Abstract

The Buffalo gold deposit is a small deposit consisting of auriferous quartz-tourmaline veins within a granodiorite stock in the Red Lake greenstone belt. This study aims to characterize the mineralizing fluids through fluid inclusion and stable isotope analyses and to compare them with those of the world-class Campbell-Red Lake deposit. Four types of fluid inclusions were recognized, including carbonic, aqueous-carbonic, aqueous, and halite-bearing aqueous, with the carbonic type being the most abundant. Raman analyses indicate that the carbonic phase mainly consists of CO2, with minor amounts of N2 and CH4, and rarely detectable H2S. The homogenization temperatures of the carbonic inclusions range from −41.7° to 30.9°C. The homogenization temperatures and salinities of the aqueous, halite-bearing aqueous, and aqueous-carbonic inclusions are 130° to 276°C and 9.7 to 23.6 wt.% NaCl equiv., 155° to 207°C and 32.9 to 42.3 wt.% NaCl equiv., and 215° to 357°C and 8.3 to 19.7 wt.% NaCl equiv., respectively. The δ18OVSMOW values of tourmaline range from 8.2‰ to 9.0‰, and those of quartz from 11.4‰ to 11.9‰, with estimated fluid temperatures from 323° to 399°C based on the quartz-tourmaline isotopic geothermometer. It is postulated that separate CO2-dominated and aqueous fluids intermittently invaded the fracture/vein system in response to fluid pressure fluctuations, with limited mixing. The CO2-dominated fluid, previously recognized in Campbell-Red Lake as the main mineralizing fluid, is inferred to have been derived from deeper parts of the crust. This deep CO2-dominated fluid reservoir might have been a common source for gold mineralization in the Red Lake greenstone belt.

Sommaire

Le gîte aurifère Buffalo est un petit dépôt composée de veines de quartz-tourmaline à l’intérieur d’un stock de granodiorite dans la ceinture de roches vertes de Red Lake. Cette étude vise à caractériser les fluides impliqués dans la minéralisation par les inclusions fluides et par les isotopes stables ainsi que de les comparer avec ceux du gîte de classe mondiale de Campbell-Red Lake. Quatre types d’inclusions ont été notes, incluant des inclusions carboniques, aqueuses-carboniques, aqueuses et aqueuses avec halite, les inclusions carboniques étant la variété la plus abondante. Les analyses par spectroscopie Raman indiquent que la phase carbonique consiste principalement en CO2, avec des quantités mineures de N2 et de CH4, et rarement avec des quantités mesurables de H2S. Les températures d’homogénéisation des inclusions carboniques vont de −41.7° to 30.9°C. Les températures d’homogénéisation et les salinités des inclusions aqueuses, aqueuses avec halite et aqueuses-carboniques sont varient de 130° à 276°C et de 9.7 à 23.6 poids.% NaCl équiv., 155° à 207°C et 32.9 à 42.3 poids.% NaCl équiv., et de 215° à 357°C et 8.3 à 19.7 poids.% NaCl équiv., respectivement. Les valeurs de δ18OVSMOW pour la tourmaline vont de 8.2 à 9.0‰, et celles du quartz de 11.4 à 11.9‰, avec une température estimée du fluide entre 323° et 399°C sur la base du géothermomètre isotopique quartz-tourmaline. Il est postulé qu’un fluide riche en CO2 et un fluide aqueux distinct ont envahi le système de fractures/veines par intermittence en réponse à des variations dans la pression des fluides, et qu’il y a eu peu de mélange. Le fluide riche en CO2 est également observé à Campbell-Red Lake ou il est le fluide minéralisant principal, il est présumé être dérivé de portions plus profondes de la croûte. Ce réservoir profond de fluides riches en CO2 pourrait être la source commune des minéralisations aurifères de la ceinture de roches vertes de Red Lake.

Introduction

The Archean Red Lake greenstone belt in northwestern Ontario is one of Canada’s foremost gold mining camps, with a total of 105 past and current gold-producing mines and occurrences within an area of about 500 km2 (Sanborn-Barrie et al., 2004). The largest gold deposit in the greenstone belt, the Campbell-Red Lake deposit, has approximately 840 t of gold (past production + reserves) and an average grade of 21 g/t Au (Chi et al., 2006), making it one of the largest and richest gold deposits in the world.

