Seeing Through Metamorphism: Geochemical and Isotopic Studies of Amphibolite Facies Mafic-Siliciclastic (Besshi-Type) Massive Sulfide Deposits at the Elizabeth Copper Mine, Vermont, USA

The geologic characteristics and origin of hydrothermally altered rocks associated with ancient volcanogenic massive sulfide (VMS) deposits have been investigated on both regional and local scales by numerous workers (e.g., Barrett and MacLean, 1999; Large et al., 2001; Piercey, 2009; Pilote et al., 2020). Most such studies have focused on geochemical and/or isotopic analyses of deposits and surrounding strata that have undergone variable degrees of post-ore metamorphism, from subgreenschist to granulite grade. Alteration zones spatially associated with VMS deposits are commonly attributed to premetamorphic hydrothermal processes causally linked to the sea-floor hydrothermal system that formed the sulfide deposits, i.e., prior to, during, and after—but still related to—sulfide mineralization. Strongly metamorphosed and intensely deformed alteration zones developed in basalts that have very high concentrations of Al, Si, Mg, Na, K, and B have not been comprehensively studied.

We present here new geologic, mineralogic, lithogeochemical, and stable isotope (δ¹⁸O, δ³⁴S) data on rocks and ores from the highly deformed and metamorphosed (middle amphibolite facies) Elizabeth copper deposit in the Vermont copper belt of east-central Vermont. This VMS deposit is one of several in the U.S. Appalachians hosted in predominantly clastic metasedimentary strata that have high contents of Na, K, and/or Al in wall rocks, including Ducktown in Tennessee, Ore Knob in North Carolina, and Gossan Lead in Virginia (Gair and Slack, 1984; Slack, 1993). Elizabeth is atypical among this group of deposits in possessing a greater diversity of Al-Si-Na-B-K enrichments in stratabound wall rocks, in addition to stratiform Fe-Cu-Zn-Mn mineralization. Our study is the first to use a full complement of field and laboratory data to constrain protolith compositions, elucidate the effects of premetamorphic subseafloor hydrothermal alteration, and present an integrated genetic model. We also apply these data to understand processes and environments in other ancient sediment-hosted VMS deposits where the nature and origin of chemically distinctive wall rocks may be enigmatic, and to modern, sediment-dominated, sea-floor hydrothermal systems in which such wall rocks are yet to be discovered.

Massive sulfide deposits of the Vermont copper belt are contained within predominantly metasedimentary rocks of the Connecticut Valley-Gaspé trough that extends ~1,500 km from southern Connecticut northward through western Massachusetts and eastern Vermont to southeastern Québec then eastward to the Gaspé Peninsula (e.g., Tremblay and Pinet, 2005). Diverse tectonic environments have been proposed for strata of this Silurian-Devonian sequence, including intercontinental extensional rift and back-arc settings, a foreland basin, and a transpressive basin (McWilliams et al., 2010; Perrot et al., 2018; Waldron et al., 2019). Rocks are mainly carbonaceous and calcareous metapelite with subordinate quartzite and quartzose marble. Stratigraphically extensive but volumetrically minor metabasaltic amphibolite occurs locally, chiefly in eastern Vermont (Fig. 1A, B). Based on U-Pb detrital zircon and igneous zircon dates, and evidence from scarce plant fossils, depositional ages for the metasedimentary strata of this extensive belt range from late Silurian to Early Devonian (McWilliams et al., 2010; Perrot et al., 2018; Karabinos et al., 2019).

The stratigraphically lowest major unit of the Connecticut Valley sequence is the Silurian-Devonian Waits River Formation that occurs mainly in the western part of the belt (Fig. 1A). This carbonate-rich unit consists mainly of calcareous metapelite, minor quartzose metalimestone and metadolostone, and sparse pelitic schist. Calcite marble is the predominant carbonate-rich rock at high metamorphic grades at and near the Elizabeth deposit.

The Standing Pond Volcanic Member of the Waits River Formation forms an elongate unit that extends discontinuously from northeastern to southeastern Vermont (Fig. 1B) to northwestern Massachusetts. This distinctive unit, of Early Devonian age (413.5 ± 0.11 Ma; Karabinos et al., 2019), occurs chiefly along the contact between the Waits River Formation and the overlying Gile Mountain Formation, but in the northern part of the copper belt is entirely within the upper part of the Waits River Formation. The thickness of the Standing Pond Volcanic Member varies widely, from only a few meters up to ~300 m. Most of the unit is fine-grained hornblende-plagioclase amphibolite, but in places it includes thin (<1-m) beds of quartzose hornblende schist, sulfidic metachert, and very fine grained quartz-spessartine rock termed coticule (Spry, 1990). Quartz-magnetite iron formation is rare, occurring locally at a few small sulfide deposits (Cookville and Pike Hill mines) as layers 5 to 20 cm thick. A lack of reported pillow structures in the Standing Pond Volcanic Member has led to the interpretation that this unit is mainly a metamorphosed mafic tuff (Hatch, 1987). Fine-grained metadiabase and rare metagabbro near the small Cookville mine (Fig. 1B; White and Eric, 1944) may represent feeder dikes for the Standing Pond Volcanic Member in that area.

The Gile Mountain Formation is stratigraphically above the Waits River Formation and occurs in both a synformal western facies and a spatially more extensive and structurally complex eastern facies (Fig. 1A). Bedding styles differ in that the younger western facies is characterized by rhythmically interbedded quartzite and pelitic schist, whereas the older eastern facies is dominated by pelitic schist with minor interbedded quartzite with only rare graded beds (Fisher and Karabinos, 1980; Hatch, 1988; McWilliams et al., 2010). Quartzite units are typically 10 to 30 cm thick. Amphibolite in thin beds (<10 cm) is rare in the western facies, but the stratigraphically lower eastern facies contains multiple thin to thick (up to ~60-m) beds of this lithology (Fig. 2). U-Pb geochronology of detrital zircons from quartzite beds of both the western and eastern facies suggests maximum depositional ages in the Early Devonian (McWilliams et al., 2010; Perrot et al., 2018; Karabinos et al., 2019).

Pelitic schists of the Gile Mountain Formation vary greatly in mineralogy. Where unaltered, most comprise a prograde assemblage chiefly of quartz, muscovite, plagioclase, and biotite, together with lesser garnet, biotite, calcite, and graphite. Trace constituents are ilmenite or rutile. Some outcrops, especially in one unit east of the Elizabeth mine, contain prominent graded beds (Rolph, 1982; Slack et al., 2001). Relatively uncommon units, including at the Elizabeth mine, are pelitic beds, <0.5 m thick, composed dominantly of muscovite, kyanite, plagioclase, and abundant graphite, with minor quartz and local garnet and staurolite. These thin aluminous beds are distinguished in the field by a dark gray color and by the presence of coarse (1- to 5-cm) euhedral crystals of kyanite.

Strata of the Connecticut Valley-Gaspé sequence have been intensely deformed by the Early Devonian Acadian orogeny (e.g., White and Jahns, 1950; Lyons, 1955; Fisher and Karabinos, 1980). Two major stages of deformation are recognized: an early pervasive isoclinal folding event and a later episode of regional arching that formed the Stafford and Pomfret structural domes (Fig. 1B). The earliest deformation, D₁, is recorded by abundant, small-scale isoclines and layer-parallel, axial-surface foliations that are both folded by more prominent, larger-scale mesoscopic folds. The second deformation, D₂, formed open to isoclinal folds coeval with the development of an axial-surface foliation that is highly variable in extent, as a function of rock type and structural locus. The final D₃ event produced the regional arching and doming and was associated with, or possibly driven by, localized intrusion of granitic bodies (Slack et al., 2001).

Both the Elizabeth and Ely VMS deposits are localized near the crests of major fold structures (synclines or anticlines) within a group of reclined, zig-zag folds in the eastern facies of the Gile Mountain Formation (White and Jahns, 1950). These folds, which are F₂ structures that developed during D₂ deformation (Offield and Slack, 1990), occur along the eastern flank of the D₃ regional arch and were tilted 10° to 30° by formation of the Strafford dome (Fig. 1B). The D₃ arching in places tightened the F₂ folds but did not develop a penetrative mesoscopic fabric. D₃ was responsible for producing most of the folds and some of the ductile thrust faults present at outcrop scale, including those that partly control the configuration of the massive sulfide layers in the district (Slack et al., 2001). The D₁ event developed fine-scale isoclinal folding, shearing, and thrusting that affected adjacent rock types very differently without producing structural repetition or interfingering relationships, especially within the massive sulfide bodies that generally remained as semicontinuous layers, although affected by extensive internal deformation.

The metamorphic history of eastern Vermont is complex and includes three prograde Acadian events and at least one retrograde event. Within the Connecticut Valley-Gaspé trough (Fig. 1A), regional metamorphic grades are broadly zoned from chlorite + biotite near the New Hampshire border, increasing westward to garnet and staurolite + kyanite, then decreasing to garnet within the Brownington syncline and farther west (Wolfe et al., 2021). Petrologic studies in the area of the Elizabeth deposit by Menard and Spear (1993, 1994) determined that M₁ metamorphism produced biotite-grade assemblages during an early nappe-stage deformation (D₁), and that M₂ developed garnet-grade assemblages during a second nappe-stage event (D₂). In their model, peak metamorphism in the area (M₃) coincided with D₃ formation of the Strafford dome, producing porphyroblastic kyanite and staurolite plus local garnet and hornblende, all of which overprint both S₁ and S₂ schistosities. The peak prograde metamorphic facies of the deposit is thus middle amphibolite. Based on ²⁰⁸Pb/²³²Th ages of monazite from the Gile Mountain Formation, this metamorphism produced kyanite during a late stage of the Acadian orogeny, in the Early Mississippian at 352.9 ± 8.9 Ma (Wing et al., 2003). Peak dome-stage D₃/M₃ conditions took place at 620° to 660°C (Spear, 2014). Retrograde metamorphism resulted in the local replacement of hornblende, garnet, and other M₁-M₃ minerals by chlorite ± biotite ± calcite ± K-feldspar ± muscovite (Menard and Spear, 1994).

Intrusive rocks of the study area include granitic plutons and lamprophyre dikes. The small granite exposures in the district (Fig. 1A, B) are Acadian bodies of probable Early Devonian age that intrude strata of the Waits River Formation. Small pegmatite and aplite dikes present in the core of the Strafford dome (not shown) cut tight to isoclinal F₁ folds in Waits River schists and are folded by F₂ folds and hence predate the D₃ deformation that produced the Strafford dome (Slack et al., 2001).

Lamprophyre occurs as dikes up to 2 m thick that cut metasedimentary rocks of the Gile Mountain Formation including wall rocks of the Elizabeth deposit (Howard, 1969). Rare in outcrop, the dikes are distinctive in drill core by having augite phenocrysts and large amygdules up to ~10 cm in diameter composed of prehnite, calcite, and laumontite. These late dikes are likely part of the regionally widespread group of Mesozoic mafic intrusions related to the White Mountain Plutonic-Volcanic Suite (e.g., McHone, 1992).

Samples were collected from outcrops, open pits, and drill cores. A small proportion of our samples (EZ prefixes) comes from underground workings studied by H.E. McKinstry and P.F. Howard in the 1950s and loaned to us by Harvard University. Several kilometers of drill core were logged for lithologic and structural information from the southern, central, and northern parts of the Elizabeth mine area, from which hundreds of representative samples were collected (DH prefixes).

Reflected- and transmitted-light microscopy of polished thin sections were used to determine mineralogy and paragenetic relationships, and to select areas for electron microscopy. Scanning electron microscopy (SEM) was done on carbon-coated thin sections in the Electron Microbeam Laboratory at the Geology, Energy & Minerals Science Center (United States Geological Survey [USGS]-Reston, VA), using a Hitachi SU-5000 FE-SEM instrument operated in high vacuum mode and fitted with a backscattered electron (BSE) detector and an Oxford Ultima 100-mm² EDS-SDD (energy dispersive spectroscope-silicon drift detector). Qualitative EDS analysis was used to aid in mineral identification. Typical operating conditions for imaging and analysis were a 15- to 20-KeV accelerating voltage, a ~4-nA beam current, and a 10-mm working distance. In some cases, these conditions were adjusted (e.g., 7 KeV, 25 KeV) to improve image resolution and/or reduce the interaction volume of generated X-rays. The mineralogy of selected samples was determined at the USGS in Reston by powder X-ray diffraction (XRD) using a PANalytical X’Pert Pro diffractometer with Cu-Kα radiation, followed by data reduction via PANalytical Highscore Plus software version 4.1 for pattern processing and semiquantitative mineralogical analysis.

Whole-rock analyses were done at both Activation Laboratories (Actlabs) in Ancaster, Ontario, and the USGS in Reston. Because of the large grain size of some samples (e.g., garnet up to 7 cm in diameter), initial preparation involved crushing rocks ~5× larger than the largest crystal. Analyses of most silicate and carbonate rocks for major, selected trace, and rare earth elements were determined by inductively coupled plasma-optical emission spectrometry (ICP-OES) on rock powders fused with lithium metaborate/tetraborate. This method ensures complete acid dissolution of resistate minerals such as zircon, monazite, xenotime, and chromite. Instrumental neutron activation analysis (INAA) was used for Br, As, Sb, Sc, Cr, and Au. Concentrations of Li, B, Ga, Cu, Ni, Co, Zn, Cd, Pb, Mo, Re, Te, and Se were acquired by high-resolution inductively coupled plasma-mass spectrometry (ICP-MS) following an aqua regia digestion of powders. Total carbon and sulfur were determined by LECO infrared analyzer and CO₂ by coulometry after digestion with 2N perchloric acid; graphitic carbon was calculated by difference of total carbon minus carbon in CO₂. Analyses were made on duplicate samples and on 8 to 12 standards. Precision and accuracy for concentrations ≥100× the minimum detection limit (MDL) generally were better than ±5% and in many cases, such as for major elements, were better than ±1%. For concentrations approximately 10× the MDL, precision and accuracy were about ±10 to 20%, depending on the method used. Details of the various analytical methods are available online at www.actlabs.com.

Some silicate and carbonate rocks were analyzed for major and trace elements and most REEs at the U.S. Geological Survey (USGS) using the methods described in Arbogast (1996). Results for most elements were obtained using inductively coupled plasma-atomic emission spectroscopy (ICP-AES) of powders fused with lithium metaborate/tetraborate prior to analysis to ensure complete acid dissolution of resistate minerals such as zircon, monazite, xenotime, chromite, and barite. Other elements including As, Co, Cr, Fe, La, Sb, Se, Th, U, and W were determined by INAA; Au, Hg, Te, and Tl were analyzed by atomic absorption spectroscopy (AAS).

