Carlin-type gold deposits (CTDs) of Nevada are the largest producers of gold in the United States, a leader in world gold production. Although much has been resolved about the characteristics and origin of CTDs in Nevada, major questions remain, especially about (1) the role of magmatism, whether only a source of heat or also metals, (2) whether CTDs only formed in the Eocene, and (3) whether pre-Eocene metal concentrations contributed to Eocene deposits. These issues are exemplified by the CTDs of the Cortez region, the second largest concentration of these deposits after the Carlin trend.

Carlin-type deposits are notoriously difficult to date because they rarely generate dateable minerals. An age can be inferred from crosscutting relationships with dated dikes and other intrusions, which we have done for the giant Cortez Hills CTD. What we term “Cortez rhyolites” consist of two petrographic-geochemical groups of siliceous dikes: (1) quartz-sanidine-plagioclase-biotite-phyric, high-SiO2 rhyolites emplaced at 35.7 Ma based on numerous 40Ar/39Ar dates and (2) plagioclase-biotite-quartz ± hornblende-phyric, low-SiO2 rhyolites, which probably were emplaced at the same time but possibly as early as ~36.2 Ma. The dikes form a NNW-trending belt that is ~6 to 10 km wide × 40 km long and centered on the Cortez Hills deposit, and they require an underlying felsic pluton that fed the dikes. Whether these dikes pre- or postdated mineralization has been long debated. We show that dike emplacement spanned the time of mineralization. Many of both high- and low-SiO2 dikes are altered and mineralized, although none constitute ore. In altered-mineralized dikes, plagioclase has been replaced by kaolinite and calcite, and biotite by smectite, calcite, and marcasite. Sanidine is unaltered except in a few samples that are completely altered to quartz and kaolinite. Sulfides present in mineralized dikes are marcasite, pyrite, arsenopyrite, and As-Sb–bearing pyrite. Mineralized dikes are moderately enriched in characteristic Carlin-type elements (Au, Hg, Sb, Tl, As, and S), as well as elements found in some CTDs (Ag, Bi, Cu, Mo), and variably depleted in MgO, CaO, Na2O, K2O, MnO, Rb, Sr, and Ba. In contrast, some high-SiO2 rhyolites are unaltered and cut high-grade ore, which shows that they are post-ore. Both mineralized and post-ore dikes have indistinguishable sanidine 40Ar/39Ar dates. These characteristics, along with published interpretations that other giant CTDs formed in a few tens of thousands of years, indicate the Cortez Hills CTD formed at 35.7 Ma. All Cortez-area CTDs are in or adjacent to the Cortez rhyolite dike swarm, which suggests that the felsic pluton that fed the dikes was the hydrothermal heat source. Minor differences in alteration and geochemistry between dikes and typical Paleozoic sedimentary rock-hosted ore probably reflect low permeability and low reactivity of the predominantly quartzofeldspathic dikes.

Despite widespread pre-35.7 Ma mineralization in the Cortez region, including deposits near several CTDs, we find no evidence that older deposits or Paleozoic basinal rocks contributed metals to Cortez-area CTDs. Combining our new information about the age of Cortez Hills with published and our dates on other CTDs demonstrates that CTD formation coincided with the southwestern migration of magmatism across Nevada, supporting a genetic relationship to Eocene magmatism. CTDs are best developed where deep-seated (~6–8 km), probably granitic plutons, expressed in deposits only as dikes, established large, convective hydrothermal systems.

Carlin-type gold deposits (CTDs or sedimentary rock-hosted disseminated gold deposits) in Nevada formed from the conjunction of many critical factors (Hofstra and Cline, 2000; Cline et al., 2005; Muntean et al., 2011; Groves et al., 2016; Meffre et al., 2016; Cline, 2018; Muntean and Cline, 2018; Muntean, 2020). Documented or interpreted factors include (1) location above the Neoproterozoic rifted continental margin, (2) primarily Paleozoic silty carbonate host rocks, (3) structural preparation by Paleozoic-Mesozoic crustal shortening, and (4) enhanced permeability because of contemporaneous extension. The deposits themselves are generally understood to have formed at shallow depths (tops of deposits ≤2 km) and at epithermal temperatures (~180°–240°C) and to be spatially and temporally associated with Eocene magmatism where geologic relationships permit such association to be documented (Sillitoe and Bonham, 1990; Seedorff, 1991; Henry and Ressel, 2000; Johnston and Ressel, 2004; Cline et al., 2005; Ressel and Henry, 2006; Muntean et al., 2011; Johnson et al., 2015; Cline, 2018; Muntean and Cline, 2018; Muntean, 2020).

Less resolved are the role of magmatism, the source of metals and fluids, and to what extent CTDs form from multiple gold-depositing hydrothermal systems separated by millions to even hundreds of millions of years. Was magmatism simply a heat source to drive hydrothermal systems or was it also an essential source of gold, other metals and metalloids, ligands, and fluids (e.g., Henry and Boden, 1998; Kesler et al., 2005; Muntean et al., 2011; Thompson, 2011; Large et al., 2016; Muntean, 2020; Sillitoe and Brogi, 2021)? Was gold derived from magmas whose ultimate metal source is the mantle (Muntean et al., 2011) or recycled from the crust? If recycled, was gold derived from the underlying Neoproterozoic clastic sedimentary sequence (Ilchik and Barton, 1997; Tosdal et al., 2003), sedimentary exhalative deposits in Devonian host rocks (Emsbo et al., 1999, 2003, 2006), or black shales (Large et al., 2011)? Are CTDs multistage, cumulative deposits with significant gold deposition ± remobilization from Cretaceous or Devonian hydrothermal systems, largely or partly driven by magmatism (Drews-Armitage et al., 1996; Groff et al., 1997; Hall et al., 2000; Groves et al., 2016; Meffre et al., 2016; Blamey et al., 2017; Henry et al., 2020)? These are not either/or questions. Magma could have transported gold derived from crustal assimilation, because Eocene magmas interacted extensively with the crust such that most silicic magmas are predominantly crustally derived (Farmer and DePaolo, 1983; King et al., 2004; Vikre et al., 2011; Watts et al., 2016b; Cousens et al., 2019). A related question is whether CTDs are distal expressions of higher-temperature, distal disseminated Au-Ag deposits that are proximal to upper crustal intrusions also associated with porphyry Cu-Mo-Au systems (Sillitoe and Bonham, 1990; Presnell and Parry, 1996; Gunter and Austin, 1998; Cunningham et al., 2004; Johnston and Ressel, 2004; Cline et al., 2005; Heinrich, 2005; Large et al., 2016; Hollingsworth et al., 2017; Cline, 2018; Muntean and Cline, 2018; Ressel et al., 2019; Muntean, 2020; Sillitoe, 2020; Sillitoe and Brogi, 2021).

If igneous activity that drove hydrothermal systems and supplied metals and fluids was the ultimate factor controlling CTD formation, why did these deposits form only in the Eocene? Northern Nevada has undergone multiple episodes of magmatism, notably in the Jurassic, Cretaceous, Eocene-Oligocene, and Miocene, and, where well dated, CTDs are restricted to the Eocene. Finally, to what extent was contemporaneous extensional faulting necessary to enhance permeability and allow hydrothermal circulation?

The Cortez region of the Battle Mountain-Eureka trend offers an important opportunity to evaluate some of the above questions. It is the second largest concentration of CTDs in the world, after the Carlin trend (Figs. 1, 2; Cline et al., 2005; Muntean et al., 2011; Muntean and Cline, 2018; Bradley et al., 2020). The Pipeline and Cortez Hills deposits have produced or contain more than 11 and 21 Moz Au, respectively (Arbonies et al., 2011), and major deposits are still being discovered (e.g., ~15 Moz Au at Goldrush-Fourmile; Creel and Bradley, 2013; Bradley and Eck, 2015; Bradley et al., 2020). Other CTDs in the Cortez area are the original Cortez mine (Radtke et al., 1987), 2.5 km northwest of Cortez Hills; Horse Canyon mine, 4 km east of Cortez Hills (Foo and Hebert, 1987); and the Gold Acres mine, 2.5 km west of Pipeline. The Cortez region is similar in important aspects to the Carlin trend, especially the presence of the same Jurassic, Cretaceous, Eocene, and Miocene episodes of magmatism and hydrothermal activity (Ressel and Henry, 2006; Henry et al., 2020).

The giant Cortez Hills CTD was discovered in 2002 and went into production in 2008 (Arbonies et al., 2011; Jackson et al., 2011; Maroun et al., 2017; Bradley et al., 2020). Devonian and Silurian silty carbonate rocks and a transecting breccia are the primary host rocks at Cortez Hills (Figs. 3, 4; Arbonies et al., 2011; Jackson et al., 2011). Cortez Hills is divided into the breccia zone, lower zone, and pediment deposit. The steeply plunging breccia zone is variably interpreted as hydrothermal (Jackson et al., 2011) or solution collapse (Maroun et al., 2017). The lower zone consists of tabular deposits that parallel bedding and a thrust fault in Silurian rocks (Fig. 4). The pediment deposit consists of gold-bearing clasts eroded off the top of the breccia zone and incorporated into middle Miocene conglomerate south of the in situ deposits.

This paper focuses on the timing and primary and alteration characteristics of a suite of rhyolite dikes, here informally named the Cortez rhyolites, that are intimately associated with and, based on our work, coeval with Cortez Hills mineralization. Whether these dikes pre- or postdate mineralization is a long-standing question (Wells et al., 1969; Rytuba, 1985; Radtke et al., 1987; McCormack and Hays, 1996; Hays et al., 2004). The most recent interpretation, largely based on preliminary results of our dating, is that the dikes were emplaced during a late ore stage (Arbonies et al., 2011). We show here that dike emplacement spanned pre- to postmineral time, which demonstrates that the Cortez Hills deposit formed at 35.7 Ma, although the dikes do not contain ore (see also Maroun et al., 2017).

Details of field and analytical methods and data reduction for 40Ar/39Ar and U-Pb dating are in Appendix Text A1. This study relies on our extensive new detailed and reconnaissance geologic mapping in the greater Cortez region (John et al., 2000, 2008; Colgan et al., 2008, 2011, 2014; Moore and Henry, 2010). At Cortez, we determined 24 new 40Ar/39Ar dates (20 sanidine, three biotite, and one sericite; all except the sericite and one biotite are on Cortez rhyolites) and five U-Pb zircon dates (all Cortez rhyolites) for this study (Table 1; App. Table A1). All 40Ar/39Ar dating was done at the New Mexico Geochronological Research Laboratory (methods in McIntosh et al., 2003; Henry et al., 2017). All dates, including published, were calculated or recalculated to an age of 28.201 Ma for Fish Canyon sanidine monitor (Kuiper al., 2008). Most sanidine dates were done between 2008 and 2011 with a single collector MAP215-50 mass spectrometer. Two biotite dates were done, and two sanidine samples reseparated and reanalyzed, with a significantly more sensitive and precise multicollector ARGUS VI instrument (Heizler, 2011; Heintz et al., 2014; Heizler et al., 2014; Hereford et al., 2016).

Uranium-lead zircon dating was done at the Stanford-U.S. Geological Survey (USGS) using sensitive high resolution ion microprobe with reverse-geometry (SHRIMP-RG) (Colgan et al., 2022). Thirty-nine samples of Cortez rhyolites were analyzed for major oxides and trace elements at Washington State University, ALSGlobal, and AGAT Laboratories (USGS contract) by methods detailed in Appendix Text A1 (App. Table A2). Backscattered images and semiquantitative chemical analyses were obtained by energy dispersive spectroscopy using the JEOL 7100FT field emission scanning electron microscope at the Mackay Microbeam Laboratory at the University of Nevada, Reno (representative mineral analyses are in App. Table A3).

