This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed.

Abstract

The Caledonian orogenic belt of northern Britain hosts some significant quartz vein-hosted gold deposits. However, as in orogenic belts worldwide, the relationship between gold mineralization and regional tectonics, magmatism, and metamorphism is a matter of debate. This is primarily due to the absence of precise temporal constraints for the mineralization. Here we report high-precision 40Ar/39Ar and Re-Os ages for the largest known gold deposit at Curraghinalt (2.7 Moz) in Northern Ireland and use these ages to constrain the regional geologic setting of the gold mineralization and establish a genetic model.

The gold resource is contained in a suite of quartz sulfide veins hosted by Neoproterozoic (Dalradian) metasediments, which have been thrust over an Ordovician island arc (Tyrone Igneous Complex). Previous studies recognized two generations of gold sulfide mineralization and we have identified a third in microshears that cut the veins. In the absence of precise geochronological data, mineralization ages from Ordovician to Carboniferous have been proposed.

We have dated muscovite (40Ar/39Ar) in quartz vein-hosted clasts of Dalradian wall rock to 459.3 ± 3.4 Ma (all 40Ar/39Ar and Re-Os ages herein are reported at the 2σ confidence level including all sources of uncertainty), an age that we interpret as representing the regional cooling path and which provides a maximum age constraint for all gold mineralization. This is consistent with the quartz veins postdating the end of main-stage deformation in the Grampian event of the Caledonian orogeny (ca. 465 Ma).

Molybdenite (Re-Os) and sericite (40Ar/39Ar) from the newly identified gold-bearing microshears (third generation of gold mineralization) yield indistinguishable Re-Os models and 40Ar/39Ar ages, with a combined age of 455.8 ± 3.0 Ma. The radioisotope ages and field evidence temporally constrain gold mineralization at Curraghinalt to the lower Late Ordovician.

Data show that the gold mineralization was emplaced during the Grampian event of the Caledonian orogeny. The ca. 10 Ma maximum possible mineralization interval (462.7–452.8 Ma) for all three episodes of gold emplacement is postpeak metamorphism and main deformation, coinciding with a period of rapid uplift and extensional tectonics following orogenic collapse. While previous studies have suggested the involvement of magmatic fluids in the deposition of the primary gold resource, the absence of magmatism throughout most of the mineralization interval and the nature of the geologic setting suggest that crustal orogenic fluids should also be considered. Overall Curraghinalt displays most of the characteristics of orogenic gold deposits but also some important differences, which may be explained by the geologic setting.

The timing of mineralization at Curraghinalt broadly coincides with the shift from compressional to extensional tectonics. The extensional regime, rapid uplift, and a crustal profile comprising metasediments overlying a still hot island arc were ideal for creating large and long-lasting hydrothermal systems deriving heat, metals, and some of the fluids from the underlying arc.

Introduction

The Curraghinalt gold deposit is hosted by quartz veins, which cut Neoproterozoic Dalradian rocks in the Sperrin Mountains, near the town of Gortin (Fig. 1). The deposit was discovered in 1983 following investigation of a soil arsenic anomaly (Earls et al., 1996) and it is now at an advanced stage of exploration. The current resource is 2.7 million ounces (Moz) Au distributed as measured: 0.02 million metric tons (Mt) @ 21.51 g/t for 10 koz, indicated: 1.11 Mt @ 12.84 g/t for 460 koz, and inferred: 5.4 Mt @ 12.74 g/t for 2.23 Moz. This makes Curraghinalt the largest known gold deposit in the British Isles.

The gold-bearing quartz veins consist of up to four generations of quartz, each deposited from a fluid of different composition and temperature (Wilkinson et al., 1999). Gold mineralization is hosted by the second and fourth generations of quartz (Earls et al., 1996; Wilkinson et al., 1999; Parnell et al., 2000). Views have changed on the genesis and timing of the first and main generations of gold mineralization. Initially it was linked to orogenic collapse during the latter part of the Grampian event of the Caledonian orogeny (Alsop and Hutton, 1993a) but later work by Hutton and others (inParnell et al., 2000) linked it to late Caledonian (Siluro-Devonian) intrusive activity. The second generation of gold has been attributed to mixing of formation waters resident in Dalradian metasediments with basinal brines during Carboniferous basin inversion (Wilkinson et al., 1999).

The “absolute” age of the gold mineralization, in particular, has been poorly constrained due to a paucity of critical geochronological data, which has restricted the development of a robust genetic model for Curraghinalt. A maximum age constraint is provided by the Dalradian wall rocks, which were deformed and metamorphosed during the Grampian event of the Caledonian orogeny in the period ca. 475 to 465 Ma (Dewey, 2005). Fault gouge 40K-40Ar ages of ca. 315 and 325 Ma from shears cutting the veins provide minimum age constraints for the gold mineralization (Earls et al., 1996; Parnell et al., 2000).

Study Aims

Despite the economic importance of quartz vein-hosted gold deposits, a number of important genetic questions remain unresolved. These include the temporal relationship between mineralization, magmatism, metamorphism, and structural events and the timing and duration of hydrothermal activity (Goldfarb et al., 2005). In this contribution we aim to address some of these outstanding issues. To do this we targeted a relatively small deposit with a large preexisting geologic and geochemical database in a well-characterized setting. Petrographic studies have been carried out on 46 polished thin sections from a representative suite of samples from seven of the ten quartz veins constituting the gold resource. This work identified sericite- and molybdenite-bearing microshears cutting the gold-bearing veins. We have dated these two minerals using the 40Ar/39Ar and Re-Os techniques, respectively, to provide a minimum age for the gold mineralization. Further, we have dated muscovite in wall-rock clasts contained within the veins, which offers the potential to obtain a maximum or actual age for the gold mineralization. We present a revised paragenesis and new geochronological data that show all the gold mineralization at Curraghinalt is Late Ordovician and was deposited in a relatively narrow time interval (ca. 10 Ma) during the latter part of the Grampian event of the Caledonian orogeny.

Regional Geology

The geology of the area surrounding Curraghinalt comprises three main groups of rocks: Dalradian metasediments in the Grampian terrane to the north of the Omagh thrust, the Tyrone Igneous Complex in the Midland Valley terrane to the south, and upper Paleozoic sediments which are widely distributed throughout these terranes (Fig. 1).

The Dalradian Supergroup comprises Neoproterozoic metasediments and basic metaigneous rocks deposited on the passive margin of Laurentia ca. 800 to 500 Ma (Strachan et al., 2002; Cooper and Johnston, 2004). To the southeast the NW-dipping Omagh thrust truncates the Dalradian rocks and separates them from the underlying Ordovician Tyrone Igneous Complex (Cooper and Mitchell, 2004). The Omagh thrust is part of a major terrane-bounding structure known as the Fair Head-Clew Bay Line, which is thought to be continuous with the Highland boundary fault in Scotland. Two gold deposits, Curraghinalt and Cavanacaw, and some minor gold mineralization follow the Fair Head-Clew Bay Line in the north of Ireland. Dalradian sedimentation was terminated by an arc-continent collision in the Ordovician (Grampian event of the Caledonian orogeny; Hollis et al., 2013a). Polyphase deformation and regional metamorphism in the Dalradian of western Ireland is dated to between ca. 475 to 465 Ma (Friedrich et al., 1999).

