Quartz Vein Formation and Deformation during Porphyry Cu Deposit Formation: A Microstructural and Geochemical Analysis of the Butte, Montana, Ore Deposit

Quartz vein stockworks in porphyry copper deposits form when overpressurized fluid above a crystallizing magma chamber invades the country rock and precipitates minerals in the fractures. The mineral content of veins and adjacent alteration envelopes changes through time as the fluids decompress and cool from near-magmatic temperature (~700°C) in the early stages to 350°C and below [1]. The existing large body of work on quartz vein microstructures in metamorphic rocks ([2] and references therein) forms a starting point for this work on high temperature quartz deformation in porphyry Cu deposits and the connection of quartz deformation microstructures to vein formation. In this contribution, we combine scanning electron microscope-cathodoluminescence (CL) imaging with electron backscatter diffraction (EBSD) to show that high temperature quartz veins in the Butte porphyry Cu-Mo deposit contain oriented undeformed euhedral crystals juxtaposed with randomly oriented deformed quartz grains that recrystallized by grain boundary migration and that recrystallization occurred at roughly the same temperature as the initial quartz precipitation.

Butte veins display three metamorphic quartz recrystallization regimes: bulging recrystallization, subgrain rotation, and grain boundary migration [3, 4]. Each regime creates characteristic microstructures that occur in different parts of the Butte vein system, though grain boundary migration is by far the most abundant. The various recrystallization microstructures correlate to different recrystallization temperatures and to deformation at different strain rates and magnitudes, as described herein.

The 65 Ma Butte porphyry Cu-Mo deposit provides an ideal study site because it has been thoroughly studied for more than 100 years [5] and because it is hosted by the homogeneous Butte Granite (Figure 1(a)) [6, 7], which eliminates geochemical and structural complexities caused by heterogeneous country rock. The veins and vein envelopes are well-characterized at Butte, as explained below, but some key observations lack explanation. Earlier investigations into barren quartz/quartz-molybdenite veins revealed two cathodoluminescence (CL) textures [8–10], hereafter referred to as “CL-euhedral” and “CL-mottled.” CL-euhedral texture consists of alternating dark and light growth zones of euhedral crystal faces, reflecting primary crystal growth from a hydrothermal fluid (Figure 2(a)). CL-mottled crystals are unevenly dappled with patches of darker CL on a background of lighter CL (Figure 2(c)). A third texture consists of wavy concentric “ghost” bands (Figure 2(b)). While CL-euhedral textures reflect crystal growth zoning, the origin of CL-mottled and ghost bands texture has been puzzling. In quartz-molybdenite veins, CL-mottled quartz is contiguous with CL-euhedral quartz. Crystals in deep barren quartz veins from drill holes that intersected the cupola region of the Butte deposit are exclusively CL-mottled. We present evidence that CL-mottled crystals form by grain boundary migration in CL-euhedral crystals and suggest that coexistence of the two textures is a consequence of preferential deformation in vein centers shortly after vein formation. We interpret the intermediate ghost band texture as relict CL-euhedral crystals that partially recrystallized, but strain ceased before the texture ripened to mottled. We argue that the preponderance of CL-mottled crystals in barren quartz veins is the result of extensive episodic deformation near the base of the deposit, directly above the magma body that underlies the magmatic cupola.

The Butte porphyry Cu-Mo ore deposit is hosted in the 76 Ma Butte Granite [11]. The Butte Granite intruded the Proterozoic Belt Supergroup, a marine sedimentary sequence, and Archean crystalline basement to the south of Butte [5]. The granitic melt likely formed when the Paleoproterozoic basement was partially melted by and mixed with mantle-derived melts when subduction in North America was happening at a low angle in the Late Cretaceous [11]. Emplacement of the Butte Granite into the cogenetic Elkhorn Mountain Volcanics was structurally controlled by reactivated Precambrian faults, which also controlled the emplacement of other intrusive members of the Boulder Batholith [12, 13] and occurred when the region was experiencing compression with the maximum compressive stress-oriented east-west. Based on geophysical data, Berger et al. [12] identified a southern feeder zone for the Butte Granite, whose structures also served to focus mineralizing fluids. Within the Butte Granite are alaskite and granoaplite dikes and sills that are the products of late high-silica granite crystallization. Following the usage of Roberts [13], alaskite refers to rocks of highly variable textures and grain size that are made up of K-feldspar, albite, and quartz. The term includes aplites, pegmatites, and the gradations between them. Alaskites occur throughout the Boulder batholith as tabular bodies and generally have sharp contacts with the Butte Granite [13]. Alaskites are interpreted as the water-rich residua of the parental Butte Granite melt [13, 14]. Granoaplites are like alaskite but are coarse-grained and porphyritic. Granoaplites have phenocrysts of plagioclase, hornblende, biotite, and quartz eyes set in a groundmass of quartz and alkali feldspar. Contacts between the Butte Granite and granoaplite are gradational.

The Butte deposit consists of early porphyry Cu mineralization known as the pre-Main Stage and subsequent Cordilleran-style base metal veins of the Main Stage. The pre-Main Stage porphyry Cu-Mo ore formed at depths of ~5-9 km [15], and the Main Stage polymetallic veins formed at shallower depths (Figure 1(b)) as fissure fillings in faults [16]. The recently examined genetic link between porphyry deposits and epithermal deposits supports the idea that the Main Stage veins result from the late evolution of the Butte porphyry copper system and are the roots of an epithermal deposit that has been eroded away [17–21].

The Butte system has two concentrically zoned porphyry Cu-Mo domes, distinguished by separate magnetite vein zones and closure of molybdenum grade contours (Figure 1(c)). Each dome is approximately 2 km wide. The western Anaconda dome is separated from the eastern Continental-Pittsmont dome by a bulb-shaped region hosting pyrite veins and pervasive gray sericitic alteration that is 1.2 km wide and extends 2 km deep below the Berkeley Pit (Figure 1(b)).