The gold mineralization in the Red Lake greenstone belt has been historically assigned to tectonic environments ranging from syn-volcanism to pre-, syn-, and late-regional deformation and metamorphism by different authors (Kerrich et al., 1981; Andrews et al., 1986; Penczak and Mason, 1997, 1999; Tarnocai, 2000). A recent study by Dubé et al. (2004) indicates that the main gold mineralization took place during the Uchian phase of the Kenoran orogeny (~2.7 Ga, see below), and minor gold mineralization occurred in the late- to post-tectonic stages. These deposits have a wide range of features that indicate they are best classified as orogenic type (Groves et al., 1998; Chi et al., 2009).

Orogenic gold deposits are characterized by mineralizing fluids of aqueous-carbonic type, approximated by the H2O-NaCl-CO2 system (Ridley and Diamond, 2000). The majority of orogenic gold deposits record fluid compositions with H2O > CO2 (Ridley and Diamond, 2000). However, a recent fluid inclusion study of the Campbell- Red Lake gold deposit (Chi et al., 2006) indicates that gold mineralization was related to an unusually CO2-rich, H2O-poor fluid, as indicated by the predominance of carbonic inclusions (without a visible aqueous phase). It is unknown how this fluid was distributed in the Red Lake greenstone belt in terms of time and space, and in particular whether or not it was limited to the deformation zone that hosts the Campbell-Red Lake deposit, or if it was limited to a certain period of time during the evolution of the greenstone belt. To evaluate these questions, it is necessary to study fluid inclusions in gold deposits outside the deformation zone hosting the Campbell-Red Lake deposit. Few fluid inclusion studies (LaKind, 1984; Chi et al., 2009) have been carried out in the Red Lake greenstone belt other than those at the Campbell-Red Lake deposit itself (Tarnocai, 2000; Chi et al., 2002, 2003, 2006). LaKind (1984) studied fluid inclusions from the Fairlie and MacAndrew prospects and the Wilmar mine: only aqueous-carbonic and aqueous inclusions were reported, and bulk fluid inclusion analyses using gas chromatography indicated H2O/CO2 molar ratios from 2.2 to 22.7. Chi et al. (2009) studied fluid inclusions in barren and auriferous carbonate-quartz veins outside the main deformation zone hosting the Campbell-Red Lake deposit (the Red Lake mine trend): carbonic inclusions were recognized as the dominant type, although more aqueous and aqueous-carbonic inclusions were present than in the Campbell-Red Lake deposit.

This paper reports results of fluid inclusion and stable isotope studies of the Buffalo gold deposit, which is hosted in a granodiorite stock emplaced after the main phase of deformation in the central part of the Red Lake greenstone belt (Fig. 1). The objective of the study is to characterize the mineralizing fluids of the Buffalo gold deposit in terms of fluid composition and pressure-temperature conditions, and to compare them with those responsible for the formation of the Campbell-Red Lake gold deposit. In particular, we aim to evaluate whether the CO2-dominated, H2O-poor fluid recognized in the eastern part of the Red Lake greenstone belt (Campbell-Red Lake deposit and the Red Lake mine trend) was present in the central part of the greenstone belt after the main phase of deformation, and if it was present, to determine what relationship it might have had to gold mineralization.

Geological Setting and Gold Mineralization

The Red Lake greenstone belt is part of the Uchi Sub-province of the Superior Province (Fig. 1). It consists of 2.99–2.89 Ga Mesoarchean volcanic-dominated rocks and 2.75–2.73 Ga Neoarchean volcanic rocks, which are separated by an angular unconformity (Sanborn-Barrie et al., 2001, 2004). The Mesoarchean rocks consist of the Balmer, Ball, Slate Bay, Bruce Channel, and Trout Bay assemblages, and the Neoarchean rocks comprised of the Confederation, Huston, and Graves assemblages (Stott and Corfu, 1991; Sanborn-Barrie et al., 2001, 2004). The supracrustal assemblages were affected by two main episodes of penetrative deformation: D1 is related to a E–W shortening stress regime from ca. 2742 to 2733 Ma, whereas D2 reflects N–S shortening related to the collision between the North Caribou terrane to the north of the Red Lake greenstone belt and the Winnipeg terrane to the south (Fig. 1) during the Uchian phase of the Kenoran orogeny, between ~2723 and 2712 Ma (Sanborn-Barrie et al., 2004; Dubé et al., 2004). Postcollisional D3 strain is locally recorded after 2700 ± 6 Ma (Sanborn-Barrie et al., 2004; Dubé et al., 2004). The supracrustal rocks have been metamorphosed to greenschist and lower amphibolite facies, and intruded by a number of granitoid intrusions ranging in age from 2731 to 2697 Ma (Fig. 1). Based on the observation that metamorphic grade increases from the central part toward the margin of the greenstone belt, which is surrounded by large granitoid intrusions, it was proposed that the metamorphism in the Red Lake greenstone belt was of contact metamorphic nature (Andrews et al., 1986; Menard et al., 1999; Tarnocai, 2000). However, as pointed out by Thompson (2003), some of the granitoid intrusions are too young or too old to be the heat source for regional metamorphism, which overlapped with major ductile deformation (D2). Therefore, although granitoid intrusions likely caused localized contact metamorphism, which overprinted regional metamorphism, the metamorphic pattern of the Red Lake greenstone belt mainly reflects the results of peak regional metamorphism during D2 (Thompson, 2003).