Comparison of 40Ar/39Ar and U-Pb zircon dates

The 40Ar/39Ar and U-Pb zircon dates do not record the same events in the evolution of igneous magmas and rocks, because different minerals become closed to loss of radiogenic daughter isotopes at different temperatures (Simon et al., 2008; Chiaradia et al., 2014). 40Ar/39Ar dates record cooling in which individual dated minerals begin to retain radiogenic 40Ar only after cooling below characteristic closure temperatures (McDougall and Harrison, 1999). Cooling is rapid for volcanic rocks and shallowly emplaced intrusions and can be measurably slower for deeply emplaced intrusions. 40Ar/39Ar dates of sanidine from dikes record the time of dike emplacement when the dikes cooled geologically instantaneously from magmatic to hostrock temperatures (~675°–760°C [Mercer, 2021] to <100°C for the rhyolite dikes of this study). Reheating of rocks after initial cooling can induce 40Ar loss. Zircons can retain radiogenic Pb at magmatic temperatures as high as 900°C, and zircon can survive as xenocrysts and retain crystallization ages even when incorporated into magma during crustal assimilation (Watson, 1996; Cherniak and Watson, 2001). Many magma chambers are interpreted to have existed for >250 ka and were possibly incrementally assembled over several million years (Reid et al., 1997; Glazner et al., 2004; Simon et al., 2008). Uranium-lead zircon dates in such situations may record the entire lifetime of the magma chamber; an average age from them should be older than 40Ar/39Ar emplacement ages. Zircons can also undergo Pb loss due to radiation damage from U and Th decay and provide anomalously young ages, even several million years younger than emplacement (Watts et al., 2016a).

Paleozoic sedimentary rocks and Paleozoic, Mesozoic, and Cenozoic deformation

The oldest exposed rocks in the Cortez region are Cambrian to Devonian carbonate and siliciclastic rocks deposited on the North American continental shelf and slope (Gilluly and Gates, 1965; Gilluly and Masursky, 1965; Wrucke, 1974; Leonardson, 2011, 2015; Pzl of Fig. 2). These strata are structurally overlain by the Roberts Mountains allochthon, a highly deformed sequence of Cambrian to Devonian, deep-marine chert, argillite, quartzite, and greenstone (Pzu). In the context of the CTDs, the two rock packages are generally, and throughout the rest of this paper, referred to as the “lower plate” and “upper plate” in reference to their respective positions relative to the Roberts Mountains thrust. Upper plate rocks were thrust eastward over the coeval lower plate rocks during the Late Devonian-Mississippian Antler orogeny (Roberts et al., 1958). Mississippian-Permian clastic and carbonate rocks of the Antler overlap sequence overlie the Roberts Mountains allochthon (e.g., Roberts et al., 1958) (Po, Fig. 2). Exposed pre-Cenozoic rocks in the area of Figure 2 predominantly consist of upper plate rocks of the Roberts Mountains allochthon, with lower-plate carbonate rocks locally exposed in structural windows through the Roberts Mountains thrust.

Leonardson (2011) documented a Mesozoic episode of E-directed thrusting that repeats the Roberts Mountains thrust and upper and lower plates around the Pipeline deposit. This event generated asymmetric anticlines at Pipeline and throughout the Cortez region, including Cortez Hills (Fig. 4), that were important in focusing flow of Carlin-type hydrothermal fluids (Leonardson, 2011, 2015; Bradley et al., 2020). Folding could have occurred in the Jurassic (Luning-Fencemaker fold-and-thrust belt and Elko orogeny; Thorman et al., 1991; Wyld, 2002; Zuza et al., 2020) or Late Cretaceous (Sevier orogeny; Dickinson, 2006).

Formation of the Eocene Elko basin around the Carlin trend is variably interpreted to result from minor extension with surfacebreaking faults (Solomon et al., 1979; Haynes, 2003; Howard, 2003; Henry, 2008, 2018; Henry et al., 2011) or from nonfault-related uplift followed by subsidence resulting from removal (rollback) of the Farallon slab (Smith et al., 2017). We favor minor extension because surface-breaking faults are documented by the first set of studies. Moreover, because the progression of magmatism—the evidence for slab rollback—continued and progressed across the Great Basin through to the present (Humphreys, 1995; Henry and John, 2013; Best et al., 2016), the rollback uplift mechanism for formation of the Eocene basins would require that similar basins form and migrate along with the magmatism, but they did not.

Major extension near Cortez began about 17 to 16 Ma, as part of regional extension across much of the Basin and Range province of western North America (Colgan and Henry, 2009). In the Cortez region, extension is expressed primarily by eastward tilting along W-dipping normal faults that locally reactivated Paleozoic or Mesozoic thrust faults (Colgan et al., 2008, 2014; Colgan and Henry, 2009). As it is elsewhere in the Basin and Range, middle Miocene extension around Cortez is partitioned into domains of high and low strain (Colgan et al., 2008, 2014). The amount of extension and tilting decreases from 100% strain (avg 40° dips) across the Caetano caldera and northwestern Shoshone Range to almost nothing in the northern Nevada rift and Cortez Mountains where middle Miocene rocks dip 5° to 10° east-southeast (Gilluly and Gates, 1965; Gilluly and Masursky, 1965; John et al., 2000; Colgan et al., 2014; this study). Cortez Hills lies in the transition between tilt magnitudes. Middle Miocene clastic rocks that overlie Cortez Hills accumulated in a half graben and dip ~20° east into the W-dipping, basin-bounding Cortez fault (Fig. 4). Similar tilts on 35 to 16 Ma rocks in single stratigraphic sections in the Caetano caldera and the northern Shoshone Range indicate negligible extension during that time. Middle Miocene extension ended by ~10 Ma, and subsequent deformation took place on widely spaced, high-angle normal faults that formed the modern ranges and basins (e.g., the Crescent fault; Fig. 3; cf. Colgan and Henry, 2009).

Igneous episodes of the Cortez region

The Cortez area has undergone five episodes of igneous activity in the Jurassic, Cretaceous, Eocene (two episodes at 40 to 39 and 36 to 34 Ma), and middle Miocene (Figs. 1, 2; Henry et al., 2020). Jurassic and Cretaceous rocks consist mostly of granitic plutons emplaced during subduction of the Farallon plate beneath North America, inboard of the main batholith belt in the Sierra Nevada (Dickinson, 2006, 2013). Jurassic intrusions around Cortez and in the northern Great Basin were emplaced between about 166 and 158 Ma, shortly following Jurassic crustal shortening and coeval with major plutonism in and immediately east of the Sierra Nevada (Dilles and Wright, 1988; Mortensen et al., 2000; Ducea, 2001; Irwin, 2003; du Bray, 2007; Barton et al., 2011; Cecil et al., 2012; Wyld and Wright, 2014; Zuza et al., 2020, 2021).

Jurassic intrusions near Cortez Hills consist of the quartz monzonitic-granodioritic Mill Canyon stock-laccolith, dikes and irregular intrusions of porphyritic rhyolite and dacite, and lamprophyre dikes. They are the southwestern part of a belt of Jurassic intrusive and volcanic rocks that occupies most of the Cortez Mountains (Figs. 24; Muffler, 1964; Wyld and Wright, 2014). Contact effects are small; marble or hornfels extend only a few meters into wall rock (Gilluly and Masursky, 1965). Uranium-lead zircon and 40Ar/39Ar dates demonstrate that the Mill Canyon and related intrusions were emplaced between about 162 and 158 Ma (Mortensen et al., 2000; Wyld and Wright, 2014).

Although known exposures of Cretaceous rocks in the study area consist of a single dacite dike, drilling has revealed three major granodiorite intrusions at ≤1-km depth, and Cretaceous zircon xenocrysts in Cenozoic rocks indicate the presence of additional unexposed intrusions (Figs. 2, 3; Henry et al., 2020). A subsurface granodiorite at Gold Acres has long been known from its adjacent hornfels, skarn, and magnetic anomaly and from drill holes around the Gold Acres CTD (Gilluly and Gates, 1965; Wrucke, 1974; Townsend et al., 2011). Uranium-lead dating of the Gold Acres intrusion demonstrates that it is ~105 Ma (Mortensen et al., 2000). Two drill holes encountered ~104 and 109 Ma granodiorites near Cortez Hills (Figs. 2, 3; Henry et al., 2020).

Eocene magmatism in the Cortez area is part of a southwestward sweep of magmatism that began in northernmost Nevada about 47 Ma and migrated continuously southwestward across the Great Basin (Henry and John, 2013). Magmatism between 47 and 40 Ma was predominantly intermediate to silicic (andesite-dacite, granodiorite, and minor rhyolite) intrusive-effusive with a few voluminous rhyolitic caldera-forming eruptions. Between 40 and 37 Ma, magma composition remained the same, forming abundant lavas and intrusions but no voluminous caldera-forming ignimbrites. At about 37 Ma, eruption of voluminous caldera-forming rhyolites resumed while intermediate effusive activity continued. The period of intense caldera-forming eruptions from ~37 to at least 17 Ma constitutes the well-known “ignimbrite flare-up” (Coney, 1978; Henry and John, 2013; Best et al., 2016). West-southwestward migration of magmatism continued into the late Cenozoic ancestral Cascade arc of the Sierra Nevada (Busby et al., 2008; Cousens et al., 2008; du Bray et al., 2014; John et al., 2015; John and Henry, 2020). The best interpretation for these patterns is that the Farallon slab subducted along the base of the continental lithosphere without any intervening asthenosphere during the Late Cretaceous and early Cenozoic, and dehydration of the slab hydrated the overlying lithosphere and primed it for future melting (Humphreys et al., 2003). During subsequent rollback or delamination beginning about 50 Ma, asthenosphere welled up to replace the slab, heated the overlying fertile, hydrated lithosphere, and generated voluminous magmas (Coney and Reynolds, 1977; Best and Christiansen, 1991; Humphreys, 1995; Henry and John, 2013; Best et al., 2016).

Published and our geologic mapping and geochronology indicate that Eocene magmatism occurred in two major episodes in the Cortez region: ~40 to 39 and ~36.3 to 33.5 Ma (John et al., 2008; Henry et al., 2020). An intense, entirely intrusive episode of predominantly granodiorite occurred in the northern Shoshone Range at 40 to 39 Ma (Fig. 2). Three major intrusions and numerous plugs of similar granodiorite intruded upper plate rocks near Hilltop, at Granite Mountain, and at Tenabo. Structural reconstruction and deposition of 36 Ma volcanic rocks on upper plate rocks nearby (Gilluly and Gates, 1965; Colgan et al., 2014) indicate shallow emplacement at ≤2 km. The largest and least altered intrusion, Granite Mountain (~8 km2), is predominantly granodiorite (Gilluly and Gates, 1965; Kelson et al., 2008). Similar bodies with more felsic and mafic variants are exposed at Hilltop (2 km2) and Tenabo (1.3 km2), where they are commonly hydrothermally altered (McCusker, 1996; Kelson et al., 2008). Numerous 40Ar/39Ar and U-Pb dates (Kelson et al., 2005, 2008; Colgan et al., 2022) indicate the intrusions were emplaced about 39.8 to 39.4 Ma. Similarity in mineralogy, composition, and age allow the three main intrusions to be apophyses of one larger body that underlies ~55 km2.

Igneous rocks emplaced from 36.3 to 33.5 Ma are predominantly rhyolite and include the major ~35.7 Ma Cortez rhyolite dike swarm—the focus of this study—and the 34.0 Ma Caetano Tuff and caldera (Colgan et al., 2008; John et al., 2008; Watts et al., 2016b). Eruption of ~1,100 km3 of crystal-rich, high- to low-SiO2 rhyolite Caetano Tuff at 34.0 Ma induced collapse of the Caetano caldera, which was initially ~20 km long × 12 to 18 km wide (Fig. 2). Major extension beginning in the middle Miocene tilted intracaldera Caetano Tuff into a series of 40° E-dipping blocks and stretched the caldera to its present day 42-km width. Tilting exposed paleodepths as deep as 5 km, including a complete section of intracaldera tuff as much as 4 km thick, cogenetic, postcollapse intrusions that are petrographically and compositionally similar to last erupted Caetano Tuff, and locally underlying, precaldera 35.2 Ma andesite and Paleozoic sedimentary rocks.