Bulk analyses of massive sulfide samples for trace elements and REEs were done by Actlabs, the USGS, and X-Ray Assay Laboratories (XRAL) in Toronto. The Actlabs data were all acquired on fused powders (lithium metaborate/tetraborate) by inductively coupled plasma-mass spectrometry (ICP-MS), except for As, Br, Cr, Co, Ir, Sc, Se, and Sb that are acquired by INAA. Results for Au, Hg, and Te, done at the USGS, are acquired by AAS. Data for Tl were obtained using AAS at XRAL.

Sulfur and most oxygen isotope compositions were determined at the USGS in Reston. Sulfide minerals were handpicked under a binocular microscope and mixed-sulfide samples were drilled out of massive sulfide; estimated purity was >95%, in some cases with minor inclusions of quartz or other sulfides. Sulfur extraction was performed by direct combustion of sulfide minerals (1,050°C for pyrrhotite, chalcopyrite, and sphalerite; 1,100°C for pyrite). Sulfur from mixed-sulfide samples was obtained by sequential chemical extraction as described by Shanks and Seyfried (1987). H₂S released by these processes was precipitated as Ag₂S, which was subsequently combusted in the laboratory with Cu₂O at 1,050°C to produce SO₂, followed by purification via vacuum distillation, and then analysis using a Nuclide 6” 60° gas-source isotope ratio mass spectrometer. The δ³⁴S values are reported in standard per mil notation (‰) relative to Vienna-Canyon Diablo Troilite (V-CDT). Reproducibility and accuracy of the sulfur isotope data are estimated to be ±0.1‰, based on multiple analyses of NBS-123 sulfide (δ³⁴S = 17.1‰).

Oxygen isotopes were determined on powders of rocks and mineral separates. The same bulk-rock powders used for whole-rock chemistry were treated with HCl to remove carbonate minerals prior to δ¹⁸O analysis. Following carbonate removal, silicate oxygen was converted to CO₂ by the ClF₃ method of Borthwick and Harmon (1982) for all but two samples. Data were obtained with a Finnigan MAT 251 mass spectrometer. The values are reported in standard per mil notation (‰) relative to Vienna-Standard Mean Ocean Water (V-SMOW). Reproducibility and accuracy of the oxygen isotope data are estimated to be ±0.1‰, based on multiple analyses of NBS 28 (δ¹⁸O = 9.6‰). Two whole-rock samples (JS-82-72 and DH 972-13) were run for oxygen isotopes at the USGS in Denver using the method of Clayton and Mayeda (1963), involving reaction with BrF₅ overnight at high temperature to produce O₂ gas, which is converted to CO₂ by reaction with hot carbon. The CO₂ was then purified cryogenically and transferred to a Pyrex container for isotopic analysis using a Micromass Optima isotope ratio mass spectrometer. Mineral separates of quartz, biotite, plagioclase, hornblende, garnet, and magnetite were drilled out of samples and/or handpicked under a binocular microscope; estimated purity was >98%, with impurities consisting of other silicate minerals and sulfides. Oxygen isotope analyses of these mineral separates used the same methods as for the whole-rock powders, except that carbonate-bearing samples were first leached with HCl prior to δ¹⁸O analysis.

Volcanogenic massive sulfide deposits of the Vermont copper belt form stratabound and mainly stratiform lenses that contain more than 50 vol % sulfides. The deposits are generally tabular to sheetlike in geometry and on a regional scale occur in three stratigraphic units (Fig. 2). The Elizabeth and Ely deposits are within the Gile Mountain Formation, whereas the Pike Hill deposits occur in the Waits River Formation; the Cookville and Orange and Gove deposits, as well as several very small deposits and occurrences, are within or adjacent to the Standing Pond Volcanic Member of the Waits River Formation. Details on the Ely, Pike Hill, and other VMS deposits of the district can be found in White and Eric (1944), Offield et al. (1993), and Slack et al. (1993, 2001). All of these VMS deposits have been termed Besshi-type after the geologically similar Mesozoic deposits in Japan (e.g., Fox, 1984; Slack, 1993) and are equivalent to the mafic-siliciclastic VMS classification of Barrie and Hannington (1999) and the pelitic-mafic classification of Franklin et al. (2005) and Galley et al. (2007). The Elizabeth mine, largest in the district, produced 2.9 Mt of ore at an average grade of 1.8 wt % Cu, 0.5 wt % Zn, and 5.6 g/t Ag intermittently from 1830 to 1958, when the mine closed (Slack et al., 2001).

The sulfide orebodies have been described in detail by McKinstry and Mikkola (1954) and Howard (1959a, b, 1969). Production was from four stratiform orebodies covering a total strike length of 3.4 km. The main (No. 1) orebody was mined from the two open pits and from underground workings to the north (Fig. 3); the No. 2 orebody represents a faulted segment of this same zone. The No. 3 orebody was exploited only at the very northern end of the mine and forms a separate stratigraphic unit. Production from a fourth orebody at the South mine apparently came from an extension of the main (No. 1) orebody. The average thickness of the ore is reported to have been about 7 m, with a maximum of nearly 20 m (McKinstry and Mikkola, 1954); the thickest sections are associated with major drag folds, particularly in the No. 1 (north) pit, in the No. 2 (south) pit, and at the South mine (Howard, 1969). The underground northern end of the deposit had an average Cu grade of 4.9 wt %, in contrast to the south pit, where the average Cu grade was 1.1 wt %. The northern end was also richer in Zn, Pb, and Ag, relative to the south pit that had comparatively higher Mn, Mo, Ni, Co, and Au (Howard, 1959b). Lenses, pods, and boudins of quartz ± minor carbonate are common within the ore zone, and in places outline folds. Veins of quartz ± tourmaline ± muscovite ± biotite ± garnet ± calcite ± zoisite ± ilmenite ± pyrrhotite ± chalcopyrite ± rutile ± kyanite (Howard, 1959a) typically cut all folds and the S₂ schistosity.

The Elizabeth deposit is located on the east side of the Strafford dome (Figs. 1, 2). Spatially associated is a series of amphibolite lenses that occur stratigraphically above five lower amphibolite horizons (Fig. 2). Based on the mapping of Howard (1959a, 1969), Rolph (1982), and our own observations, the structure that contains the deposit can be traced northwestward around the dome, into the axial zone of one of the prominent F₂ zigzag folds, the Old City syncline (White and Jahns, 1950). At the south end of the deposit, the tabular main orebody defines a set of tight, second-order folds in the core of a larger syncline. From there northward for more than 3 km, the orebody also is tabular but mainly occupies the steep limb and crest of a single syncline-anticline pair. That structure, in part, was referred to by previous workers (McKinstry and Mikkola, 1954; Howard, 1969; Rolph, 1982) as the Elizabeth syncline, which we interpret as one of a series of second-order folds in the core of the Old City syncline (Slack et al., 2001).

A geologic plan map of the Elizabeth deposit (Fig. 3A) shows the close association of amphibole-rich rocks with all of the ore zones and of plagioclase-rich rocks with ore in the north pit. The longitudinal section (Fig. 3B) illustrates the north-plunging geometry of the ore zone. A simplified map of the south pit, prior to mining, delineates the main sulfide orebody and two prominent F₂ fold structures within hanging-wall strata (Fig. 4). Geologic cross sections of the south and north pits indicate that faults mark the boundary of massive sulfide with wall rocks (Fig. 5A, B), including structural removal of the footwall in the former section. The main F₂ fold is well developed at the north end of the north pit as a tight to nearly isoclinal structure (Fig. 5C), which strikes approximately N5°E, plunges gently (12°–15°) to the north, and has an axial surface that dips moderately steeply (about 65°) to the east. Farther north, underground level maps and cross sections based on mine faces and drill core data, made by geologists of the Vermont Copper Company in the 1950s, show the F₂ fold that contains the main orebody and enclosing wall rocks as a broad open structure (Figs. 5D, 6; see also Howard, 1969, plate 25). Near the Gile Mountain-Waits River contact northwest of the Elizabeth mine, this fold is also open and trends northwest, with steep axes that plunge down the dip of the axial surface (White and Jahns, 1950).

Several thrust faults have been inferred along the contacts of massive sulfide lenses. The most important is on the east side of the south pit, where the footwall to massive to semimassive pyrrhotite is in direct contact with unaltered beds of quartzite and intercalated pelitic schist. This footwall lithologic sequence contrasts with that exposed on the west side of this south pit and the north pit, where outcrops are coarse garnet-mica schist and calcareous coticule, respectively (Fig. 5A, B). Howard (1969) shows additional examples of inferred thrust faults that cut the massive sulfide ores and wall rocks. Detailed structural logs made during our study of drill cores (e.g., DH 1006 and 1008) document myriad small folds defined by deformed quartz veins and by the repetition of units over short intervals. Unresolved is the location of hanging-wall strata on the east side of the deposit (Figs. 4, 5), a problem that may be solved by the presence of a hidden thrust fault, which is not observed on the surface due to the presence of mine waste and soil cover.

The Elizabeth deposit is east of and stratigraphically above the Standing Pond Volcanic Member of the Waits River Formation and lower strata of the Gile Mountain Formation (Fig. 2). Rocks of the Standing Pond Member are characterized by fine-grained needles of hornblende in a plagioclase-rich matrix, locally with thin lenses composed of epidote ± quartz (Fig. 7A) and stratiform coticule beds generally <1 m thick at the stratigraphic top of the amphibolite. Overlying clastic metasedimentary rocks of the Gile Mountain Formation, stratigraphically below the Elizabeth deposit area, typically consist of interlayered pelitic schist and minor quartzite, in many places with abundant stratabound quartz veins (Fig. 7B).

Figure 8 shows a reconstructed lithostratigraphic column for the Elizabeth mine sequence based on outcrops in the open pit and data from numerous drill cores. Although the complexities of two or more episodes of shearing and thrust faulting preclude the delineation of a totally unambiguous sequence, some features suggest that the overall lithostratigraphy was not appreciably disrupted during Acadian deformation. First, several distinctive marker metasedimentary wall rocks (e.g., footwall quartzite unit; Fig. 7D) have been traced for more than 700 m along strike, from the middle of the south pit to the north end of the north pit. Second, as documented below, rocks that occur within our interpreted stratigraphic footwall are greatly depleted in Na relative to unaltered metabasalt, whereas those in the inferred hanging wall locally contain high Na concentrations; such patterns of footwall Na depletion and hanging-wall enrichment are common in VMS deposits (e.g., Barrett and MacLean, 1999; Piercey, 2009). Third, Mn-rich coticules occur in both the interpreted stratigraphic footwall and hanging wall, as in numerous VMS deposits worldwide (Spry, 1990; Spry et al., 2000; Dubé et al., 2007). Overall, we conclude that our reconstructed stratigraphy, or “pseudostratigraphy” (Marshall, 1990), is a valid approximation of the lithologic succession in the Elizabeth mine area for documenting and interpreting geochemical and isotopic features of the wall rocks and sulfide ores.

The Elizabeth mine sequence includes amphibolites and other rock types that generally occur within about 30 m stratigraphically of the orebody. This sequence corresponds to the hanging-wall and footwall amphibolite and the “westwall amphibolite” of McKinstry and Mikkola (1954) and here includes all intervening rock types. Rock types that enclose the mine sequence, to the east and west, are common within the surrounding Gile Mountain Formation, dominated by pelitic schist with minor quartzite. Note that in the following discussion, we restrict the term amphibolite to rocks having 50 vol % or more hornblende, subordinate plagioclase, and little or no quartz (i.e., true amphibolite); we do not follow the usage of McKinstry and Mikkola (1954), who classified amphibolite as containing as little as 10 vol % hornblende. It is important to note that most of the lithologically atypical rock units within the mine sequence, described below, have not been traced in adjacent drill cores for more than ~100 m, suggesting that these localized units are discontinuous along strike and are related to the sulfide deposits. The distinctive rock types studied here correspond in general to those produced during the first two stages of conformable rock alteration in the model of Howard (1959b, 1969).

Footwall rock types: The base of the Elizabeth mine sequence consists of medium- to coarse-grained, calcareous hornblende schist and lesser gneiss, together with several other rock types, including true amphibolite. Based on mapping in the area of the open pits and the work of Howard (1969), the thickness of this true amphibolite unit varies from <1 m to as much as 20 m. Hornblende ranges in size from needles a few millimeters in length to large crystals up to 10 cm long. Within fine-grained varieties, the hornblende needles commonly define two foliations, S₁ and S₂; the coarse crystals form porphyroblasts that cut S₁, and in some cases, S₂. Characteristic is hornblende schist or gneiss (Figs. 7C, 9A), in which euhedral hornblende crystals (~10–30 vol %) are in a matrix of plagioclase and lesser calcite, with minor clinozoisite, biotite, muscovite, or pyrrhotite, and sparse magnetite and rutile or ilmenite (Fig. 10A). Calcite also commonly occurs as large porphyroblasts that in places make up 30 vol % of the rock. Amphibolite is uncommon in this unit, forming layers of needle amphibolite <1 m thick (Fig. 9B) or rare thin (<30-cm) intervals of a different, fine-grained variety consisting of hornblende and prominent plagioclase phenocrysts accompanied by minor dolomite, in which albite replaces calcic plagioclase phenocrysts (Fig. 10B). Closely associated with the needle amphibolite in many places are layers <20 cm thick of coticule, in this setting comprising fine-grained, euhedral spessartine-rich garnet in a quartz ± calcite ± biotite ± hornblende ± pyrrhotite matrix (Fig. 9C). Reconnaissance electron microprobe (EMP) analyses indicate that the garnets contain a major almandine component in addition to spessartine. In many drill cores, coticules are in direct contact with needle amphibolite. Stratigraphically above the coticule units is a distinctive coarse garnet-mica schist that contains almandine-rich garnets up to 6 cm in diameter, with a matrix chiefly of plagioclase, muscovite, biotite, and local chlorite (Fig. 10C).

A thin overlying unit of disseminated to semimassive sulfide occurs locally along the east walls of the north and south pits. This unit may be an extension of the No. 3 orebody, described by Howard (1969), that was only mined underground from coordinates 14200N to 16700N. We have not attempted to trace this sulfide unit on the surface because of cover by mine waste and the effects of D₁ and D₂ thrust faulting that make it difficult to discriminate between the No. 3 and No. 1 orebodies.