Gold Mineralization in the Red Lake Greenstone Belt

A number of gold occurrences and deposits occur in the Red Lake greenstone belt, mainly in the supracrustal rocks, and some in the granitic intrusions (Fig. 1). Gold mineralization took place mainly during D2, broadly synchronous with peak regional metamorphism and extensive granitic plutonism (Dubé et al., 2004), and minor gold mineralization occurred during D3 (Dubé et al., 2004). According to Andrews et al. (1986), most important gold deposits occur along major D2 deformation zones (Fig. 1), although these deformation zones were not confirmed in more recent work by Sanborn-Barrie et al. (2004). Most gold deposits are associated with quartz-sulfide (arsenopyrite, pyrite, pyrrhotite) veins that crosscut and/or replace carbonate veins as well as carbonate and potassic alteration (disseminated sericite/muscovite and some biotite; Dubé et al., 2004; Sanborn-Barrie et al., 2004). According to Parker (2000), the most important gold mineralization (including the Campbell-Red Lake deposit) is located in areas of extensive ankerite alteration (and veining), termed “proximal ferroan-dolomite alteration”, although numerous gold occurrences are also found in areas of “distal calcite alteration” and beyond (Fig. 1).

The Buffalo Gold Deposit

The Buffalo gold deposit is located at the southern margin of a granodiorite intrusion, the Dome stock (about 7 km × 6 km), in the central part of the Red Lake greenstone belt (Fig. 1). The deposit was mined from 1981 to 1982, and produced 1656 ounces of gold at an average grade of 0.11 ounces per ton (Pettigrew, 1999). It is one of several small gold deposits or occurrences hosted within the Dome stock, including the Red Lake Gold Shore and Skookum Gold mines (Sanborn-Barrie et al., 2004). The Dome stock has a zircon U-Pb age of 2718 ± 1 Ma (Corfu and Andrews, 1987), which is within the duration of D2. Lacking a pervasive foliation, the intrusion postdated the main cleavage-forming stage of D2 and associated metamorphism, but predated the termination of D2 at 2712 Ma (Dubé et al., 2004; Sanborn-Barrie et al., 2004).

The Dome stock intruded the Mesoarchean Balmer assemblage, near the intersection of the regional NE-trending Flat Lake–Howey Bay deformation zone and NW-trending Pipestone Bay–St. Paul Bay deformation zone of Andrews et al. (1986; Fig. 1). Several large xenoliths (up to 3 × 10 m) of basaltic, sedimentary, and gabbroic rocks, highly sheared and aligned east–west, are included in the granodiorite intrusion. The deformation within the xenoliths was mainly produced during the early phase of D2, i.e., predating the intrusion of the Dome stock at 2718 Ma. A number of aplite dikes, also oriented east–west, were emplaced within the granodiorite intrusion (Fig. 2). These dikes, together with the xenoliths and the granodiorite, were subjected to east–west shearing, producing a weak east–west foliation in the granodiorite and folding of the aplite dikes in the vicinity of the Buffalo deposit. This east–west foliation, which is consistent with the D2 deformation pattern (Sanborn-Barrie et al., 2004), was probably produced during the late phase of D2, when deformation was localized rather than pervasive.