Pre-Main Stage and Main Stage alteration and mineralization types and relationships have been extensively documented elsewhere [1, 13, 22, 23]. The earliest veins of the potassic series are quartz-chalcopyrite-pyrite veins with biotite, K-feldspar, and sericite alteration envelopes (“EDM” veins) [22], which include biotite crackles connected to biotite breccias surrounding the apices of quartz porphyry dikes [1]. Biotite crackles are microscopic quartz veinlets with haloes of wall rock altered to biotite, muscovite, and K-feldspar; biotite breccias consist of subangular-subrounded wall rock clasts in a matrix of biotite+quartz+K-feldspar+pyrite [23]. EDM veins vertically grade through a series of alteration types that form because of a single parent fluid cooling and interacting with wall rock at various water-rock ratios [1]. These sulfide-bearing quartz veins are cut by barren quartz and quartz-molybdenite veins. The term “barren” here is a field term that refers to veins with only trace amounts of molybdenite. Barren quartz veins characterize the base of the deposit, forming extensive stockwork zones. Some barren quartz veins grade upwards into quartz-molybdenite veins. The youngest pre-Main Stage veins are pyrite-quartz veins with gray sericitic (GS) alteration in which igneous plagioclase and K-feldspar are replaced by quartz, muscovite, and pyrite and igneous biotite and hornblende are altered to mixtures of muscovite, pyrite, quartz, Mg-chlorite, and rutile [1].

A barren quartz vein stockwork is typical in the root zones of many porphyry Cu-Mo deposits [24, 25] and is accompanied by intense silicification, also known in the literature as “quartz flooding,” in the region above the cupola. Intense silicification refers to the addition of large amounts of nonvein quartz to the host rock groundmass, likely by the infiltration of quartz-supersaturated fluids. Barren quartz and quartz-molybdenite veins generally lack alteration envelopes, although thin envelopes containing biotite or K-feldspar occur in the deepest of samples [1]. In this study, we focus on barren quartz from the deep zone of intense silicification intersected by a deep drill hole where quartz-molybdenite veins are nearly absent (sample location shown by the yellow star in Figure 1(b)) and on quartz-molybdenite veins from a shallower region exposed in the Continental Pit where barren quartz veins are rare (sample location shown by the red star in Figure 1(a)). We also describe a quartz-rich Main Stage vein (sample location shown by the blue star in Figure 1(a)). These samples were chosen because they are representative of the major types of quartz veins in the Butte system and because they represent the spatial and temporal changes in quartz precipitation coeval with deposit formation.

Samples are from drill core from deep drill hole 1A (DH-1A, an angled hole that intersected the barren quartz stockwork) and from highwall exposures on the C-East pushback of the Continental Pit. Samples from the pit were collected while mapping the active pit in 2014 in cooperation with Montana Resources. The drill core was sampled during logging of the historic core and is currently housed by Montana Resources at the Montana Bureau of Mines and Geology at Montana Tech. Petrographic descriptions of thick and thin sections were made with an optical microscope. The sections were scanned at high resolution (3200 dpi) in both cross-polarized and plane-polarized light using a Nikon slide scanner. Grain boundary microstructures were identified from thin sections using a petrographic microscope (Figure 3). Representative samples of barren quartz stockwork, quartz-molybdenite veins, and a monogenetic Main Stage vein were analyzed in greater detail using methods described below. All analyses were carried out in the Center for Advanced Materials Characterization in Oregon (CAMCOR) at the University of Oregon, except for electron backscatter diffraction data, which was acquired at the University of Lausanne.

Thick sections were cleaned with ethanol, coated with ~50 nm of carbon, and imaged with an FEI Quanta 200 FEG environmental scanning electron microscope (SEM) at 15 keV with a beam current of 40 nA. Grayscale cathodoluminescent, secondary electron (SE), and backscattered electron (BSE) images were collected. Major phases were identified using energy dispersive spectroscopy (EDS).

To enable comparisons of the different types of microscopic images we correlated images precisely using PictureSnapApp (downloadable software from Probe Software, Inc.), wherein high resolution visible light scans of thin sections can be calibrated with the sample in the SEM chamber so that the scans can be used for navigation.

Trace elements can enhance or dampen quartz CL intensity depending on how they distort the crystal lattice. A detailed study of correlations between trace elements and CL intensity of quartz from Butte and other hydrothermal systems was done by Rusk et al. [10], who found that for high temperature quartz (>350°C), CL intensity correlates positively with Ti concentration, and that for low temperature quartz (<350°C), CL intensity correlates negatively with Al concentration. Titanium solubility in quartz is a function of pressure and temperature [26] but can also be influenced by growth kinetics [27, 28].

CL brightness in an image depends on the gain and offset of the CL detector set by the operator to either emphasize grayscale midtones in the image or to enhance contrast. For any given quartz CL image, it is generally acceptable to interpret the brighter regions as having higher Ti and the darker regions as having lower Ti. However, the correlation between the two is nonlinear, making it difficult to quantify the Ti from CL by interpolation between CL-bright and CL-dark regions [9]. It is likewise misguided to compare CL intensities of different images acquired at different operating conditions. In other words, “bright CL” in one image does not necessarily correspond with the “bright CL” of another.

Electron backscatter diffraction data (EBSD) was collected on a composite quartz-molybdenite vein, BMA-023-16Fa. The sample was polished overnight on a vibratory polisher with 0.05 μm silica grit and a highly napped cloth. Prior to analysis, it was carbon coated with 20 nm of carbon. Automatically indexed electron backscatter diffraction patterns were collected with an Oxford Instruments Symmetry detector running the Aztec 5.1 software on a Tescan Mira field emission SEM at the University of Lausanne with a 25 kV accelerating voltage and a beam current of 4 nA. The sample was tilted to 70° and analyzed at a working distance of 20-25 mm. Patterns were acquired at a step size of 10 μm. Quartz constituted more than 90 vol.% of indexed phases, accompanied by minor calcite, sericite, and molybdenite. Nonindexed points were grain boundaries, unidentified inclusions, or polishing artefacts such as scratches. Indexing and subsequent cleaning of the data was done using with the Aztec software. EBSD data was plotted and analyzed with MTEX [29].