Coticules up to ~1 m thick occur near the inferred base of the main orebody in the stratigraphic footwall within or closely associated with needle amphibolite and hornblende-rich rocks (Fig. 6). Characteristic are layers and laminations composed of fine-grained quartz and spessartine-rich garnet with minor apatite and pyrrhotite, local calcite and biotite, and traces of pyrrhotite and/or chalcopyrite (Fig. 7E). Contacts of these coticules with altered wall rocks are typically gradational, whereas contacts with relatively unaltered needle amphibolite are sharp. In outcrops on the west side of the north pit and in drill cores, the coticules contain up to 20 vol % calcite and/or biotite. Garnet is distinctive in thin section as small euhedral orange grains, typically <1 mm in diameter, containing inclusions of quartz, apatite, and calcite (Fig. 10D). Reconnaissance electron microprobe analyses of these garnets show up to 13.5 wt % MnO (Sp_30.4Al_50.6Gr_10.6Py_7.2Uv_0.1). Several occurrences of this fine-grained orange garnet have been found in outcrops up to 100 m west of the mine and in drill cores, with garnet forming abundant (~5–10 vol %) disseminated grains in otherwise typical clastic metasedimentary rocks of the Gile Mountain Formation. For comparison, garnets in pelitic schists of this formation distal from the Elizabeth deposit have low Mn contents, with X_Sp ~0.10 (Menard and Spear, 1994; this study).

Locally, beneath the main orebody is a massive dolomitic plagioclase rock (Fig. 7F, 9D). This unit, up to 5 m thick, is best exposed along the east wall at the south end of the north pit and is considered equivalent to the feldspathic rock mapped by White (1943) and White and Eric (1944) in the same area (Figs. 3, 6). Characteristic is a massive texture with a weakly expressed foliation. Major minerals are calcic plagioclase and porphyroblastic carbonate (Fe dolomite + lesser calcite), together with minor chlorite, pyrrhotite, and biotite (Fig. 10E). At the contact with massive sulfide, this feldspathic rock locally contains abundant (70–80 vol %) fine-grained clinozoisite in multiple seams and lenses <1 cm thick. Thin beds (<1 m thick) of calcite marble have an uncertain stratigraphic position.

An additional footwall rock type identified in one drill core (DH 1006) is a different plagioclase-rich unit of plagioclase + mica ± quartz ± garnet ± actinolite ± calcite ± sulfide (pyrrhotite ± chalcopyrite) schist (Figs. 9L, 10L). Megascopically, this assemblage superficially resembles those in many clastic metasedimentary rocks of the Gile Mountain Formation, but generally lacks quartz, although ~10 to ~40 vol % occurs in some samples as determined by XRD. A few samples from this unit have minor (<5 vol %) hornblende that forms randomly oriented blades. The unit occurs in multiple intervals up to ~3 m thick interlayered with calcareous amphibolite, and locally within pelitic schist (Fig. 6). An important feature is a distinctive signature of immobile trace element and REE concentrations (discussed below) that warrant designation as a separate rock type. The unit also occurs in the stratigraphic hanging wall in one area of the south pit, interlayered with calcareous amphibolite, but appears to be more common in the footwall part of the mine sequence.

Hanging-wall rock types: At the stratigraphic top of the orebody is a distinctive layered tourmalinite (Fig. 7G). This unit, varying in thickness up to nearly 2 m, is especially prominent between the two open pits and at the southern end of the south pit (Fig. 4). Principal minerals are quartz, brown tourmaline, coarse albite, and minor pyrrhotite, chalcopyrite, tremolite, and pale green muscovite (Figs. 9E, 10F). Electron microprobe analyses indicate that the tourmaline has a dravite composition (Taylor and Slack, 1984); the muscovite has up to 0.67 wt % Cr₂O₃ based on reconnaissance EMP data. A less-common facies of this unit consists of pale green muscovite and quartz with minor brown tourmaline and sparse chlorite, pyrrhotite, and chalcopyrite. Within this facies are rare lenses <10 cm thick composed of quartz, muscovite, and abundant (~30–50 vol %) fine-grained graphite (Fig. 7H).

Above the tourmalinite is the tremolite-phlogopite schist unit of McKinstry and Mikkola (1954). This unit is widespread in the walls of the open pits, in underground workings to the north (Howard, 1969), and in drill cores (this study), forming discontinuous lenses 1 to 5 m thick (Fig. 7I). Logging of cores indicates that the tremolite-phlogopite schist occurs chiefly on the stratigraphic hanging wall of the main orebody, but is also present locally in the ore zone and the footwall. In places, this rock type forms irregular dike-like bodies up to 0.5 m wide that cut both S₁ and S₂ schistosities. The tremolite-phlogopite schist therefore is not everywhere a true stratigraphic unit, although at least locally it clearly was deformed together with adjacent massive sulfide lenses (see Howard, 1969, fig. 17). In addition to tremolite (and commonly actinolite), this schist also typically contains brown tourmaline and thin seams of plagioclase together with lesser calcite, chlorite, pyrrhotite, and rutile (Figs. 9F, 10G). Wollastonite was reported by McKinstry and Mikkola (1954) but is not confirmed by our study.

The next higher unit in the mine sequence is a laminated plagioclase-rich rock with minor biotite and porphyroblastic garnet and hornblende (Fig. 7J). Thicknesses vary from 1 to 3 m. Characteristic are thin (1- to 3-mm) laminae of albite-oligoclase bordered by seams of biotite; the laminae can be traced through the garnet and hornblende porphyroblasts that overgrow the dominant S₂ schistosity (Fig. 9G, H). Where such porphyroblasts are absent, this rock is a laminated, plagioclase-rich granofels. The matrix locally contains biotite, muscovite, tourmaline, staurolite, pyrrhotite, apatite, ilmenite, and rutile (Fig. 10H). This unit is correlated with the base of the westwall amphibolite of McKinstry and Mikkola (1954) and is likely the altered amphibolite unit described by Howard (1959b, 1969).

The western wall of the north pit and northern to central parts of the south pit are made chiefly of a coarse garnet-mica schist (Figs. 4, 7K). This unit is equivalent to the main part of the westwall amphibolite as described in early studies and is visually distinctive. Contacts with amphibolite are gradational over ~0.5 to 1.0 m. The coarse garnet-mica schist varies in thickness from several meters to as much as 10 m and contains large, almandine-rich garnets up to 8 cm in diameter. Hornblende porphyroblasts up to 10 cm long occur in places; many are altered to an assemblage of biotite + chlorite ± carbonate. The matrix consists mainly of muscovite, biotite, and plagioclase with local epidote (Figs. 9I, 10I); the muscovite is bright apple green, especially in contact with garnet, due to small amounts of Cr (Howard, 1959b). Quartz is volumetrically sparse (<3 vol %), typically forming disseminated grains with plagioclase in thin, discontinuous laminae <1 mm wide and 3 to 5 mm long (Fig. 10J). Cordierite was reported in this unit by McKinstry and Mikkola (1954), but our petrographic work has not confirmed its presence. Within short (<10-cm) intervals in drill core, the matrix locally contains 10 to 20 vol % staurolite (Fig. 9J) accompanied by abundant white mica (margarite and minor paragonite based on SEM-EDS), together with sparse amounts of zoned, clear to dark purplish-blue corundum (Fig. 9K) and sillimanite replaced by muscovite (Fig. 10K). Accessory phases include carbonate, clinozoisite-epidote, and pyrrhotite. Also, in places within the garnet-mica schist are thin intervals of calcareous ± biotite amphibolite (Fig. 7L) and calcite-biotite coticule.

Approximately 0.5 km east of the mine is a single stratabound tourmalinite within clastic metasedimentary rocks of the Gile Mountain Formation (Fig. 2). This discontinuous unit, up to 1 m thick, is composed predominantly of tourmaline and quartz with locally abundant graphite and minor pyrrhotite. No other stratabound tourmalinites have been found in strata beyond local wall rocks of the Elizabeth deposit.

Postdeformational features in the mine area include the lamprophyre dikes described above and planar quartz-tourmaline veins. These veins typically dip steeply, cut all metamorphic fabrics, and lack sulfide minerals. The only timing constraint on the veins is a minimum age of Early Devonian (late to post-Acadian orogeny).

The most complete mineralogical description of the orebodies is that of McKinstry and Mikkola (1954). Pyrrhotite is the most abundant mineral, occurring both as massive sulfide and disseminations in many of the wall rocks. The chief ore mineral is chalcopyrite, which typically shows mutual intergrowths with pyrrhotite; it reportedly occurs in greater amounts in the northern part of the mine relative to the southern part. Pyrite is uncommon and generally forms distinctive cubes as much as 5 cm in diameter in massive pyrrhotite, or rarely massive aggregates. Sphalerite is a minor constituent characteristically disseminated throughout the ore. Atypical are lenses of nearly pure massive sphalerite as much as 0.3 m thick (McKinstry and Mikkola, 1954, p. 27), and a thin, sphalerite-rich marble ~5 cm thick. Other sulfides known in trace amounts are galena, molybdenite, cubanite, tetrahedrite-tennantite, and valleriite. The volumetrically most important gangue minerals are quartz, plagioclase, tourmaline, calcite, and rutile; tourmaline locally constitutes up to 20 vol % of pyrrhotite-rich massive sulfide (Taylor and Slack, 1984). Notably, the sulfide zones typically contain fragments or inclusions of folded and foliated wall rocks (White and Eric, 1944; Howard, 1969) that define a characteristic “durchbewegung” texture, which is a common feature in highly deformed massive sulfide deposits worldwide (e.g., Vokes, 1969; Marshall and Gilligan, 1989; Jansson et al., 2018; Matt et al., 2020).

Whole-rock geochemical data for all types of low-sulfide silicate and carbonate samples are available in Appendix 1, and in Slack et al. (2024a), that also includes metadata.

Unaltered amphibolites in the Vermont copper belt: Numerous mafic metaigneous rocks from the Standing Pond Volcanic Member of the Waits River Formation and the Gile Mountain Formation were analyzed for a complete suite of elements. Representative analyses are presented in Table A1 and Appendix 1. Samples of the Standing Pond Member come from outcrops; those from the Gile Mountain Formation are from both outcrops and drill cores at the Elizabeth mine. One amphibolite sample each from the Ely and Pike Hill mines was also analyzed. Several samples of metadiabase and one fine-grained metagabbro, and one sample of epidote-rich amphibolite, come from outcrops of the Standing Pond Member. With rare exception, all of these samples are unmineralized based on very low contents of Cu (≤ 5 ppm), Zn (<100 ppm), and total S (<0.01 wt %).

Petrologic discrimination diagrams using immobile trace elements indicate tholeiitic (Y vs. Zr) and non-arc tholeiitic (Cr vs. Ti) compositions (Fig. 11A, B). Other binary plots (Ta/Yb vs. Zr/Yb; Ta/Yb vs. Th/Yb) and ternary plots (Zr-[Ti/100]-[Y × 3]; Th-[Tb × 3]-[Ta × 2]) are consistent with protoliths of normal mid-ocean ridge basalt (N-MORB) for most samples, although some outliers exist (Fig. 11C-F). No analyses in our database indicate compositions similar to those of ocean island basalts or continental tholeiites.

Ternary (Fig. 12) and binary (Fig. 13) plots compare whole-rock data for altered wall rocks from the Elizabeth deposit to unaltered metabasalt (including metadiabase and metagabbro) from the Standing Pond Volcanic Member, to relatively unaltered metabasalt (amphibolite) from the Elizabeth deposit, and to unaltered clastic and chemical metasedimentary rocks from the Gile Mountain Formation.

Clastic and chemical metasedimentary rocks of the Gile Mountain Formation: Bulk analyses of unaltered clastic metasedimentary rocks were obtained as a baseline for evaluating geochemical changes in Elizabeth wall rocks during VMS mineralization and later metamorphic fluid flow. Assignment here of an unaltered status to these samples is based chiefly on the absence of veins, sulfides, or tourmaline. This screening protocol yields 17 unaltered samples in three Elizabeth drill cores hundreds of meters from known sulfide accumulations, and 12 samples in regional outcrops >1 km from the deposit (Table A1; App. 1).

On ternary plots, the whole-rock data mostly define restricted fields. An Fe-Al-Mg plot shows relatively low proportions of Mg (Fig. 12A). Variable enrichment in Ca is evident in a Ca-Al-Mg plot (Fig. 12B), the greater proportions of Ca mainly reflecting minor to abundant Mg-poor calcite; these calcite-rich samples correspond to the calcic pelitic schists described by Menard and Spear (1993, 1994). An (Na × 2)-Al-(Fe + Mg) plot (Fig. 12C) documents significant enrichment of Na and depletion of Fe and Mg for most of the quartz-albite-tourmaline rocks, relative to the bulk of data for the other altered wall rocks of the mine sequence. Major enrichment in Si is shown on an Si-Al-(Fe + Mg) plot (Fig. 12D) by all of the tourmaline-rich rocks except for one muscovite-rich sample. Figure 12E illustrates a range of proportions of K on a (K × 2)-Al-(Fe + Mg) plot. A Ca-(Mn x 5)-(Fe + Mg) plot (Fig. 12F) indicates that four samples of visually unaltered clastic metasedimentary rock from the mine area proportionately have anomalously high Mn contents, like the coticules, relative to the main compositional field for this rock type.

Altered mafic metavolcanic and related wall rocks at the Elizabeth deposit: Whole-rock analyses of 54 samples of variably altered mafic metavolcanic rocks indicate diverse major element compositions, including high concentrations of Al, Mg, Fe, or Ca in many samples and high Si or Mn in several others (App. Table A2; App. 1). Results for different rock types generally cluster in compositionally different fields (Fig. 12), which relative to unaltered metabasalts of the Elizabeth deposit and Standing Pond Volcanic Member are (1) calcareous amphibolite (hornblende + plagioclase + calcite), which is proportionally enriched in Ca; (2) laminated albite + muscovite + garnet + hornblende ± biotite rock that is relatively enriched in Al; (3) carbonate + quartz ± white mica ± sulfide schist, which has higher Ca; (4) biotite + quartz + plagioclase schist that is relatively enriched in Al; (5) plagioclase + mica ± garnet ± quartz ± calcite ± sulfide schist, which also is very aluminous (up to 25.4 wt % Al₂O₃) and in some samples has high Fe or Mn; (6) white mica + staurolite ± garnet ± sillimanite ± corundum schist that is greatly enriched in Al (up to 32.5 wt % Al₂O₃); (7) coarse garnet + white mica ± biotite schist, which is similarly enriched in Al and depleted in Mg and Ca; (8) tremolite-actinolite + phlogopite ± tourmaline schist that is enriched in Mg but depleted in Ca; and (9) tourmaline-rich rocks that are enriched in Al, Na, Si, and B relative to the other rock types of the mine sequence (Fig. 12A-D; 0.3–3.2 wt % B₂O₃). Compared to average N-MORB, most of these wall rocks have proportionately lower Mg (Fig. 12A), higher Na (Fig. 12C), and higher K (Fig. 12E).