The deposit lies within the “distal calcite alteration zone” of Parker (2000) and Sanborn-Barrie et al. (2004), but minor ankerite occurs in mineralized veins. Gold is associated with two sets of quartz-tourmaline-pyrite-(chalcopyrite)- carbonate veins oriented NW and NE (Fig. 2). The NEtrending veins are generally very narrow (≤3 cm), whereas the NW-trending veins are up to ~10 cm wide and commonly crosscut the former (Fig. 3a). The two sets are interpreted to be coeval and conjugate, and to have resulted from D3 deformation (Pettigrew, 1999), but it is also possible that they were formed in late D2 deformation (after intrusion of the Dome stock at 2718 Ma, but before the end of D2 at 2712 Ma). Within the veins, quartz and tourmaline occur in bands parallel to the vein margin (Fig. 3b) or locally in different segments along strike of a vein (Fig. 3c). The quartz-tourmaline veins are characterized by an alteration halo of albite-tourmaline-ankerite (weathered to pink on the surface) surrounding the tourmaline-rich portion of the veins, which is absent where quartz is in direct contact with the granodiorite (Fig. 3d). Sulfides and gold mineralization occur both in the alteration halo and in the vein. The gold grades correlate positively with sulfide contents, both in the vein and the alteration envelope (Pettigrew, 1999). The sulfides and associated native gold are disseminated in the veins and alteration envelopes, and they also occur with Ag-Au-Bi tellurides along microfractures in the veins. These relationships were interpreted to indicate that mineralization in the alteration envelopes was coeval with the emplacement of the veins, whereas that in the veins was later than the vein formation (Pettigrew, 1999). However, it is more plausible that mineralization in both the veins and alteration envelopes was introduced by the same fluids that precipitated the main vein minerals in multiple episodes, with sulfides and gold being precipitated later than quartz and tourmaline in a given episode during the process of vein formation.

Study Methods

Eight doubly polished sections (~100 μm thick) were prepared from samples of quartz-tourmaline-gold veins and the alteration envelope (locations shown on Fig. 2). These sections were studied first for petrography (including transmitted and reflected light) to establish the relative timing of minerals, and then used for fluid inclusion microthermometry and Raman spectroscopy. Separated (hand-picked and checked under microscope) quartz and tourmaline samples were analyzed for oxygen isotopes for the purposes of estimating the temperature of vein filling using isotopic geothermometers, and the isotopic composition of the fluids responsible for vein formation.

Fluid inclusion microthermometry was carried out on a Linkam THMS 600 heating/freezing stage at the University of Regina. The stage was calibrated with synthetic fluid inclusions of H2O (ice melting temperature = 0.0°C; critical temperature = 374.1°C) and H2O-CO2 (CO2 melting temperature = −56.6°C). The accuracy is about ±0.2°C for measurements below 31°C, whereas that of total homogenization temperatures is around ±1°C. The fluid inclusion assemblage (FIA) method (Goldstein and Reynolds, 1994) was used to constrain the validity of the microthermometric data: a variation of ≤20°C for total homogenization temperature and ≤5°C for CO2 homogenization temperature is considered acceptable in this study. For randomly distributed and isolated fluid inclusions, the FIA method cannot be strictly applied, and microthermometric data from such inclusions were constrained by neighboring inclusions: a very large difference in homogenization temperature between neighboring inclusions was considered as indicating that the inclusions might have experienced post-entrapment modification (Chi and Lu, 2008).

The compositions of volatiles of the carbonic phase in fluid inclusions were analyzed with a RM2000 Renishaw laser Raman spectroscope (excitation laser wavelength = 514 nm) at the Geofluid Laboratory of the University of Regina. The objective used for analysis is 100×. Silica capillaries filled with gas mixtures of known composition, prepared at the United States Geological Survey using a method developed by Chou et al. (2005), were used as standards for quantification of the gas composition.

Fluid salinities, densities, and isochores were calculated using the CLATHRATES and FLUIDS software packages of Bakker (1997, 2003). Fluid properties of carbonic and aqueous-carbonic inclusions were calculated with the equations of Duan et al. (1996), and those of aqueous inclusions with the equations of Zhang and Frantz (1987). The isochores were calculated with the equations of Duan et al. (1992, 1996) for carbonic inclusions, Bowers and Helgeson (1983) and Bakker (1999) for aqueous-carbonic inclusions, and Knight and Bodnar (1989) for aqueous inclusions.

