and Japan where we recognize colloid aggregation textures. A subset of these deposits is discussed in the text

Full size image

Bonanza ores in both regions are characterized by banded, open space-filling vein textures, with electrum as the principal Au phase, whereas Ag is hosted by sulfides, sulfosalts, and selenides, although naumannite (Ag2Se) and aguilarite (Ag4SeS) are often the most abundant phases (Shimizu et al. 1998; Saunders et al. 2008). Quartz, opal, chalcedony, calcite, quartz-replacing calcite, and adularia are the common associated vein minerals, and chalcopyrite is the most abundant base-metal sulfide mineral associated with Au-Ag minerals (Fig. 2a). A hallmark of these ores is their banded nature, reflecting multiple Au-Ag depositing events (Fig. 2b). Highest-grade bands contain fractal dendrites of electrum-opal (Fig. 2c, d), naumannite-aguilarite, and locally, chalcopyrite, which have been interpreted to have formed by aggregation of metallic nanoparticles (Saunders 1990, 1994, 2012). Although the deposits considered here are ≤∼16 Ma, such gold nanoparticle aggregation textures in epithermal ores have also been recognized in Cambrian deposits (Foley and Ayuso 2012).

Fig. 2
figure 2

a Photomicrograph (reflected light) showing typical rich Au-Ag “ginguro” ore from the Hishikari deposit, Kyushu, Japan. Mineral abbreviations: el electrum, cpy chalcopyrite, nm naumannite, qtz quartz. b Photograph of multiple electrum and opal bands from the Sleeper deposit, Nevada (field of view is 7 cm). c Photomicrograph (transmitted and reflected light) of a fractal electrum dendrite in opal and quartz (gray) from the Sleeper deposit (from Saunders 2012). d SEM image of a fractal dendrite similar to the one shown in (c) of electrum exposed by dissolving host silica with HF. This texture is interpreted to form by aggregation of physically transported electrum nanoparticles and crystals (Saunders 1994, 2012)

Full size image

Materials and methods

Data presented here are from petrographically well-characterized ore and gangue minerals from bonanza gold-silver ores, including the Sleeper, Midas, National, Buckskin National deposits (Nevada), Silver City district (Idaho), and the Koryu and Hishikari deposits (Japan) (Tables 1 and 2). The high-grade ores and banded nature of the ores (formed by deposition of progressive layers along vein walls over time) allowed for a precise determination of mineral relationships. In particular, ore and gangue minerals chosen for isotopic analysis came from bands that contained abundant Au-Ag minerals. This was accomplished by microdrilling of individual minerals or by dissolution and disaggregation of bulk samples by cold or hot hydrogen fluoride (HF). In the latter procedure, minerals were hand-picked using a binocular microscope to separate them. Minerals for analysis were then dissolved for Cu and Pb isotopic analyses with a multicollector inductively coupled plasma mass spectrometry (ICP-MS) or by standard combustion techniques for S isotopes. Because of the abundance of chalcopyrite in close association with Au-Ag minerals in these bonanza ores, it was targeted for our S isotope analyses and also much of the Cu isotope analyses. However, the ore mineral electrum and naumannite-aguilarite typically contain enough Cu and Pb (>10 ppm) to allow measurement of their Cu and Pb isotopic ratios. Similarly, physically separated gangue minerals (quartz, opal, adularia, or quartz + adularia) also contain enough Pb for isotopic analysis.

Table 1 Pb isotope data for electrum and gangue from the northern Nevada epithermal deposits
Full size table
Table 2 Sulfur and copper isotope data for epithermal ores from the northern Great Basin (USA) and Japan
Full size table

Cu isotope measurements on chalcopyrite were conducted using similar techniques as described in Mathur et al. (2010). Electrum and naumannite were dissolved in Teflon beakers with 8 ml of ultrapure aqua regia at 140 °C for 24 h or until complete dissolution occurred. Concentrations of copper were measured from these solutions on an ICP-MS to determine the amount of solution to dry for purification by ion exchange chromatography. Concentrations ranged from 12 to 220 ppm Cu in electrum and naumannite, and a volume that contained 1 μg of copper was dried and used for ion exchange chromatography. The exchange chromatography, mass spectrometry preparation, and measurement methods are identical to Mathur et al. (2010).