A representative quartz-molybdenite vein (sample BMA-018-16Fa) with no macroscopic evidence of rebreaks, minor amounts of molybdenite, and macroscopically massive quartz was selected for Ti measurements by EPMA. On this sample, 175 Ti analyses on four CL-euhedral and two CL-mottled grains were collected on a Cameca SX100 electron microprobe equipped with five wavelength dispersive spectrometers in CAMCOR at the University of Oregon. The beam was set to a 15 keV energy, 100 nA current, and 5 μm diameter. Ti Kα was measured on all spectrometers, and aggregate mode was turned on. The counting time for Ti Kα was 425 seconds for quartz, MgO, NiO, and MnO reference materials. Intensity data were corrected for time dependent intensity loss (or gain) using a self-calibrated correction for Ti Kα. Mean atomic number (MAN) background intensity data were calibrated, and the Bremsstrahlung continuum absorption was corrected for Ti Kα (see Donovan et al. [30] for details). Unknown and reference material intensities were corrected for deadtime. Reference material intensities were corrected for drift over time. A quantitative blank correction with the quartz reference material as a known unknown was applied. The Main Stage vein was also analyzed, but all points were below detection limit.

In addition to the quartz-molybdenite and Main Stage veins, two samples of barren quartz stockwork from DH-1A, samples 11052-6121 and 11052-6019 (the last four numbers correspond to the depth of the drill hole the samples are from in feet), were analyzed for Ti, Fe, K, and Al. For these runs, Ti was measured simultaneously on a reference material and on a large pentaerythritol crystal. The beam had 20 keV energy, 200 nA current, and 20 μm diameter. Reference materials were synthetic TiO₂ for Ti Kα, orthoclase MAD-10 for Al Kα and K Kα, and magnetite U.C. #3380 for Fe Kα. Counting times were 450 seconds for all elements, and a MAN background was used.

The stockwork veins in drill core DH-1A consists of 0.1-7 cm thick, massive, barren quartz. In any given ~10 m of core there are quartz veins with sharply defined planar edges and others with undulose edges (see Appendix (available here)); both coexist with veins that have diffuse gradational contacts that bleed into the country rock. Much of the country rock is intensely silicified or “quartz flooded.” Altered wall rock, whether quartz porphyry or Butte Granite, is altered to gray sericite mineral assemblages and contains, in order of decreasing abundance, quartz, sericite, disseminated pyrite, and trace chalcopyrite. Molybdenite occurs as small (<5 mm) cryptocrystalline blebs near the edges of veins or dispersed throughout the regions of recrystallized quartz.

Barren quartz veins in the stockwork display textures described by Urai et al. [31], and references therein, indicating complete recrystallization, as explained below. Quartz crystals are anhedral, show no crystallographically preferred orientations (CPOs) in the thin section with the quartz plate inserted, and do not exhibit a shape-preferred orientation (Figure 3). Recrystallized quartz grains range in size from 0.05 to 0.2 mm. Grain boundaries are characterized by lobate margins which are broadly indicative of grain boundary migration recrystallization (Figure 3(c)), but features indicative of subgrain rotation (Figure 3(b)) and bulging recrystallization (Figure 3(a)) are also present. There are also small grains that have “foam textures”—quartz with curved grain boundaries and ~120° triple junctions, which are indicative of static recrystallization [2]. Other deformation features include undulose extinction, slightly misoriented domains within crystals, and subgrains decorating grain boundaries. Recrystallized grains contain anhedral fluid inclusions that are randomly distributed throughout (Figure 4(b)). Primary pores are abundant along grain boundaries.

Quartz-molybdenite veins are planar and are ~0.1 to 13 cm thick. The veins lack alteration envelopes (see Appendix A2). The walls of smaller, monogenetic quartz-molybdenite veins are lined with 0.1-1 cm CL-euhedral cockscomb quartz, and vein centers are filled with equigranular quartz. The $c$-axes of cockscomb quartz crystals are perpendicular to the vein wall, which can be easily seen under the petrographic microscope because the orientation of host crystals is shown by $c$-axis-parallel-aligned euhedral fluid inclusions (Figure 4(a)) [32–34]. The fluid inclusions are types B35 and B60, meaning that, on average, about 35% or 60% of the volume of the fluid inclusion is occupied by a bubble at room temperature (Figure 4; compare Rusk et al. [15]). Composite veins record many rebreaks (similar to syntectonic metamorphic veins) and are much more common than monogenetic quartz veins. Composite veins are basically many monogenetic veins layered together.

Like barren quartz veins, three different grain boundary textures occur in quartz-molybdenite veins (Figure 3). Equigranular quartz grains in vein centers range in size from 20 to 200 μm in diameter, but most are 20 to 70 μm and many have straight grain boundaries. The smaller crystals (20-70 μm) do not display undulose extinction. Adjacent to these small, nonundulose crystals are larger crystals (100-200 μm along the longest observable axis) that do exhibit undulatory extinction. The grain boundaries of larger grains without neocrysts are serrate or lobate, indicative of grain boundary migration recrystallization. Some recrystallized quartz has a polygonal texture. All recrystallized quartz types contain many small, irregularly shaped fluid inclusions such as those shown in Figure 4(b) as well as unidentified mineral inclusions and planar arrays of secondary fluid inclusions. The anhedral forms of these inclusions are characteristic of fluid inclusions that have been broken and dispersed throughout a crystal during grain boundary migration recrystallization [35]. Where inclusions were intersected during polishing, pock marks are left behind on the polished surfaces.

In larger quartz-molybdenite veins, those exceeding 7 cm thickness, the quartz is distinctly banded with molybdenite (see Appendix A2). Much of the molybdenite is microcrystalline, and it most commonly occurs in thin, discrete layers along the edges of veins and edges of bands within large veins. Some individual molybdenite grains are visible with a hand lens and rarely with the unaided eye. The molybdenite is flecked through fine-grained quartz or is in nearly monomineralic layers. In smaller veins, the molybdenite layers are predominantly planar, but in larger veins, the molybdenite layers are arcuate and smeared through the interior of the quartz vein, creating a marbled appearance (see A2 in Appendix). Microscopic flecks of molybdenite are also dispersed throughout quartz-rich regions. Layer thicknesses range from ~0.2 to 2 cm in the large, banded veins.