The oxygen isotope analysis of quartz and tourmaline was done at the Laboratory for Stable Isotope Science, University of Western Ontario. The accuracy of measurement of the δ18OVSMOW standard is better than ±0.17‰, and the precision is better than ±0.14‰. Fluid temperatures were calculated using the quartz-tourmaline oxygen isotope geothermometer (Kotzer et al., 1993), and fluid oxygen isotope compositions were calculated from the quartz-water fractionation equation of Zheng (1993).

Petrography and Paragenesis

The samples studied are from quartz-tourmaline veins and their alteration halos. The fact that the alteration halo is always associated with tourmaline-rich portions of the veins (Fig. 3d) indicates that the wall-rock alteration is contemporaneous with tourmaline precipitation in the vein.

The alteration halo consists mainly of albite (>80%) and significant amounts of tourmaline and ankerite, with minor ilmenite and pyrite. Ankerite occurs as irregular veinlets and patches replacing albite, followed by disseminated pyrite and minor chalcopyrite also replacing albite. Tourmaline occurs in irregular patches replacing albite and ankerite, and is commonly accompanied and locally replaced by pyrite, chalcopyrite, and ilmenite. Ilmenite is locally altered to leucoxene, which also occur as veinlets cutting through pyrite crystals. In our samples, we have not seen native gold in the altered wall rocks, but it was reported by Pettigrew (1999) that gold occurs as 2–3 μm inclusions in carbonate and albite. No gold was observed in pyrite in this study, nor by Pettigrew (1999). However, based on the positive correlation between gold grades and sulfide contents in the wall rocks, it is inferred that sulfide precipitation was a likely cause of gold deposition. This indirect reasoning puts gold in a position with or immediately after sulfides in the paragenesis. The sequence of alteration mineral formation in the halo is therefore albite → ankerite → tourmaline → pyrite (minor chalcopyrite), gold → ilmenite → leucoxene.

Within the veins, quartz and tourmaline occur as alternating bands parallel to the vein (Fig. 3b) or as segments alongstrike of the vein (Fig. 3c). Tourmaline generally occurs as fibrous crystals at high angles to the vein wall. Elsewhere, different portions of the veins are composed of massive tourmaline or quartz. Massive tourmaline is locally crosscut by quartz veinlets, which in turn contain a new generation of tourmaline occurring as rosettes growing on the wall of the veinlets (Fig. 3e). In massive quartz, tourmaline crystals are scattered and commonly broken into multiple segments (Fig. 3f). These relationships suggest that quartz and tourmaline are broadly contemporaneous, and were precipitated in multiple episodes. The massive quartz in the veins consists of coarse crystals, which commonly show deformation features, including undulose extinction, deformation lamellae, subgrains, and aggregates of fine grains. However, quartz veinlets sandwiched in tourmaline layers are much less deformed. Minor amounts of sulfides (mainly pyrite, minor chalcopyrite) are disseminated in various parts of the veins, replacing quartz and tourmaline. Locally, large amounts of pyrite occur in the central part of the veins, crosscutting quartz and tourmaline. No visible gold was found in the veins in our thin sections, but Pettigrew (1999) reported grains of gold and associated tellurides occurring as 5–50 μm inclusions in pyrite crystals, and in microfractures in quartz and tourmaline. Although local crosscutting relationships suggest that sulfides, gold, and tellurides postdate quartz and tourmaline, it does not necessarily indicate that sulfides and gold mineralization took place after the termination of the vein formation. It is more likely that sulfides and gold were introduced into the veins (microfractures) between different episodes of vein increment, and thus sulfides, gold, and tellurides can be considered broadly contemporaneous with quartz and tourmaline, forming from the same hydrothermal system in multiple episodes.

Fluid Inclusion Studies

Fluid inclusions were studied in quartz from the quartz-tourmaline veins; fluid inclusions are rare in tourmaline and most of them are not workable (too small or leaked). Isolated fluid inclusions (Figs. 4a–c) and those in small clusters (Fig. 4d) within least deformed quartz crystals (i.e., those closely associated with and protected by tourmaline) are considered primary or pseudosecondary inclusions, and represent fluids from which quartz (and tourmaline), and by inference sulfides and gold, were precipitated. Fluid inclusions in long-healed fractures or trails (Figs. 4e–h) are treated as secondary inclusions. Because the vein formation consists of multiple crack and seal events, not all secondary inclusions represent fluids trapped after vein formation; many of them might have trapped the same fluids as primary inclusions in other parts of the veins. Most fluid inclusions occur randomly; they cannot be objectively classified as either primary or secondary, and are treated as “undetermined”.