There are three main lithologic groups in the rocks to the south of the Omagh thrust, the Tyrone Central Inlier, the Tyrone Plutonic Group, and the Tyrone Volcanic Group (Fig. 1). The last two constitute the Tyrone Igneous Complex (Cooper and Mitchell 2004). The Tyrone Central inlier is composed of psammitic and semipelitic paragneisses, known as the Corvanaghan Formation (Cooper and Johnston, 2004) that are closely associated with syntectonic leucosomes (U-Pb, ca. 467 Ma, Chew et al., 2008) and are cut by post-tectonic pegmatites. Biotite cooling ages imply that the Tyrone Central inlier was metamorphosed and deformed prior to ca. 468 Ma (Chew et al., 2008). Detrital zircon age profiling suggests an upper Dalradian Supergroup affinity for the Tyrone Central inlier and as such it is interpreted to represent part of an outboard segment of Laurentia, possibly detached as a microcontinent prior to arc continental collision (Chew et al., 2008). The Tyrone Plutonic Group represents the uppermost portion of a ca. 484 to 480 Ma suprasubduction zone ophiolite accreted with the ca. 475 to 469 Ma Tyrone Volcanic Group island arc onto an outboard microcontinental block prior to ca. 470 Ma during the Grampian event of the Caledonian orogeny (Cooper et al., 2008, 2011; Hollis et al., 2012, 2013a, b). Various arc-related granitoids intruded the Tyrone Volcanic Group following a reversal of subduction polarity and have been dated at ca. 470 to 464 Ma (Cooper et al., 2011).

The Grampian orogeny resulted in crustal thickening, nappe structures, and metamorphism (Cooper and Johnston, 2004). Collapse of the orogenic pile followed during the interval ca. 470 to 450 Ma, accompanied by exhumation, extension, and partial melting (Alsop and Hutton, 1993a; Flowerdew et al., 2000; Clift et al., 2004). Final closure of the Iapetus Ocean occurred during the mid-Silurian and resulted in another orogenic event, the Scandian event of the Caledonian orogeny, marked in the north of Ireland by magmatism and further deformation and metamorphism (Kirkland et al., 2013).

Curraghinalt: Geology and Mineralization

In detail the country rocks hosting the gold veins mirror the regional deformation picture (Fig. 2). The dominant structure is the Sperrin nappe and the veins lie on the inverted, NW-dipping limb of this fold (Alsop and Hutton, 1993b; Fig. 3). Peak metamorphism to upper greenschist facies was reached during D3, which is the final compressional phase of the Grampian event. D3 in the Sperrin Mountains marked the emplacement from the northwest of the Dalradian Supergroup over the Tyrone Igneous Complex and Tyrone Central inlier along the Omagh thrust. This event appears to overlap with the intrusion of a suite of arc-related plutons into the Tyrone Volcanic Group between ca. 470 to 464 Ma (Cooper et al., 2008, 2011; Hollis et al., 2012, 2013b), the youngest of which is dated to ca. 464 Ma. This provides a maximum age constraint of ca. 464 Ma for the gold mineralization at Curraghinalt since the quartz veins were not affected by D3. D3 was followed closely by regional-scale extensional shearing attributed to orogenic collapse and in the Sperrin Mountains, this was accompanied by NE-trending quartz veins (Alsop and Hutton, 1993a).

The reported gold resource is distributed within a swarm of 10 NW-SE-oriented and steeply NE dipping quartz veins that are mainly hosted by semipelites of the Dalradian Argyll Group (Fig. 4A). The limits of the veins are mainly defined by drilling and their full extent along strike and downdip is unknown. The resource is close to the contact between the Glengawna (graphitic pelites, semipelites, and psammites) and Mullaghcarn (psammites and semipelites) Formations (Earls et al., 1996; Lawther and Moloney, 2012; Fig. 4B). Two generations of gold mineralization have been identified (Wilkinson et al., 1999; Parnell et al., 2000). The first generation (Q2) is the most economically important and this gold contains small quantities of silver and is accompanied by arsenopyrite, tellurides, and most of the pyrite, whereas the second (Q4) contains more silver and is accompanied by most of the Sb-As sulfosalts, chalcopyrite, and barite and all of the carbonates (Fig. 5A).

Two major E-W-trending and steeply dipping shear zones cut the veins: the Kiln and Crows Foot shears (Fig. 4A). These are part of a 1-km-wide belt of faults that can be traced for several kilometers (Fig 4B). They are aligned with a reentrant in the Omagh thrust zone and are consistent with a lateral ramp in the footwall of the thrust (Parnell et al., 2000). Some of the veins and the Kiln shear are exposed in an adit driven to investigate the Curraghinalt mineralization below surface (Fig. 4A). Dextral movement on the shears is thought to be responsible for the formation of the vein-hosting structures (Clifford et al., 1992; McCaffrey and Johnston, 1996).

A major structural control on the vein-filling structures has been attributed to repeated movements of the Omagh thrust over the footwall ramp. Thus, Parnell et al. (2000) proposed episodic E-SE-directed thrusting of the Dalradian over the Tyrone Igneous Complex over a prolonged period of time (Ordovician to Carboniferous), which led to reactivation of structures hosting the veins and emplacement of the different generations of mineralization. It was suggested that the main (first) generation of gold mineralization was related to Late Caledonian (Siluro-Devonian) movements. In contrast, Alsop and Hutton (1993a) proposed that the gold mineralization (undifferentiated) was related to fluid release during NW-directed downdip extension in the hanging wall of the Omagh thrust. This was linked to collapse of the Sperrin nappe in the latter stages of the Grampian event of the Caledonian orogeny. The veins are cut by NE-trending, NW-dipping normal faults (Parnell et al., 2000), which may be linked to further collapse of the Sperrin nappe, or a later unrelated extensional episode.

Petrographic Study of the Sericite- and Molybdenite-Bearing Microshears

Field relationships

The microshears have been observed in situ in the Curraghinalt adit only, where they cut the T17 vein (Figs. 4A, 6). Here the shears are thin (ca. 1 mm) and dark gray. They are subvertical and strike roughly east-west and as such they are essentially parallel to the Kiln shear, which cuts the T17 vein at this point. Whether this is due to rotation of the T17 vein within the Kiln shear zone (G. Earls, pers. commun.) or is a primary orientation remains unknown. Their orientation in core and the deposit-wide orientation is also currently unknown.

The microshears have been identified in six out of seven quartz veins sampled and in 22 out of the 46 polished thin sections we examined. It is uncertain whether the shears are restricted to veins. While they can be readily identified megascopically in quartz veins, they are difficult to see in pelitic wall rock. However, the presence of numerous Mo-rich zones in wall-rock intersections in cores suggests that they are present.

Mineralogy

The microshears range in thickness from 0.05 to 1 mm and consist predominantly of fine-grained (10–30 μm) sericite with patchy and locally abundant molybdenite, minor pyrite, and rare carbonate, chalcopyrite, electrum (at. % Au 65.0–78.9), a Bi-Ag telluride, a Ti oxide, zircon, and apatite (Fig. 7). Quartz adjacent to the sericite may be locally fine grained and recrystallized, reflecting granulation during brittle shearing. There are some mineralogical similarities between the microshears and the two stages of sulfide and precious metal mineralization recognized by Parnell et al. (2000), such as pyrite, chalcopyrite, a Bi telluride, and electrum. However, there are also important differences (i.e., the presence of molybdenite and sericite and the absence of quartz).