The Main Stage vein (sample BMA-020-16Fa) from the easternmost Continental Pit (blue star in Figure 1(a)) is 2 cm wide and contains quartz that lacks undulatory extinction and recrystallization microstructures. The wall rock is pervasively altered to a white argillic assemblage that includes residual abundant chalcopyrite and primary biotite. The vein is symmetrically zoned from an outermost 2 mm layer of microcrystalline milky quartz, overlain by a <1 mm layer of clearer microcrystalline quartz, and a third <1 mm layer of milky quartz from which macroscopically euhedral 0.3 to 2 mm comb quartz crystals project into vugs in the center of the vein, some of which contain chalcopyrite, pyrite, and bornite—a common Main Stage assemblage.

Barren quartz vein samples from DH-1A are entirely CL-mottled. Th quartz-molybdenite vein from the eastern Pittsmont dome (sample BMA-018-16Fa; red star in the Continental Pit area of Figure 1(a)) has the following symmetrically developed sequence of textures from the edges to the centers: wall rock fracture surface, a 0.01 mm layer of CL-mottled crystals, a 0.5-2 mm layer of cockscomb CL-euhedral quartz crystals, and a central layer of CL-mottled grains (Figure 5). Oscillatory zones are continuous from one crystal to the next along cockscomb layers (Figure 5). Most growth zones in CL-euhedral crystals are 10 to 15 μm thick, but they range in thickness from 25 μm to <1 μm, and most CL-bright bands have sharp boundaries against adjacent CL-dark bands (Figures 2(a) and 5).

Some CL-euhedral crystals have mottled edges (Figure 5). CL-mottled crystals ranging in size from 10 to 30 μm fill the centers of veins and dominate the volumetric bulk. Some crystals in the center zone have a “ghost band” CL texture that is visually intermediate between CL-euhedral and CL-mottled (Figure 2(b)), wherein what appear to be relict bright bands of CL-euhedral texture have been modified by diffusion. Isolated CL-euhedral crystals with mottled edges can be found in vein centers surrounded by CL-mottled crystals.

The Main Stage quartz vein displays breccia and cockscomb quartz textures (Figure 6). Layers of cockscomb quartz exhibit euhedral oscillations that are unambiguously continuous from one crystal to the next (labelled as “oscillatory zones that follow continuously from one crystal to the next” in Figure 6(f) and seen also in the upper portion of Figure 6(e)). In the center of the vein are smaller euhedral quartz crystals with banding patterns in their cores. These cores of single crystals are encrusted in rims whose banding patterns track from one crystal to the next (Figure 6 panels). In addition, the vein center contains clasts that appear to be rip-ups of an early generation of dark CL-homogeneous quartz (Figure 6(e)). The interpretation of these clasts as rip-ups is supported by the presence of larger clasts of preexisting veins that are incorporated into Main Stage veins at hand sample and outcrop scales. These CL-homogeneous clasts are overgrown with CL-euhedral crustiform quartz that appears to match the encrusting quartz otherwise in the center of the vein (Appendix section A4).

Other textures include sharply defined, angular domains of mixed and patchy CL-bright and dark quartz (Figures 6(c)–6(e)), hereafter referred to as CL-chaotic (see also Appendix section A4). Heterogeneity in the CL-bright domains occurs as CL-gray feathers or wavy regions with slightly less CL intensity. In some regions, CL-bright domains are plumose or flame-like. Regions of CL-euhedral quartz occur sporadically within chaotic regions. It is notable that, despite the complexity and small scale of the CL-chaotic texture, it is not formed of clusters of microcrystals. In other words, in the thin section (Figure 6(a)), CL-chaotic crystals are optically continuous and relatively large (100–500 μm) while CL-bright and dark domains in that same optically continuous crystal are much smaller (20-80 μm).

In all quartz veins imaged, there are ubiquitous late-stage microfractures that cut CL-mottled and CL-euhedral grains alike (Figures 2, 5, 6, and 7; see also Rusk et al. [10]). Similarly, CL-dark rinds are common along the grain boundaries of CL-mottled crystals (Figure 7). Because these fractures cut both CL-mottled and CL-euhedral crystals, they are later than both of them and not further considered here.

An EBSD map (Figure 8(a)) of a composite quartz-molybdenite vein from the Continental Pit (BMA-023-16Fa; red star in Figure 1) shows the distribution of randomly oriented CL-mottled crystals along with a preserved layer of CL-euhedral crystals. Figure 8(b) shows a thin section in cross-polarized light with a layer of cockscomb quartz outlined in red. That same layer is outlined by an inset in quartz orientation maps (Figures 8(a) and 8(c)), a map showing the grain orientation spread (Figure 8(d)), and is shown in CL in Figure 8(e). The correlated misorientation diagram indicates the presence of Dauphiné twins, as shown by a large peak at 60° (Figure 8(f)). The uncorrelated misorientation diagram shows that the overall texture of the mapped region is close to random (Figure 8(g)).

Overall, the EBSD data shows that there are small layers of quartz where the crystals are oriented subparallel to one another but that the majority of vein quartz is randomly oriented.

We analyzed CL-mottled crystals and bright and dark growth bands of CL-euhedral textures and found the largest range in Ti concentrations in CL-euhedral crystals of the quartz-molybdenite vein from the Continental Pit (BMA-018-16Fa). Ti concentrations in four crystals (⁠$n=132$⁠) range from 17 to 323 ppm in the bright domains with a mean of $80±63ppm$ Ti and 16 to 40 ppm Ti in the dark domains with a mean of $25±8ppm$ Ti. Two crystals have CL-euhedral cores that grade into CL-mottled edges with slightly elevated Ti concentrations compared to contiguous, entirely CL-mottled crystals. These CL-mottled crystals contain 23 to 47 ppm Ti with a mean of $31±6ppm$ Ti. Mottled edges of crystals with CL-euhedral cores have between 7 and 45 ppm Ti with mean of $29±9ppm$ Ti. Quartz in barren quartz veins, all of which is CL-mottled, ranges from 99 ppm Ti to below the detection limit and has a mean concentration of $37±19ppm$ Ti (⁠$n=79$⁠). Analytical uncertainty on individual data points is a maximum of ±4 ppm Ti, and variation due to measurement error is small compared to the spread of Ti concentrations. Ti concentrations, errors, and sampling statistics are shown in Figure 9 and are available for download in the Electronic Materials. All Ti measurements from the Main Stage sample BMA-020-16Fa are below the detection limit of 7 ppm (⁠$n=43$⁠), which confirms the below-detection microprobe Ti results for Main Stage quartz that Rusk et al. [10] reported.