Whole-rock data for relatively immobile components (Al₂O₃, TiO₂, Zr, Cr, Th, Sc) and related ratios (Sc/Th, Cr/Zr) illustrate contrasts among samples of Elizabeth mine wall rocks, relatively unaltered metabasalt and metadiabase of the Standing Pond Volcanic Member, and unaltered clastic metasedimentary rocks of the Gile Mountain Formation (Fig. 13). The wall rocks are characterized by lower Zr/Al₂O₃ ratios (Fig. 13A), mostly lower TiO₂ contents (Fig. 13B), higher Cr for a given TiO₂ content (Fig. 13C), uniformly lower Th like the samples of the Standing Pond Volcanic Member (Fig. 13D), and generally higher Sc/Th and Cr/Zr ratios (Fig. 13E). In contrast, a plot of the alteration index [100 × (K₂O + MgO)/(K₂O + MgO + Na₂O + CaO)] vs. the chlorite-carbonate-pyrite index [100 × (MgO + FeO)/(MgO + FeO + Na₂O + K₂O)] shows considerable scatter, including trends for samples of the tremolite-phlogopite schist and quartz-albite-tourmaline rocks toward chlorite + pyrite and albite end members, respectively (Fig. 13F). In this study, metabasalt sample DH 972-13 is considered the least altered among all metabasaltic amphibolites analyzed from the mine area. This assignment is based on broad similarity in whole-rock composition to average MORB (White and Klein, 2014), in particular the concentrations of MgO, CaO, Na₂O, and K₂O. Texturally, this sample uniquely retains plagioclase phenocrysts (Fig. 10B) and has trace element compositions similar to that of unaltered N-MORB (Figs. 11, 12). Relative to this least altered metabasalt, most altered metabasaltic samples in the mine sequence are enriched in Al and Na and depleted in Mg, except for the carbonate + quartz ± white mica ± sulfide schist and the tremolite-actinolite + phlogopite ± plagioclase ± tourmaline schist that are relatively enriched in Ca and Mg, respectively.

Elizabeth massive sulfides and chemical metasedimentary rocks: Whole-rock analyses of nine samples of representative massive sulfide from the Elizabeth deposit are listed in Appendix Table A3. Analyses of other massive sulfides from the Ely, Pike Hill, and Gove deposits are presented in Slack et al. (2001). All Elizabeth samples have high Fe and generally high to locally very high Cu and Zn contents (up to 20.6 and 26.2 wt %, respectively). Also present in some samples are appreciable Ag (up to 99 ppm), Au (up to 0.85 ppm), Cd (up to 1,500 ppm), Co (up to 786 ppm), Mn (up to 5,600 ppm), Mo (up to 420 ppm), and Se (up to 106 ppm). Low concentrations are evident for As, Ba, Bi, Hg, Ni, Pb, Sb, Sn, W, Te, In, and Tl in nearly all analyzed samples, compared to typical VMS deposits worldwide (e.g., Monecke et al., 2016). One exception is sphalerite-rich sample EZ-1023 that contains 15.7 ppm In.

Results are also presented for chemical metasedimentary rocks from the Elizabeth mine area and the Standing Pond Member of the Waits River Formation. Included in this category are 10 samples of fine-grained spessartine + quartz ± biotite ± calcite ± sulfide rock (coticule). Whole-rock analyses show Fe-rich compositions with uniformly low proportions of Mg and mostly low Ca, with consistently high proportions of Mn (Fig. 12A, B, F). With one exception, all coticule samples have higher Al₂O₃/TiO₂ ratios than the field of unaltered clastic metasedimentary rocks of the Gile Mountain Formation (Fig. 13A).

Clastic and chemical metasedimentary rocks: Shale-normalized (SN) REE + Y data for the clastic metasedimentary rocks show relatively flat patterns with abundances of ~0.6× to 1.6× PAAS (average post-Archean Australian shale) for 12 samples from regional outcrops and 17 samples in drill core from the Elizabeth mine (Fig. 14A, B). The range of PAAS-normalized La_SN/Yb_SN ratios is similar for both sample groups (0.65–1.14 and 0.63–1.08, respectively). Although no significant Ce anomalies exist in the regional samples or those from the Elizabeth mine (Ce/Ce* = 0.95–1.04, excluding 0.82 for one sample), all but one sample display negative Eu anomalies including a low Eu/Eu* value of 0.18 for a regional sample of graphitic chlorite-mica schist collected from outcrop (sample LW-84-18); other regional samples have moderate negative values of 0.50 to 0.77. The one exception in this group is a graphitic metapelite with a moderate positive Ce anomaly of 1.68. Note that in this study, the Eu/Eu* values for all samples are calculated on a chondrite-normalized (CN)basis, because this method provides a better measure of Eu anomalies than shale normalization (Sindol et al., 2022).

Unaltered amphibolites in the Vermont copper belt: Chondrite-normalized data for REE + Y show abundances of 5× to 30× chondrite, with La_CN/Sm_CN and La_CN/Yb_CN ratios of 0.33 to 0.97 and 0.41 to 1.44, respectively, both of which largely mirror that of average N-MORB (Fig. 14C). Based on major element and immobile trace element contents (Ti, Cr, Sc), fine-grained amphibolite sample DH 972-13 with relict plagioclase phenocrysts (Fig. 10B) is considered the least altered metabasalt in the Elizabeth mine sequence (Figs. 12, 13).

Altered mafic metavolcanic and related wall rocks at the Elizabeth mine: REE data for these compositionally diverse rock types are broadly similar in terms of abundance levels (3× to 20× chondrite) and La_CN/Sm_CN and La_CN/Yb_CN ratios of 0.29 to 0.73 and 0.33 to 1.33, respectively (Fig. 14D). The range of these ratios is similar to that shown by minimally altered metabasalts (hornblende-plagioclase “needle amphibolite”) in the Elizabeth mine sequence and those of the Standing Pond Volcanic Member (Fig. 14C). No significant chondrite-normalized Ce anomalies are evident (Ce/Ce* = 0.88–1.12), but some samples have small negative or small to moderate positive chondrite-normalized Eu anomalies (Eu/Eu* = 0.62–0.79 and 1.25–2.02, respectively). Data for seven samples from Elizabeth drill core (plagioclase + mica ± quartz ± garnet ± actinolite ± calcite ± sulfide schist) that visually resemble surrounding clastic metasediment are compositionally very different. On a PAAS-normalized plot, these seven samples differ greatly from the unaltered clastic metasedimentary rocks at the Elizabeth deposit in having very low contents of light REEs (LREEs) and very low La_SN/Sm_SN ratios of 0.08 to 0.22, uniformly small negative Ce anomalies (Ce/Ce* = 0.75–0.89), and a wide range of Eu/Eu* values from 0.31 to 1.08, not including one sample (DH 1006-71) with an extremely large negative Eu anomaly of 0.05 (Fig. 14B). However, on a chondrite-normalized plot (Fig. 14D), no depletion of LREEs is evident relative to the least altered and highly altered metabasalts. Unlike the relatively low La_CN/Sm_CN ratios that characterize the altered metabasalts, three samples of tourmaline-rich rock (quartz + albite + tourmaline ± sulfide) show different patterns with La_CN/Sm_CN ratios of 0.81 to 1.52, including low LREE abundances of 1.2 to 1.8× chondrite in one sample.

Elizabeth mine coticules, calcite marble, and massive sulfides: PAAS-normalized REE data for coticules and one calcite marble are presented in Figure 14E. Abundances for the seven coticules vary from 1 to 5× PAAS with La_SN/Yb_SN ratios of 0.49 to 0.77, uniformly small negative Ce anomalies (Ce/Ce* = 0.67–0.71), and no significant Eu or Y anomalies. All of these negative Ce anomalies are valid, using the shale-normalized Pr/Pr* v Ce/Ce* discriminant plot of Bau and Dulski (1996). Data for the calcite marble from drill core, containing minor quartz and muscovite, indicate REE abundances of 0.2 to 0.7× PAAS but without calculated Ce, Eu, or Y anomalies (App. 1).

PAAS-normalized REE data obtained for the massive sulfide samples (Fig. 14F) display abundances of 0.005 to 0.1× La_SN/Yb_SN ratios of 0.28 to 0.60. Samples having varying proportions of pyrrhotite, chalcopyrite, and pyrite, with minor sphalerite in some cases, are characterized by small to moderate negative Ce anomalies (Ce/Ce* = 0.64–0.96), but on the basis of PAAS-normalized Pr/Pr* values of less than 1.05, are false and instead reflect positive La anomalies (Bau and Dulski, 1996); also shown are small negative or positive Eu anomalies (Eu/Eu* = 0.54–1.25). Y/Ho ratios range from 20.0 to 28.0, excluding sample EZ-1023 with a higher ratio of 37.6. Sample EZ-1023, sphalerite- and calcite-rich, has relatively high contents of heavy rare earth elements (HREEs), a very low PAAS-normalized La/Yb ratio (0.01), and a moderate positive Y anomaly (Y_SN/Ho_SN = 1.39); this sample also displays a true moderate negative Ce anomaly (Ce/Ce* = 0.57) and a small positive Eu anomaly (Eu/Eu* = 1.25). Pyrrhotite-rich sample EZ-1041 is also distinctive in showing a flat REE pattern (La_SN/Yb_SN = 1.00), a true small negative Ce anomaly (Ce/Ce* = 0.69), a large positive Eu anomaly (Eu/Eu* = 3.98), and a large negative Y anomaly (Y_SN/Ho_SN = 0.11). Mineralogical residence of the minor contents of REEs in these massive sulfide samples is unknown, but may be chiefly in iron sulfides (Zeng et al., 2015), with lesser amounts in metamorphic magnetite after hydrothermal iron oxyhydroxides (cf. Mills and Elderfield, 1995), in hydrothermal or diagenetic apatite, and in diagenetic or detrital REE-rich minerals like monazite, allanite, and xenotime.

Sulfur isotope values of 63 sulfide mineral separates (pyrrhotite, chalcopyrite, sphalerite, pyrite) from massive and semimassive sulfide of the Elizabeth deposit range from 4.3 to 9.3‰ (Fig. 15; App. Table A4). Disseminated pyrrhotite and chalcopyrite (n = 18) in wall rocks at the Elizabeth and South deposits display a narrower range of 6.7 to 8.8‰. For comparison, δ³⁴S values for pyrrhotite, chalcopyrite, and sphalerite from massive and semimassive sulfide from the Ely (n = 18), Pike Hill (n = 30), Gove (n = 14), and Cookville (n = 4) deposits range from 6.0 to 9.1, 1.0 to 4.9, 2.0 to 4.2, and 2.2 to 3.7‰, respectively. Disseminated pyrrhotite (n = 9) and pyrite (n = 2) in amphibolite from the Standing Pond Volcanic Member have mostly lower sulfur isotope values of –8.8 to 2.8‰. Calculations of possible temperatures based on coexisting mineral pairs (pyrite-pyrrhotite, pyrrhotite-chalcopyrite, and sphalerite-chalcopyrite; App. Table A4) mostly yield disequilibrium values based on sulfur isotope fractionation factors compiled and analyzed in Seal (2006). However, a handful of mineral pairs give temperatures (°C) that may represent hydrothermal depositional conditions (238, 269, 311, 355, 372); one sample pair gives a temperature of 553°C, which might be interpreted as metamorphic reequilibration. In contrast, most of the sample pairs either give extremely high temperatures, indicating very little fractionation between mineral pairs, or show reversals of the expected fractionation. In general, most of the sulfide minerals are not in equilibrium and must have formed at different times or temperatures or from different fluids.

Oxygen isotope analyses of 74 whole rocks from the Elizabeth deposit and surrounding units including the Gile Mountain Formation, Waits River Formation, and Standing Pond Volcanic Member of the Waits River Formation are listed in Appendix Table A5 and plotted in Figure 16. Among Elizabeth samples, the δ¹⁸O value of one “needle” amphibolite is 10.7‰; values for five coarse hornblende-plagioclase ± calcite amphibolites are 10.7 to 11.5‰. Plagioclase-phyric amphibolite DH 972-13 has δ¹⁸O of 11.1‰. Various other types of altered wall rocks at the Elizabeth mine (n = 15) have oxygen isotope values of 10.7 to 13.4‰. Five coticules (quartz + spessartine garnet ± calcite ± biotite ± pyrrhotite) range in δ¹⁸O from 10.7 to 11.9‰. Three siliceous tourmaline-rich rocks (quartz + tourmaline + albite + sulfide) have values of 12.9 to 14.3‰. Results for 22 unaltered clastic metasedimentary rocks in drill cores and from open pits vary from 10.9 to 13.4‰. In contrast, drill core intervals of the plagioclase-rich rocks (n = 6) that appear mineralogically and texturally similar to unaltered clastic metasedimentary host rocks range in δ¹⁸O from 11.0 to 11.8‰. For comparison, whole-rock data for regional samples from this study area display the following δ¹⁸O values: (1) unaltered clastic metasedimentary rocks of the Gile Mountain Formation (n = 13), 10.4 to 13.7‰; (2) “needle” amphibolites of the Standing Pond Volcanic Member (n = 4), 7.9 to 9.1‰; (3) quartz-spessartine coticule rocks from the Standing Pond Member (n = 1), 12.5‰; and (4) metadolostone of the Waits River Formation (n = 5), 16.1 to 21.7‰. The average whole-rock δ¹⁸O value for the altered plagioclase-rich rocks at the Elizabeth deposit (11.4 ± 0.35‰) is slightly lower than that for the clastic metasedimentary rocks sampled from regional outcrops (12.6 ± 0.98‰).