Paragenesis

A detailed study of the 45 polished thin sections (Table 1A) provided by Dalradian Gold plus 1 polished thin section (A3929) selected by ourselves, using combinations of transmitted light (TL), reflected light (RL), backscanning electron microscopy (BSEM), and cathodoluminescence (CL), was carried out to characterize and establish precisely the paragenetic position of the microshears. During the course of this work the deposit paragenesis of Earls et al. (1996) and Parnell et al. (2000) was confirmed. Some of the BSEM and the entire quantitative mineral analyses were carried out in the Department of Geology and Petroleum Geology, University of Aberdeen, using an ISI-ABT55 SEM fitted with a LINK ANALYTICAL AN10/55S system. Quantitative analyses were acquired and processed with the LINK ZAF4/FLS program. The combined BSEM and CL studies were carried out at the Imaging Spectroscopy and Analysis Centre (ISAAC) facility of the University of Glasgow with an FEI Quanta 200F field emission SEM using a KE developments panchromatic CL detector.

An important feature of the earlier studies by Wilkinson et al. (1999) and Parnell et al. (2000) is that, using combined CL and BSEM analysis, four generations of quartz (Q1-Q4) were recognized within the deposit. Two of these (Q2 and Q4) have been linked to episodes of gold sulfide mineralization (Wilkinson et al., 1999, Parnell et al., 2000). Our combined CL/BSEM analysis focused on four polished thin sections from a main vein (106.16) and an unnamed minor vein (Table 1B), which showed critical relationships between the microshears and key minerals in the deposit paragenesis.

All four sections showed four generations of quartz generally similar in character to those described by Wilkinson et al. (1999) and Parnell et al. (2000; Fig. 8A, B). In all sections Q1 and Q2 are volumetrically the most important. These two generations of quartz show some differences from these earlier studies in terms of the clast to cement ratio and the presence of zoning. While Q1 is only seen as clasts, as reported by Wilkinson et al. (1999) and Parnell et al. (2000), the proportion of Q2 cement varies significantly, so that some breccias are clast supported and may contain minimal amounts of “dark” (in CL) Q2 cement, whereas others are more fully cemented and Q2 may be the dominant component. Also, Q1 clasts showing oscillatory zoning are common and some sector zoning may be present (Fig. 8A, B). Q3 occurs in veinlets cutting Q1 clasts and Q2 cement. Under CL imaging, it can show concentric zonation but does not have the brecciated appearance of Q1 (Fig. 8A). The final generation of quartz (Q4) shows the brightest luminescence of all and cuts all previous generations of quartz and early pyrite (Fig. 8A).

Various crosscutting relationships show that the microshears not only postdate gold-bearing pyrite of the first generation of gold mineralization (Q2) but importantly, also quartz, chalcopyrite, and carbonates of the second generation of gold mineralization (Q4) (Figs. 9, 11–13). Second generation chalcopyrite was distinguished on the basis of contained angular inclusions of arsenopyrite (Fig. 10) (unique to stage 1 mineralization, Parnell et al., 2000) and also inclusions of electrum with at. % Au in the range 69.2 to 77.9 (characteristic of second generation gold, Parnell et al., 2000). Fe-rich carbonates are characteristic of the second generation of gold mineralization and continued crystallizing after the other minerals (Parnell et al., 2000, fig. 8). We have observed carbonates (siderite) not only being cut by the microshears (Fig. 13A) but also rare ferroan dolomite veins cutting the microshears (Fig. 13B). No sulfide-bearing veins have been observed cutting the microshears.

These relationships show clearly that the microshears are a new third generation of sulfide and gold mineralization and, accordingly, we present a revised paragenesis (Fig. 5A, B). Thus, by dating the microshears, a minimum age for the first two generations of gold mineralization can be obtained.

Dating the Gold Mineralization

During the petrographic study we searched for minerals with which to directly date the gold mineralization. The only possibilities found were pyrite and chalcopyrite but their Re contents were too low (<0.5 ppb) for determination of precise Re-Os ages. As a result we decided to date metamorphic muscovite in wall-rock clasts enclosed within the quartz veins in the hope that the 40Ar/39Ar systematics of the muscovite had been reset by the thermal pulse associated with the first generation of gold (below). Constraints on the maximum age of mineralization are already provided by the age of regional metamorphism, and deformation of the vein wall rocks, and by a D3 deformed granite located within the Tyrone Igneous Complex (ca. 464 Ma, discussed above). Dating of the sericite-molybdenite shears (third gold phase) would provide a minimum age for the first and second generations of gold mineralization (the main resource). This approach will allow for temporal constraints to be placed on the duration of mineralization.

Estimate of maximum age of mineralization

The gold-bearing quartz veins contain clasts of pelitic wall rock (Figs. 14, 15) as well as of quartz (Q1). The pelitic clasts are largely composed of muscovite that shows a strong foliation, probably corresponding to a composite S2-S3 fabric developed over much of the Sperrin Mountains (Alsop and Hutton, 1993a). Minerals from the first two generations of gold mineralization fill the spaces between different clasts within the veins and thus demonstrate that clast formation predated these mineralization events. Providing there has been no thermal disturbance to the muscovite 40Ar/39Ar systematics then an 40Ar/39Ar age would provide a maximum age for gold mineralization, the muscovite 40Ar/39Ar age potentially recording the time at which the terrane cooled from peak metamorphic temperatures (ca. 475–465 Ma, discussed above) through the muscovite closure temperature (Tc) window (ca. 400°C). However, we know that the maximum ambient fluid temperature (Tmax) reached during the first generation of gold mineralization was ca. 400°C (Parnell et al., 2000) and as such, we have to consider the possibility that the 40Ar/39Ar system will have been reset and potentially that the muscovite may record the actual age of this generation of mineralization (i.e., the 40Ar/39Ar system will postdate the age at which the terrane cooled through the muscovite closure temperature). The degree of resetting will be dependent on the duration of Tmax as well as the effective diffusion dimension (i.e., crystal radius) of the muscovite Ar diffusion domain (e.g., Mark et al., 2008; Harrison et al., 2009). Note that the maximum fluid temperatures of the second stage of gold mineralization are significantly below the muscovite closure temperature and thus later resetting of the system is unlikely.

Estimate of minimum age of mineralization

Dating of sericite and molybdenite in the sericite-molybdenite shears that cut Q1–Q4 will not only provide a direct age for the newly discovered third generation of gold mineralization, but will also constrain a minimum age for the first two generations of gold mineralization. Note that shearing is a mechanical process and on the scale observed within the polished thin sections this process will have had no impact on the 40Ar/39Ar systematics of the muscovite within the wall-rock clasts (Kirkpatrick et al., 2012).

Samples for radioisotopic dating

All samples for dating (Re-Os and 40Ar/39Ar) were obtained from core and hence have good geologic context as well as petrographical control. Sericite-molydenite shears were located during the petrographic study and also by examining Mo-rich intersections identified from assay data. The majority of the shears found by the latter method lacked sufficient molybdenite for Re-Os dating, although most had sufficient sericite for 40Ar/39Ar dating. Only one sample (B1633) had sufficient quantities of both minerals for a dual analysis using both techniques. Suitable wall-rock clasts within quartz veins were identified during the petrographic study and sections/ wafers of rock were prepared for dating. The four shears selected for dating are from two of the main veins (Sheep Dip and 106-16) and have similar characteristics (grain size, thickness, mineralogy) to those seen in all the examined samples (Table 2). Specifically, we dated a shear with coexisting sericite and molybdenite and a shear containing sericite only from the Sheep Dip vein, and molybdenite and sericite from separate shears cutting vein (106-16).