Ti concentrations in quartz can be used to estimate temperatures of quartz crystallization. In a previous paper, we showed that the Huang and Audétat [27] calibration yields temperatures that are consistent with independent estimates for samples from Butte and several other well-constrained igneous systems (see Figure 1 of Acosta et al. [28]). We assume the fluid is at rutile saturation (i.e., the activity of TiO₂ is fixed at unity) because rutile is abundant in barren quartz and quartz-molybdenite veins. All temperature estimates reported herein are calculated using a lithostatic pressure of 2.5 kbar based on the fluid inclusion analyses of Rusk et al. [15] and Roberts [13]. Calculations at a hydrostatic pressure of 0.63 kbar yield temperature estimates that are ~83°C lower.

Previous temperature estimates of barren quartz/quartz-molybdenite veins range from 500 to 700°C using three different thermometers. Roberts [13] used coexisting ilmenite-hematite pairs to constrain the temperature to 550–700°C. Brimhall [23] used K-Na partitioning in K-feldspar and muscovite to constrain the temperature to 550-700°C as well. Field et al. [36] used S isotopes of coexisting sulfide mineral pairs to estimate quartz-molybdenite temperatures of 500-630°C.

Combining the Ti-in-quartz Mercer and Reed [37] dataset with ours gives an average of 48 ppm Ti and a temperature of 660°C for barren quartz veins. We note that the Mercer and Reed [37] dataset is made up of seven barren quartz samples taken from four different drill holes, and so, the variability in their data is probably due in part to natural variations in the temperatures of formation of individual veins.

The quartz-molybdenite vein from the Continental Pit gives temperatures ranging from 499 to 907°C (Figure 9). The highest temperatures are from CL-euhedral bright bands and are unrealistic given the fluid temperature estimates outlined above, indicating that these bands have Ti in excess of the equilibrium concentration. In contrast, an average Ti of 25 ppm in CL-euhedral dark bands corresponds to 600°C. Averages of 29 ppm Ti for mottled edges and 31 ppm Ti for mottled crystals yield respective temperatures of 612°C and 619°C. An average of 37 ppm Ti from mottled quartz in barren quartz veins from DH-1A corresponds to a temperature of 635°C. It is unlikely that the deformation that resulted in CL-mottled textures occurred at a temperature significantly higher than that of quartz precipitation. Barren quartz/quartz molybdenite veins are thus estimated to have formed and deformed at temperatures between 600°C and 635°C.

The maximum Ti concentration of 99 ppm in barren quartz corresponds to a near-magmatic temperature of 741°C at 2.5 kbar, which is a reasonable upper limit given that some deep barren quartz veins adjacent to the cupola have aplitic centers and contain a mixture of fluid and melt inclusions (not examined in this study) [13, 15]. Overall, temperature estimates for CL-mottled and the dark bands of CL-euhedral quartz made using the Huang and Audétat [27] calibration of the TitaniQ thermobarometer are in good agreement with independent temperature constraints.

The strain that transforms CL-euhedral to CL-mottled can be interpreted using the substantial body of work on quartz recrystallization and how different processes are expressed in measurable quantities such as crystal shapes, crystal size distributions, and crystallographic preferred orientations (CPOs).

Bulging recrystallization occurs between 250°C and 400°C and is a common outcome of high strain rates [4]. As a bulge ripens, it becomes a discrete small grain, so this regime produces a bimodal grain size distribution wherein larger crystals are mantled by new small crystals (core-and-mantle structures; Figure 3(a)). Subgrain rotation occurs between 400°C and 500°C [31] and is characterized by straight grain boundaries and, in some instances, a bimodal crystal size distribution. Grain boundary migration is active above about 500-600°C and can result in coarse crystals with lobate and serrate grain boundaries [4, 38]. Grain boundary migration in an isotropic stress field is also the mechanism for quartz annealing/polygonization, whereby equilibrium foam structures characterized by curved grain boundaries and 120° triple junctions form [39]. Regardless of whether grain boundary migration created lobate grain boundaries or foamy grain boundaries, we find that all CL-mottled grains have grain boundary recrystallization microstructures.

Recrystallization changes initial crystal size distributions as a function of recrystallization mechanism and differential stress. During recrystallization, some grains grow at the expense of others, which can produce an equilibrium grain size distribution that may be finer- or coarser-grained than the initial one. Recrystallization can result in grain size reduction and subsequent mechanical weakening [40, 41] but can also result in grain coarsening and strength hardening [41]. However, interpreting recrystallization crystal size distributions in natural samples is difficult because of overprinting during cooling, a potential dependence on water activity, misinterpretation of the dominant recrystallization mechanism, the presence of secondary minerals that can pin otherwise mobile grain boundaries, and heterogeneous strain accommodation [2].

Grain boundary sliding, the movement of grains relative to one another, may accompany quartz recrystallization [42, 43] at low strain rates and high temperatures. Grain boundary sliding can also operate at low stresses in dilatant systems where it can be accommodated by cataclasis and particulate flow [42]. The sliding may prevent the formation of a crystallographically preferred orientation (CPO) or destroy an existing CPO [44, 45] and may randomize grain boundary misorientation axes [46]. The presence of pores along grain boundaries and small grain sizes can facilitate grain boundary sliding [2, 47].

The initial crystal size distribution also exerts a control on material strength during ductile deformation. Typically, materials that are coarser grained are stronger than materials that are finer grained which are less cohesive. The opposite can also be true when deformation occurs mostly as dislocations moving through a crystal; grain boundaries stop dislocations, so dislocation movement is limited by the density of grain boundaries (Hall-Petch strengthening).