Middle Miocene igneous rocks in the Cortez region are predominantly basaltic andesite to andesite lavas and dikes and one rhyolite intrusion, the “bimodal basalt-rhyolite assemblage” of John (2001), that erupted along the NNW-trending northern Nevada rift (Fig. 2). Rift magmatism is part of Yellowstone hot-spot magmatism (Zoback et al., 1994; John et al., 2000, 2003; Camp et al., 2013). Formation of the rift at ca. 17 Ma coincided with the onset of significant, locally major, extension in this part of the Basin and Range province, although the rift itself underwent only minor extension (Zoback et al., 1994; John et al., 2000; Colgan and Henry, 2009; Colgan, 2013).

Field and petrographic characteristics

The Cortez rhyolites are a suite of porphyritic rhyolite dikes, sills, and minor lava domes and pyroclastic rocks that form a nearly 40-km-long NNW-trending belt centered around Cortez Hills and extending north to Tenabo and south to the northern Simpson Park Mountains (Figs. 25; Table 1). The Cortez dikes consist of two petrographic types that mostly form two compositional types: a finely porphyritic, high-SiO2 rhyolite and a more coarsely porphyritic, low-SiO2 rhyolite. The two petrographic types probably intruded in a single pulse at 35.7 Ma, but some low-SiO2 rhyolites could have intruded as early as ~36.2 Ma. Our field, geochemical, and geochronological data address this question, which bears on the character and duration of the magmatic heat and possible metal source for Cortez Hills and other CTDs of the area.

The rhyolites form a complex array in and around Cortez Hills, north to the original Cortez mine, and south to the northeastern margin of the Caetano caldera and beneath Quaternary cover. The rhyolite belt is about 8 km wide near Cortez, where the easternmost body is a thick dike or lava dome that crops out in Horse Canyon 2 km west of Goldrush. Blocks of rhyolite are common as megabreccia in Caetano Tuff in the eastern part of its caldera, which indicates that they were common at least that far west. Cortez rhyolite dikes are abundant around Tenabo (Wrucke, 1974; this study), and one dike and possibly related rhyolite lava domes in the Fye Canyon volcanic rocks (Fig. 2) are exposed in the northern Simpson Park Mountains. The overall distribution suggests a NNW-striking belt of dikes from Tenabo to the northern Simpson Park Mountains, but postdike cover in Crescent Valley and south of Cortez precludes determining the full distribution (Fig. 2).

Individual Cortez rhyolites commonly follow bedding or low-angle structures, then step abruptly across bedding along high-angle structures, and then again follow bedding (Fig. 5A-C). The rhyolites were emplaced along the same structural conduits as the ore-bearing fluids (Arbonies et al., 2011; Jackson et al., 2011). Bodies also commonly bifurcate and pinch and swell (Figs. 4, 5A). Interiors of dikes crystallized to a mix of quartz and K-feldspar. Matrix K-feldspar is compositionally similar to sanidine phenocrysts except generally lacking detectable Ba, which ranges from ~0.4 to 1.4 wt % in phenocrysts (App. Table A3). Common marginal vitrophyres that vary from preserved glass to altered to smectite attest to rapid cooling during emplacement. Spherulites and snowflake texture in highly altered rocks indicate the former presence of many more vitrophyres (Fig. 5H). A 300- to 400-m-diameter megabreccia block of variably pumiceous lithic tuff—petrographically part of the Cortez rhyolites—in the Caetano caldera demonstrates that some Cortez rhyolite erupted. Several dikes encountered in drill holes around Cortez Hills transition upward from massive and flow banded into pumiceous breccias with rounded clasts of dike and Paleozoic rocks up to 80 cm in diameter in a matrix of finer clasts of both types and phenocrysts (Fig. 5D). These pyroclastic breccia dikes crop out in several areas between the Fortress and Copper faults along the Caetano caldera margin (Fig. 3) and probably were shallow subsurface feeders to small pyroclastic eruptions.

The predominant petrographic type of Cortez rhyolite is high-SiO2 rhyolite that contains 10 to 22% phenocrysts, mostly of quartz and sanidine up to 2 mm across, 1 to 4% plagioclase, 0.5 to 2% biotite, and sparse hornblende preserved in one vitrophyre (Fig. 6A, B). High-SiO2 rhyolites are concentrated around Cortez Hills and occur throughout the entire dike distribution from Tenabo to the northern Simpson Park Mountains. They are variably altered, ranging from strongly altered and mineralized to apparently unaltered but hydrated vitrophyres. Sanidine survived almost all alteration, but plagioclase is altered to kaolinite and calcite in most samples, and biotite is altered to smectite in many. Apatite, zircon, and xenotime (allanite?) are present in both petrographic types. Melt inclusions are common in both quartz and sanidine phenocrysts. Mercer (2021) reported melt inclusions in quartz are 10 to 160 µm in diameter and have compositions similar to our whole-rock analyses of high-SiO2 rhyolite.

Less abundant low-SiO2 rhyolites contain 15 to 20% phenocrysts, mostly of plagioclase up to 6 mm long, 3 to 4% biotite, ≤2% quartz, and sparse hornblende; sanidine is absent (Figs. 5C, 6C, D). These crop out or are encountered in drill holes in and within about 500 m of the Cortez Hills deposit. Most examples are highly altered, with destruction of plagioclase, biotite, and hornblende phenocrysts. One dike in drill hole DC189 appears to be zoned from a more feldspar-rich, quartz-poor border (DC189-2181) to a more quartzrich interior (DC189-2196); all feldspars are destroyed throughout this dike.

Geochemistry

Variable alteration complicates interpretation of chemical composition. However, the use of unaltered rocks and of immobile elements allows good characterization (Figs. 7, 8). We also compare the Cortez rhyolites with unaltered Caetano Tuff rhyolites to help understand the former’s primary composition. The Cortez rhyolites are high K and magnesian (calc-alkaline) with other typical volcanic-arc characteristics (e.g., Y/Nb; Fig. 7E), similar to all mid-Cenozoic rocks of Nevada (Christiansen and McCurry, 2008; Henry and John, 2013; Best et al., 2016; Watts et al., 2016b). Plots of SiO2 versus other elements that are traditionally used to characterize igneous rocks illustrate significant alkali mobility (Fig. 7). Although total alkalis (Na2O + K2O) in Cortez rhyolites mostly overlap with those of unaltered Caetano Tuff, plots of individual alkalis demonstrate major loss of Na2O and variable addition or loss of K2O. Probably all Cortez rhyolite samples have lost some Na2O. A small proportion of nonhydrothermally altered Caetano Tuff samples have lost Na2O (Fig. 7D).

A plot of SiO2 versus immobile TiO2 shows two distinct groups that match the petrographic types (Fig. 8). A high-TiO2 group (0.33–0.38 wt %) matches the plagioclase-biotite-quartz-phyric rocks and has normalized SiO2 contents from 72.8 to 76.7 wt %. A low-TiO2 group (≤0.12 wt %) matches the sanidine-bearing rhyolites and has SiO2 contents from 67.8 to 86.3 wt %. The wide range and overlap in SiO2 between the two types reflect loss or addition of other major oxides and possibly minor SiO2 mobility. The SiO2 content of a least altered sample of the high-TiO2 group (H09-1, 72.8 wt % SiO2, 0.38 wt % TiO2; plagioclase and biotite largely preserved) and comparison with unaltered Caetano Tuff indicate that the high-TiO2 type is low-SiO2 rhyolite. The six least altered, low-TiO2 rocks have SiO2 ranging from 75.1 to 79 wt % and averaging 76.9 wt %, demonstrating that they are high-SiO2 rhyolites. TiO2 in these six samples varies from 0.047 to 0.116 wt %, which demonstrates a compositional range within the high- SiO2 rhyolites. Without the effects of alteration, the SiO2-TiO2 trend of Cortez rhyolites would match that of Caetano Tuff, which also shows a range from low to high SiO2 (John et al., 2008; Watts et al., 2016b). We therefore call the two Cortez rhyolite types high- and low-SiO2 rhyolites despite using TiO2 as a primary discriminator.

Plots of TiO2 versus other elements better depict the two petrographic-chemical types and their overall compositional range (Fig. 8). Linear correlations of TiO2 with P2O5 and several trace elements (Zr, La, and Eu depicted here) indicate the latter were also immobile. Both types show a perceptible range in TiO2, from 0.334 to 0.382 wt % for the low-SiO2 type and 0.021 to 0.116 wt % for high-SiO2 type, with most ≤0.08 wt %. Covariation of plotted elements with TiO2 indicates that the variations are related to differentiation. Decreases in TiO2, P2O5, Zr, La, Ba, and Eu are consistent with fractionation resulting from crystallization of magnetite-ilmenite (Fe-Ti oxides), apatite, zircon, allanite, sanidine, and feldspar, respectively. This pattern is best illustrated by Eu in high-SiO2 rhyolites, which decreases from 0.83 to 0.29 ppm with decreasing TiO2 (Fig. 8F). The most evolved high-SiO2 rhyolites have the lowest concentrations of light rare earth elements (LREEs), highest concentrations of heavy rare earth elements (HREEs), and largest Eu depletions. The linear trends in composition are typical of Eocene-Oligocene rhyolites of the Great Basin and consistent with fractionation of the observed major and minor phases (Christiansen and McCurry, 2008; Watts et al., 2016b; Mercer, 2021). The nonlinearity in Eu between low-SiO2 rhyolites and the most TiO2-rich high-SiO2 rhyolites indicates the latter are not simple differentiates of the former.

Two samples, DC215-1938 (Fig. 6A, B) and H12-162 from the Simpson Park Mountains, are similar to high-SiO2 rhyolites but distinguished by higher TiO2 (0.12% wt %) as well as higher Zr, Ba, and LREEs and lower SiO2 (Fig. 8; App. Table A2). DC215-1938 also has the only hornblende found in high-SiO2 rhyolites.

Two samples that plot either off the linear trends or at very high TiO2 have undergone intense kaolinite alteration and probable mass loss (alteration is discussed below). DC281-2465 of low-SiO2 rhyolite has 0.559 wt % TiO2 and 20.70 wt % Al2O3 (not shown on Fig. 8A). DC189-2255 of high-SiO2 rhyolite has 0.142 wt % TiO2 and 31.48 wt % Al2O3, normalized SiO2 and Al2O3 total 99.3 wt %, and X-ray diffraction and scanning electron microscopy analyses reveal that it is almost entirely quartz and kaolinite. Chondrite-normalized REE patterns of both samples parallel but are much higher than those of less altered samples, suggesting major mass loss, probably 50 to 70% in DC189-2255 and 25 to 55% in DC281-2465 based on REE ratios. Eu, which would not be affected by alteration, suggests that DC189-2255 was a typical high-SiO2 rhyolite probably with ~0.07 wt % TiO2 and 77 wt % SiO2 before alteration. A few percent of small (~24–100 µm) spherulites reside in an extremely fine, quartz-kaolinitic matrix of DC189-2255, indicating a former vitrophyre that probably was more reactive than crystalline rock (Fig. 5F). DC281-2465 probably was close in composition to, or possibly slightly less silicic than, other low-SiO2 rhyolites, because it lacks quartz phenocrysts.