Muscovite within three wall-rock clasts hosted by two different gold-bearing veins (DAL 49 and DAL 6) were targeted for in situ UVLAMP (e.g., Mark et al., 2010b) 40Ar/39Ar dating (Fig. 15). The clasts are muscovite rich and show a foliation typical of the pelitic wall rocks (Fig. 14B). The muscovites show no trace of alteration or recrystallization.

Two samples of molybdenite-bearing shears (A8406 and B1633) were selected for Re-Os analysis (Fig. 7, Table 2). Two sericite-bearing shears were also targeted for in situ UVLAMP (e.g., Mark et al., 2007) 40Ar/39Ar dating within two different gold-bearing veins (DAL 33 and DAL 42; (Fig. 16, Table 2). A third fine-grained sericite- and molybdenite-bearing shear from a different vein (B1633, Fig. 7) was also analyzed. The sericite in this sample was sufficiently abundant for harvesting using standard mineral preparation techniques (e.g., Mark et al., 2010a) and subsequent 40Ar/39Ar incremental step-heating (e.g., Mark et al., 2011a).

Analytical Methods

Re-Os dating

Two molybdenite separates from sample A8406 were obtained. One separate was isolated using traditional mineral separation protocols, e.g., crushing, magnetic Frantz separation, heavy liquids, and water flotation (Selby and Creaser, 2004). Although several tens of milligrams were isolated the molybdenite in sample A8406 was extremely fine grained and thus much of the molybdenite remained either enclosed in the quartz vein or was lost in the fine fraction during sieving or washing of the sample. A second mineral separate was achieved utilizing the HF isolation approach (Lawley and Selby, 2012). The latter uses concentrated HF at room temperature for the digestion of the silicate material hosting the molybdenite. Molybdenite from sample B1633 was also isolated using the HF method.

The Re-Os analytical protocol follows that described by Selby and Creaser (2001), with a slight modification to the isolation protocol of Re. An aliquot of molybdenite doped with a known amount of tracer solution comprising 185Re and normal Os isotope composition was loaded into a carius tube with a 1/3 mL mix of concentrated HCl and HNO3 and sealed, and then heated to 220°C for 24 hours Os was isolated from the acid solution using solvent extraction with CHCl3 and further purified using microdistillation. The Re was isolated using solvent extraction by NaOH and Acetone, and then further purified using anion HNO3/HCl chromatography. The Re and Os fractions were analyzed for their isotope compositions using negative Thermal ionization mass spectrometry (N-TIMS, Creaser et al., 1991; Volkening et al., 1991) using a Thermo Electron TRITON mass spectrometer. Measurements were made statically using the Faraday Cups. The measured Re and Os isotope ratios were oxide corrected offline. The data were corrected for fractionation. Analytical uncertainties are propagated and incorporate uncertainties related to Re and Os mass spectrometer measurements, blank abundances, and isotopic compositions, spike calibrations, and reproducibility of standard Re and Os isotope values. Procedural blanks conducted during the period of the molybdenite analysis are negligible relative to the Re and Os abundances measured in the samples (Re = 1.1 ± 0.3 ppt, Os = 0.5 ± 0.3 ppt, 187Os/188Os = 0.23 ± 0.05; n = 3). In-house reference solutions run during the analysis (Re Std. = 0.59763 ± 0.00108; DROsS = 0.16084 ± 0.00003; n = 3) are similar to long-term reproducibility data reported by Lawley and Selby (2012, and references therein). The Re-Os ages are presented as model ages from the simplified isotope equation [t = ln(187Os/187Re + 1)/λ], where t = model age, and λ = 187Re decay constant and assumes no initial radiogenic Os. Inclusion of decay constant uncertainty and reporting of data with 2σ uncertainty allows for direct comparison of the Re-Os ages with the 40Ar/39Ar ages.

40Ar/39Ar dating

Samples for in situ UVLAMP 40Ar/39Ar dating were prepared as doubly polished fluid inclusion wafers using the evolved approaches of Mark et al. (2005, 2006, 2008). Sericite for incremental step-heating was prepared using the methodology outlined in Mark et al. (2010b). Briefly, samples were crushed gently in a mortar and pestle and subjected to magnetic separation. The sericite-bearing fraction was run down a shaking table and a relatively pure sericite split collected. Subsequently clean grains were handpicked under a binocular microscope. All samples (wafers and separates) were cleaned in acetone and deionized water. They were parcelled in high-purity Al disks for irradiation. Standards Fish Canyon sanidine (28.02 ± 0.16 Ma, Renne et al., 1998), GA1550 biotite (98.79 ± 0.96 Ma, Renne et al., 1998) and Hb3gr hornblende (1073.57 ± 5.31 Ma, Jourdan et al., 2006) were loaded adjacent to the samples to permit accurate characterization of the neutron flux (J parameter). Samples were irradiated for 2,700 min in the Cd-lined facility of the CLICIT facility at the OSU TRIGA reactor. Standards were analyzed on a MAP 215-50 system (described below briefly and in more detail by Ellis et al., 2012)—Fish Canyon sanidine was analyzed by CO2 laser total fusion as single crystals (n = 20), GA1550 (n = 20) was also analyzed by CO2 laser total fusion, and Hb3gr was step-heated using a CO2 scanning laser (n = 5; Barfod et al., 2014). Using GA1550 the J parameter was determined to a precision approaching 0.1% uncertainty. Using the J-parameter measurements from GA1550 ages were determined for Fish Canyon sanidine and Hb3gr. The ages overlapped at the 68% confidence (1σ) with the ages reported by Renne et al. (1998), showing the J parameters determined from GA1550 to be accurate.

Wafers were loaded into an ultra-high-vacuum (UHV) laser cell with an SiO2 window. In situ UVLAMP Ar extraction was conducted using a New Wave UP-213-nm UV laser system (described in Moore et al., 2011). Raster pits, 50 × 50 × 5 μm2 (amounts of ablated material approximately 1,250 μm3) were made in mineral surfaces to extract the Ar isotopes. All gas fractions were subjected to 180 s of purification by exposure to two SAES GP50 getters (one maintained at room temperature, the other held at ca. 450°C). A cold finger was maintained at −95.5°C using a mixture of dry ice (CO2[S]) and acetone. Ion beam intensities (i.e., Ar isotope intensities and hence ratios) were measured using a MAP 215-50 mass spectrometer in peak jumping mode. Measurements were made using a Balzers SEV-217 electron multiplier. The system had a measured sensitivity of 1.12 × 10−13 mol/V. The extraction and cleanup, as well as mass spectrometer inlet and measurement protocols and data acquisition were automated. Blanks (full extraction line and mass spectrometer) were made following every two analyses of unknowns. The average blank ± standard deviation (n = 28) from the entire blank run sequence was used to correct raw isotope measurements from unknowns. Mass discrimination was monitored by analysis of air pipette aliquots after every five analyses of unknowns (n = 13, 7.21 × 10−14 mol 40Ar, 40Ar/36Ar = 289.67 ± 0.63).