A brief review of quartz slip features here provides a basis for interpretations explained below. At low strain rates and temperatures below ~650°C, basal slip (perpendicular to the $c$-axis) is the dominant slip system in quartz [48–50]. At higher temperatures and strain rates, $c$-slip (slip parallel to the $c$-axis) is the dominant slip system. A given crystal may deform plastically by either of these slip systems in a deviatoric stress field depending on its orientation. The resolved shear stress, the effective shear acting on a slip plane, is maximized when σ₁ is 45° from the slip plane and minimized when σ₁ is perpendicular to the slip plane [51]. For the relevant temperature range (~650°C), quartz could slip along either the basal or prism planes. A crystal is said to be in an easy glide orientation if it is oriented in a given stress field such that the resolved shear stress on a given slip plane is higher than the critical resolved shear stress, which is the effective shear needed to activate glide on a given slip plane. A crystal is said to be in a hard glide orientation if it is oriented in a given stress field such that the resolved shear stress on a given slip plane is less than the critical resolved shear stress.

The only twins found in the EBSD data are Dauphiné twins (Figure 8(f)). Dauphiné twinning is defined by a 180° axis rotation about the $c$-axis [0001] of quartz, and it has no effect on the orientation of the crystallographic axes of the crystals, meaning that adjacent Dauphiné-twinned crystals are in the same hard- or easy-glide positions relative to one another and deform identically because the orientations of glide systems are the same in both [52, 53]. The existence of Dauphiné twins in cockscomb quartz layers does not alter the response to an applied shear stress but can be used to place additional constraints on the temperature of quartz growth.

Dauphiné twins can be primary growth twins, mechanical twins, or transformational twins. Mechanical twins form during deformation. Transformational twins form during cooling through the α–β transition temperature. In β quartz, m-faces are absent or extremely poorly developed, and the r- and z-faces become crystallographically and morphologically equivalent [54]. Cockscomb quartz Dauphiné twins did not form by cooling because we can see the trace of prism faces and distinguish between r- and z-faces in Butte vein quartz. Because the α–β transition is displacive, meaning that it is just a slight change in bond angle and requires no breaking or making of covalent bonds, CL textures are unaffected by the phase change. Thus, the cockscomb quartz formed below the α–β transition temperature, which corresponds to 638°C at 2.5 kbar or 590°C at 0.62 kbar [55]. Since quartz recrystallizes at the same (or similar) temperature as growth (see below), the Dauphiné twins in recrystallized quartz are likely a combination of growth and mechanical twins.

The CL-dark bands record near equilibrium growth, and the high-Ti bright bands reflect excursions away from equilibrium, as supported by the following observations: (1) CL-dark euhedral bands have roughly the same amount of Ti from one dark band to the next, whereas Ti concentrations in bright growth bands are higher and wider-ranging (Figure 9). (2) The TitaniQ temperature estimates for dark bands and CL-mottled crystals are in good agreement with each other and with independent temperature estimates using other geothermobarometers. (3) Ti concentrations in dark bands are similar to those in nearby recrystallized CL-mottled crystals. This last observation is discussed in more detail in the section below.

The unrealistically high temperature estimates of CL-bright quartz bands suggests that they contain Ti in excess of the equilibrium value. Fast growth is expected to increase the uptake of Ti [27, 28], and one possibility is that CL-bright bands form during pressure fluctuations that are relatively short-lived but cause fast quartz growth.

It is accepted that CL-euhedral quartz grains are primary crystals precipitated from a hydrothermal solution, but the origin of coexisting CL-mottled crystals is less obvious. CL-mottled crystals display undulose extinction in the thin section and have grain boundary forms indicative of recrystallization, indicating that CL-mottled quartz may form by deformation of CL-euhedral quartz with consequent smearing (ghost band textures are discussed in a separate section below) or obliteration of growth zonations. This raises the question: why do some CL-euhedral crystals escape deformation?

Cockscomb quartz forms by the process of geometric selection, also known as competitive grain growth, which has been studied extensively by geologists and chemists [56–60]. In the process of geometric selection, an initial nucleation along a substrate creates a layer of randomly oriented crystals nucleated on the fracture surface (Figure 10(a), step 1) [60, 61]. Crystals that happen to be oriented with $c$-axes perpendicular to the fracture surface outgrow the others because their tips protrude into the cavity where the supply of silica from the fluid is most readily available (Figure 10(a), step 2), resulting in a layer of large, subparallel crystals (Figure 10(a), step 3) [62, 63]. This creates cockscomb quartz, which can form from fluids at low degrees of quartz supersaturation that undergo either slow or mild fluctuations in growth conditions such as pressure or temperature [59, 64, 65].

In contrast to the edges of quartz-molybdenite veins being cockscomb quartz, the vein centers are made of anhedral, equigranular, CL-mottled quartz. The equigranular quartz in vein centers may precipitate in situ but not as epitaxial overgrowths on cockscomb quartz, or the quartz grains may nucleate higher or lower in the vein fluid, then accumulate in the vein center. None of these would result in a crystallographically preferred orientation in the centers of veins; i.e., vein centers would have initially had crystals in both hard and easy glide orientations (Figure 10(a), step 4).

We infer that after the veins were filled but before they cooled, recrystallization at high temperature deformed CL-euhedral quartz in vein centers to CL-mottled quartz. Our data showing that Ti concentrations in CL-mottled crystals is roughly the same as those in the dark bands of adjacent CL-euhedral crystals indicates that deformation temperatures were nearly the same as the temperature of primary crystallization of CL-dark bands. Recrystallization was dominated by grain boundary migration at low differential stress. Grain boundary migration reset quartz towards the equilibrium Ti concentration for the recrystallization temperature. Volume diffusion of Ti through quartz is very slow [66, 67], but strain-induced grain boundary migration can reset Ti in quartz towards a new equilibrium value at a rate that is much faster [68–70] especially in fluid-rich systems such as Butte [71]. Lower temperature recrystallization mechanisms—subgrain rotation and bulging recrystallization—can reset Ti to lower equilibrium values in certain conditions [71] but are generally much less likely to do so [71, 72]. Haertel et al. [73] found that synkinematic quartz veins that underwent retrograde bulging and subgrain rotation recrystallization during exhumation at temperatures less than 350°C preserved Ti concentrations and gave thermometric estimates that were a close approximation of the temperature of original vein formation. Thus, we infer that the CL-mottled texture and attendant Ti resetting occurred during high temperature grain boundary migration recrystallization and that subsequent lower temperature bulging and subgrain rotation recrystallization must have occurred during vein cooling and exhumation but did not further facilitate Ti loss and equilibration.