Oxygen isotope data were also obtained on mineral separates of quartz, biotite, garnet, plagioclase, amphibole, and magnetite from a variety of rock types at the Elizabeth deposit and vicinity (App. Table A5). Most analyses are of quartz (n = 12), which for Elizabeth samples show δ¹⁸O values from 11.9‰ (laminated albite-muscovite-hornblende-garnet rock) to 15.8‰ (tremolite-phlogopite schist); for comparison, quartz separates from six samples of unaltered clastic metasedimentary rocks of the Gile Mountain Formation, from drill cores and open pits, have values of 12.8 to 14.1‰. Biotite in one laminated albite-muscovite-hornblende-garnet rock has δ¹⁸O of 7.6‰, and three pelitic and one biotite-rich schist in the mine sequence have δ¹⁸O of 8.3 to 9.1‰. Analyses of garnet (n = 2) show δ¹⁸O values of 13.6‰ (pelitic schist of Gile Mountain Formation) and 14.3‰ (calcareous coticule). Biotite-rich quartzite and tremolite-phlogopite-albite schist in the mine sequence have hornblende and tremolite δ¹⁸O values of 9.4 and 10.5‰, respectively; two calcareous amphibolites in the mine sequence have δ¹⁸O of 9.9 and 10.0‰. One plagioclase separate from a quartz-albite-tourmaline rock has a δ¹⁸O value of 13.3‰.

Oxygen isotope data are also presented here for coexisting quartz and magnetite in quartz-magnetite iron formation from the Cookville and Pike Hill mines in the northern part of the Vermont copper belt (Fig. 1). The metamorphic zone in these areas is kyanite, like that to the south at and near the Elizabeth and Ely mines. Mineral separates of one dump sample from the Cookville mine yield δ¹⁸O values for quartz and magnetite of 13.97 and 4.60‰, respectively, and in a second dump sample 13.72 and 6.62‰, respectively. Drill core from the Pike Hill mine has δ¹⁸O values of 16.15‰ for quartz and 7.58‰ for coexisting magnetite.

The very thick succession of predominantly clastic and carbonate-rich metasedimentary rocks of late Silurian to Early Devonian age within the Connecticut Valley-Gaspé trough (Fig. 1A) has been interpreted as rift facies deposits that accumulated in a gradually deepening basin (e.g., Tremblay and Pinet, 2005; McWilliams et al., 2010). Regionally extensive metabasalts within the basin, including amphibolites of the Standing Pond Volcanic Member of the Waits River Formation and correlatives, are compositionally diverse in that exposures in southeastern Vermont and southeastern Québec are subalkaline to alkaline, within-plate tholeiites (Hepburn, 1991; Tremblay and Pinet, 2005), whereas those in the Vermont copper belt in east-central Vermont are subalkaline tholeiites, predominantly of N-MORB affinity, lacking a within-plate petrologic signature (Slack, 1994; Fig. 11). Considered overall, on a regional scale, these compositions document tholeiitic mafic volcanism in an extensional rift environment associated with two principal types of basaltic magmatism. Our whole-rock geochemical data (Fig. 13A) further indicate that the mafic metaigneous rocks of the Standing Pond Volcanic Member, including predominant amphibolite together with metadiabase and metagabbro at the Cookville mine, record various degrees of Fe-Ti fractionation (e.g., Clague and Bunch, 1976). The comparatively low-Ti metabasalts of the Elizabeth mine sequence may reflect a residual magma source following earlier eruption of the stratigraphically lower high-Ti basalts of the Standing Pond Volcanic Member. However, other explanations for these high-Ti basalts are possible, including the nature and extent of partial melting of the mantle source (Hou et al., 2011). Occurrence of these basalts on and near the boundary of the Waits River and Gile Mountain formations (Figs. 1, 2) also suggests that this mafic metavolcanic unit marks a key paleotectonic transition from an intercontinental back-arc basin to a foreland basin (see Tremblay and Pinet, 2005; McWilliams et al., 2010; Perrot et al., 2018). Tectonic discriminant diagrams of Bhatia and Crook (1986), using the immobile trace elements La, Th, Sc, and Zr (not shown), yield data for the Waits River Formation that plot mainly in the passive margin field, whereas those for the Gile Mountain Formation fall nearly exclusively in the continental arc field. These results suggest that the older sediments of the Waits River Formation were sourced from passive margin strata to the west, and the younger Gile Mountain sediments came mainly from one or more accreted volcanoplutonic arcs to the east (see van Staal et al., 2021).

The geologic setting and distinctive grain size of the coarse garnet-mica schist in the Vermont copper belt have fostered studies that proposed an origin involving regional metamorphic fluid flow within a sedimentary protolith. Critical to these arguments are locations of the schist units. The distribution shown in the early mapping of Doll (1944) was used by Barnett and Chamberlain (1991), Skelton (1996), and Vyhnal and Chamberlain (1996) as the basis for detailed isotopic and petrologic studies, on the premise that the coarse garnet-mica schist units are confined to the hinges of F₂ folds. However, this premise is incorrect because this schist also occurs on the limbs of the folds west of the Ely mine (Fig. 2; White and Eric, 1944; White and Jahns, 1950) and on the east side of the Strafford dome over a strike length of 4.8 km, from the South mine to north of the Elizabeth mine (White and Eric, 1944, plate 1). Equally important is that nearly all of the garnet-mica schist units are associated with or occur along strike from VMS deposits or prospects, including the Elizabeth, Orange and Gove, and Walker deposits. Doll (1944) also described coarse garnet-mica schist with abundant pyrrhotite and chalcopyrite on McMaster Hill, where no sulfide prospects are known.

Regionally the coarse garnet-mica schist averages ~10 m in thickness and, with rare exception, occurs adjacent to amphibolite in the Standing Pond Volcanic Member (Doll, 1944) and the Gile Mountain Formation at the Elizabeth mine (Fig. 2). In both settings, the coarse garnet-mica schist is typically gradational into amphibolite and contains large (5- to 8-cm) garnets in a mica-rich groundmass. Some occurrences in the Standing Pond Member have minor amounts of quartz (Barnett and Chamberlain, 1991) in contrast to the Elizabeth deposit, where quartz is modally insignificant (<0.5 vol %) as sparse disseminated grains in thin laminae (Fig. 10J). In both settings, the coarse garnet-mica schist is highly aluminous, including at Elizabeth (22.1–25.4 wt % Al₂O₃; App.1). No whole-rock analyses of samples from the Standing Pond Member settings have been published, and none was analyzed in this study. However, based on literature descriptions and our optical and SEM petrography, all of the coarse garnet-mica schist units are very similar texturally and mineralogically, with the exception of the quartz present in some of the Standing Pond Member occurrences.

A fundamental issue in evaluating the origin of the coarse garnet-mica schist is the nature of the protolith. Previous studies that investigated this rock type all inferred a sedimentary precursor within the upper part of the Waits River Formation (Barnett and Chamberlain, 1991; Skelton, 1996; Vyhnal and Chamberlain, 1996). None of these studies reported whole-rock geochemical data, but it is clear from our analyses that the coarse garnet-mica schist in the Elizabeth mine sequence (Fig. 7K) is a highly altered tholeiitic metabasalt and not a clastic metasedimentary rock, based on relationships among immobile trace elements Sc, Cr, Zr, and Th, especially the uniformly high Sc and Cr contents but low Zr and Th contents (and high Sc/Th ratios) relative to data fields for metasedimentary samples of the Gile Mountain Formation (Fig. 13B-E). Given similarities in geologic setting, texture, and mineralogy, we suggest that the stratigraphically lower occurrences of coarse garnet-mica schist along the Standing Pond Member contact also have a basaltic protolith. This interpretation has the advantage of more readily explaining the oxygen isotope data of Barnett and Chamberlain (1991) and Vyhnal and Chamberlain (1996), who reported whole-rock δ¹⁸O values that require pervasive alteration by a low-¹⁸O and high-temperature (>500°C) fluid in order to account for a decrease of 3‰ adjacent to the Standing Pond Volcanic Member, relative to a postulated carbonate-rich sedimentary protolith within the Waits River Formation (δ¹⁸O = 16.1–21.7‰ in the study area; App. Table A5). A larger-scale study by Stern et al. (1992) found similar ¹⁸O depletions of 2 to 4‰ in carbonate adjacent to the Standing Pond Member contact (and regionally next to Waits River Formation contacts), attributing this pattern to synmetamorphic fluid flow focused along the axes of major antiforms and the Strafford and Pomfret domes. However, if the protolith is instead tholeiitic basalt, as we document here for occurrences at the Elizabeth deposit, then by analogy, the difference in oxygen isotope values for the needle amphibolite samples from the Standing Pond Member (δ¹⁸O 7.9–9.1‰; App. Table A5) is an increase of ~2 to 3‰ relative to average MORB (5.7‰; White and Klein, 2014). We attribute this increase to albitization (spilitization) by sea-floor hydrothermal fluids based on data for modern and ancient spilitized basalts (Mengel and Hoefs, 1990; Hernandez-Uribe et al., 2020). This model is consistent with our whole-rock δ¹⁸O values of 11.4 and 12.6‰ for two coarse garnet-mica schist samples from the Elizabeth deposit, suggesting formation of the schist by VMS-related hydrothermal alteration (Beaudoin et al., 2014), as discussed below. Support for this interpretation comes from the study of Vyhnal and Chamberlain (1996), which emphasized that the nappe-stage garnets in the coarse garnet-mica schist overprint a local 3‰ oxygen isotope gradient at the Waits River-Standing Pond contact, and that this gradient is folded by the nappes. Their proposed mechanism involving premetamorphic, layer-parallel fluid flow during diagenesis is consistent with our model, except that data for the Elizabeth deposit samples presented here suggest that the premetamorphic fluid flow event was part of a sea-floor hydrothermal alteration (~150°–250°C) and not a lower-temperature (<150°C) diagenetic process. If our VMS-related hydrothermal origin for a basaltic protolith of the coarse garnet-mica schist is correct, it suggests that exploration potential for undiscovered VMS deposits may exist within and near large outcropping exposures of this schist unit on the northwest side of the Strafford dome (Fig. 2).

The whole-rock geochemical data presented here are used as a foundation for discriminating the protoliths of the intensely altered wall rocks that occur at the Elizabeth deposit. Relatively immobile trace elements, including Sc, Cr, Ti, Zr, and Th, are used in this context to distinguish tholeiitic metavolcanic from clastic metasedimentary rocks, the former characterized by higher Sc, Cr, and Ti and lower Zr and Th compared to the latter (Figs. 8, 13; App. 1). Using these geochemical criteria and the approach of MacLean and Barrett (1993), the least altered tholeiites are needle amphibolite sample DH 1006-61 and plagioclase-phyric amphibolite DH 972-13. These two samples show different degrees of Fe-Ti fractionation (Fig. 13A), DH 972-13 having Cr vs. TiO₂ values coherent with those of nearly all of the altered wall rocks of the mine sequence (Fig. 13C), excluding two samples of white mica + staurolite ± sillimanite ± corundum schist. Based on these geochemical relationships, DH 972-13 is considered representative of mafic volcanic protoliths of nearly all of the altered wall rocks and is used herein as the least altered, baseline composition for calculating mass changes; a limitation to our study is the lack of other suitable samples from the Elizabeth deposit having an appropriate petrologic signature to include in this least altered category. Clastic sedimentary rocks are ruled out as possible protoliths because all of the altered wall rocks (excluding coticules) have uniformly lower Th contents and higher Sc/Th ratios, relative to unaltered pelitic schists of the surrounding Gile Mountain Formation (Fig. 13D-E).

Calculated mass changes presented here rely on the use of sample DH 972-13 as the least altered metabasalt in the Elizabeth mine sequence. This sample was chosen because unlike the other hornblende-plagioclase amphibolites—including needle amphibolites—it contains primary plagioclase phenocrysts (Fig. 10B). Also, the bulk composition of DH 972-13 shows minimal Fe-Ti fractionation compared to some other metabasalts in the mine area and those regionally within the Standing Pond Volcanic Member, plotting along the trends of Al₂O₃ vs. TiO₂ and Cr vs. TiO₂ for nearly all of the altered metabasalts in the mine sequence (Fig. 13A, B). Other relatively unaltered metabasalts there, such as DH 1006-61, show variable Fe-Ti fractionation (Fig. 13C) and are thus considered inappropriate for use in these mass-change calculations. Based on relationships observed in mine exposures and drill cores, including a stratabound geometry for the alteration assemblages, metasomatism that produced the mass changes is attributed largely if not entirely to sea-floor and subseafloor hydrothermal processes, in general agreement with the conclusions of McKinstry and Mikkola (1954) and Howard (1959b, 1969).

Appendix Table A6 lists ranges and means of Al-normalized molar element ratios for six key metabasaltic rock types of the mine sequence. Use of Al as the normalizing immobile element is justified here on the basis of strong correlations (not shown) among 45 samples for Al-Ti (R² = 0.89) and Al-Cr (R² = 0.82). Average mass changes (in percent) are shown in Figure 17 and listed in Table 1. Briefly, these calculations mostly indicate average major additions (>50%) of Mn, Na, and K and major losses (>50%) of Mg and Ca. Changes in Si, Ti, and Fe are mostly negligible (±5%). Data for three samples of siliceous tourmaline-rich rocks (not shown) indicate average major increases for Si (+254%) and Na (+137%), major decreases for Fe (–85%), Mg (–85%), Ca (–83%), and Mn (–74%), a minor decrease for K (–19%), and no change for Ti (0.0%). Volume changes are not considered here, but mass losses are typically associated with volume losses, whereas mass gains are linked to volume gain (e.g., MacLean and Barrett, 1993; Pilote et al., 2020).

With one exception, Al-normalized mass changes for Ti among the different lithologic groups are negligible to very small (<5%) thus suggesting limited mobility of Al and Ti during regional metamorphism. An appreciable increase in Ti determined for the aluminous white mica + staurolite ± garnet ± sillimanite ± corundum schist (+9.4%) implies a net mass gain for this generally immobile element, although this result is misleading because data for two of these three samples indicate Fe-Ti fractionation (trend A-Aʹ; Fig. 13C) with increased Ti contents relative to the main trend (B-Bʹ) for the other altered wall rocks considered here. Also, our use of a single precursor composition for all altered mafic rocks may not be valid, such that a small amount of primary variation in Ti existed prior to VMS-related alteration. Aluminum was clearly mobile within the Elizabeth mine sequence, as evidenced by several occurrences in the south pit of late quartz-kyanite veins that cut all foliations, hence indicating that this Al mobility occurred during a late- to postmetamorphic process. The formation of similar quartz-kyanite veins in the central Alps of Switzerland and on the Isle of Unst in Scotland have been attributed to the involvement of alkali Al silicate aqueous complexes, with quartz and kyanite precipitation linked to large fluid-rock ratios and time-integrated fluid fluxes, coupled with fluid cooling and a net decrease in pH (Beitter et al., 2008; Bucholz and Ague, 2010). The findings at Unst also suggest that the Al required for kyanite formation did not come from the local vein selvages but instead was derived regionally (Bucholz and Ague, 2010). By analogy, the Al in the rare quartz-kyanite veins at the Elizabeth deposit may have come from a larger source area that included surrounding strata of the Gile Mountain Formation.