The sericite separate was step-heated using a CO2 laser (approx. 500°–1,500°C, optical pyrometer measurements). Extracted gases were subjected to 300 s of purification by exposure to two SAES GP50 getters (one maintained at room temperature, the other held at ca. 450°C). A cold finger was maintained at −95.5°C using a mixture of dry ice (CO2[S]) and acetone. Ion beam intensities (i.e., Ar isotope intensities and hence ratios) were measured using a GVI ARGUS V noble gas mass spectrometer in “true” multicollection mode (Mark et al., 2009). Faraday cups (1011 ohm 40Ar, 1012 ohm 39–36Ar) were used to make measurements. The system had a measured sensitivity of 7.40 × 10−14 mol/V. The extraction and cleanup, as well as mass spectrometer inlet and measurement protocols and data acquisition were automated. Blanks (full extraction line and mass spectrometer) were made following every two analyses of unknowns. The average blank ± standard deviation (n = 15) from the entire blank run sequence was used to correct raw isotope measurements from unknowns. Mass discrimination was monitored by analysis of air pipette aliquots after every five analyses of unknowns (n = 16, 7.32 × 10−14 moles 40Ar, 40Ar/36Ar = 299.76 ± 0.23).

All Ar isotope data were corrected for backgrounds, mass discrimination, and reactor-produced nuclides and processed using standard data reduction protocols (e.g., Mark et al., 2005) and reported according to the criteria of Renne et al. (2009). The atmospheric argon isotope ratios of Lee et al. (2006), which have been independently verified by Mark et al. (2011b), were employed. The decay constants of Steiger and Jaeger (1977) were used as an intermediary step in the calculation of 40Ar/39Ar ages. The Berkeley Geochronology Center software (MassSpec) was used for data regression. 40Ar/39Ar ages presented relative to the standard ages of Renne et al. (1998) and the decay constants of Steiger and Jaeger (1977) include only analytical uncertainties (termed “analytical precision) reported at 2σ uncertainty level to allow for the internal comparison of our 40Ar/39Ar ages at the highest precision. Following determination of weighted average, 40Ar/39Ar ages for each sample of the data were reprocessed and ages were determined relative to the statistical optimization model of Renne et al. (2010, 2011) and are reported including full systematic uncertainties (termed “full external precision”) at the 2σ level. This final step allows for accurate comparison of the 40Ar/39Ar and Re-Os ages using the most comprehensive set of physical constraints of any of the various 40Ar/39Ar calibrations in use today, as highlighted by Mark et al. (2014), Renne et al. (2013), and discussed in Renne et al. (2014). This approach is discussed further in the “Results” section below.

Results

All Re-Os isotope measurements are presented in Table 3 and the 40Ar/39Ar data are presented in Appendix File DM1.

Shear molybdenite, Re-Os

Data for sample A8406 (A and B) yield two Re-Os model ages that are indistinguishable (455.8 ± 5.5 and 454.1 ± 5.5 Ma, analytical precision only). The Re and 187Os abundances for aliquots A and B are 1.6 and 3.5 ppm, and 7.5 and 16.7 ppb, respectively. The two analyses show variable abundances of Re and 187Os owing to quartz contamination of aliquot A as discussed in the “Analytical Methods” section above. Aliquot B was a pure molybdenite separate. We have taken a weighted average of the two Re-Os ages (± analytical precision) and subsequently propagated the systematic decay constant uncertainty into the age data. The data yield a weighted average Re-Os age of 455.0 ± 3.9/4.4 Ma (2σ, analytical/full external precision).

Sample B1633 possessed a higher abundance of Re (15.2 ppm) and 187Os (73.8 ppb). The sample yielded an Re-Os age of 459.1 ± 2.2/2.7 Ma (2σ, analytical/full external precision). Sample B1633 is indistinguishable at the 2σ level from the weighted mean Re-Os age of sample A8406.

Wall-rock clast-hosted muscovite in quartz veins, UVLAMP 40Ar/39Ar

In situ laser ablation analyses (samples DAL 6A, DAL 6B, and DAL 49) of the muscovite within wall-rock clasts entrapped in the quartz veins (Fig. 15) yielded a normally distributed data population with an age of 455.2 ± 1.4 Ma (analytical precision, MSWD 1.2, n = 12; Fig. 17). All analyses had greater than 90% radiogenic 40Ar (40Ar*). The data plotted on an isotope correlation plot yield an inverse isochron-data plot close to the x axis (39Ar/40Ar) and give an age that is indistinguishable from the weighted average age of the data population, but define a low-precision y axis reading that is indistinguishable from the atmospheric 36Ar/40Ar ratio (Lee et al., 2006).

Sericite in shears, UVLAMP 40Ar/39Ar

Sericite in shears from within two quartz veins (samples DAL 42 and DAL 33) were analyzed using in situ UVLAMP 40Ar/39Ar dating (Fig. 16). Data yielded weighted average 40Ar/39Ar ages of 449.1 ± 4.0 Ma (analytical precision, MSWD 0.3, n = 6) and 449.4 ± 4.4 Ma (analytical precision, MSWD 0.5, n = 8), respectively. There was more atmospheric contamination in the shear sericite than the wall-rock muscovite, as evidenced by relatively low 40Ar* yields of between 50 and 80%. The lower 40Ar* yields however allowed for more spread of the data on the isotope correlation plots with both samples defining inverse isochrons with ages that are indistinguishable from the weighted average ages, and initial trapped components of atmospheric composition (Lee et al., 2006). The mean age from all analyses of sericite in the sheared quartz veins is 449.2 ± 3.0 Ma (analytical precision, MSWD 0.5, n = 13; Fig. 17).

Sericite in shears, incremental heating 40Ar/39Ar

Sericite from the same shear as molybdenite in sample B1633 was incrementally step-heated. The data yield a plateau with greater than 80% 39Ar (n = 7) and an age of 420.6 ± 1.4 Ma (analytical precision, MSWD 1.6; Fig. 18). The first two age steps step up from ca. 320 Ma to the plateau, suggesting that there has been some disturbance to the sample at this time. When cast on an isotope correlation plot the plateau steps define an inverse isochron with age indistinguishable from the plateau age and an initial trapped 40Ar/36Ar component indistinguishable from atmospheric argon ratios (Lee et al., 2006). Note that the plateau age is much younger than the coeval Re-Os age for this sample (discussed later).

40Ar/39Ar ages relative to modern standard ages and decay constants (Renne et al., 2010, 2011)

The 40Ar/39Ar method is a relative dating technique with all ages referenced back to a standard of known age and decay constant. Renne et al. (2010, 2011) published an optimization model that used constraints from 40K activity, K-Ar isotope data, and pairs of 238U-206Pb and 40Ar/39Ar data as inputs for estimating the partial decay constants of 40K and 40Ar*/40K ratio for Fish Canyon sanidine. This calibration has reduced systematic uncertainties (i.e., 40Ar/39Ar accuracy) to less than 0.25%. To present the most accurate ages for the dated samples and to produce Re-Os comparable 40Ar/39Ar data we have run all ages through the optimization model (the model spread sheet obtained directly from Paul Renne, Berkeley Geochronology Centre). Data are reported at the 2σ uncertainty level as ± analytical/systematic uncertainties. The following ages are used throughout the remaining text. Their robustness and equivalence to the Re-Os ages are discussed below: Wall-rock muscovite (DAL 49 and DAL 6): 459.3 ± 2.8/3.4 Ma, sericite from shears (DAL 33 and DAL 42): 453.3 ± 3.0/3.6 Ma, and sericite from shear (B1633): 424.4 ± 2.8/3.8 Ma.