For cockscomb quartz layers, basal glide would occur parallel to the vein edge (and perpendicular to the $c$-axis) and prism glide would be perpendicular to the vein edge (and parallel to the $c$-axis). Cockscomb quartz is thus a layer of hard glide-oriented crystals with respect to the expected paleostresses (Figure 10(b) , step 5) and also with respect to both possibly active slip systems. Crystals sandwiching the cockscomb quartz were initially in a variety of hard and easy glide orientations. As stresses acting on the vein fluctuate from local, magma chamber-related stresses to regional stresses, easy glide-oriented crystals will preferentially accommodate deformation. Glide fails to occur in cockscomb crystals because the shear stress on the potentially active glide planes of cockscomb quartz is less than the critical stress needed for slip to occur. In contrast, quartz crystals in layers above and below the layer of cockscomb quartz are initially in a combination of hard and easy glide orientations, because no primary growth process could have produced a strong CPO in these layers. At the onset of deformation, easy glide-oriented quartz grains deform first. As they deform around crystals in hard glide orientations, the hard glide-oriented crystals rotate into easy glide orientations until they too deform plastically [74]. Thus, strain is preferentially accommodated in the noncockscomb quartz layers that sandwich cockscomb quartz, leaving the cockscomb quartz relatively intact (Figure 10(b), step 5). Stated another way, the cockscomb quartz layer is rheologically stronger than the noncockscomb quartz layers. A similar conclusion was reached by Toy et al. [75] in their investigation of the Alpine Fault Zone mylonites from New Zealand. Toy et al. [75] found that quartz crystallographic orientations developed in early deformation events controlled which slip systems were activated during subsequent deformation and, thus, whether or not the early fabric was preserved or overprinted. Preferential deformation due to crystal orientation has been shown to occur in peridotite [76], metallurgical samples [77], and deforming ice. For example, polycrystalline aggregates of ice, which is also in the hexagonal crystal system and is generally accepted to deform analogously to quartz, undergo preferential deformation of easy glide-orientated crystals in glaciers [78, 79]. Marmo and Wilson [80] showed that a vein made of oriented columnar ice crystals in hard glide orientations remained undeformed during glacial flow as strain was accommodated in the surrounding easy glide-oriented ice. Bestmann et al. [71] also showed that the crystallographic orientation of quartz in metamorphic veins determined whether a crystal persisted as a relatively unstrained porphyroblast or was deformed into ribbon or mylonitic quartz. These results highlight the role that growth fabrics can have on rock strength during ductile deformation.

Another possible reason for the difference in relative strengths between cockscomb quartz and vein interiors could be starting grain size. However, because recrystallization modified the crystal size distribution of vein interiors via grain boundary migration, we cannot speculate as to what the initial crystal size distribution was and how it influenced subsequent heterogeneous strain accommodation. CL-euhedral cockscomb crystals are much larger than CL-mottled crystals, but this could be due to grain size reduction of CL-mottled crystals during recrystallization and cannot be used as evidence that CL-mottled crystals were initially larger or smaller than CL-euhedral ones.

Barren quartz veins, which occur deeper in the system and grade upward into quartz-molybdenite veins, are completely lacking in CL-euhedral crystals. It is possible that cockscomb quartz did not form at depth. It is also possible that these veins lack cockscomb quartz because they underwent more recrystallization than their shallow counterparts owing to their proximity to the cupola. More intense deformation of barren quartz veins would also explain the observation that many of them have undulating vein forms—initially planar veins could have undergone weak folding during cupola inflation and deflation cycles.

When cupola fluids are overpressurized, there is a localized zone around the cupola where the stress field is dominated by cupola-imposed stresses. The wall rock adjacent to the cupola is also hot and relatively ductile owing to its proximity to the magma chamber. Far from the overpressurized cupola, stress is regional, which was compressive with a horizontal σ₁ in Butte [16]. Between the cupola stress field and the far-field tectonic forces is a zone where the stress is a summation of the two. In the proximity of this zone, stress orientation would fluctuate (Figure 10(b), step 5) as the cupola inflates, hydrofractures, deflates, and reinflates [81]. Thus, veins emplaced in the zone of combined regional and local stresses undergo localized deformation as cupola stresses change. This may be accompanied by collapse of the deflated cupola or subsequent rotation of the vein as a block during later fluid expulsions.

To summarize, early quartz veins in Butte formed by the following:

• (1)

  Hydrofracturing rock around the magma chamber cupola to create dilatant fractures filled with fluid

• (2)

  Growth of cockscomb quartz by geometric selection (Figure 10(a), steps 1-3) wherein crystals with $c$-axes perpendicular to the wall rock substrate outgrow the others, creating a layer of parallel crystals

• (3)

  Solidification of the center of the vein (Figure 10(a), step 4) where euhedral, homogeneously nucleated, and randomly oriented crystals jam or settle in the centers of veins, thereby plugging the vein

• (4)

  Grain boundary migration and dissolution-precipitation recrystallization of veins in the cupola and its halo (Figure 10(b), step 5)

  • (a)

    Barren quartz veins are abundant in and next to the cupola where high temperatures enable ductile strain, causing veins to recrystallize to CL-mottled textures

  • (b)

    Quartz-molybdenite veins are exterior to the cupola where the stress regime is produced by a combination of regional and cupola pressurization effects. In contrast to the cupola, this zone is less ductile and transient shear stresses arise from cupola inflation and deflation. After a vein is mostly solidified, noncockscomb layers of quartz in the vein accommodate stress by rotating and deforming plastically, converting CL-euhedral textures to CL-mottled textures. Strain preferentially afflicts crystals in the centers of veins, sparing relatively strong layers of cockscomb quartz at the vein edges. The randomly oriented crystals lining the vein margin also recrystallize, leaving only the cockscomb layers intact

• (5)

  Local low-temperature recrystallization occurs as the pluton and its hydrothermal system cooled. Subgrain rotation and bulging recrystallization do not affect Ti concentrations in quartz

The lack of a CPO in recrystallized, CL-mottled vein interiors could be explained in two ways. One is that the initially random crystal configuration in vein interiors stayed random, i.e., that the majority of recrystallization occurred in an isotropic stress field or in a stress field where the magnitude of deviatoric stresses were small. This is our preferred hypothesis because the textures do not indicate large differential stresses. An alternative hypothesis is that grain boundary sliding in a dilatant, fluid-flushed system destroyed or prevented the formation of an extensive CPO.