Average mass changes for the altered metabasaltic rock types at the Elizabeth deposit show uniform increases in Na and K together with a general increase in Mn and decreases in Mg and Ca (Fig. 17). The greatest overall increases are represented by Mn and K. Enrichment in Na is also widespread for individual samples, relative to inferred least altered metabasalt DH 972-13 (2.62 wt % Na₂O) and metadiabase and metagabbro of the Standing Pond Volcanic Member (1.80–2.66 wt % Na₂O). For comparison, the average Na₂O content of MORB is 2.52 ± 0.41 wt % (White and Klein, 2014). Whole-rock Na₂O contents of relatively unaltered needle amphibolite in the mine sequence (3.40–3.56 wt %) and in the Standing Pond Volcanic Member (2.80–4.63 wt %) are attributed mainly to sea-floor spilitization based on averages of 3.35 to 5.82 wt % Na₂O for unmetamorphosed spilites (Hekinian, 1971; Lugović et al., 1991; Gürsu and Göncüoğlu, 2005). Whole-rock Na₂O contents that characterize the majority of altered metabasalts in the mine sequence, from 3.0 up to 6.0 wt % (App. 1), may record early spilitization followed by VMS-related hydrothermal alteration. Significantly, albite-rich wall rocks, mainly containing >6 wt % Na₂O and more than 50 vol % albite (or oligoclase), are well documented in VMS deposits elsewhere including Joma and Skorovass in Norway (Olsen, 1980; Reinsbakken, 1980), Gossan Lead in Virginia (Gair and Slack, 1984), Bergslagen in Sweden (Rickard, 1988; Frank et al., 2019), Snow Lake in Manitoba (Bailes et al., 2016), Hellyer in Tasmania (Gemmell and Fulton, 2001), and Aguas Teñidas in the Iberian Pyrite Belt (Gisbert et al., 2021). At the Elizabeth deposit, similar sodic compositions occur within the hanging-wall laminated albite-rich rock (Figs. 7J, 9G-H) and commonly in the plagioclase + mica ± quartz ± garnet ± calcite ± sulfide schist (Fig. 9L).

The major enrichment of K in all lithologic groups (Fig. 17) is expressed by abundant biotite and lesser muscovite. Margarite and in places paragonite are volumetrically significant in the aluminous staurolite-white mica-sillimanite-corundum unit. Biotite is typically foliated in the laminated albite-muscovite-hornblende-garnet rock (Figs. 7J, 9H) and unfoliated in the coarse garnet-mica schist (Figs. 7K, 9I). In the latter example, biotite replaces hornblende that cuts the S₂ schistosity, indicating a late- or postpeak metamorphic introduction of K into this rock, which is likely part of the retrograde chlorite ± biotite ± calcite ± K-feldspar ± sericite assemblage in the area (Menard and Spear, 1994). For the former paragenesis, the K source is attributed to footwall clastic sediments of the Gile Mountain Formation leached by VMS-related hydrothermal fluids and to seawater (Lydon, 1988; Hannington, 2014), whereas the latter paragenesis likely reflects K remobilized in retrograde metamorphic fluids by the breakdown at depth of the assemblage staurolite + biotite + quartz (e.g., Prenzel and Abart, 2009). Discriminating the contributions of K from VMS-related vs. retrograde metamorphic processes cannot be done without more detailed studies, but the scarcity of sericitized plagioclase in the muscovite-rich matrix of the Na-rich rocks suggests that this K-rich matrix predates Acadian regional metamorphism and hence likely formed by sea-floor hydrothermal processes.

Manganese is enriched in all common altered rocks of the Elizabeth mine sequence. Excluding Mn contained in the coticules, we suggest that these enrichments are linked to reduced, Mn-bearing hydrothermal fluids that altered precursor mafic volcanic rocks, producing proportional Mn contents in several lithologic groups (e.g., coarse garnet-mica schist; plagioclase + mica ± garnet ± quartz ± calcite ± sulfide schist) that are greater than those of the other wall rocks and relatively unaltered metabasalts of the mine sequence (Fig. 12F). Also of note are the appreciable Mn enrichments in four samples of visibly unaltered clastic metasedimentary rocks of the Gile Mountain Formation in the hanging wall of the deposit, suggesting that exhalative Fe-Mn oxyhydroxides continued to be deposited after formation of the sulfide orebody and subseafloor alteration of basaltic wall rocks.

Loss of Mg characterizes all key units of the mine sequence except the hanging-wall tremolite-phlogopite schist. Calcium also shows a net mass loss in all but one of these units, mirroring the behavior of Mg. Considered relative to least altered metabasaltic protolith sample DH 972-13, this systematic mass loss of Mg is unusual compared to the Mg enrichment that characterizes the alteration zones of most VMS deposits (e.g., Franklin et al., 2005; Piercey, 2009). One possible explanation for this pattern involves a subseafloor process in which unevolved Mg-rich seawater was only a minor component of the ascending hydrothermal fluids that altered primary tholeiitic basalts on and beneath the sea floor prior to, during, and after VMS mineralization.

Needle amphibolites of the Standing Pond Volcanic Member have δ¹⁸O values of 7.9 to 9.1‰, which are appreciably higher than average unaltered MORB (5.69 ± 0.33‰; White and Klein, 2014). In the Elizabeth mine sequence, needle amphibolite DH 1006-63 and plagioclase-phyric amphibolite DH 972-13 have δ¹⁸O values of 10.7 and 11.1‰, respectively (App. Table A5), the increases relative to MORB likely reflecting both spilitization and hydrothermal alteration. Even higher oxygen isotope values characterize altered metabasalts of the mine sequence, up to 14.2‰ (Fig. 16). Among all 26 silica-poor altered metabasalts of the mine sequence analyzed for both δ¹⁸O (10.7–13.4‰) and SiO₂ (37.70–56.34 wt%), no statistically significant correlation exists (R² = 0.20). Overall, the high δ¹⁸O values for these altered basalts are consistent with low-temperature hydrothermal alteration at approx. 150° to 250°C and under high water/rock ratios, following earlier models by Munhá and Kerrich (1980) for basalt in the São Domingos deposit in the Iberian Pyrite Belt of Portugal; Brauhart et al. (2000) for andesite-basalt in the Panorama VMS deposit in Western Australia; and Beaudoin et al. (2014) for basaltic andesite in the LaRonde Penna VMS deposit in Québec. Based on oxygen isotopes, similar low temperatures of 150° to 250°C were proposed by Gemmell and Fulton (2001) for the formation of quartz-albite and sericite assemblages in the basalt-hosted alteration zones of the Hellyer deposit in Tasmania. Higher-temperature (>250°–350°C) alteration of precursor basalt at the Elizabeth deposit would in contrast have produced δ¹⁸O values lower than that of pristine MORB by ~2 to 4‰ (see Roberts et al., 2003; Holk et al., 2008; Taylor et al., 2014), which are not recognized in our study, as well as large positive Eu anomalies that are absent in the sulfide-poor wall rocks (Fig. 14D). For reference, the oxygen isotope value of coeval seawater in the Early Devonian was approx. –4 to –2‰, compared to 0‰ today (Jaffrés et al., 2007). Relatively high δ¹⁸O and δ¹¹B values for tourmaline from Elizabeth massive sulfide and the hanging-wall quartz-albite-tourmaline unit (10.5 to 11.4 and –14.2 to –13.1‰, respectively) suggest formation by a mineralizing fluid in which boron was derived predominantly from footwall clastic sediments (Palmer and Slack, 1989).

Oxygen isotope values for quartz and magnetite separates from the Cookville and Pike Hill mines to the north of Elizabeth (Fig. 1B) provide insights into post-ore regional metamorphic conditions in the area. Using the quartz-magnetite oxygen isotope fractionation factor of Chiba et al. (1989), temperatures of 546° and 668°C are obtained for the two Cookville samples, and 584°C for the Pike Hill sample. The 668°C temperature is much higher than the D₂ peak temperature of 525° ± 35°C, determined by Menard and Spear (1994) and Ashley et al. (2013) for a pelitic schist ~5 km southwest of the Ely mine, at the same staurolite-kyanite metamorphic grade. These higher temperatures may reflect the thermal effects of unexposed Acadian (Early Devonian) granitic intrusions near these mines. The other two quartz-magnetite pairs are within error of the peak metamorphic temperatures when errors in the quartz-magnetite calibration are included (Chiba et al., 1989). Calculations of temperatures based on quartz-biotite, quartz-albite, quartz-hornblende, and quartz-garnet oxygen isotope fractionations do not give temperatures consistent with peak metamorphic or hydrothermal temperatures (App. Table A5).

Figure 18 is a ternary cation Fe + Mg-Si-Al plot of whole-rock compositions of least altered and altered metabasalts from the Elizabeth deposit and vicinity. Relative to most of the former and to average N-MORB, data for most samples show trends toward chlorite or sericite, the latter evident especially from results for the aluminous white mica-rich schist. A separate trend mainly from albite to quartz is represented by the quartz-albite-tourmaline rocks. Based on these cation data, we interpret the premetamorphic protoliths to have been dominated by chlorite and/or sericite that formed chiefly by hydrothermal processes, a hypothesis supported by the recognition of such assemblages in the alteration zones of many VMS deposits that have not undergone high-grade metamorphism like Elizabeth (e.g., Franklin et al., 2005; Piercey, 2009). Data for several samples of the tremolite-phlogopite (± plagioclase ± tourmaline ± sulfide) schist trend toward amesite (Mg-rich chlorite), consistent with the common abundance of Mg-chlorite in the alteration zones of unmetamorphosed VMS deposits worldwide. This phlogopite-rich unit and other hanging-wall rock types that contain minor biotite, such as the coarse garnet-mica schist (Figs. 7K, 9I), lack trends towards the biotite end member (Fig. 18). No samples plot in the quartz + chlorite field; on this basis, there is no evidence for a premetamorphic, quartz-chlorite alteration assemblage.

The large net mass changes calculated for wall rocks with basaltic protoliths in the Elizabeth mine sequence (Fig. 17) are interpreted to reflect alteration linked to VMS mineralization. Previous work by McKinstry and Mikkola (1954) and Howard (1959b) related this alteration to epigenetic ore formation after local basaltic volcanism. McKinstry and Mikkola (1954) proposed that both mineralization and alteration took place after folding and metamorphism. Howard (1959b) distinguished three main stages of alteration, the first two conformable to the ore zones and the third discordant. In his model, Stage I formed outer biotite and inner sericite subzones prior to sulfide mineralization, not only at the Elizabeth deposit but also adjacent to amphibolites elsewhere within the Gile Mountain Formation and the Standing Pond Volcanic Member. Stage II was linked to ore formation at Elizabeth and also developed outer biotite and inner sericite subzones. Stage III, associated with post-ore faults and shear zones, consists of outer chlorite and inner sericite zones. Howard (1959b) attributed all three of these alteration stages, and sulfide mineralization, to synmetamorphic processes.

The extensive mass changes evident in hanging-wall strata (Fig. 8) require a post-ore timing if the massive sulfides formed on the sea floor by syngenetic processes and not as subseafloor replacements, based on the geologic and geochemical arguments presented below. In particular, the pronounced enrichments of K in all key lithologic units and Na in the hanging-wall laminated, albite-rich rock are consistent with formation by semiconformable alteration beneath the sea floor (see Galley, 1993; Doyle and Allen, 2003), preferentially focused within permeable basaltic tuffs (Fig. 19). However, at least some of this K enrichment clearly occurred during metamorphism, based on the common replacement of hornblende by biotite in the coarse garnet-mica schist unit of the Elizabeth deposit (Fig. 7K) and of hornblende by biotite and chlorite in the laminated albite-hornblende-garnet rock (Fig. 9G), as part of a regional retrograde assemblage of chlorite ± biotite ± calcite + K-feldspar ± sericite recognized by Menard and Spear (1994). The Mn enrichments may be contemporaneous, but also could reflect exhalative chemical precipitates that formed after major sulfide mineralization and during pauses in local mafic volcanism. Relatively impermeable cap rocks have been invoked by other workers to explain subseafloor confinement of VMS-related hydrothermal fluids, and subseafloor formation of massive sulfides and semiconformable alteration zones (Doyle and Allen, 2003; Hannington et al., 2005; Jones et al., 2006; Piercey et al., 2015), but no convincing cap rocks are evident at the Elizabeth deposit. The very siliceous (up to 85.0 wt % SiO₂; App. 1) facies of the tourmaline-bearing unit in the hanging wall directly above the massive sulfides (Fig. 8) is a plausible candidate, yet this unit is restricted to the area south of the north open pit, based on early studies (McKinstry and Mikkola, 1954; Howard, 1969) and on our logging of drill cores. Coticules in the upper part of the mine sequence may be a better choice, based on apparent greater stratigraphic continuity, except that this Mn-rich chemical sedimentary rock is not stratigraphically directly above the highest occurrence of altered metabasalt. Moreover, if alteration of the hanging-wall basalts occurred soon after sulfide mineralization, precursors of the siliceous tourmaline-bearing unit and the coticules likely were, at most, poorly lithified. On balance, therefore, no lithologic units in the uppermost part of the Elizabeth mine sequence are considered suitable as cap rocks that would explain the intensely altered metabasalts present in the stratigraphic hanging wall of the deposit.

The presence in the mine sequence of distinctive rock types such as the dolomitic plagioclase rock and laminated albite-rich rock that extend for at least 100 m along strike is consistent with data from other VMS deposits in which semiconformable alteration zones may be up to tens of kilometers in length (Galley, 1993). At the Elizabeth deposit, Howard (1969) recognized alteration zones in amphibolite up to 20 m wide and up to 100 m long that are continuous around fold noses and thus formed prior to deformation. However, some distinctive rock types such as the coarse garnet-mica schist lack continuity even between drill holes only ~50 m apart, thus suggesting a process of laterally discontinuous alteration governed by irregular zones of low vs. high permeability (Doyle and Allen, 2003). For example, in the São Domingos VMS deposit in the Iberian Pyrite Belt, sodic enrichment like that at Elizabeth is preferentially focused in high-permeability zones such as flow-top breccias, coarse volcaniclastic rocks, and the margins of tuffs and sills (Munhá and Kerrich, 1980). A similar process likely existed at the Elizabeth deposit.