Discussion

Timing of terrane cooling through the muscovite Tc postpeak metamorphism

Most geochronological data for the Grampian event of the Caledonian orogeny in the British and Irish Caledonides comes from western Ireland and northeastern Scotland. Studies in western Ireland (closest to Curraghinalt) indicate that peak metamorphism and the main deformational episodes occurred during a brief period of time between ca. 475 to 465 Ma (e.g., Friedrich et al., 1999). High-temperature conditions lasted no longer than emplacement of the Oughterard granite which marks the end of Grampian magmatism in Connemara, western Ireland (U-Pb zircon: 463 ± 3 Ma) and thereafter the cooling history was relatively simple with surface temperatures being reached by ca. 443 Ma (Friedrich et al., 1999).

Friedrich et al. (1999) analyzed micas by 40Ar/39Ar dating, their data indicating that southern Connemara cooled from 460° to 350°C between 460 to 450 Ma. There is a large database of mica cooling ages from Connemara (Elias et al., 1988; Clift et al., 1996), the Ox Mountains in northwest Ireland (Flowerdew et al., 2000) and northeastern Scotland (Dempster, 1985; Dempster et al., 1995). Soper et al. (1999) noted that most of these fall in the range 465 to 440 Ma and provide further evidence of the regional cooling history. Closer to Curraghinalt a single 40Ar/39Ar biotite cooling age of 468 ± 1 Ma from the Tyrone Central inlier suggests that cooling was more rapid in this part of the orogen (Chew et al., 2008). At first inspection our wall-rock muscovite clast age (459.3 ± 2.8/3.4 Ma) agrees well with the cooling path for southern Connemara which, like Curraghinalt, is part of the Grampian terrane but is clearly different to that of the Tyrone Central inlier, which lies in the Midland Valley terrane (Fig. 19). We can envisage our 40Ar/39Ar data supporting one of two different scenarios.

Scenario 1

The simplest solution (Occam’s Razor) and our favored interpretation is that the wall-rock mica 40Ar/39Ar age of 460 Ma from within the vein-supported clast is recording the time that the terrane cooled through the muscovite closure temperature of 400°C following peak metamorphism. This scenario provides a maximum age constraint to the gold mineralization. We note for later discussion and in support of this scenario that the temperature for the wall rocks at ca. 460 Ma is close to that of the first generation of gold as determined by fluid inclusion data (Wilkinson et al., 1999; Parnell et al., 2000).

Scenario 2

Given that the metamorphic temperatures reached in southern Connemara (upper amphibolite facies) were higher than the Sperrin Mountains (upper greenschist facies), differences in the cooling histories could be expected. In this case we have to allow for the possibility that the terrane actually cooled to below the muscovite argon closure temperature before 460 Ma and it was the hot fluids associated with the first generation of gold mineralization that reset the 40Ar/39Ar age to ca. 460 Ma. This would mean that our 40Ar/39Ar age for the wall-rock mica within the vein-supported clast actually dates the first generation of gold mineralization. The combination of the maximum temperature (400°C) reached and the likely maximum duration of the fluid-driven thermal pulse (Q2, 0.1–1 Ma, which is based on the cooling time of a small single-phase pluton and the duration of a mesothermal orogenic hydrothermal system; Cathles et al., 1997; Weatherly and Henley, 2013) would have impacted the Ar systematics of the muscovite. To interrogate this we have run a series of DIFFARG models (Wheeler, 1996; Mark et al., 2008) to show the effect of a 400°C pulse of 0.1, 0.5, and 1 Ma duration (covering all likely worst-case scenarios for the duration of the mineralization episode) on the age of a 460 Ma muscovite crystal with an appropriately defined effective diffusion dimension. We measured some of the muscovite crystals (Fig. 14B). The width of the crystals (short axes) ranges from 10 to 30 μm with an average of around 20 μm. For a 20-μm-sized muscovite crystal (i.e., effective diffusion dimension of 20 μm) exposed to a 400°C pulse for 0.1, 0.5, and 1 Ma, the 40Ar/39Ar ages are completely reset (Fig. 20). In fact, exposure to this temperature for less than 0.01 Ma actually resets the system fully.

Although plausible scenario 2 requires reheating of the region by pulsing of hot fluids through the fault-fracture system shortly after cooling from temperatures in excess of 500°C (upper greenschist facies), and we feel there is no need to invoke such a mechanism at this time given the coincidence between muscovite closure temperature, terrane temperature, and fluid inclusion data detailed in scenario 1. It is important to note that although we favour scenario 1, both scenarios effectively yield a maximum age for the first generation of gold mineralization.

Combined 40Ar/39Ar and Re-Os age for the sericite-molybdenite shears

The Re-Os and 40Ar/39Ar ages for the molybdenite and sericite, respectively, from samples DAL 33, DAL 42, A8406 A-B, and B1633 are equivalent at the 2σ uncertainty level. Therefore, we have determined a combined Re-Os and 40Ar/39Ar age for the sericite-molybdenite shears by calculating the mean age ± standard deviation (reported at the 2σ uncertainty level). The combined age is 455.8 ± 3.0 Ma (allowing for all sources of random and systematic uncertainty).

Significance of age data to timing and duration of mineralization

As discussed above, the most likely interpretation of our clast-hosted muscovite age is that it represents a regional cooling age and provides a maximum age for the first generation of gold mineralization. The age of the third and last generation of gold mineralization is provided by the combined Re-Os and 40Ar/39Ar ages, which yield a weighted average age of 455.8 ± 3.0 Ma for sericite and molybdenite coexisting with electrum in microshears cutting the first two generations of gold mineralization. This age further provides a minimum age constraint for the first two generations of gold mineralization, which includes the main gold resource and constrains the maximum possible duration of mineralization to a window of ca. 10 Ma (i.e., 462.7–452.8 Ma; Fig. 21).

Temporal correlations between mineralization and regional events

These new chronological data show that mineralization closely followed the end of the main tectonomagmatic events (ca. 465 Ma) of the Grampian event of the Caledonian orogeny (Fig. 22). There is good evidence that at this time uplift, exhumation and orogenic extensional collapse were taking place in the Sperrin Mountains and elsewhere in the British and Irish Caledonides. Mica 40Ar/39Ar cooling ages from the British and Irish Caledonides have been interpreted to support rapid exhumation and cooling of the nappe pile (Friedrich et al., 1999; Soper et al., 1999; Flowerdew et al., 2000). Collapse of the orogen soon after peak metamorphism and main-stage deformation is indicated by extensional structures in the Sperrin Mountains that can be traced for at least 100 km along strike into southern Donegal (Alsop and Hutton, 1993a) and detachment faulting in western Ireland (Clift et al., 2004). Simultaneously, Dalradian sediments in Connemara were rapidly exhumed, possibly at rates of at least 0.7 km/Ma (Power et al., 2001). The radioisotopic ages obtained herein from a well-characterized paragenetic sequence indicate that the main gold resource at Curraghinalt was emplaced during this regional extensional regime, as originally envisaged by Alsop and Hutton (1993a). It has been proposed that the vein-filling structures were formed during compressional movements on the Omagh thrust over a footwall ramp (Parnell et al., 2000). However, the timing of mineralization indicates that the vein structures were more likely formed as the ramp area became a zone of extension during hanging-wall collapse.