Cortez rhyolites are mostly compositionally similar to Caetano Tuff but show greater scatter because of greater alteration (Figs. 7, 8). Least altered Cortez rhyolites have the same range in SiO2 and plot with unaltered Caetano Tuff for many major oxides and trace elements. The two suites overlap on most plots of TiO2 and immobile elements (Fig. 8). A notable difference is in TiO2 versus Al2O3, where many Cortez rhyolites have undergone kaolinitic alteration and variable mass loss. Plots of Zr and Nb versus TiO2 illustrate that crystallization of zircon depleted both suites in Zr, but that the most evolved Caetano Tuff rocks are highly enriched in Nb (also in Rb, Y, U, HREEs, and Ta and strongly depleted in TiO2, Ba, and Sr; Watts et al., 2016b). This divergence in trend indicates either that Caetano magma underwent more extreme differentiation or that some phase(s) incorporated Nb and these other elements in Cortez rhyolites so that it did not undergo the extreme enrichment; ilmenite could have incorporated Nb and Ta (Mercer, 2021). A plot of Y versus Nb also illustrates this point (Fig. 7E). All except one Cortez rhyolite sample have low Y and Nb concentrations and plot in the volcanic arc field, whereas a small volume of highly differentiated Caetano Tuff magma is greatly enriched in these elements and plots in the within-plate field. The one high-Y-Nb Cortez rhyolite is sample DC189-2255, which is entirely altered to quartz and kaolinite and has undergone major mass loss, leading to apparent Y-Nb enrichment. The lack of very low Eu contents in Cortez rhyolites compared to Caetano Tuff also indicates lesser overall differentiation, particularly lesser feldspar fractionation (Fig. 8F).

The similarities in composition between Cortez rhyolites and Caetano Tuff allow high- and low-SiO2 Cortez rhyolites to be genetically related to each other by processes similar to those that generated the compositional range in the tuff: mantle melt generation, major crustal assimilation (70–80% in the Caetano Tuff), fractional crystallization, and magma mixing (Watts et al., 2016b). Cortez rhyolites have undergone less overall differentiation, and the proportion of crustal involvement is unknown. Given their importance for Carlin-type gold mineralization, further investigation of the geochemistry and petrogenesis of the Cortez rhyolites is warranted.

Age

Twenty 40Ar/39Ar sanidine (all on high-SiO2 rhyolites), two 40Ar/39Ar biotite (one each on high- and low-SiO2 rhyolites), and five SHRIMP U-Pb zircon (two high-SiO2 and three low-SiO2 rhyolites) dates tightly constrain the time of high-SiO2 rhyolite emplacement to 35.7 Ma but are more complex about the time of low-SiO2 rhyolite emplacement (Table 1; Figs. 9, 10). Dated high-SiO2 rhyolites span the dike distribution from Tenabo (n = 2) through the Cortez area (n = 17, including 10 mineralized) to the northern Simpson Park Mountains (n = 1). The weighted mean age and range of all 20 sanidine samples are 35.71 ± 0.05 Ma (all ± are 2σ) and 35.59 ± 0.15 to 35.84 ± 0.07 Ma; the mean and range of 17 dates from around Cortez Hills are 35.71 ± 0.05 Ma and 35.59 ± 0.15 to 35.77 ± 0.07 Ma. The 0.25 and 0.18 Ma age ranges mostly reflect analytical uncertainty and incomplete removal of minor alteration products (clay minerals and partial dissolution along cleavage planes) by HF leaching during mineral separation.

Sanidine from two samples (strongly mineralized CHUE017-843 and unaltered H07-69) was reseparated and analyzed in 2021. The new analyses illustrate the much greater precision and accuracy of analyses on the Argus VI multicollector mass spectrometer (Fig. 9A, B). Additionally, the two dates are separated by only 0.03 Ma, whereas the older, single-collector analyses differed by 0.12 Ma. The sanidine dates unequivocally place time of high-SiO2 dike intrusion at 35.7 Ma.

A biotite date of 36.13 ± 0.10 Ma on sample DC215-1938 is significantly older than the 35.60 ± 0.07 Ma date on coexisting sanidine (Table 1). Biotite should also record rapid cooling during emplacement but can be affected by recoil and contain excess 40Ar to yield anomalously old 40Ar/39Ar dates (Smith et al., 2003; Lipman and McIntosh, 2008; Bachmann et al., 2010; Hall, 2013). This result, as well as a biotite date on a low-SiO2 rhyolite (see below), is important for age interpretation.

The 40Ar/39Ar sanidine dates on two dikes at Tenabo are the oldest determined—35.83 ± 0.08 and 35.84 ± 0.07 Ma (H10-21, H10-22; Table 1)—although within uncertainty of the overall mean. Sanidine from the dike in the Simpson Park Mountains yielded 35.71 ± 0.01 Ma (H12-162).

Uranium-lead zircon dates on two high-SiO2 rhyolites—08-DJ-125 (unaltered vitrophyre from a megabreccia block in the Caetano caldera) and DC189-2255 (completely altered to quartz and kaolinite)—are 37.64 ± 0.30 and 36.89 ± 0.41 Ma, respectively (Table 1; Fig. 10). Both dates are outside the range of sanidine dates on high-SiO2 rhyolites, and the date on 08-DJ-125 is far outside analytical uncertainty of coexisting sanidine 40Ar/39Ar date of 35.72 ± 0.03 Ma (sample H07-69 collected from the same outcrop as 08-DJ-125). The zircon date is also at a time of no known magmatism in the Cortez region (Henry et al., 2020). The significance of these anomalous U-Pb dates is uncertain. The older date cannot be reconciled with the sanidine date by either a long-lived magma chamber or analytical uncertainty. Regardless, it does not record emplacement age, because it does not match the sanidine date.

Low-SiO2 rhyolites lack sanidine, so we dated one biotite by the 40Ar/39Ar method and zircons in three samples by U-Pb. Biotite and zircon dates on least altered sample H09-1, which has fresh biotite and plagioclase only partly altered to calcite, are 36.30 ± 0.04 and 35.94 ± 0.27 Ma, respectively (Table 1). The biotite date is distinctly older than the sanidine dates on high-SiO2 rhyolites, whereas the zircon date is analytically indistinguishable. Two of the three low-SiO2 rhyolite zircon dates (including 36.15 ± 0.35 and 35.75 ± 0.40 Ma on DC281-2460 and DC289-2224, respectively; Table 1; Fig. 10) overlap with the sanidine dates from high-SiO2 rhyolites, and the date on DC281-2460 is only slightly outside analytical uncertainty. About two-thirds of zircon grains from DC281-2460 cluster between 36 and 35 Ma. A distinct tail of older grains extends to 38.80 ± 0.94 Ma. The entire population of 30 analyses has a reasonable mean squared weighted deviation (MSWD; App. Text A1), so no analyses were eliminated. However, eliminating even the three oldest individual dates, which are outside 2σ uncertainty of 36.15 ± 0.35 Ma, brings the apparent age down to 35.98 ± 0.27 Ma.

The evidence for recoil and/or excess 40Ar in biotite from DC215-1938 and documented anomalously old biotite 40Ar/39Ar dates (Smith et al., 2003; Lipman and McIntosh, 2008; Bachmann et al., 2010; Hall, 2013) indicates the 36.30 ± 0.04 Ma biotite date on H09-1 does not record emplacement time. Dates on biotite from Oligocene volcanic rocks in the San Juan Mountains, Colorado, are commonly several hundred thousand years older than dates on coexisting sanidines (Lipman and McIntosh, 2008). Based on all these observations, we conclude the following: (1) neither biotite date records emplacement, (2) zircon dates are erratic and do not always record emplacement, and (3) emplacement of the low-SiO2 rhyolites most likely was coeval with the high-SiO2 rhyolites at 35.7 Ma but could have been as early as about 36.2 to 36.1 Ma. Apparent zonation between low-SiO2 border phase and high-SiO2 interior in one dike in DC189 supports coeval emplacement.

Trace element patterns in zircon are also distinctive (Fig. 11). The three low-SiO2 samples occupy a separate field at higher EuN/EuN* (chondrite-normalized [Sun and McDonough, 1989] europium anomaly where EuN* = (SmN× GdN)1/2) and lower Hf than the two high-SiO2 samples. EuN/EuN* decreases slightly with increasing Hf in all samples consistent with coeval crystallization of zircon and plagioclase/sanidine. The nonoverlap between the two suites suggests the high-SiO2 rhyolites are not direct differentiates of the low-SiO2 rhyolites, consistent with the nonlinearity in whole-rock Eu (Fig. 8F).

Most Cortez rhyolites are altered, as indicated by replacement of plagioclase by kaolinite and calcite and of biotite by smectite, calcite, and marcasite (Fig. 12). Half of dated samples are distinctly mineralized with introduction of sulfide minerals and modest enrichment in Carlin-type elements Au, As, Sb, Hg, and Tl, as well as enrichment of Ag, Bi, Cu, and Mo in some CTDs (Fig. 13A, B). Three types of alteration affected the dikes: glass hydration, alteration of glass to smectite, and Carlin-type alteration (Table 2). The temperatures of the first two and their relationship to Carlin-type alteration are unresolved.

Immediately following emplacement, the Cortez rhyolites consisted of unaltered phenocrysts of plagioclase, sanidine, biotite, quartz, and rare hornblende in a predominantly crystalline quartz-K-feldspar-plagioclase matrix (Fig. 6). The matrix of thin dikes and margins of thicker dikes commonly quenched to glass (Fig. 5A-C). Glass hydration, probably soon after emplacement, had no effect on phenocrysts (Table 2). All phenocrysts including hornblende in high-SiO2 rhyolite—hydrated vitrophyre sample DC215-1938 from a dike that cuts and postdates the Cortez breccia deposit (Fig. 4)—are preserved, and no alteration minerals are present (Fig. 6A, B). However, DC215-1938 has 7.2 ppm Tl (the highest of any sample) and anomalous As (46 ppm; App. Table A2). Given the lack of any mineralogic evidence for hydrothermal alteration, we interpret the high Tl to result from postmineralization mobilization from surrounding ore. High concentrations of Au, As, Sb, Hg, Tl, and S in clay-filled fractures in middle Miocene sedimentary deposits above the Carlin trend demonstrate this process (Cluer, 2012). A least altered lowSiO2 rhyolite (H09-1) dike from outcrop has plagioclase only weakly altered to calcite and unaltered biotite (Fig. 6C, D). Hydration does appear to have variably removed Na2O. Whether hydration was related to Carlin-type alteration is unknown. Hydration of vitrophyres by low-temperature groundwater has long been known to induce substantial Na2O loss commonly accompanied by K2O gain (Jezek and Noble, 1978; Ellis et al., 2022), although hydration by epithermal fluids could also remove Na2O.

Vitrophyric margins of dikes commonly are altered to smectite. Describing dikes at the Cortez mine, Radtke et al. (1987, p. 321) stated, “Margins (of dikes) … have undergone argillic alteration and are weakly mineralized.” Smectite alteration of formerly glassy margins of syn-ore Deep Star rhyolite dikes is interpreted to have occurred late during the ore-forming stage (Heitt et al., 2003). Hydrothermal fluids may have altered vitrophyres to smectite at Cortez, although low-temperature diagenesis can also alter rhyolitic glass to smectite (Christidis and Huff, 2009).

Carlin-type alteration is marked by variable phenocryst and matrix alteration, removal of many elements, introduction of several sulfide minerals, and introduction of characteristic Carlin-type trace elements (Fig. 13; Table 2). Alteration consists of replacement of plagioclase in phenocrysts and matrix of both high-and low-SiO2 rhyolites by kaolinite and calcite and of biotite by smectite, calcite, and marcasite (Fig. 12). The presence of marcasite in former biotite phenocrysts suggests sulfidation of Fe. Plagioclase phenocrysts are more commonly altered than biotite phenocrysts; plagioclase is altered in all samples in which biotite is altered, but several samples with altered plagioclase have fresh biotite (Fig. 12A, B, E, F). Sanidine is unaltered except in sample DC189-2255, which is entirely altered to quartz and kaolinite (Fig. 5H).