Chalcopyrite samples were used to measure S isotopic ratios. Samples were weighed, mixed with V2O5, CuO, and quartz, and then milled. Samples were then placed in a combustion tube, were thoroughly dried, and then heated in a furnace at 1050 °C. H2O and CO2 were removed from SO2, which was then analyzed for S isotope ratios in a mass spectrometer at the University of Georgia.

The fine-grained nature of electrum from the Sleeper deposit made it impossible to physically separate enough of it for Pb isotope analyses. Thin (2–3 mm) slices were cut perpendicular to banding, and ore and gangue minerals were subsequently separated chemically. Sample cleaning and Pb separation was done under clean lab environment at the Department of Geological Sciences at the University of Florida. The separated vein pieces were cleaned by sonication with Optima-grade acetone multiple times then rinsed with 4× H2O. After the acetone cleaning, each piece was transferred to a precleaned Teflon vial and was cleaned overnight with warm Optima-grade 3.5 N HNO3. The nitric acid was discarded, and the vein piece was again rinsed multiple times with 4× H2O. Concentrated Optima-grade HF was added to the clean vein sample and left overnight on the hot plate with the Teflon vial closed to dissolve the silicate fraction. The following day, the resultant HF solution was transferred to a clean Teflon vial. The remaining electrum looked clean and free of any silicates, but as a precaution, was subjected to another leaching with clean Optima-grade HF. The second batch of HF was added to the first, and all of the HF was evaporated to dryness. This HF residue constitutes the gangue fraction that was subsequently processed for Pb isotope analysis. In brief, Pb from the silicate (HF) fraction was purified on Dowex 1×-8 resin in traditional 1 N HBr medium; details are provided in Kamenov et al. (2013). The separated clean electrum after the double HF attack was rinsed multiple times with 4× H2O, then leached overnight in warm 3.5 N HNO3, then rinsed again with 4× H2O and dissolved with aqua regia. Pb was separated from the electrum also following protocol described in Kamenov et al. (2013). In brief, the resultant aqua regia solution was evaporated to dryness, and the electrum residue was dissolved in 2 ml 6 N HCl. The 6 N HCl solution was loaded on columns packed with Dowex 1× 8 resin and the elutant was collected in a clean Teflon vial. The gold-free solution was then evaporated to dryness, and Pb was further purified with 1 N HBr. Pb isotopic compositions for all samples were determined on an Nu-Plasma MC-ICP-MS with Tl normalization following a method described in Kamenov et al. (2004). Long-term analysis of the NBS 981 Pb isotope standard yielded the following average values: 206Pb/204Pb = 16.937 (±0.004, 2σ), 207Pb/204Pb = 15.490 (±0.003, 2σ), and 208Pb/204Pb = 36.695 (±0.009, 2σ).

Results

The use of copper isotopes in ore deposit research is a new and evolving field (Li et al. 2010; Mathur et al. 2010; Wall et al. 2011; Asadi et al. 2015). Cu isotopic ratios were measured on the three principal (and coexisting) phases in these ores (e.g., Fig. 2a): electrum, naumannite, and chalcopyrite. Results indicate δ65Cu values range between approximately −2.5 and +3.5 ‰, with most values falling between −1 and +1 ‰ (Fig. 3). The latter range is very similar to δ65Cu values for porphyry copper deposits which are generally interpreted to have formed from magmatic fluids, metals, and sulfur (Li et al. 2010; Asadi et al. 2015).

Fig. 3
figure 3

Histogram showing δ65Cu in ‰ from western USA (northern Great Basin) and Japanese epithermal ores. Data for “mantle rocks” and iron meteorites from Bishop et al. (2012) and for porphyry deposits from Li et al. (2010) and Asadi et al. (2015). Zero per thousand (0 ‰) is the value assigned to the NIST copper standard