Based on composition and phase assemblages at room temperature, fluid inclusions are classified into four types: carbonic (type c), aqueous-carbonic (ac), aqueous (a), and halite-bearing aqueous (a+h), with type c being the most abundant. Some of the type c and type ac inclusions contain a tiny (<<1 μm) opaque mineral, which is considered to be a daughter mineral because it occurs consistently within a fluid inclusion assemblage or among neighboring inclusions. However, the opaque mineral is too small to be identified microscopically or with Raman. The ratio of the volume of the carbonic phase to the total volume of type ac inclusions ranges from 35% to 90%. Type ac inclusions with large carbonic/total ratios (mostly ≥60%) homogenize into the carbonic phase or by critical behavior, whereas those with small carbonic/total ratios (mostly ≤40%) homogenize into the aqueous phase. Many type ac inclusions decrepitated before homogenization.

Type a, ac, and c inclusions occur both as primary and pseudosecondary inclusions (Figs. 4a–d) and as secondary inclusions (Figs. 4e–g). Type a+h (Fig. 4h) inclusions occur in trails or are scattered, the latter being considered as undetermined. Based on these relationships and the interpretation that gold was introduced before termination of vein formation, it is inferred that type a, ac, and c fluid inclusions represent fluids present during vein formation and gold mineralization. The fluid represented by type a+h inclusions might also be present during vein formation, but was mainly entrapped after termination of vein formation. However, the different types of fluid inclusions were not found coexisting within individual FIA, and so the different fluids might not be in physical contact and in chemical equilibrium during vein formation and gold mineralization.

The fluid inclusion microthermometric results are summarized in Table 1 and illustrated in Figures 5 and 6. Detailed microthermometric data are listed in the Appendix, where the data for individual inclusions are presented for isolated and randomly distributed fluid inclusions, whereas the average for individual fluid inclusion assemblages is listed in order to avoid bias toward FIAs containing numerous fluid inclusions with similar microthermometric attributes (Goldstein and Reynolds, 1994; Chi and Lu, 2008). Each data point in Figures 5 and 6 represents an individual inclusion (for isolated and randomly distributed inclusions) or average of an FIA. The means in Table 1 were calculated from these data points.

From Table 1 and Figures 5 and 6 it can be seen that for a given type of fluid inclusions (type a, ac, a, or a+h), the microthermometric attributes are broadly similar between different occurrences (primary-pseudosecondary, secondary, and undetermined), although some differences have also been noted. For type c and type ac inclusions, primary-pseudosecondary inclusions appear to have lower CO2- melting temperatures than secondary and undetermined ones (Figs. 5a and c). The lowest CO2-homogenization temperatures (−45° to −20°C) appear to be mainly found in type c inclusions of undetermined origin (Fig. 5b). For type a and a+h inclusions, secondary inclusions show a narrower range of homogenization temperatures than primary-pseudosecondary and undetermined ones (Figs. 6a and c), as is also true for halite-melting temperatures for type a+h inclusions (Fig. 6d). These observations are interpreted to indicate that the fluids present during the precipitation of vein minerals (preserved as primary-pseudosecondary and some of the undetermined inclusions) were also circulating in microfractures (preserved as secondary and some of the undetermined inclusions) between different episodes of vein increment, but the activity of the fluids (especially aqueous ones) might have outlasted the vein formation (preserved as secondary inclusions).

Raman spectroscopic analyses of type c inclusions and the carbonic phase of type ac inclusions (Table 2) indicate the carbonic phase is mainly composed of CO2 (87.7 to 100 mole %) with minor amounts of N2 (0 to 10.5 mole %), and CH4 (0 to 3.9 mole %). Minor amounts of H2S (0.1 to 0.9 mole %) were detected in two inclusions. The average compositions of type c inclusions and the carbonic phase of type ac inclusions are similar, with average CO2, CH4, N2 compositions for type c being 97.7 mole %, 0.7 mole %, and 1.6 mole %, and for type ac being 97.8 mole %, 0.2 mole %, and 2.0 mole % (Table 2).

Oxygen Isotopes of Quartz and Tourmaline

The analytical results of oxygen isotope compositions of tourmaline and quartz are listed in Table 3. The δ18OVSMOW values of tourmaline range from 8.2‰ to 9.0‰, and those of quartz fall in a narrow range from 11.4‰ to 11.9‰ (Table 3).