Kaolinite in both matrix and altered plagioclase phenocrysts is mostly fine grained (≤10 µm). However, former plagioclase phenocrysts in low-SiO2 sample DC281-2465 are divided into domains of coarse (20–50 µm) and fine kaolinite (Fig. 12D). Calcite in plagioclase phenocrysts is nearly 100% Ca with less than 1 at. % Mg or Fe. Calcite in matrix is variably enriched in Mg and Fe to as much as 12 and 7 at. %, respectively.

SiO2 contents of altered high-and low-SiO2 rhyolites vary greatly and probably primarily reflect loss of other major oxides (Fig. 7; App. Table A2). Minor mobility of SiO2 is shown petrographically by growth-banded botryoidal quartz filling cavities, disseminated quartz in matrix, and fibrous quartz (initially chalcedony?) overgrowths on quartz phenocrysts.

Sulfide mineralogy

Identified sulfide minerals are marcasite, pyrite, arsenopyrite, arsenian pyrite (with Sb), minor pyrrhotite, and late pyrite and realgar along fractures (Figs. 4, 5G, 14; App. Table A3). In our most thoroughly investigated sample, CHUD131-265 (21 ppb Au, the second highest Au content of analyzed dikes, and 1,440 ppm As, the highest value), marcasite forms porous cores surrounded by pyrite (Fig. 14A, B). Arsenopyrite forms abundant, fine-grained overgrowths on marcasite and locally on pyrite (Fig. 14B, C, E). Arsenian pyrite occurs in two forms that overgrew pyrite and arsenopyrite in sample CHUD131-265. Elongate and zoned fingers (~1 × 10 µm) overgrew pyrite crystals with porous marcasite cores (Fig. 14B, C). Botryoidal arsenian pyrite as much as ~30 µm in diameter overgrew (replaced?) pyrite in an area with nearby bladed marcasite (Fig. 14D-F). We have not found arsenian pyrite in contact with marcasite, but because arsenian pyrite is younger than pyrite, which is younger than marcasite, arsenian pyrite is later. Pyrrhotite forms a few 10- to 15-µm inclusions in pyrite associated with the porous marcasite (Fig. 14A).

Sample CHUE017-843 (Fig. 4; 9 ppb Au, 1,035 ppm As), the first core sample that showed obvious mineralization of pyrite-realgar– coated fractures, has similar porous marcasite overgrown by pyrite and arsenopyrite. Only a few small (<1–5 µm) grains of arsenian pyrite were found. These minerals overgrew unaltered matrix K-feldspar. Late, coarse-grained pyrite and realgar coat fractures in sample CHUE017-843 but were not found in the matrix (Fig. 5G). Arsenian pyrite also was found in DC189-2097, which also has arsenopyrite, and in DC281-2465.

The arsenian pyrite fingers are zoned from As- and Sb-rich cores (as much as 7.7 wt % As and 1.1 wt % Sb) to As-poor rims (0.4 wt % As) with undetectable Sb. Antimony correlates positively with As. All analyses of botryoidal arsenian pyrite have high As (5.5–7.7 wt %) and higher Sb (1.6–3.1 wt %) than in the finger-type arsenian pyrite.

Effect of alteration on major oxides

Plot of the molar ratios (2Ca + Na + K)/Al versus K/Al demonstrate compositional changes consistent with mineralogic alteration (Fig. 15; Warren et al., 2007). The least altered samples of Cortez rhyolites and Caetano Tuff overlap with the compositions of unaltered rhyolites of the 25 Ma Elevenmile Canyon caldera (Colgan et al., 2017). All suites illustrate the moderately alkalic character of Great Basin rollback rocks. The least altered Cortez rhyolites have lower (2Ca + Na + K)/Al values than most Caetano Tuff samples because of lower Na2O contents (≤3 wt % vs. ~4 wt % in Caetano Tuff). We interpret the low Na2O contents in the least altered Cortez rhyolites and some Caetano Tuff samples to result from processes other than Carlin-type alteration, including hydration of vitrophyres, because Caetano Tuff did not undergo Carlin-type alteration (John et al., 2011). Cortez rhyolite samples in which only plagioclase is altered overlap with samples in which both plagioclase and biotite are altered. Replacement of plagioclase by kaolinite reduced Na2O contents to ~1 to 1.5 wt %. Kaolinitization also removes Ca, but calcite deposition apparently led to an increase in CaO in some samples (Fig. 13A). Destruction of biotite released K, but the total biotite content of the rocks was probably insufficient to measurably affect the K/Al ratio. Most mineralized Cortez rhyolite samples plot with the plagioclase- and plagioclase-biotite–altered rocks, because plagioclase ± biotite were also altered in the mineralized rocks. Six Cortez samples plot at lower (2Ca + Na + K)/Al and K/Al. Three samples have (2Ca + Na + K)/Al values of about 0.5 and K/Al between 0.18 and 0.29. These samples have particularly low Na2O (0.34–0.72 wt %). Two of these samples have <3 wt % K2O, which suggests K2O loss, although both have unaltered sanidine; alteration likely removed K2O from the matrix. Three samples plot near the kaolinite composition because of their near 0 wt % CaO, Na2O, and K2O contents. Sample DC189-2255, which is nearly completely altered to quartz and kaolinite, has the highest Au content (37 ppb) of the analyzed dikes (App. Table A2).

Alteration other than Carlin-type mineralization has affected alkalis (Fig. 15). An unaltered Cortez rhyolite and one Caetano Tuff sample plot at high K/Al (>0.6) with three Cortez rhyolites—two mineralized and one with altered plagioclase and fresh biotite. All these samples have >7 wt % K2O and 1.2 to 2.2 wt % Na2O, likely resulting from glass hydration in the unaltered samples. Even the mineralized samples may have undergone glass hydration before or during the earliest part of Carlin-type alteration. Caetano Tuff in the western part of the caldera has undergone intermediate argillic (plagioclase replaced by kaolinite or smectite; sanidine unaltered) and advanced argillic (both plagioclase and sanidine replaced by kaolinite; minor alunite) alteration (Fig. 2; John et al., 2011). These Caetano samples are also nearly totally depleted in CaO, Na2O, and K2O and plot at or near the kaolinite composition (Fig. 15).

Geochemical enrichment

Log isocon diagrams of representative mineralized high-and low-SiO2 rhyolites show that they are modestly enriched in Au, Hg, Ag, Bi, Sb, Cu, Mo, Tl, As, and S and variably depleted in MgO, CaO, Na2O, K2O, MnO, Rb, Sr, and Ba (Fig. 13). Ore-stage pyrite at Cortez Hills is also enriched in these elements, except Mo (Maroun et al., 2017). Lack of published data on ore composition hinders comparison with the rhyolites, but available data indicate the rhyolites are much less enriched in Au, As, and Hg compared to ore. The highest analyzed Au content of our Cortez rhyolite samples is 37 ppb. Bradley et al. (2020, p. 347) report “arsenic values routinely exceeding 1,000 ppm and commonly >1 wt %, and mercury values from 100 to >1,000 ppm” in ore. In contrast, maximum and average contents of mineralized Cortez rhyolites are 1,440 and 217 ± 406 ppm (As; all 1s) and 9.4 and 0.8 ± 2.2 ppm (Hg) (App. Table A2). Maximum and average values of other Carlin-type elements are 0.52 and 0.16 ± 0.13 ppm (Ag), 71.5 and 18.1 ± 19.6 ppm (Sb), 0.50 and 0.12 ± 0.14 (Bi), 5.3 and 1.5 ± 1.2 ppm (Tl), and 7.8 and 2.3 ± 2.3 ppm (Cu). All of these are low compared to Carlin-type ore in general (Radtke, 1985; Cline et al., 2005).

Cortez rhyolites underwent Carlin-type alteration but are not ore

That many Cortez rhyolite dikes have undergone the same Carlintype alteration as the Paleozoic host rocks at Cortez Hills is demonstrated by similar alteration mineralogy, especially sulfide mineralogy including arsenian pyrite, and characteristic Carlin-type geochemical enrichment and depletion (Figs. 1214). Nevertheless, none of the Cortez dikes analyzed by us constitute ore; the highest Au content is 37 ppb (0.001 oz/ton; App. Table A2).

The presence of kaolinite and marcasite in Cortez rhyolites indicates moderately acidic, moderate-temperature fluids (pH ~3–3.5 [<~5 for marcasite], T <240°C; Murowchick and Barnes, 1986; Murowchick, 1992; Hofstra and Cline, 2000), similar to Carlin-type fluids in general. The change from biotite altered to smectite to unaltered biotite suggests the acidic fluid increased in pH as it migrated through and reacted with the quartzofeldspathic dikes. The progression from pyrrhotite to marcasite to pyrite to arsenopyrite + pyrite to arsenian pyrite indicates increasing sulfur fugacity and/or cooling, variable arsenic fugacity, and possibly increasing pH. Late pyrite + realgar fracture coatings probably reflect increasing sulfur and oxygen fugacity and/or cooling (Hofstra and Cline, 2000, their fig. 16).

Although kaolinite alteration is common in Carlin trend deposits (Kuehn and Rose, 1992; Hofstra and Cline, 2000; Ressel et al., 2000a, b; Heitt et al., 2003) and at Getchell (Cassinerio and Muntean, 2011), kaolinite is minor in the Paleozoic silty carbonate host rocks at Cortez Hills, and whether it formed as part of the Carlin hydrothermal system is unclear (Maroun et al., 2017). Kaolinite replaces feldspars in Paleozoic host rocks at Pipeline, and Cortez Hills host rocks have undergone argillization (Bradley et al., 2020), but it has not otherwise been reported as a part of Carlin-type deposits of the Cortez region. As pointed out by Hofstra and Cline (2000), argillization is prominent in igneous, metamorphic, and siliciclastic rocks but much less so in carbonate rocks, which likely explains the presence of kaolinite in Cortez rhyolites and its paucity or absence in the Paleozoic carbonate rocks. Maroun et al. (2017) summarize kaolinite occurrence in deposits of the Carlin and Getchell trends and the disparate interpretations that it was or was not part of Carlintype systems. They conclude that several factors, especially changes in temperature or pH, influence whether kaolinite is present. Low Al content of the carbonates is undoubtedly also important. We also suggest that uncertainty about when kaolinite formed during the long diagenetic and alteration history of Lower Paleozoic host rocks, including supergene oxidation, complicates interpretation (e.g., Folger et al., 1996, 1998; Cline et al., 2005).

Both Cortez rhyolites and the Paleozoic carbonate host rocks of the Cortez Hills deposit (as well as Carlin-type deposits in general) contain marcasite, pyrite, and arsenian pyrite (Cline et al., 2005; Maroun et al., 2017; Cline, 2018; Muntean and Cline, 2018). The rhyolites are distinctive in having abundant arsenopyrite, which is present and contained significant Au in Paleozoic host rocks in the adjacent Cortez mine (Wells and Mullens, 1973) and present in the Getchell trend (A.A. Longo, writ. commun., 2021). However, it has not been reported at Cortez Hills (Maroun et al., 2017) and is absent or uncommon in most Carlin-type deposits (e.g., Radtke, 1985). Arsenopyrite is considered more characteristic of higher-temperature, distal disseminated (“Carlin-style”) deposits (Cline et al., 2005). Arsenopyrite, especially as overgrowths on pyrite, is present in the few Carlin-type deposits that are predominantly or partly hosted by igneous or other noncarbonate host rocks. Notably, the Beast deposit on the Carlin trend is entirely hosted by synmineralization 37.6 Ma rhyolite dikes and breccias in which arsenopyrite is common (Ressel et al., 2000a; Johnson et al., 2015). Arsenopyrite is also present in mineralized 39 Ma dacite dikes at Meikle (Ressel et al., 2000b), mostly in igneous host rocks at Goldstrike (de Almeida et al., 2010), in exo- and endoskarn around the Jurassic Goldstrike intrusion at Deep Star (Fleet and Mumin, 1997), in the telescoped polymetallic vein-porphyry-style, distal disseminated Carlin-type Cove deposit (Johnston et al., 2008), and in the predominantly quartzite host rocks at Marigold (Fithian et al., 2018), although Marigold was previously classified as a distal disseminated deposit associated with a porphyry system (Cline et al., 2005). These occurrences suggest arsenopyrite is characteristic not only of higher-temperature distal disseminated deposits but also of lower-temperature Carlin-type deposits in noncarbonate host rocks.