Full size image

Copper isotope values also overlap with δ65Cu values of “mantle” rocks and metallic meteorites (Bishop et al. 2012; Fig. 3). Further, all three of these phases yield similar results from the same samples indicating a certain degree of isotopic equilibrium between ore minerals and a similarity of δ65Cu values between principal districts studied here (Fig. 4). Given our current limited understanding of Cu isotope fractionation during the formation of a number of different ore deposit types, we do not have sufficient data at this point to fully characterize all types of copper or copper-bearing deposits. However, data presented here show that Cu isotope signature of Cu-bearing epithermal minerals is surprisingly similar to those from the center of porphyry copper (e.g., formed at much higher temperature). All δ34S values measured in this study were only from NGB epithermal ores, and all values determined were from chalcopyrite, as it is by far the most abundant sulfide mineral in the Au-Ag-rich bands in the ores. At epithermal temperatures (∼150–300 °C), δ34S of chalcopyrite is essentially equal to the δ34S value of aqueous sulfide in the ore-forming fluids (Seal 2006). Results (Fig. 5) show δ34S ranging from ∼−7 to +4 ‰ with about half of the values falling between −1 and +1 ‰, values that are generally interpreted to be indicative of a principal magmatic source of sulfur (Seal 2006). Due to extensive published sulfur isotope data from the two Japanese deposits and the more reconnaissance nature of our copper isotope investigations there, we did not analyze δ34S values for those ores. However, published data from the high-grade “ginguro” bands from Hishikari (−1 to +2 ‰; Shikazono 1999) and Koryu (−3.9 to +2.7 ‰, mean ∼0 ‰; Shimizu et al. 1998) are similar to NGB epithermal ores (John et al. 2003; Fig. 5), as are recent δ34S data from epithermal ores in New Zealand (Christie et al. 2007), Indonesia (Yuningsih et al. 2011), and Kamchatka (Takahashi et al. 2012). Plotting both Cu and S isotope values measured on the same chalcopyrite grains from the NGB epithermal ores also shows a range of values similar to “magmatic” porphyry copper ores (Fig. 6).

Fig. 4
figure 4

Histogram subsets of δ65Cu data for specific districts, showing mineral phases that Cu isotope data were measured on. Mineral abbreviations are the same as in Fig. 2

Full size image
Fig. 5
figure 5

Histogram showing δ34S in per thousand from the northern Great Basin (USA) epithermal ores; all data are from chalcopyrite

Full size image
Fig. 6
figure 6

Plot of δ34S versus δ65Cu data for six chalcopyrite samples from epithermal deposits in the northern Great Basin for which both parameters were measured. The highlighted box is the typical range of δ34S and δ65Cu values for porphyry copper deposits

Full size image

In the process of conducting our earlier Pb isotopic investigations (Kamenov et al. 2007), we observed data that suggested that the Pb isotopic composition of electrum was different from adularia codeposited with the gold. In particular, here, we evaluated that finding at the Sleeper deposit (Fig. 7). Pb isotope data show disequilibrium between the gold ore and the silicate fraction that cannot be explained with in situ U and Th decay. The observed trend in Pb-Pb isotope space suggests distinct sources for gold and gangue minerals in the epithermal hydrothermal system (Fig. 8). The silicate minerals show elevated 207Pb/204Pb indicating that the elements composing the gangue are most likely leached from local sedimentary or metamorphic rocks. In contrast, the gold ore shows lower 207Pb/204Pb indicating derivation from a likely deeper, mafic magmatic source, supporting the S and Cu isotope data. The observed Pb isotope disequilibrium between gangue and electrum occurs at the millimeter to centimeter scale within the epithermal vein.

Fig. 7
figure 7

Photograph of banded bonanza ore from the Sleeper deposit, exhibiting at least seven different electrum-rich bands, which was used for the Pb isotope investigation described in the text; sample numbers correspond to those listed in Table 1

Full size image
Fig. 8
figure 8

Plot of lead isotope data for Sleeper deposit (NV) electrum and gangue from this work compared to northern Nevada epithermal ores and gangue data from Kamenov et al. (2007). “Nevada Au” includes data from Jumbo, National, Seven Troughs and Sleeper and “Nevada gangue” includes data from Jumbo, National, Sleeper, New Alma, and Sandman. Arrows point to the three Sleeper electrum samples and the one Sleeper gangue sample reported in Kamenov et al. (2007). Bar in the upper right corner represents the two standard errors of the Pb isotope measurements

Full size image

Discussion

Recent research has shown that the subduction process can prepare the lithospheric mantle for later “fertile” magma partial melts to form in the mantle, which then can rise through the crust and trigger ore formation (Richards 2009; Saunders and Brueseke 2012). Thus, a postsubduction mantle plume (western USA) or rift-triggered partial melting (Japan) can apparently trigger partial melting of this “fertilized” lithospheric mantle and can lead to volcanism and related epithermal ore formation.