Early workers proposed a syn- to postmetamorphic epigenetic origin for the massive sulfide deposits in the Elizabeth mine area and elsewhere in the Vermont copper belt (McKinstry and Mikkola, 1954; Howard, 1959a, b, 1969). The Elizabeth deposit is noteworthy for having an extensive strike length of 3.4 km, from the South mine northward to the south and north pits (Fig. 3) and farther north as documented in underground workings and extensive diamond drilling (Howard, 1969). The overall geologic setting of the Elizabeth deposit, and others in this copper belt, is very similar to the settings of ancient Besshi-type (or mafic-siliciclastic; Barrie and Hannington, 1999) deposits worldwide (Fox, 1984; Slack, 1993). Association in the Vermont copper belt of relatively large deposits at Elizabeth to only one amphibolite unit, stratigraphically above five others in the lower part of the Gile Mountain Formation (Fig. 2), indicates that this strongly mineralized horizon was linked to special volcanogenic and hydrothermal processes.

Several lines of evidence suggest that the stratiform Fe-Cu-Zn sulfide deposits at the Elizabeth mine formed largely, if not wholly, by chemical precipitation in marine bottom waters and not by subseafloor replacement (cf. Piercey, 2015; Piercey et al., 2015). First, the 3.4-km length of the deposit is unlikely to record replacement of strata below the sea floor, which would require sulfide precipitation along one continuously permeable horizon over a significant distance. Second, whole-rock analyses of representative samples of massive sulfide from the Elizabeth deposit show uniformly low contents of Sc (≤11.9 ppm), Cr (≤61 ppm), Zr (≤8 ppm), and Th (≤0.6 ppm), except for one atypical massive pyrite sample that has 110 ppm Cr (App. Table A3). The values for Sc and Cr are considered too low to record subseafloor replacement of a mafic tuff precursor, although the large density increase and mass gain resulting from this process are complicating factors. Replacement of permeable clastic sediment is also considered unlikely, given uniformly low contents of Zr (≤15 ppm) and Th (≤0.22 ppm) in the massive sulfide samples. Third, no remnants of a mafic tuff precursor—or other rock types—have been found within the massive sulfide samples studied here, excluding the many clasts of folded and foliated wall rocks that occur throughout much of the orebody (Howard, 1969), forming a “durchbewegung” texture attributed here to post-ore deformation and remobilization of massive sulfide with accompanying incorporation of adjacent wall rocks (see Vokes, 1969; Marshall and Gilligan, 1989). Lastly, although the Y/Ho ratios of most samples are close to the chondritic value of 27 (App. Table A3) and thus suggest a rock-buffered source, sphalerite-rich sample EZ-1023 has a much higher ratio of 37.6, which indicates a major component of oxidized seawater with Y/Ho >44 (Lode et al., 2015). This sample and typical massive sulfide sample EZ-1041 (pyrrhotite + minor chalcopyrite) both have negative Ce anomalies (0.57 and 0.69, respectively), thus further supporting at least periodically oxygenated bottom waters during VMS mineralization. Subseafloor replacement in contrast would likely have anoxic pore waters due to the required abundance of H₂S. Both of these samples also display positive Eu anomalies (1.25 and 3.99, respectively) that record mineralization at >250°C under reducing conditions (e.g., Sverjensky, 1984). Preservation of these contrasting redox signatures in the same samples likely reflects minimal dilution of the hydrothermal fluid by seawater (Barrett et al., 1990; Zeng et al., 2015). The presence of negative Eu anomalies without Ce anomalies in two of the other massive sulfide samples (EZ-46, EZ-1031) is attributed here to transient anoxia in bottom waters of the local basin.

Elevated Mn contents of coticules in the stratigraphic footwall and hanging wall of the Elizabeth deposit (up to 2.8 wt % MnO) suggest derivation of the manganese predominantly from submarine-hydrothermal fluids as documented in modern VMS systems (e.g., German and Von Damm, 2003; Hannington, 2014). The common occurrence of coticules within or adjacent to true amphibolite (Fig. 6) is consistent with a link to local basaltic volcanism. Whole-rock geochemical data indicate that the coticules closely associated with true amphibolite are not metasomatized basalt, based on uniformly low contents of Cr (≤130 ppm) determined for eight samples (App. 1), although the occurrence of minor hornblende in a few coticule intervals in drill core suggests the local presence of a small mafic igneous component. In addition, the average Zr/Hf ratio of the coticules (41.7 ± 1.60, n = 8) is much higher than that of the least altered amphibolites of the mine sequence (31.8 ± 0.88; n = 3) but is very similar to the average of unaltered clastic metasedimentary rocks (regional samples) of the Gile Mountain Formation (41.2 ± 1.70; n = 13), consistent with a predominant clastic detrital source of zircon in the coticules. Higher Al₂O₃/TiO₂ ratios for all but one of the coticules relative to the field for clastic metasedimentary rocks of the Gile Mountain Formation (Fig. 13A) suggest that the coticules formed from an especially clay-rich protolith that is not represented among our samples of the Gile Mountain Formation. Abundant quartz in the coticules, as well as spessartine-hosted Si and Al (from clays), are also considered detrital in origin, a premise supported by whole-rock δ¹⁸O values of 10.7 to 12.5‰ that fall within the range determined for unaltered clastic metasedimentary rocks of the Gile Mountain Formation (Fig. 16). Deposition of the coticule precursors is believed to involve syngenetic precipitation of hydrothermally derived, aqueous manganese and iron oxyhydroxides that mix with clays in clastic sediment, together with detrital quartz, followed by transformation during low-grade regional metamorphism into a fine-grained, spessartine-quartz rock (Spry, 1990; Spry et al., 2000; Romer et al., 2011). This process resulted in the formation of coticule sediments prior to and after VMS mineralization (Figs. 8, 19). The origin of the calcite marble in the mine sequence is unknown but may be hydrothermal like that in the low-temperature chimneys and mounds of Guaymas basin (Peter and Shanks, 1992). This origin could also explain the uncommon, high-grade sphalerite and sphalerite-calcite lenses in the Elizabeth ore zones that might represent deformed clasts derived from chimneys or mounds.

The stratiform tourmaline-rich rocks in the stratigraphic hanging wall would conventionally be attributed to exhalative hydrothermal processes in this type of basalt-dominated sequence (Slack, 2022). However, mineralogical and whole-rock geochemical data presented here strongly suggest an origin by pervasive metasomatism of basalt. First, most of these rocks contain abundant albite (Figs. 7G, 9E), which is rare in tourmalinites. Second, and more compelling, are concentrations of Al₂O₃, TiO₂, Cr, Th, together with Sc/Th and Cr/Zr ratios, which fall within or near data for altered metabasalts of the mine sequence and not within or near the field for unaltered clastic metasedimentary rocks of the Gile Mountain Formation (Fig. 13A-E). These results are evident for both highly siliceous (81.9–85.0 wt % SiO₂) and aluminous (15.2 and 18.3 wt % Al₂O₃) varieties of the tourmaline-rich rocks (App. 1). Depletion of Al₂O₃, TiO₂, Cr, and Sc in the highly siliceous varieties, compared to an inferred basaltic precursor, can be explained by major mass gain involving the addition of significant silica during subseafloor alteration and mineralization. Unlike the other altered metabasalts of the mine sequence that lack evidence of REE enrichment or depletion, relative to an inferred N-MORB protolith (compare Fig. 14C and D), the tourmaline-rich rocks have undergone appreciable LREE mobility, including minor enrichment in two samples and major depletion in one (Fig. 14D). The abundance of boron in these rocks (up to ~3 wt % B₂O₃ based on 30 vol % tourmaline) and relatively high δ¹¹B values for tourmaline in the stratiform quartz-albite-tourmalinite and massive pyrrhotite samples (–13.1 and –14.2‰, respectively) are attributed to extensive reaction of mineralizing hydrothermal fluids with clay-rich clastic sediments in the deep footwall, based on models for modern sediment-hosted VMS systems in which elevated B contents and similar δ¹¹B values record derivation of the boron predominantly from marine clays (see Palmer and Slack, 1989; Trumbull et al., 2020; Slack, 2022).

Locally abundant graphite in the hanging-wall quartz-albite-tourmaline unit (Fig. 7H) is rarely reported in chemical sedimentary rocks associated with ancient massive sulfide deposits. However, well-documented modern examples, without tourmaline, are known in Guaymas basin (Gulf of California) and Escanaba trough (northeast Pacific Ocean), where hydrothermal petroleum is widespread and abundant in surficial sediments and active hydrothermal mounds (Simoneit and Lonsdale, 1982; Gieskes et al., 1988; Simoneit, 1988; Kvenvolden et al., 1994). Organic geochemical and carbon isotope studies of this petroleum indicate a source from immature detritus in the clastic ± carbonate sediments that fill Guaymas basin and Escanaba trough, and with which metalliferous hydrothermal fluids interacted during ascent to the sea floor (Simoneit, 1988; Schoell et al., 1990). We propose that a similar process formed the abundant graphite in the quartz-albite-tourmaline unit at the Elizabeth deposit, given the common presence in unaltered clastic metasedimentary footwall strata of thin beds of graphitic pelitic schist with up to 1.3 wt % noncarbonate (graphitic) carbon (App. 1).

Constraining the origin of highly metamorphosed VMS deposits like Elizabeth is a major challenge that requires careful documentation of the protoliths of strongly altered wall rocks. The key is distinguishing effects of premetamorphic from synmetamorphic processes with a goal of seeing through the metamorphism (e.g., Marshall and Spry, 2000). Early studies of the Elizabeth deposit proposed epigenetic replacement models for sulfide mineralization and local wall-rock alteration—especially sericitization—occurring during or after Acadian deformation and metamorphism (McKinstry and Mikkola, 1954; Howard, 1959a, b, 1969). Our field and geochemical work in contrast suggests that the ores and wall-rock alteration formed prior to deformation and metamorphism, respectively on and beneath the Early Devonian sea floor.

Extensive publications exist on multiply deformed and highly metamorphosed (middle amphibolite facies or above) VMS deposits, but few concentrate on the protoliths of altered wall rocks. Studies that have focused on this issue include those on the Montauban deposit in Québec (Bernier and MacLean, 1993), the Manitouwadge district in Ontario (Pan and Fleet, 1995), the Vihanti-Pyhäsalmi district in Finland (Roberts et al., 2003), the Prieska deposit area in South Africa (Theart et al., 2011), the Bergslagen district in Sweden (Jansson et al., 2017; Frank et al., 2019), and the Bon Ton, Cotopaxi, and other deposits in Colorado (Berke et al., 2023). For most of these deposits, as well as deposits studied elsewhere, the metamorphic mineralogy of metavolcanic host rocks can generally be used as a first-order guide as to whether the precursors are mafic or felsic (e.g., modal abundance of amphibole vs. quartz), followed by detailed geochemical analyses augmented in some cases by oxygen isotope analyses to provide robust constraints on the protoliths. However, in the case of the Elizabeth deposit, only small amounts of mafic minerals occur in the metamorphic assemblages in some wall rocks, such as the footwall dolomitic plagioclase rock and the plagioclase-mica-garnet-quartz-sulfide schist, yet immobile trace element systematics provide compelling evidence for their origin as tholeiitic basalt. On this basis, we advise caution in assuming that the metamorphosed wall rocks of VMS deposits with minimal or no mafic minerals have clastic sedimentary (or felsic volcanic) protoliths and argue that high-precision data for immobile trace elements like Sc, Cr, Zr, and Th are required in order to properly distinguish the precursor rock types.

Combined geologic, geochemical, and isotopic studies presented here provide strong evidence for the formation of the Elizabeth massive sulfide deposit and the origin of enclosing wall rocks. The geometry of the orebody, with a total length of 3.4 km and width of less than ~500 m (unfolded based on cross sections using drill core data), suggest that mineralization occurred in an elongate trough on the sea floor. No reports are known in the mine area of stretching lineations that could explain this sulfide geometry entirely by post-ore deformation, but some rocks there display boudinage (McKinstry and Mikkola, 1954; Howard, 1969). Also, outside of the Vermont copper belt, strata of the Connecticut Valley-Gaspé trough locally show evidence of high strain and tectonic flattening, including stretched conglomerate pebbles and sheath folds (Currier and Jahns, 1941; Walsh et al., 2020). Given these deformational effects, it is not possible to define exact primary dimensions of the Elizabeth deposit, although a highly elongate geometry is considered likely based on the absence of intense stretching of rocks in the mine area.

Possible modern analogs of the Elizabeth and other VMS deposits of the Vermont copper belt are on sediment-covered ridge crests. Examples include Guaymas basin in the Gulf of California, and Escanaba trough and Middle Valley in the northeast Pacific Ocean. However, an important caveat is that these modern deposits chiefly form mounds and not sheet-like sulfide deposits despite being contained mostly in graben structures (Koski et al., 1985, 1988; Koski, 1990; Zierenberg et al., 1993; Houghton et al., 2004). The deep copper replacement zone at Middle Valley (Zierenberg et al., 1998) is stratabound based on limited drilling but is not considered a valid analog for the Elizabeth deposit because it lacks coticules with negative Ce anomalies and no positive Eu anomalies that suggest low-temperature deposition under oxic conditions, in contrast to the positive Eu anomalies (and no Ce anomalies) that would likely exist in a Cu-rich, high-temperature replacement deposit.

The metalliferous deposits of the Red Sea also need to be considered as possible analogs, especially given their sheetlike morphology and range of sulfur isotope values (Shanks, 2001) that are broadly like those of the Elizabeth deposit and many other Besshi-type deposits (Slack, 1993, and references therein). The Atlantis II brine-related metalliferous sediments were deposited from an extensive (~14- × 6-km), 200-m-thick brine pool with dissolved Cl content about 8× seawater values (~27 wt %). Metalliferous mud deposits as thick as 20 m occur as finely layered, extremely fine-grained, gelatinous sediments with ≥90 wt % water. Upper and peripheral (flanking edges of the brine pool) deposits predominantly contain Fe and Mn oxyhydroxides and silicates with only minor pyrite, sphalerite, and chalcopyrite (Shanks and Bischoff, 1980; Laurila et al., 2015). Deeper in the metalliferous deposits, upper and lower sulfide zones (each ≤4 m thick) are separated by an iron and manganese oxide zone up to 11 m thick. Upon dewatering, compaction, and lithification, the Atlantis II Deep deposits would be extensive and sheetlike. The southwest basin of the Deep is the locus of current fluid venting and sulfide deposition and is roughly 4 × 3 km, analogous to the extent of the Elizabeth deposit. The upper and peripheral Fe and Mn oxyhydroxide deposits may be precursor analogs to the Elizabeth coticule rocks. The supersaline hydrothermal fluids in the Atlantis II Deep provide a likely source for the abundant albitization of basalts, as observed for the Elizabeth mine sequence. Although metaevaporites have not been recognized in the Vermont copper belt, Rich (1979) found numerous halite-bearing fluid inclusions in synmetamorphic quartz veins within the Waits River Formation in southeastern Vermont; scapolite is also reported from metasedimentary localities in this area (www.mindat.org) and in the Waits River Formation in northeastern Vermont (Hall, 1959; Léger and Ferry, 1993). These occurrences of saline fluid inclusions and scapolite suggest the former presence of evaporites in the Waits River Formation (e.g., Moine et al., 1981; Warren, 2016; Morrissey and Tomkins, 2020). If metaevaporites also existed in this formation stratigraphically below the Elizabeth deposit, they could have served as a source of high-salinity fluids to form a brine pool, like in the modern Red Sea.