Enrichment in Carlin-type trace elements, although minor, further demonstrates that the Cortez rhyolites were altered by the same hydrothermal system as the Cortez Hills deposit. Cortez rhyolites show the same element enrichment and depletion as in the 37.6 Ma rhyolite ore host at the Beast deposit (Ressel et al., 2000a; the Cortez Hills CTD has no published trace element data to compare). Lack of major enrichment in Cortez rhyolites probably reflects their low permeability—especially if Carlin-type alteration postdates smectite-alteration of marginal vitrophyres (Fig. 5A-C)—and the low reactivity of the quartzofeldspathic rhyolites. Low FeO content could also be a factor (<1 wt % FeO in high-SiO2 rhyolites; App. Table A2), although the unmineralized Wenban Limestone and Roberts Mountains Formation have equally low Fe contents (Wells et al., 1969).

Both Cortez rhyolites and the lower Paleozoic silty carbonates that are the primary host rock in most Nevada Carlin-type deposits including Cortez Hills have undergone the same Carlin-type alteration. However, the Paleozoic rocks have also undergone burial diagenesis and Jurassic, Cretaceous, and two Eocene episodes of intrusion and hydrothermal alteration (Maroun et al., 2017; Henry et al., 2020). These events introduced or altered minerals and added or depleted chemical constituents that are unrelated to the Carlin-type system; determining what is related to the Carlin system can be difficult. Cortez rhyolites provide a simpler system to evaluate the effects of Carlin-type alteration (Table 2).

The Cortez Hills deposit formed at 35.7 Ma

Individual Cortez rhyolite dikes vary from highly altered and mineralized to unaltered, which indicates their emplacement spanned the time of Carlin-type mineralization at Cortez Hills. Weighted mean dates of unaltered samples are indistinguishable from those of mineralized and variably altered samples (Table 1). The mineralogic and geochemical characteristics of mineralized dikes indicate that they underwent the full CTD hydrothermal system, not simply the late stages as initially interpreted based on the presence of pyrite-realgar veinlets in a few dikes (Arbonies et al., 2011; Jackson et al., 2011; Maroun et al., 2017; Muntean, 2020; Mercer, 2021). All phenocryst phases, including the only identified hornblende, are unaltered in vitrophyre sample DC215-1938 (35.64 ± 0.05 Ma; Table 1), despite it intruding high-grade ore in the breccia zone; its intrusion postdates ore. The lack of alteration of DC215-1938 vitrophyre contrasts sharply with intense quartz-kaolinite alteration of former vitrophyre DC189-2255 (Fig. 5H). Premineralization intrusion of DC215-1938 into the high-grade ore zone without mineralogic signs of alteration seems extremely unlikely.

The 40Ar/39Ar dates demonstrate the Cortez Hills hydrothermal system was active ca. 35.7 Ma. These dates only constrain the duration of mineralization as brief.

The Carlin-type hydrothermal system at Cortez Hills may have lasted no more than 45 ka. Hickey et al. (2014b) interpreted apatite fission-track data to show that hydrothermal flow at the giant Betze-Post CTD on the Carlin trend lasted 15 ka to no more than 45 ka, and Maroun et al. (2017) suggested this duration is applicable to Cortez Hills. The brief duration of Cortez Hills mineralization indicated by 40Ar/39Ar dates is consistent with this interpretation but does not require it. Mineralization predates the 34.0 Ma Caetano caldera, because caldera megabreccia contains blocks of mineralized upper plate sedimentary rocks and jasperoid with barite rosettes. These blocks were probably shed from the top of the Cortez Hills deposit during caldera collapse (Pixie, Fig. 3; John et al., 2008).

A similar 35.7 Ma age of mineralization at the Goldrush-Fourmile deposit is also likely. The only igneous rocks in the vicinity of Goldrush are 35.7 Ma Cortez rhyolites, subsurface Cretaceous granodiorite, and the Jurassic Mill Canyon granodiorite and lamprophyre dikes. Mineralization is presumably related to one of those events. Eocene limestone that underlies 34.0 Ma Caetano Tuff ~400 m east of Cortez rhyolite in Horse Canyon (Fig. 3) could be spring deposits formed by CO2 released during decalcification of Paleozoic limestone by the Cortez Hills or Goldrush hydrothermal systems. The Eocene limestone locally has gossanous hematitic layers enriched in As, Ba, Hg, Mo, Mn, Sb, and W (App. Table A2).

The Gold Acres and Pipeline deposits: Cretaceous hydrothermal activity but Eocene CTDs?

The giant Pipeline and nearby Gold Acres deposits are adjacent to the belt of Cortez rhyolites (Fig. 2), so also could have formed at 35.7 Ma. However, as we show here, these deposits remain undated. Despite extensive oxidation that destroyed primary characteristics, both deposits are interpreted to be CTDs—based on geochemical association (high Au/Ag [>50 at Pipeline], As, Hg, Sb, Tl, and W) and the appropriate lower plate, silty carbonate host rocks that were decalcified, argillized, and silicified—and to be Eocene (Rytuba, 1985; Hays and Foo, 1991; Foo et al., 1996; Blamey and Norman, 2000; Cline et al., 2005). However, the only dated hydrothermal activity is Cretaceous.

Mineralization at Gold Acres is divided into three spatially superimposed episodes from oldest to youngest—Mo-W–bearing skarn, Cu-Pb-Zn-Ag-As–bearing skarn, and Carlin-type (Wrucke et al., 1968; Wrucke and Armbrustmacher, 1975; Armbrustmacher and Wrucke, 1978; Wrucke, 1985; Hays and Foo, 1991). The skarns are interpreted to be Cretaceous because they are in upper and lower plate rocks adjacent to the Gold Acres intrusion (Fig. 2; Wrucke and Armbrustmacher, 1975). Potassium-argon dates on sericite from drill core in the intrusion (95.9 ± 1.9 Ma) and from a strongly altered dike in the Gold Acres deposit (96.6 ± 1.9 Ma) (Silberman and McKee, 1971) are consistent with a Cretaceous age but are significantly younger than the dates on the intrusion (104.8 ± 0.2 Ma, U-Pb zircon, Mortensen et al., 2000; 101.2 ± 2.0 Ma, K-Ar biotite, Silberman and McKee, 1971). The younger K-Ar dates could indicate partial Ar loss from reheating or hydrothermal activity ~8 to 10 Ma after the intrusion.

In the Gold Acres CTD, gold occurs with iron oxides after pyrite in a shear zone breccia in lower plate Roberts Mountains Formation immediately below the Roberts Mountains thrust (Gilluly and Gates, 1965; Wrucke et al., 1968; Wrucke and Armbrustmacher, 1975; Hays and Foo, 1991). Wrucke and Armbrustmacher (1975) interpreted CTD mineralization to have occurred during the waning stages of the Cretaceous hydrothermal system. However, Wrucke (1985) stated that argillized dikes at Gold Acres are similar to those at Cortez and suggested Gold Acres might be the same age as Cortez. In contrast, we now recognize that the argillized dikes are not the same as the Cortez rhyolite dikes.

The Pipeline deposit is hosted by lower plate carbonate rocks in the crest of an anticline (Leonardson, 2011). The lack of igneous rocks at Pipeline precludes even relative dating. However, Blamey and Norman (2000) and Blamey et al. (2017) investigated quartz-sericite-pyrite veins containing macroscopic gold that underlie Pipeline and interpreted them to be feeders to the deposit. Fluid inclusion microthermometry and gas analysis indicate vein formation at 300°C and 2 kb lithostatic pressure or ~8-km depth. A disturbed sericite 40Ar/39Ar spectrum declined slightly from ~110 to 90 Ma over the first 85% of 39Ar release, then dropped sharply to almost 0 Ma. Although Blamey et al. (2017) favored a total gas age of ~93 Ma age for vein formation, the data allow an age indistinguishable from that of the 105 Ma Gold Acres intrusion. Four samples of mixed diagenetic and epigenetic illite from wall rock showed gently to steeply climbing spectra, with the flattest parts between 110 and 100 Ma, probably reflecting degassing from a mix of diagenetic and epigenetic illite and 39Ar recoil and again allowing an ~105 Ma age. Blamey et al. (2017) interpreted true ages of epigenetic illite formation to be ~5 to 10 Ma younger than total gas ages of 99.6, 100.3, 106.6, 132.9 Ma because of recoil. These data allow formation of epigenetic illite at the same time as the Gold Acres intrusion. Quartz-sericite-pyrite veins with visible gold are unlike Carlin-type mineralization, and the temperature and depths are greater than for Carlin-type deposits, so these dates only demonstrate Cretaceous hydrothermal activity, not that the Pipeline CTD is Cretaceous.

Retention of mostly Cretaceous ages of sericite and illite does not preclude a younger, possibly Eocene age of hydrothermal activity to generate the Gold Acres and Pipeline CTDs. The low temperatures of Carlin-type systems (≤240°C) are insufficient to induce measurable radiogenic Ar loss from these minerals even over 100,000-year time spans (Folger et al., 1996, 1998; Hofstra et al., 1999; Hall et al., 2000).

Based on apatite fission-track (AFT) dating, Arehart and Donelick (2006) interpreted a 38.7 ± 2.0 Ma age of the Pipeline deposit, which we disregard for the following reasons. This age is a weighted mean of seven of 32 dates “within or near the Pipeline system [that] yield model AFT ages near 40 Ma” (Arehart and Donelick, 2006, p. 32) and that apparently are within ~5 km of the deposit. However, 25 other fission-track dates in the same area range from 12.8 ± 4.7 to 66.9 ± 4.1 Ma. Arehart and Donelick attribute ages younger than ~40 Ma to variable resetting by middle Miocene activity of the northern Nevada rift, which lies ~10 km to the east of Pipeline (Fig. 2). However, the overall ages show no spatial relationship to the deposit or the northern Nevada rift.

Eocene overprinting of Cretaceous mineralization around Gold Acres and Pipeline is possible but cannot be demonstrated from any data, and the age of these deposits is undetermined. The nearest known Eocene intrusions are the ~39 Ma granodiorite and 35.8 Ma Cortez-type rhyolites at Tenabo, 6 km north of Gold Acres (Fig. 2).

The relationship of CTDs to contemporaneous magmatism and extension

Based on our dating of mineralized to unaltered 35.7 Ma Cortez rhyolite dikes and interpretations of brief ≤45 ka duration of Carlintype systems (Hickey et al., 2014b; Maroun et al., 2017), the Cortez Hills and Cortez deposits formed at 35.7 Ma, contemporaneous with the major pulse of rhyolitic intrusion. Cortez Hills and Cortez are probably the most tightly temporally constrained CTDs. We favor a similar age for the recently discovered Goldrush and Fourmile deposits, given that they lie only 6 km from Cortez Hills in an area with 35.7 Ma dikes, although Miocene, Cretaceous, and Jurassic intrusions are also nearby. Pipeline may never be precisely dated, but it lies next to the belt of Cortez rhyolite dikes (Fig. 2), which suggests it is also within a thermal aureole of the underlying intrusion that fed the dikes.

Here we discuss (1) to what extent Jurassic, Cretaceous, or older Eocene mineralization contributed to CTDs of the Cortez region, (2) the tie of Cortez rhyolites to an underlying larger pluton, (3) the possible relationship to contemporaneous extension, (4) depth of formation of CTDs, and (5) implications of regional age patterns of magmatism and CTDs.