The Japanese and NGB bonanza epithermal ores are enriched in Se, thus Lindgren (1933) would have classified the deposits considered here as “gold-selenide” epithermal ores. Saunders (2012) noted that these ores are also enriched in copper due to their abundance of chalcopyrite in highest-grade ores, and interpreted the Cu enrichment (along with gold) to result from a mafic magmatic source. In fact, at Hishikari, gold shows a stronger statistical correlation with copper than either silver or selenium (Nagayama 1993). The present study indicates a predominant magmatic origin for Cu and S. This is consistent with previous Pb isotope data on electrum from NGB mid-Miocene epithermal ores that indicated a primitive (mafic) magmatic origin for gold (Kamenov et al. 2007); Re-Os isotope data from these ores suggest a similar origin (Saunders et al. 2014). Further, the Pb isotopic disequilibrium between ore and gangue minerals in NGB epithermal ores indicates distinct sources for the two materials. O and H isotopes indicate a predominant meteoric water source for ore-forming solutions in the Japanese deposits of Hishikari (Hayashi et al. 2000; Faure et al. 2002) and Koryu (Shimizu et al. 1998), as well as the NGB ores (John et al. 2003; Saunders et al. 2008), an interpretation which is generally consistent with fluid inclusion microthermometric data (Izawa et al. 1990; Shimizu et al. 1998; John et al. 2003). As discussed above, textures in NGB and Japanese bonanza epithermal ores show evidence of silica and metallic nanoparticle physical transport and aggregation (Saunders 1990, 1994, 2012). Thus, any model for the genesis of these ores must account for (1) bonanza ore textures that demonstrate nanoparticle transport and aggregation; (2) isotopic evidence that metals and sulfur had a predominant magmatic source; (3) the Pb isotopic disequilibrium for ore and gangue minerals; and (4) that the ore-forming solutions were predominantly heated meteoric waters.

The idea that gold (and silver) nanoparticles can form in epithermal ore-forming systems perhaps dates back about a century, stemming from Lindgren’s research on the National deposit, Nevada (Lindgren 1915, 1933). More recently, evidence from ore textures in bonanza epithermal ores in western USA (Saunders 1990, 1994, 2012, and references therein) and elsewhere (Marinova et al. 2014) have provided additional support to the concept. Further, gold nanoparticles have been identified in the Champagne Pool, New Zealand (Pope et al. 2005), and metallic-sulfide nanoparticles also occur entrained in fluids discharging from seafloor hydrothermal vents (Yucel et al. 2011). Gold nanoparticles have been observed depositing from volcanic vapors (Meeker et al. 1991; Fulignati and Sbrana 1998; Taran et al. 2000; Yudovskaya et al. 2006); Hough et al. (2008) describe other environments where gold nanoparticles occur. Textural data from western USA bonanza ores show that aggregation of nanoparticles of Au-Ag alloy, Ag-Se-S phases, and chalcopyrite is an important ore-forming process (Saunders 2012).