In contrast to typical unaltered basalts at Middle Valley with ~14 wt % Al₂O₃, hydrothermally altered basalts there locally have very high Al₂O₃ contents up to 20.8 wt %, attributed to enrichment due to the leaching of other components (Stakes and Franklin, 1994). This occurrence may be a modern, but limited, analog of the much more pervasive process that we invoke for the highly altered metabasalts at the Elizabeth deposit. The lack of reported atypical wall rock types at these modern sediment-hosted systems, like for Elizabeth, may reflect limited deep drilling and a smaller net mass flux of hydrothermal fluid relative to that involved in formation of the Elizabeth deposit. The absence of an elongate morphology for the modern sulfide deposits, as observed for Elizabeth, is unexplained but could record a shorter time span of mineralization.

Our data collectively suggest that most if not all of the Fe-Cu-Zn sulfides of the Elizabeth deposit formed as chemical precipitates on the sea floor. This process likely occurred within at least partly (or episodically) oxygenated bottom waters, as evidenced by the presence of negative Ce anomalies in two samples of massive sulfide (Fig. 14F). These anomalies, together with the lack of negative Ce anomalies in most other massive sulfide samples analyzed and the occurrence of uniformly negative anomalies in coticules of the mine sequence, are consistent with deposition in a stratified water column (Fig. 19). We suggest that during VMS mineralization, the depth position of the redoxcline fluctuated due to the generation of anoxic bottom waters by the hydrothermal venting of reductants, including Fe²⁺, Mn²⁺, H₂S, H₂, and CH₄. Effects of such a process are supported by computer-based calculations by Barnes et al. (2020) using the 3-D Earth system model (cGENIE), integrated with physicochemical parameters of VMS-related hydrothermal systems in modern marine basins.

The occurrence of semiconformable alteration zones in the footwall and hanging wall of the Elizabeth deposit attests to subseafloor metasomatism and leaching of tholeiitic basalts. Reduced fluids are implicated on the basis of many small to moderate positive Eu anomalies (App. 1). Subseafloor alteration possibly was promoted locally by early formation of impermeable cap rocks such as the highly siliceous facies of the tourmaline-bearing unit that occurs in the hanging wall immediately above the orebody (Fig. 8). Within most of the other altered metabasalts of the mine sequence, original tholeiitic basalt was leached principally of Ca and Mg (Fig. 17) via acidic hydrothermal fluids under high fluid-flux conditions. This leaching of Mg is in contrast to whole-rock geochemical data for modern sediment-dominated systems like Middle Valley and ancient VMS deposits in which footwall basalts typically lost Ca and Na but gained Mg (Stakes and Schiffman, 1999; Teagle and Alt, 2004; Galley et al., 2007). This major difference and the common gain of K and locally Na within the Elizabeth samples suggest that the precursor basalt there was altered by hydrothermal fluids having low contents of Ca and Mg, but high K and in part Na. We suggest that this unusual composition reflects previous chemical reactions and buffering of the hydrothermal fluids by clastic sediments in the deep footwall, which in unaltered surface samples locally contain abundant muscovite and plagioclase (App. 1; Menard and Spear, 1993). This hypothesis is supported by the experimental study of Bischoff et al. (1981) involving the hydrothermal alteration of graywacke by seawater at 200° and 300°C, in which reaction at the higher temperature quantitatively removed nearly all of the Mg, thus producing a Mg-undersaturated fluid capable of leaching this component from stratigraphically higher basalt. Additional support comes from the presence of a major net loss of Mg in one sample of deep footwall basalt in the predominantly sediment-hosted sea-floor hydrothermal system at Middle Valley (Stakes and Schiffman, 1999, fig. 6B). In our model, the hydrothermal fluids rose along synsedimentary faults and were channeled along contacts and within permeable basaltic tuffs (Fig. 19), but without a significant component of entrained Mg-rich seawater. In some cases, alteration by the hydrothermal fluids produced very aluminous bulk compositions that, during later Acadian metamorphism, formed abundant garnet and staurolite comparable to the selvages of amphibolite facies quartz-kyanite veins studied by Ague (2011). Domains of highly aluminous compositions in the altered metabasalts at the Elizabeth deposit were transformed into abundant white mica and staurolite, minor sillimanite, and rare corundum during this same regional metamorphic event. The lateral extent of the altered metabasalts there, at least 3 km along strike, is similar to the results of modeling studies for the modern sea-floor hydrothermal systems of Lau basin in the southwest Pacific Ocean, where high cumulative water/rock ratios calculated for a subseafloor permeable basalt unit extend laterally for as much as 7 km (Schardt and Large, 2009).

No vent sites or feeder zones have been identified at the Elizabeth deposit, but the much higher average Cu grade of 4.9 wt % reported for ore underground in the northern part of the mine, relative to the 1.1 wt % average grade in open pits to the south (Howard, 1959b), may indicate the presence of one or more unrecognized hydrothermal conduits at the north end. On a regional scale, VMS mineralization and broadly coeval wall-rock alteration at the Elizabeth deposit are preferentially linked to unfractionated tholeiitic metabasalts, in contrast to the Ti-rich metabasalts that appear to dominate the Standing Pond Volcanic Member, within which sulfide deposits are very small compared to the Elizabeth deposit. The presence in the stratigraphic footwall of the deposit of a distinctive unit of massive quartzite with interlayered pelitic schist (Fig. 7D), which is uncommon in surrounding strata of the Gile Mountain Formation, suggests that this unit records a high-energy sedimentary environment linked to local extensional faulting in the basin, prior to basaltic volcanism and sulfide mineralization. An alternative explanation involving a local turbidite facies in the mine sequence is considered unlikely, because the footwall quartzite is massive and texturally very different from the well-graded turbidite unit in the Gile Mountain Formation that occurs ~1 km east of the mine (Rolph, 1982; Slack et al., 2001).

Sulfur isotope data for sulfide minerals in the Elizabeth and other deposits of the Vermont copper belt (Fig. 15; App. Table A4) suggest that the contained sulfur was derived from varying proportions of that leached from tholeiitic basalt and from coeval seawater sulfate. Disseminated sulfides (pyrrhotite ± pyrite) in amphibolites of the Standing Pond Volcanic Member with low δ³⁴S values, mainly from 0.0 to 1.7‰, are within the range of igneous sulfur in modern MORB (0.06 ± 1.66‰; White and Klein, 2014), suggesting a predominant igneous (basaltic) source of primary iron sulfide minerals. Exceptions are two pyrrhotite samples with very low values of –8.8 and –5.0‰ that may reflect remobilization of an isotopically light bacterial sulfide component leached from underlying strata of the Waits River Formation, following the model used to explain similar values in modern VMS deposits of Guaymas basin in the Gulf of California (Peter and Shanks, 1992; Shanks, 2001). A δ³⁴S value of 2.8‰ in one pyrrhotite sample likely records a mixture of predominant igneous sulfur and a minor proportion of coeval (Early Devonian) seawater sulfate that had an average δ³⁴S value of ~20‰ (Claypool et al., 1980). Pyrrhotite and chalcopyrite separates from the small Gove and Cookville deposits that are within or adjacent to the Standing Pond Volcanic Member have slightly higher δ³⁴S values from 2.0 to 4.2‰, and 2.2 to 3.7‰, respectively. The Pike Hill deposit, hosted mainly in calcareous pelitic schist of the Waits River Formation near the contact with the Standing Pond Member, has pyrrhotite, chalcopyrite, and sphalerite values of 1.0 to 4.9‰, which, excluding one value of 6.4‰, are similar to those for Gove and Cookville samples and to values for modern volcanic-hosted deposits on mid-ocean ridges (e.g., Shanks and Seyfried, 1987; Woodruff and Shanks, 1988). The Ely and Elizabeth deposits, hosted in the Gile Mountain Formation, have generally higher δ³⁴S values up to 9.3‰ for pyrrhotite and chalcopyrite; several sphalerites show the same range. These sulfur isotope values are typical of modern sea-floor hydrothermal systems in sediment-covered ridge crests such as Middle Valley and Escanaba trough that are hosted within siliciclastic turbidites but show a close relationship to mafic volcanism (Böhlke and Shanks, 1993; Zierenberg and Shanks, 1994). The higher δ³⁴S values in ancient sediment-hosted deposits like Elizabeth and Ely, relative to those of purely basalt-hosted deposits, likely reflect the involvement of significant amounts of seawater sulfate that were reduced either thermochemically or by organic carbon in the sedimentary section below the hydrothermal systems (McDermott et al., 2015). Hydrothermal fluids generated in sediment-free, mid-ocean ridge systems only have sufficient reducing capacity to raise δ³⁴S values to ~4‰ (Janecky and Shanks, 1988; Shanks, 2001, 2014). As a result, the higher values that characterize the Ely and Elizabeth deposits require an additional reducing agent such as ferrous iron in minerals or organic carbon in footwall clastic sediments, the latter component supported by the presence of abundant graphitic carbon in some clastic metasedimentary samples of the Gile Mountain Formation (App. 1).

Boron and lead isotopes also provide insights into the origin of the VMS deposits of the Vermont copper belt. Limited data for boron isotopes in tourmaline from the district (Palmer and Slack, 1989) include results for two samples from the Elizabeth deposit (δ¹¹B = –14.2 and –13.1‰) that indicate a boron source predominantly from clays originally present in clastic metasedimentary rocks of the Gile Mountain Formation, as does a value of –15.4‰ for tourmaline from the Ely deposit; one tourmaline from the stratigraphically lower Pike Hill deposit (Fig. 1B) has δ¹¹B of –13.8‰ that also suggests a mainly clastic sediment source for the boron (Palmer and Slack, 1989; Trumbull et al., 2020), although in this case from the Waits River Formation. This interpretation of a clastic source for the boron is consistent with the above model invoked to explain the lack of Mg enrichment in the Elizabeth wall rocks via reaction and buffering of the hydrothermal fluids with clastic sediments in the deep footwall of the deposit and minimal, if any, contribution of entrained Mg-rich seawater. Lead isotope values of ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, and ²⁰⁸Pb/²⁰⁴Pb for single galena separates from Elizabeth (17.858, 15.474, and 37.403, respectively) and Pike Hill (17.965, 15.582, and 37.753, respectively) are relatively unradiogenic compared to other VMS deposits of the northern Appalachians, consistent with a lead source derived predominantly from deep Grenvillian (~1.1 Ga) basement rocks that underlay Silurian-Devonian strata of the Vermont copper belt during sulfide mineralization (Slack et al., 2024b).

The spectrum of lithologically and chemically distinctive wall rocks at the Elizabeth deposit appear unique among other mafic-siliciclastic (Besshi-type) VMS deposits of the Appalachians and Caledonides. If such wall rocks are indeed absent from the myriad other deposits of this type elsewhere in the orogen, and are not overlooked or misidentified, then the Elizabeth deposit serves as a valuable end member for better understanding subseafloor hydrothermal alteration of basalts associated with massive sulfide mineralization. This proposal could be applied to future studies of deep drill cores of potential modern analogs in Guaymas basin, Middle Valley, and Escanaba trough.

We are especially indebted to our late USGS colleague Terry Offield for his careful work in deciphering local and regional structure. Thanks are also due to the late Malcolm Annis for mine area mapping and logging of drill cores, to former USGS colleagues Cynde Sears, Paul Atelsek, and Charles Connor, and to several field assistants for logging of myriad cores. Annis and Gary Cygan (formerly USGS) made a topographic base of the open pits via plane table and alidade that was used for geologic mapping. John Jackson of the USGS provided X-ray diffraction data for rocks; the late Mary Mrose (USGS) identified mineral grains by X-ray powder camera methods. Damon Bickerstaff, Paul Atelsek, and Brett Valentine (all USGS) helped with sample organization, electron microprobe analyses, and SEM imaging, respectively. Rick Moscati and Julia McIntosh of the USGS provided oxygen isotope data for two samples. We also thank Joe Kowalik who collected several samples of amphibolite from the Standing Pond Volcanic Member for analysis.

Gratitude is expressed to the late Leonard Cook of Norwich, Vermont, who granted full access to mine properties and drill cores that was invaluable to our research. Geologic mine plans and cross sections, made in the 1940s and 50s by staff of the Vermont Copper Company, were loaned by the late John Lyons and the late Half Zantop (both Dartmouth College). We also thank the Harvard University Mineralogical Museum for donating massive sulfide and other mine samples from the H.E. McKinstry collection for study. Greg Walsh (USGS) provided important guidance on regional geology. Bob Seal and Jane Hammarstrom (both USGS) offered detailed comments and suggestions on an early version of the manuscript. Very thorough and constructive journal reviews by Tim Barrett and Jan Peter, and extensive comments by Associate Editor Paul Spry, helped clarify and focus the final version and are greatly appreciated.

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

John F. Slack is an emeritus scientist at the U.S. Geological Survey (USGS) in Reston, Virginia. Since retirement from the USGS in 2016, he has been an adjunct professor at Memorial University of Newfoundland in St. John’s, Newfoundland. He received his Ph.D. degree from Stanford University in 1976. His research has focused on the geology, mineralogy, and geochemistry of diverse types of stratabound mineral deposits, chiefly volcanogenic massive sulfide and sedimentary exhalative ores, plus iron formations, phosphorites, and metalliferous black shales. Recently, his efforts have been mainly on critical mineral deposits and potential in New England. He is a Fellow of the Society of Economic Geologists and the Geological Society of America and a member of the Society for Geology Applied to Mineral Deposits, the Mineralogical Society of America, and the Geochemical Society. He received the Distinguished Service Award of the U.S. Department of the Interior in 2016.