Based on the interpretation that quartz and tourmaline are broadly contemporaneous in the veins and the assumption that both minerals were in equilibrium with the parent fluids, the temperatures of the fluids were calculated from the oxygen isotope fractionation between tourmaline and quartz mineral pairs. Using the tourmaline-quartz fractionation equation of Kotzer et al. (1993), the calculated fluid temperatures are 323°C, 358°C, and 399°C. Using the quartz-water fractionation equation of Zheng (1993) and the calculated temperature of the mineral pairs, the calculated δ18OVSMOW values of the fluids were 5.6‰, 6.0‰, and 7.1‰, respectively (Table 3).

Discussion and Conclusions

The main objective of this study was to compare the small Buffalo gold deposit with the giant Campbell-Red Lake deposit in terms of fluid composition and pressure temperature conditions of gold mineralization. In particular, we wished to investigate whether or not the waterpoor, CO2-dominated fluid recognized in the Campbell- Red Lake deposit was also present in the Buffalo gold deposit, and if so, what role it might have played in gold mineralization. This study indicates that carbonic inclusions are indeed abundant in the Buffalo gold deposit, although aqueous and aqueous-carbonic inclusions are also commonly found, which is different from Campbell-Red Lake where aqueous and aqueous-carbonic inclusions are rare (Chi et al., 2006). Furthermore, the halite-bearing fluid inclusions common in the Buffalo deposit do not occur in the Campbell-Red Lake deposit. The similarities and differences between the two deposits, and their significance for gold mineralization models in the Red Lake greenstone belt, are further discussed below.

Carbonic inclusions without a visible aqueous phase are not uncommon in orogenic gold deposits. According to the statistics of Ridley and Diamond (2000), 38% of orogenic gold deposits contain carbonic inclusions. However, only a few deposits contain predominantly carbonic inclusions with almost no other types of inclusions, including the world-class Ashanti (Schmidt Mumm et al., 1997) and Campbell-Red Lake gold deposits (Chi et al., 2006). The origin and significance of the carbonic inclusions have been a subject of debate. Although carbonic inclusions might be produced by fluid phase separation (e.g., Robert and Kelly, 1987; Guha et al., 1991) and/or preferential leakage of H2O relative to the carbonic components after entrapment (Hollister, 1988, 1990; Bakker and Jansen, 1994; Johnson and Hollister, 1995), these mechanisms cannot satisfactorily explain why so few aqueous inclusions were trapped and/ or preserved in these cases. Therefore, the predominance of carbonic fluid inclusions is interpreted to reflect a CO2-dominated fluid system (Schmidt Mumm et al., 1997; Chi et al., 2006, 2009). The CO2-dominated fluid in the Campbell- Red Lake deposit was interpreted to have originated from granulite facies metamorphism occurring at deeper levels of the crust, and was channeled upward along major deformation zones. Water-dominated fluids from the country rocks were largely prevented from entering the veins by the positive pressure of the CO2-dominated fluid until after the main phase of gold mineralization (Chi et al., 2009).

In the case of the Buffalo gold deposit, the carbonic inclusions are also interpreted to represent a separate fluid phase present during vein formation and mineralization, rather than indicating phase separation or being the result of post-entrapment preferential leakage of water from aqueous-carbonic inclusions, based on the fact that the carbonic inclusions are not associated with aqueous-carbonic inclusions and show consistent homogenization temperatures within individual fluid inclusion assemblages. Furthermore, the abundance of the carbonic inclusions is also interpreted to indicate the predominance of a CO2-dominated fluid in the veining system, possibly linked to the regional deformation zones near the deposit (Fig. 1). The type a and type a+h fluid inclusions are interpreted to represent various aqueous fluids from the country rocks, and the type ac inclusions represent mixtures of the deeply derived CO2-dominated fluid and the aqueous fluids from the country rocks. The mixing of the different fluids might not have been complete as discussed below.