There is difficulty in reconciling data that show epithermal ore-forming solutions were predominantly shallow meteoric water with our isotopic data indicating that metals and sulfur have a magmatic origin. Saunders et al. (2008) attempted to explain this disparity for NGB ores by invoking a two-hydrothermal solution model: (1) a shallow geothermal system driven by the magmatic heat source at depth which leaches major elements from surrounding country rocks required to make the gangue minerals in veins and (2) episodic incursions of magmatic components into the shallower geothermal system, which causes the banded nature of ore and repeated deposition of ore minerals over time. That model was largely based on geologic and hydrothermal conditions at present-day Yellowstone National Park, USA (Lowenstern and Hurwitz 2008). At Yellowstone, a shallow rhyolitic magma chamber is interpreted to be the heat source for the world’s largest geothermal system, but deeper basaltic magmas apparently contribute magmatic volatiles (including sulfur) episodically to the shallow geothermal system (Lowenstern and Hurwitz 2008). So, if magma bodies at depth are the source of metals, metalloids, and sulfur (and selenium) that are incorporated into the shallow epithermal ores, how does this occur? One hypothesis is that volatile metals and sulfur are transported to the epithermal environment (and mix with the shallow geothermal waters) in low-density magmatic “vapors” that contribute magma-sourced metals but little water (Hedenquist and Lowenstern 1994; Heinrich et al. 2004; Williams-Jones and Heinrich 2005). Would nucleation of Au-Ag-Cu minerals directly in the shallow, low-temperature, epithermal setting preserve magmatic isotopic signatures for those minerals? It seems more likely that Au-Ag-Cu phases nucleated as nanoparticles at depth and were physically transported to the epithermal environment. In the process, higher-temperature magmatic Cu, Pb, and S isotopic signatures were preserved, and ores formed by particle aggregation along vein walls. Saunders (1994) proposed that Au-Ag nanoparticles formed directly in the epithermal environment in response to boiling of ore-forming solutions. Isotopic data presented by Kamenov et al. (2007) and in this study make that interpretation less likely and support a deeper origin for the nanoparticles.

Conclusions

The first Cu isotope data from definitive ore-stage Au-Ag minerals in epithermal ores (and Cu and S isotopes on associated chalcopyrite) indicate a deeper magmatic source of metal(loids) and sulfur in these shallow LS bonanza ores. It is likely that this is not just the case for bonanza ores, but may also apply to other epithermal ores or even to other hydrothermal gold deposits. The bonanza ores provide the context for interpreting the timing of ore- and gangue-mineral deposition, show that ore mineral precipitation occurred episodically over time, and provide abundant (and large) minerals for isotopic analysis. Our new Pb isotope data, building on the previous work of Kamenov et al. (2007), show that ore and closely associated gangue minerals have a different source of Pb. This provides further support to the model of Saunders et al. (2008) that a deeper magmatic source of metal(loids) and sulfur is mixed with a shallow geothermal fluid that is leaching major elements (e.g., K, Al, Na, and Si) from the surrounding shallow host rocks to make the principal vein gangue minerals. This conclusion has implications for fluid inclusion and isotope studies that have focused on using the gangue minerals for analysis, if those minerals do indeed have principally different sources.

The isotopic data presented here are consistent with episodic introduction of metal(loids) and sulfur from a deeper magmatic source into the shallow epithermal-geothermal setting. Present-day geothermal systems are analogs of epithermal systems, but the implications from the data presented here are that (1) it is highly unlikely that geothermal systems consisting of only heated meteoric water form epithermal ore deposits, and (2) a necessary component for making an epithermal ore deposit is the presence of magma with the right composition below the epithermal system, tectonic history, and structural setting in the crust. The LS epithermal ores of the western USA (NGB) and Japan formed in tectonic settings that allowed for partial melting of a fertile lithospheric mantle, generation of metal S-rich basaltic partial melts, which rose to shallow levels in the crust and triggered ore formation (and often bimodal volcanism). Thus, the tectonic setting and the composition of the mafic magmas are perhaps what should be evaluated and focused on for exploration for new epithermal deposits.

From the exploration standpoint, it is irrelevant if the deeper magmas episodically introduced ore constituents as “vapors” or physically transported nanoparticles that probably formed from cooling and condensing of such vapors along (or both processes in concert) with the upward rise of the magmatic fluids. Nanoparticle aggregation likely provided an efficient mechanism to facilitate the characteristic high grades of these ores, but not the total amount of metal endowments present in the deposits. In the epithermal setting, cooling and boiling of the epithermal solutions can lead to further metal and silica deposition at the same time as existing silica and nanoparticles are aggregating along vein walls. Thus, in summary, a minor contribution of magmatic fluids mixing with the shallow meteoric water-dominated geothermal system would not appreciably change the O and H isotopic composition of the ore-forming system, but that small fraction of magmatic fluid contributes essentially all the metal(loid)s and sulfur to the shallow ore-forming solutions. Implications of these results profoundly affect exploration strategies for these ores, which previously often focused on exploring for fossil geothermal systems. Instead, exploration should focus on the geotectonic settings that produce the right type of magmas capable of releasing metal-rich fluids necessary to form epithermal ores and deeper deposits as well.