We speculate that mottled edges of cockscomb CL-euhedral crystals (Figure 4) form because the lattices of adjacent crystals do not align during competitive grain growth, yielding irregular boundaries with a large surface free energy [58]. Impingement of one crystal into the next can also generate intracrystalline strain (the so-called force of crystallization). This phenomenon of growth boundary mismatch is caused by variation in relative growth rates of individual faces as the crystals impinged on one another [58]. Those boundaries then recrystallize during localized grain boundary migration, driven by surface free energy minimization or minimization of intragranular strain, or via dissolution-precipitation recrystallization. If recrystallization operates by grain boundary migration, then motion would be driven by strain within the grain being consumed which would explain the lobate grain boundary and apparently unidirectional motion of grain boundaries.

We consider three possible mechanisms for the formation of the ghost band texture (Figures 2(b) and 7(c)): (1) each blurred layer formed via precipitation from quartz-saturated grain boundary fluids during recrystallization, (2) pipe diffusion of initially CL-euhedral crystals blurred the once sharp boundaries between growth layers, and (3) grain boundary migration of initially CL-euhedral crystals blurred the once sharp boundaries between growth layers.

Because it is reasonable to assume that the hydrothermal vein-sealing quartz is initially CL-euhedral, we prefer the interpretation that ghost band textures form by way of Ti diffusion through quartz and subsequent blurring of euhedral growth bands. Although volume diffusion of Ti through quartz is too slow to blur the CL-euhedral bands [67], Ti diffusion through quartz can be mediated by grain boundary migration or through pipe diffusion. In the former case, Ti is redistributed through the quartz structure as grain boundaries sweep through. Fluid inclusion sizes and shapes are modified concurrently, resulting in the irregular appearance of fluid inclusions seen in recrystallized, CL-mottled quartz (Figure 3(b)) [35]. In the latter case, Ti preferentially diffuses along interconnected dislocations. This process is called “pipe diffusion” [82] and has been documented as blurring CL growth bands in zircon [83]. Because the ghost band texture is found in crystals that have equilibrium “foam” grain boundary geometries, we favor the interpretation that the texture forms as grain boundary migration facilitated Ti diffusion in an initially CL-euhedral crystal. Thus, the ghost band texture is intermediate between the two end-member textures of unrecrystallized CL-euhedral and completely recrystallized CL-mottled. We speculate that if deformation had continued, the ghost band texture would become CL-mottled texture as Ti continued to diffuse and homogenize in the grain.

Holness and Watt [84] and Piazolo et al. [39] reported ghost band textures in contact metamorphic quartz from the Ballachulish contact aureole. Both studies concluded that the texture of subparallel bands formed during grain boundary migration, wherein each bright band represents a single growth stage. In this framework, the grain boundary migrated in a direction perpendicular to the ghost bands. Holness and Watt [84] report that the country rock is CL-mottled but also note that it underwent regional metamorphism. It is possible that the ghost bands they document formed as quartz dissolved or melted (both studies report evidence of partial melting) due to heating in the contact aureole and then reprecipitated as oscillatory overgrowths on relict CL-mottled grains as the contact aureole cooled. Ghost band texture was also reported in a shallow porphyry deposit by Müller et al. [25], but the relict cores are CL-euhedral. In this case, it is apparent that the diffuse bands are concentric around CL-euhedral cores. If the ghost bands were quartz growth events happening during recrystallization, this would require that all of the grain boundaries migrated concentrically outwards for adjacent crystals, which is impossible because a grain boundary can only advance at the expense of a neighboring crystal.

CL-mottled textures are conspicuously absent from Main Stage veins imaged in this study and by Rusk et al. [10]. The maximum temperature of formation of Main Stage veins is ~350°C [15], so perhaps CL-mottled texture can only form during higher temperature recrystallization. The abundance of CL-chaotic texture in Main Stage quartz and its absence in pre-Main Stage quartz suggest that the mechanism of its formation is also a lower temperature process. It is plausible that CL-chaotic texture is formed by way of fast growth as sucrosic quartz crystals ripened and impinged on one another. CL-chaotic texture bears a striking resemblance to the plumose quartz texture identified in epithermal deposits, commonly alongside colloform quartz. Both plumose quartz and colloform quartz are thought to form as chalcedony converts to quartz [65, 85, 86]. Further work is required to understand the formation of the CL-chaotic texture.

This contribution shows how vein quartz strain and recrystallization modify primary growth textures, including those in quartz stockworks that are characteristic of the bottoms of porphyry Cu deposits. We find that ductile deformation at lithostatic pressures is common at the base of the deposit, providing a potential constraint for better understanding of the architecture of dissected or more poorly understood porphyry systems. Recrystallized barren quartz and quartz-molybdenite veins record the conditions of deformation. Our findings show that Ti-in-quartz concentrations can provide meaningful estimates of quartz growth and deformation temperatures, but textural observations are key to accurate interpretation.

The microprobe Ti measurements used to support the findings of this study are included within the supplementary information file, along with additional CL and hand sample images.

The authors have no conflicts of interest to declare.

This research was supported by the National Science Foundation grant no. EAR1524665. We thank John Dilles for thoughtful feedback. We are thankful to Montana Resources and the Washington Company for allowing access to the mine and samples. We are especially grateful to Steve Czehura and Amanda Griffith for their help, cooperation, and insight. Access to historic core was provided the Montana Bureau of Mines and Geology. We thank the staff of CAMCOR for their assistance with data collection, especially John Donovan and Julie Chouinard. We thank Kyle Eastmann for his assistance in sample acquisition. Mark Pearce was particularly helpful with respect to the presentation and interpretation of EBSD data and suggested we look into pipe diffusion. We thank Josephine Moore for her assistance with EBSD data collection.