There is no well-documented magmatism in the Sperrin Mountains or northern Ireland that spans the interval of mineralization (462.7–452.8 Ma) or correlates with the precisely determined timing for the third phase of mineralization. However, there is an overlap between the upper end of this interval and the youngest (464 ± 2 Ma) intrusion (Pomeroy granite) of a suite of arc-related Tyrone Volcanic Group plutons (Cooper et al., 2008, 2011; Hollis et al., 2012, 2013b; Figs. 1, 21). Therefore, it is conceivable that subduction-related activity may have continued beyond ca. 464 Ma in the Tyrone arc after it was over-ridden by the Laurentian margin (Earls et al., 1996). Farther afield (200 km from Curraghinalt) in the orogen, there is magmatism that correlates with the window of mineralization. The Oughterard granite in western Ireland has been dated at ca. 462 Ma (Friedrich et al., 1999) and in northeast Scotland the S-type Kennethmont granite has been dated at ca. 458 Ma (Oliver, 2001).

There is a clustering of Rb-Sr pegmatite ages ca. 455 Ma in the Ox Mountains in northwestern Ireland and the Tyrone Central inlier (Flowerdew et al., 2000; Chew et al., 2008) but the significance of these is uncertain owing to problems interpreting Rb-Sr ages. Those in the former area may represent resetting due to a later regional fluid circulation event (M. J. Flowerdew, pers. commun.). This interpretation could be applied to the latter and may be linked to a late orogenic switch from compressional to extensional tectonics (discussed below). Despite some doubt about the exact age of the pegmatites there is a spatial and likely genetic connection between their intrusion and late orogenic crustal extension (Alsop and Hutton, 1993a; Flowerdew et al., 2000).

Sedimentation was occurring in the Pomeroy forearc basin that bounded the accreted Tyrone Igneous Complex by the Late Ordovician and possibly earlier (Cooper and Mitchell, 2004). The sediments are exposed only in a small inlier about 25 km southeast of Curraghinalt (Fig. 1) and so the full areal and stratigraphic extent of the basin is unknown. The oldest known sediments belong to the Soudleyan substage (ca. 454–452 Ma; Mitchell, 1977), which overlaps with the lower end of the mineralization interval.

Sources of fluids, heat, metals, and sulfur in the Curraghinalt mineralizing system

Using two independent geochronological systems in conjunction with a detailed reappraisal of the paragenesis, this study has established that all three generations of gold mineralization at Curraghinalt were deposited in the Late Ordovician (462.7–452.8 Ma) at a time of uplift, exhumation, and regional extension of the Grampian orogenic belt. Armed with these new radioisotopic ages and observations and data from previous studies we now reexamine some existing theories regarding the sources of heat, fluids, metals, and sulfur in the mineralizing system.

Fluids and heat

We cannot rule out the involvement of magmatic fluids in the first generation of gold mineralization (Parnell et al., 2000) but there is an alternative scenario that we consider more plausible. The case for a magmatic fluid rests on fluid inclusion data that identified a fluid with a generally higher temperature (<420°C), higher salinity (10 wt % NaCl + KCl equiv), and low CO2 content (ca. 15 wt %) than typical metamorphic fluids (Parnell et al., 2000). However, metamorphic or highly exchanged meteoric fluids with these characteristics could have been generated in the geologic setting identified in this study. For example, deeply penetrating meteoric fluids with similar temperatures have been described from the Southern Alps in New Zealand (Menzies et al., 2014), where there is no known igneous activity. The unusually high temperatures found at a shallow level here were likely achieved as a result of rapid uplift bringing hot rocks close to the surface.

A similar situation may have prevailed in the Irish segment of the Grampian orogenic belt at this time where high uplift rates are indicated by a concentration of mica cooling ages (above) and by fluid inclusion data, suggesting uplift rates of at least 0.7 km/Ma (Power et al., 2001). Considering scenario 1 for the 40Ar/39Ar data presented above, we also know that the crust was hot (close to 400°C, muscovite argon closure temperature) at the time of mineralization and therefore capable of producing crustal fluids at a similar temperature to the first generation of gold. It is also possible that the underlying arc was still hot given that the age of the youngest known arc-related granite at ca. 464 Ma overlaps within uncertainties the age of mineralization (Fig. 21) and provided an important heat source during the mineralization interval.

These considerations offer an alternative explanation for the unusual combination of high temperatures and shallow emplacement depth found at Curraghinalt (trapping pressures mainly <1 kb and epithermal vein textures are consistent with an emplacement depth of <10 km, Parnell et al., 2000). Also, since metamorphic dehydration reactions may be driven by late orogenic rapid uplift (Yardley and Cleverly, 2013) hot CO2-rich metamorphic fluids might be anticipated at similar shallow depths and mixing of these with meteoric fluids could generate the CO2 levels observed in the Curraghinalt fluids. Furthermore the moderate fluid salinities found in the Q2 mineralization (Wilkinson et al., 1999) could be generated by metamorphism of the Dalradian sediments. Shelf sediments, such as those in the Dalradian sequence, can yield fluids with salinities similar to those seen at Curraghinalt (Yardley and Graham, 2002).

In contrast to these fluids, the sources of the fluids for the second generation of gold are well constrained and involve a mixture of formation waters and basinal brines (Wilkinson et al., 1999). However, these fluids are of Ordovician rather than Carboniferous age as proposed by Wilkinson et al. (1999). The source of the basinal fluids is unknown but they may have been derived from a forearc basin, such as that occurring at Pomeroy to the southeast of Curraghinalt (Fig. 1). Another possibility is that the brines were squeezed from the overridden Tyrone Volcanic Group by compression (e.g., Mark et al., 2007). This could explain the abundance of Cu in this generation of mineralization. The east-west shears at Curraghinalt (Fig. 4) are possible fluid conduits for migration of such brines.

The origin of the fluids responsible for Q1 and the sericite- and molybdenite-bearing microshears is poorly constrained at present. A metamorphic fluid has been suggested for Q1 (Parnell et al., 2000).

The timing of the hydrothermal activity broadly coincides with the switch from compression to extension during orogenic collapse and this may have triggered the release of gold-bearing fluids (Alsop and Hutton, 1993a). At Curraghinalt this was reflected in a switch from SE-directed compressional movements on the Omagh thrust to NW-directed hanging-wall collapse and perhaps also the strike-slip faulting responsible for the formation of the gold vein-hosting structures. It is a common feature of hydrothermal systems that the shift from high-angle reverse to strike-slip faulting is favorable for a high fluid flux and mineralization (e.g., Goldfarb et al., 1991).