Did older mineralization contribute to ~35.7 Ma CTDs? Are CTDs of the Cortez region the products of single, probably magmatically driven Eocene hydrothermal systems or of multiple systems with significant gold either deposited during or recycled from Cretaceous, Jurassic, or Devonian mineralization or Paleozoic black shales? In northern Nevada, models for composite gold deposition and recycling have been best developed in the Getchell and Carlin trends, respectively. Groff et al. (1997, p. 601) interpreted that “gold mineralization occurred in five stages between 95 and 42 Ma,” and the Getchell and Twin Creeks deposits are composites of 83 and 42 Ma mineralization. Emsbo et al. (1999, 2003, 2006) interpreted major sedimentary exhalative (sedex) gold deposition in the Devonian sedimentary rocks that host CTDs in the Carlin trend and recycling of sedex gold by heated meteoric water during the Eocene to form the CTDs in the trend. Large et al. (2011) proposed that Carlin-type and orogenic gold deposits generally recycle gold from black shales.

The spatial superposition of Jurassic, Cretaceous, and ~40 Ma Eocene intrusions and Au-bearing but mostly Ag-base metal-rich deposits in the Cortez region is permissive evidence for a contribution of gold from these older deposits. Cortez Hills and Goldrush are adjacent to the Jurassic Mill Canyon and related intrusions, to subsurface Cretaceous granodiorite intrusions, and to dated Jurassic and Cretaceous hydrothermal alteration (Gilluly and Masursky, 1965; Arbonies et al., 2011; Jackson et al., 2011; Bradley and Eck, 2015; Henry et al., 2020). Jurassic lamprophyre dikes are common in the Goldrush orebody. Pipeline and Gold Acres are adjacent to the Cretaceous Gold Acres intrusion and overlap with dated Cretaceous alteration (Hays and Foo, 1991; Foo et al., 1996; Leonardson, 2011; Blamey et al., 2017).

Known Jurassic and Cretaceous deposits in the Cortez region are small, Ag-base metal rich, and Au poor (Mill Canyon and Cortez Ag district; Fig. 3; Barton et al., 2011; Henry et al., 2020). Such deposits could not have contributed significant amounts of gold to giant deposits like Cortez Hills or Goldrush, despite their spatial overlap. The most Au-rich older deposits are associated with the ~40 Ma episode of intrusive-hydrothermal activity near Hilltop and Tenabo (Figs. 1, 2).

Despite the spatial associations, no evidence yet exists that Jurassic or Cretaceous hydrothermal activity contributed to the endowments at Cortez Hills and Goldrush. Pipeline (-Gold Acres) is the only Cortez area CTD for which a Cretaceous contribution has been interpreted, and Cretaceous hydrothermal activity that generated Au-bearing quartz-sericite-pyrite veins below Pipeline is well documented (Blamey et al., 2017). However, these veins are distinctly unlike veins in CTDs, and pervasive oxidation of Pipeline precludes determining its primary characteristics or evaluating almost any genetic question.

A contribution from Devonian sedex mineralization also cannot be demonstrated. Lower plate rocks that host sedex mineralization in the Carlin trend (Emsbo et al., 2003) are present in the Cortez region, but no sedex mineralization has been identified here. Sedimentary barite deposits are abundant in upper plate rocks in the Cortez area and Carlin trend (Papke, 1984; Cline et al., 2005; Fig. 2); however, these deposits in the Cortez region lack gold.

The superposition of multiple ages and types of igneous and hydrothermal systems guarantees difficulty in assessing older contributions to Eocene deposits. Carlin-type hydrothermal fluids are too low temperature (180°–240°C) to reset older illite, muscovite, or other hydrothermal minerals, which means that they will mostly retain their original K-Ar or 40Ar/39Ar age (Folger et al., 1996, 1998).

Ties of Cortez rhyolites to a large, deep granitic(?) pluton: The many Cortez rhyolite dikes require a large felsic intrusion at depth to feed the dikes, as do similar dikes in the Carlin trend (Ressel and Henry, 2006). This larger intrusion, not the small-volume dikes, provided the heat to drive Carlin-type mineralization, despite their abundance. Silicic magmas represented by the dikes require large magma chambers to develop, either by differentiation of mantlederived, basaltic magma, by crustal melting or, more commonly, by a combination of crustal melting-assimilation and differentiation (Hildreth, 1981; Barnes et al., 2001; Watts et al., 2016b). The distribution of Cortez rhyolite dikes and their attitudes in the subsurface suggest the dikes arose from what is now a solidified pluton centered beneath Cortez Hills with a north-northwest to south-southeast elongation that would allow dikes to reach Tenabo and the northern Simpson Park Mountains (Fig. 2). Although no dikes are present at Pipeline and only Jurassic dikes have so far been found at Goldrush, the extent of the pluton indicated by Cortez-type rhyolite dikes suggests that both deposits could have formed from large hydrothermal systems related to that pluton.

We suggest the top of this pluton is at a depth of 6 to 10 km based on the evidence of detailed, upward-continued aeromagnetic data for plutons at that depth in the Carlin trend (Ressel and Henry, 2006), a preferred model of CTD generation (Muntean et al., 2011), and by comparison with the >5- to 7-km deep plutons beneath modern Italian geothermal systems that are interpreted to be analogues of Carlin-type systems (Sillitoe and Brogi, 2021). More directly, volatile contents of melt inclusions from four high-SiO2 Cortez rhyolite dikes suggest trapping at depths as great as 9 km and are interpreted to reflect the presence of a pluton source for the rhyolites at ~4- to ≥9-km depth (Mercer, 2021). The superposition of Jurassic, Cretaceous, Eocene, and middle Miocene rocks (especially the strongly magnetic northern Nevada rift) may preclude using magnetic data to detect an underlying Eocene pluton at Cortez. Either granite (>70 wt % SiO2), which is compositionally like the Cortez rhyolite dikes, or granodiorite (65–70 wt % SiO2), which is like the Jurassic, Cretaceous, and ~40 to 39 Ma Eocene intrusions of the region, is possible. Either composition could generate rhyolitic dikes as direct offshoots of a granitic magma or as differentiates of a granodiorite.

Unexposed plutons are common in northern Nevada: A major enigma for the association of CTDs with Eocene magmatism is the absence of Eocene rocks of any kind in the Getchell trend, despite some of the strongest geochronological evidence that the Getchell and Twin Creeks CTDs are Eocene (Fig. 16; Groff et al., 1997; Hall et al., 2000). Data from the Cortez area and from other areas of Eocene intrusions or CTDs provide a potential resolution. Drill holes, zircon xenocrysts, and 40Ar/39Ar dates and thermochronology of alteration minerals indicate at least seven major Cretaceous intrusions in the subsurface of the Cortez region despite outcrop consisting of a single dacite dike (Fig. 3; Henry et al., 2020). Drill holes demonstrate that three plutons are within 1 km of the present surface. Zircon xenocrysts in Eocene or Cretaceous dikes demonstrate the presence of unexposed Jurassic and/or Cretaceous plutons at Marigold (Fithian et al., 2018), at Lone Tree (Holley et al., 2019), and in east-central Nevada (Gottlieb and Miller, 2013). No other areas have had extensive U-Pb dating of xenocrysts, but these data suggest that unexposed plutons are common.

Basic geologic data further support the likelihood of subsurface Eocene intrusions. In most areas of CTDs, the only Eocene exposures are felsic dikes, accompanied by coeval felsic volcanic rocks in some: Cortez Hills, Jerritt Canyon, the northern Carlin trend, Lone Tree, Long Canyon in the Pequop Mountains, Eureka, and Northumberland (Fig. 16). The presence of felsic dikes and locally derived volcanic rocks requires underlying source plutons despite their nonexposure. This is best demonstrated by the aeromagnetic evidence for deep, 6- to 8-km plutons in the Carlin trend (Ressel and Henry, 2006). In the Pequop Mountains, ~40 Ma rhyolite dikes only intrude Cambrian rocks at stratigraphic depths 1 to 1.5 km below the Long Canyon deposit, which has abundant Jurassic dikes but no known Eocene intrusions (Bedell et al., 2010; Milliard et al., 2015; Zuza et al., 2018, 2019). The dikes are only exposed and their existence known because the Pequop Mountains were tilted ~40° and deeply eroded post-Long Canyon mineralization. Similarly, the composite ~37 Ma Harrison Pass pluton and related dikes in the Ruby Mountains are only exposed because of much later extension, tilting of about 50°, and subsequent erosion (Fig. 1; Burton, 1997; Barnes et al., 2001; Colgan et al., 2010). The top of the pluton reached Ordovician strata (Burton, 1997). Dikes propagated 1 km higher into Silurian strata at paleodepths of ~4 km. The Getchell trend has undergone negligible Cenozoic extension and tilting, so paleodepths as seen in the Pequop Mountains and at Harrison Pass are not exposed. We suggest the geologic evidence discussed here supports the presence of Eocene intrusions beneath the Getchell trend.

Is coeval normal faulting required? Extension (normal faulting) is commonly interpreted to be important in enhancing permeability for CTDs (Hofstra et al., 1999; Cline et al., 2005; Emsbo et al., 2006; Muntean et al., 2011; Rhys et al., 2015). The history of Cenozoic extension in the northern Great Basin is contentious (summarized in Henry et al., 2011; see also Best and Christiansen, 1991). Major extension that began ~17 to 16 Ma throughout the northern Basin and Range province is widely accepted (cf. Colgan and Henry, 2009). Moderate Eocene extension is recognized near the Carlin trend (Fig. 1).

In the Cortez area, Eocene magmatism and hydrothermal activity are interpreted to be accompanied by regional extension, although specific faults active at the time and synmineralization faulting have not been identified (Arbonies et al., 2011; Jackson et al., 2011; Leonardson, 2011; Bradley et al., 2020). Near but outside the Cortez area, minor faults active at ~33 Ma have been mapped in the Tobin Range (west of the Fish Creek Mountains in Fig. 1; Gonsior and Dilles, 2008), and minor extension has been interpreted at about 25 Ma in the Pinon Range (Lund-Snee et al., 2016; Fig. 1). Neither time coincides with mineralization at Cortez Hills.

Based on the absence of unconformities between Eocene and younger deposits or of thick clastic sedimentary deposits older than middle Miocene, we found no measurable extension before ~16 Ma around Cortez (Colgan et al., 2008, 2014; Colgan and Henry, 2009). Within the Caetano caldera, undated but pre-35.2 Ma conglomerate, the overlying 34.0 Ma Caetano Tuff, and ash-flow tuffs as young as 25 Ma are equally east tilted. The oldest ~16 Ma Miocene sediments in the caldera, which accumulated in half-graben basins, are conformable or in slight (~10°) angular unconformity with underlying Caetano Tuff. Middle Miocene sediments that overlie the Cortez Hills deposit are as old as 15.3 Ma (our oldest tephra date there; John et al., 2008; Colgan et al., 2011) and dip as much as 23° east. The nearby intracaldera Caetano Tuff dips ~40° east. The difference in tilt probably reflects pre-15.3 Ma, middle Miocene extension and tilting.

In the northern Shoshone Range, undated but pre-35.8 Ma conglomerate and volcanic rocks as young as 30.6 Ma are equally east tilted (Colgan et al., 2014). The pre-~35 Ma conglomerates throughout the Cortez region, the oldest Cenozoic sedimentary deposits, are thin (10–20 m) fluvial sequences that accumulated in W-flowing paleovalleys (Henry, 2008). In contrast, the middle Miocene sediments are as much as 2 km thick and progress upward from coarse fluvial gravel to fine lacustrine deposits, attesting to their accumulation in extensional half grabens (Colgan et al., 2008). The 34 to 25 Ma ash-flow tuffs flowed down these paleovalleys without interruption, which precludes significant faulting during that interval.