Metals and sulfur

Published S and Pb isotope data are consistent with the underlying Tyrone volcanic arc being not only a source of heat and fluids as discussed above but also S and Pb by leaching (Parnell et al., 2000; Standish et al., 2014). These elements could be derived from the widespread precious and base metal enrichments and volcanogenic massive sulfide and porphyry-style mineralization hosted by the arc (Clifford et al., 1992; Hollis et al., 2014). There is also isotopic evidence that some of the Pb may have been derived from the Dalradian metasediments hosting the mineralization (Standish et al., 2014). The enrichments of Mo, Bi, and Te at Curraghinalt suggest the involvement of granitic rocks in the metal reservoir and a possible source, again by leaching, are the abundant granites in the Tyrone Igneous Complex. Differences in the mineralogy between the three generations of gold mineralization (Fig. 5) may be attributed in part to changes in the metal reservoirs. Such changes would be expected in an active tectonic setting and might include episodic movements along the Omagh thrust, allowing access to different portions of the Tyrone Igneous Complex.

Evidence for post-Ordovician activation of fault structures

The age of 424.4 ± 2.8/3.6 Ma obtained for fine-grained sericite in sample B1633 is notably younger than the coeval molybdenite dated at 459.1 ± 2.2/2.7 Ma from the same sample. There is localized magmatic activity (trachy-andesite dikes) occurring at this time within the western Sperrin Mountains (Cooper et al. 2013) but this localized magmatic activity is unlikely to have disturbed the 40Ar/39Ar system in our samples as the nearest known dike is more than 10 km from Curraghinalt. However, it is possible that later fault movements have disturbed the 40Ar/39Ar system in sample B1633 with the age of 424.4 ± 2.8/3.6 Ma correlating broadly with final mid-Silurian closure of the Iapetus Ocean in northern Ireland (Chew and Stillman, 2009; Cooper et al., 2013). Possibly the stresses set up by the collision reactivated some of the shears and induced a grain-size reduction of the sericite. Grain- size reduction is known to reset the 40Ar/39Ar system and could be responsible for the younger 40Ar/39Ar age seen in this sample (Henderson et al., 2011). The coeval molybdenite dated at 459.1 ± 2.2/2.7 Ma is unlikely to have been affected by this process, partly due to the high closure temperature of the Re-Os system in molybdenite (Suzuki, 1996; Selby et al., 2002).

The 40Ar/39Ar step-heating spectrum for sample B1633 (Fig. 18) also shows a disturbance at ca. 320 Ma. This age is close to fault gouge K-Ar ages (ca. 315 and ca. 325 Ma, Fig. 21) in the area reported by Earls et al. (1966) and Parnell et al. (2000), lending support to the suggestion that these fault structures were reactivated during the Variscan orogeny.

Is Curraghinalt an orogenic gold deposit?

Given that Curraghinalt may not be intrusion related, we consider whether it is an orogenic gold deposit as defined in Goldfarb et al. (2005). The following features support this alternative classification. The ore grade is high. It shows a close association with a convergent plate margin. It postdates peak metamorphism and was emplaced into rocks affected by greenschist facies metamorphism during the later stages of still ongoing regional deformation. The temperature, salinity, and CO2 content of the fluids responsible for the main gold resource are within the ranges attributed to these deposits (Goldfarb et al., 2005), as is the metal assemblage of As, Bi, Sb, and Te (although W is lacking). Nevertheless, there are some unusual features: (1) the combination of a high temperature and shallow depth of emplacement, which, in the absence of an obvious magmatic source, is likely explained by a rapid rate of uplift (2) the Cu content is high but this is linked mainly to the later incursion of basinal brines rather than the fluids responsible for the main gold resource; and (3) the enrichments of Mo, Bi, and Te are more typical of intrusion-related gold deposits but could be explained in this case by leaching of granites in the underlying Tyrone island arc. We conclude that these unusual features can be explained by the geologic setting and that Curraghinalt is probably an orogenic gold deposit.

A genetic model for Curraghinalt

Many aspects of the Curraghinalt mineralization such as the mineralogy, geochemistry, fluid composition, and structural controls that are crucial for developing robust genetic models have been established by previous studies (Earls et al., 1996; Wilkinson et al., 1999; Parnell et al., 2000). The main contributions of this study toward understanding the genesis of Curraghinalt have been to refine the paragenesis and, more importantly through dating of the mineralization, to establish precisely its regional geologic setting. From these data a conceptual genetic model may be constructed for Curraghinalt, which can be used to inform exploration programs for similar gold deposits in the Caledonides and collision zones elsewhere.

The salient features of the setting are the following: (1) a collision zone where a continental margin has been obducted over an island arc; (2) following rapid late orogenic uplift the crustal profile at Curraginalt during the mineralization interval consisted of a thick sequence of metasediments overlying a still hot and fertile (mineralized) arc; and (3) subsequent orogenic collapse and extension resulted in increased crustal permeability. This created an ideal setting for developing large and long-lasting hydrothermal systems and the emplacement of metalliferous mineralization, i.e., multiple sources of heat, fluids, and metals in the arc and an overlying permeable crust charged with crustal orogenic fluids where fluid mixing and precipitation of metals could occur. This ideal setting may explain why Curraghinalt is the largest known gold deposit in the British and Irish Caledonides.

Conclusions

A third and new generation of gold mineralization has been identified at Curraghinalt occurring in microshears cutting the first two phases of gold-sulfide mineralization. The shears contain mainly sericite with minor pyrite and molybdenite, trace electrum, and a BiAg telluride.

All gold mineralization at Curraghinalt formed in the ca. 10 Ma interval (462.7–452.8 Ma) during the Late Ordovician. It closely followed peak metamorphism associated with the Grampian event of the Caledonian orogeny and is temporally linked with an extensional setting following orogenic uplift and collapse.

The late orogenic setting offers an alternative classification for the mineralization and we suggest that Curraghinalt is more likely an orogenic- rather than intrusion-related gold deposit. In the former scenario crustal orogenic fluids are alternatives to magmatic fluids and rapid uplift could explain the unusual combination of high temperatures and shallow emplacement of the first generation of mineralization at Curraghinalt. Brines occurring in ore fluids responsible for the second generation of mineralization may have been sourced from the Tyrone Igneous Complex or from a nearby forearc basin.

Data from this and previous studies are consistent with a genetic model whereby a combination of the extensional setting, rapid uplift, and a hot island arc underlying a thick sequence of metasediments favored the formation of large, long-lasting hydrothermal systems with metals, sulfur, and fluids derived in part from the arc.

This study highlights the importance of obtaining precise geochronological data for vein-hosted gold mineralization at an early stage of a study so that its geologic setting can be accurately identified and a robust genetic model constructed.

Acknowledgments

We are grateful to Dalradian Gold Ltd. for providing the sections for petrographic analysis, geochemical data, and general support. We would also like to thank the following: John Still, Alison Sandison, and Jenny Johnston of the School of Geosciences, University of Aberdeen, for assistance with the SEM studies (JS) and with preparing figures (AS and JJ); NERC for ongoing funding of the Argon Isotope facility at SUERC; Jim Imlach and Ross Dymock at SUERC for technical assistance; and Martin Lee at the School of Geographical and Earth Sciences at the University of Glasgow for use of the SEM/CL equipment. The paper has benefitted significantly from comments by the official reviewers and unofficial reviews by Garth Earls, Jamie Wilkinson, Mark Cooper, and Adrian Boyce, and detailed conversations with Ian Alsop (structural geology of the Sperrins) and Nyree Hill and Gawen Jenkin (gold mineralization in the Caledonides). The authors are entirely responsible for the conclusions expressed.

*
A supplementary Appendix to this paper is available at http://economicgeology.org/ and at http://econgeol.geoscienceworld.org/.

Figures and Tables

Supplementary data