We conclude that most CTDs did not form during major faulting. Minor faulting that did not generate angular unconformities or thick basin-filling sedimentary deposits cannot be eliminated. The Eocene in northern Nevada followed major Cretaceous contraction and preceded major middle Miocene extension (Best and Christiansen, 1991; Dickinson, 2006; Colgan and Henry, 2009; Long et al., 2014). The Eocene stress state may have been mildly tensional, enough to allow tensional fractures to form or reactivate existing faults and enhance permeability but insufficient to generate measurable extension or disrupt the dike or hydrothermal systems. Hickey et al. (2014a) concluded that hydrothermal flow at Pipeline followed lowangle Paleozoic or Mesozoic thrust faults, although minor Eocene extensional reactivation may have enhanced permeability.

Depth of CTDs and the Eocene-Paleozoic unconformity: Relationships between the Cortez Hills CTD, its Paleozoic host rocks, and Eocene volcanic and sedimentary rocks demonstrate that Cortez Hills formed at ≤1 km and provide further evidence for the shallow formation of CTDs (Henry and Ressel, 2000; Nutt and Hofstra, 2003; Cline et al., 2005; Ressel and Henry, 2006; Hickey et al., 2014a; Cline, 2018). Eocene ~35 Ma volcanic or sedimentary rocks rest directly on, or contain clasts of, lower plate Devonian Wenban Limestone near Cortez Hills (John et al., 2008; this study), locally recognized Devonian Horse Canyon Formation at Goldrush (Rodeo Creek Formation equivalent; Jackson et al., 2011), and Devonian-Silurian Roberts Mountains Formation in the northern Simpson Park Mountains (McKee and Conrad, 1994, Steininger et al., 2017; NuLegacy Gold Corporation, 2020). Therefore, the major host rocks for CTDs in the Cortez region were exposed at the time of mineralization. The upward transition of Cortez rhyolites from coherent dikes to pyroclastic breccias indicates near-surface emplacement.

The uppermost part of the Cortez Hills deposit was probably exposed when the Caetano caldera formed at 34 Ma. Abundant, large (to 50 m diam) to small pieces of caldera wall rock fell into the caldera during tuff eruption and caldera collapse and are surrounded by intracaldera tuff (megabreccia). Most of these blocks are unmineralized upper plate Paleozoic rocks, but a small gold resource at Pixie consists of megabreccia composed of mineralized upper plate rocks (Fig. 3). A few blocks of lower plate Wenban Limestone and jasperoid occur along the northeastern margin of the caldera ~2 km southwest of the Cortez Hills deposit. The blocks probably slumped from outcropping upper plate rocks above the deposit and from immediately underlying Wenban Limestone in the deposit. Caldera collapse occurred ~1.7 Ma after ~35.7 Ma mineralization, leaving only that amount of time to erode the deposit.

Correlation between Eocene magmatism and CTDs in northern Nevada: Here we evaluate the spatial and temporal relationship between CTDs and Eocene magmatism (Figs. 1, 16; Table 3). Because rollback magmatism migrated south-southwest through time, a magmatic relationship predicts that CTDs should also young to the south-southwest coincident with magmatism. While dating of igneous activity is mostly straightforward, direct dating of CTDs is difficult because associated minerals are rarely amenable to existing dating methods (Arehart et al., 1993, 2003; Groff et al., 1997; Hall et al., 2000; Ressel and Henry, 2006). Rare exceptions are galkhaite, an Hg sulfosalt present in ore at Getchell and Rodeo in the northern Carlin trend (Arehart et al., 2003), which can be dated by the Rb-Sr method, and adularia, which has been found only at Twin Creeks (Groff et al., 1997; Hall et al., 2000). All other age constraints rely on dating crosscutting igneous rocks. As was formerly the case at Cortez, whether the igneous rocks are mineralized is commonly controversial, so dikes may only provide an equivocal upper or lower age limit.

Known, probable, or possible ages of CTDs generally decrease from north to south, coincident with the timing of magmatism (Fig. 16). Well to reasonably well dated northern CTDs are the oldest: Twin Creeks at 42.5 ± 0.2 Ma and Marigold at ~41 Ma (Groff et al., 1997; Hall et al., 2000; Huff et al., 2016; Fithian et al., 2018). Lone Tree, a deposit with both distal disseminated and CTD characteristics (e.g., Ag/Au <2), is bracketed between 40.95 ± 0.06 Ma, the age of mineralized rhyolite dikes, and 40.57 ± 0.74 Ma, the age of adularia whose paragenetic relationship to gold is not established (Holley et al., 2019). Jerritt Canyon formed at ≤41.3 ± 0.1 Ma, the age of a mineralized basalt dike (Hofstra et al., 1999). Doby George, the northernmost but little-studied CTD, is either older or younger than an unmineralized 43.76 ± 0.06 Ma ash-flow tuff (H14-134; Tables 1, 3) that overlies mineralized Paleozoic siliciclastic rocks. LaPointe et al. (1991) and Anderson (2008) interpreted mineralization to be prevolcanic, which would make Doby George the oldest known CTD. In contrast, O.D. Christensen (Hardrock Mineral Exploration, Inc., oral commun., 2014) interprets that mineralization postdates the tuff, which is unaltered because hydrothermal fluids did not reach, or had little effect on, the feldspathic tuff.

Carlin-type deposits in the northern Carlin trend are older than those in the southern trend. Rodeo is 39.8 ± 0.6 Ma, an Rb-Sr date on galkhaite (Chakurian et al., 2003), and Meikle is 39.8 ± 1.6 Ma, an 40Ar/39Ar isochron date on illite (Ressel et al., 2000b). At the Beast deposit 6 km to the south, the concordance of 40Ar/39Ar dates on sanidine (37.82 ± 0.12 Ma) and biotite (37.79 ± 0.14 Ma) from an orehosting rhyolite dike (Ressel et al., 2000a) indicates that hydrothermal alteration at Beast occurred at 37.8 Ma (Johnson et al., 2015).

In the Rain-Railroad area of the southern Carlin trend, porphyry, skarn, carbonate replacement, and small, generally Ag rich Carlintype mineralization accompanied shallowly emplaced, 38.5 to 37.4 Ma intrusions (Ressel and Henry, 2006; Henry et al., 2015; Jackson et al., 2015; Koehler et al., 2015; Ressel et al., 2015; Hollingsworth et al., 2017). CTDs at Rain, Emigrant, and North Bullion are <38.46 ± 0.20, <37.5 ± 0.8, and <38.46 ± 0.25 Ma, respectively, based on dates of mineralized dikes (S. Garwin in Longo et al., 2002; Ressel and Henry, 2006; Henry et al., 2015). Mineralization at North Bullion is 2.5 km north of and possibly driven by hydrothermal circulation related to the 38.2 Ma composite Bullion granodiorite intrusion. Four 40Ar/39Ar dates on hydrothermal muscovite, biotite, and adularia around the intrusion range from 38.15 ± 0.11 to 37.83 ± 0.03 Ma (Henry et al., 2015).

The southernmost CTDs are the youngest: 35.7 Ma at Cortez Hills, possibly 37.28 ± 0.21 Ma based on equivocally hypogene or supergene altered dacite breccia at the Lookout Mountain CTD at Eureka (Long et al., 2014), and ≤35.45 ± 0.05 Ma based on a weakly altered rhyolite dike in the Main deposit at Northumberland (H09-127; Table 1). The temporal trend supports the interpretation of a genetic tie to magmatism (Muntean et al., 2011). We conclude that Eocene magmatism unequivocally was the heat source, but our data do not address the source of gold, other metals, or fluids. The observation of low Au, Cu, As, Sb, and Te concentrations in melt and pyrrhotite (sulfide melt) inclusions in Caetano Tuff (Watts et al., 2013, 2014) suggests Caetano magma either was not a good Au source for CTDs or the metals were removed from magma before eruption of the tuff.

The giant Cortez Hills Carlin-type deposit and the nearby Pipeline and recently discovered Goldrush and Fourmile deposits lie in an area that underwent major Jurassic, Cretaceous, Eocene (40–39 and 36–34 Ma), and middle Miocene magmatism. Cortez Hills was intensely intruded by a swarm of mostly high- and a few low-SiO2 rhyolite dikes, which we term Cortez rhyolites. The dikes form a 6- to 8-km-wide and ~40-km-long NNW-trending swarm centered on Cortez Hills. Abundant 40Ar/39Ar sanidine dates show that the high-SiO2 rhyolites were emplaced in a brief interval at 35.7 Ma. The low-SiO2 rhyolites most likely intruded at the same time. Whether these dikes were pre- or postmineralization has long been debated. Our work shows that the dikes spanned the time of mineralization: they vary from strongly altered and mineralized to unaltered despite cutting high-grade ore. In the most altered dikes, plagioclase phenocrysts were replaced by kaolinite and calcite, biotite was replaced by smectite, calcite, and marcasite, and sanidine was unaltered. In the most intensely altered dikes, sanidine was also replaced by kaolinite, and the rocks consist almost entirely of quartz and kaolinite. Mineralized dikes contain marcasite, pyrite, arsenopyrite, and arsenian pyrite and are slightly to moderately enriched in typical Carlin-type elements Au, Hg, Sb, Tl, As, and S, as well as Ag, Bi, Cu, and Mo. However, no dikes are ore, probably because they were relatively impermeable, unreactive, and Fe poor. These characteristics demonstrate that Cortez rhyolites have undergone the same alteration and geochemical enrichment that generated the deposit and support published interpretations that CTDs form during brief time spans. Cortez Hills is the most precisely dated CTD in Nevada.

The dikes almost certainly rose from a deeper (top at probably 6–10 km), silicic (probably granitic) pluton that was the heat source for mineralization. The other large CTDs of the Cortez region lie in or within ~3 km of the Cortez swarm of rhyolite dikes and may also have formed at the same time and from related hydrothermal system(s).

Based on our new age constraints at Cortez Hills and best age estimates of CTDs throughout northern Nevada, formation of CTDs migrated southwestward through time, closely mimicking the southwestward migration of slab-rollback magmatism. Contemporaneity of CTDs and magmatism strongly supports proposed genetic links between the two. Eocene magmatism was the heat source, but our data do not address the source of gold, other metals, or fluids.

Discussions with many geologists, especially Mike Ressel, have greatly aided and influenced our understanding of magmatism and mineralization. These include John Muntean, Jean Cline, and Al Hofstra; Kevin Creel, Meghan Jackson, Tom Chapin, Tom Whittle, and David Arbonies at Barrick (now Nevada Gold Mines); Bob McCusker at Coral Resources; and Jeff Blackmon at Newmont (now Nevada Gold Mines). This research was supported in part by the USGS National Cooperative Geologic Mapping Program via STATEMAP agreement G09AC00116 by contracts between Barrick Cortez Gold Mines and the University of Nevada, Reno, Nevada Bureau of Mines and Geology, and the USGS. John, Colgan, and Watts were supported by the USGS Mineral Resources and National Cooperative Geologic Mapping Programs. We also thank Lisa Peters, Rich Esser, Jake Ross, and Nels Iverson for their considerable help with 40Ar/39Ar dating at the New Mexico Geochronology Research Laboratory. Joe Wooden, Jorge Vasquez, and Matt Coble are thanked for help with SHRIMP U-Pb dating. Reviews of an earlier version by John Dilles, Jean Cline, and Celestine Mercer and of this version by Dilles, Eric Christiansen, and Associate Editor Tony Longo greatly improved this report. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.

Chris Henry is a research geologist with the Nevada Bureau of Mines and Geology, University of Nevada Reno. Chris works on applied geologic problems in magmatic, tectonic, and mineralization processes using detailed geologic mapping supplemented by geochemistry and geochronology, particularly precise 40Ar/39Ar dating. He received a B.S. degree from Caltech and a Ph.D. from the University of Texas at Austin, both very long times ago. He spent 20 years with the Texas Bureau of Economic Geology before moving to Reno in 1993. Chris is now semiretired, which means, he says, “I am no longer paid, but geology is fun so I keep doing it.”

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

Supplementary data