The homogenization temperatures of the type a inclusions (130° to 276°C) are relatively low compared to the homogenization temperatures of type ac inclusions (215° to 357°C). Most of these temperatures are lower than the temperatures calculated from the quartz-tourmaline oxygen isotope geothermometer (323° to 399°C). Because most of the type a and ac and some type a+h inclusions are interpreted to have been entrapped during vein formation (including secondary inclusions entrapped between different episodes of vein increment), their trapping temperatures are likely between 323° and 399°C. This temperature range is slightly lower than that in the Campbell-Red Lake deposit (350° to 550°C; Chi et al., 2009). The fact that homogenization temperatures are lower than the inferred temperature of vein formation suggests that the aqueous fluids were not vapor-saturated at the P-T conditions of entrapment, and a “pressure correction” needs to be applied to the homogenization temperatures to obtain trapping temperatures. Because aqueous and carbonic inclusions are broadly contemporaneous, it follows that the two fluids were in limited physical contact in most cases and were not saturated with each other, otherwise the homogenization temperatures should be equal to the temperature of vein formation.

The hydrothermal system might have been dominated by the carbonic fluid, and the intermittent invasion of the aqueous fluid from the country rocks might have resulted in limited and incomplete mixing with the carbonic fluid. Mixing of different proportions of carbonic and aqueous fluids is reflected by type ac inclusions with variable carbonic/total volume ratios. Mixing of large amounts of carbonic fluid with small amounts of aqueous fluid might have produced a carbonic-aqueous fluid with large carbonic/ total ratios. Conversely, mixing of large amounts of aqueous fluid with relatively small amounts of carbonic fluid might have resulted in an aqueous-carbonic fluid with small carbonic/total ratios. The variation of the homogenization temperatures of the type ac inclusions might be related to variable degree of mixing, and that of the type a and a+h inclusions is likely caused by fluctuation of fluid pressures.

The isochores of type c inclusions suggest a large range of fluid pressure for the temperature range from 323° to 399°C, as indicated by the shaded area in Figure 7. The higher segment of this pressure range more likely represents the correct pressure conditions, whereas the lower segment might have resulted from fluid inclusion density re-equilibration (Vityk and Bodnar, 1998; Klein and Fuzikawa, 2010). It is difficult to draw a line between the two segments, but it is tentatively inferred that the area above the lowest isochore of aqueous inclusions (a2 in Fig. 7), which are more resistant to re-equilibration than carbonicbearing inclusions, might represent the pressure range during the vein formation and mineralization, i.e., from ~680 to ~4580 bars (hatched area in Fig. 7). The isochores of most type a, ac, and a+h inclusions would intersect with the isochores of type c inclusions in the temperature range from 323° to 399°C (Fig. 7), and their relatively low homogenization temperatures can be explained by the pressure effect. For type a+h inclusions, the trapping temperatures must be greater than the halite-melting temperatures (220° to 349°C), and the required “pressure correction” to the halite-melting temperatures is relatively small.

Therefore, based on the fluid inclusion data collected in this study, the Buffalo gold deposit is inferred to have formed from interaction of two fluid systems. The first fluid system, represented by the carbonic fluid inclusions, is dominated by CO2 and derived from the deep crust, whereas the second fluid system, represented by type a and a+h inclusions, consists of aqueous fluids present in the country rocks. It is unknown whether the CO2-dominated fluid was initially water free or contained some water (<20 mole %) and experienced phase separation to produce the water-free carbonic phase (Chi et al., 2006, 2009). The two fluids might have invaded the vein/fracture network intermittently, possibly in response to pressure fluctuations within each fluid system, resulting in limited mixing (represented by the type ac inclusions). Although no visible gold was observed in our samples, indirect evidence suggests that gold was introduced by the same fluids that precipitated the quartz and tourmaline veins. Therefore, the abundance of carbonic fluid inclusions is interpreted to suggest, as in the Campbell-Red Lake deposit, that the CO2-dominated fluid played a major role in gold mineralization.

From this study and previous studies (Chi et al., 2006, 2009), it appears that the CO2-dominated fluid was present in gold deposits in different parts of the Red Lake greenstone belt, including those in the eastern part (the Red Lake mine trend) and the central part (Buffalo). This fluid was not limited to the main deformation period (D2, the Red Lake mine trend) either, because the Buffalo deposit was formed in late- to post-D2. This observation suggests that there was a common reservoir of CO2-dominated fluid beneath the Red Lake greenstone belt, which might have lasted over a prolonged period of time and furnished fluids for gold mineralization across the greenstone belt. The size of the gold deposits depends on the flux and duration of the CO2-dominated fluid, which in turn is related to the hydraulic connectivity and sustainability between the site of mineralization and the source reservoir. In the case of the Buffalo gold deposit, the flux of the deep CO2-dominated fluid system might have been minor compared to the Campbell-Red Lake deposit, thereby explaining the lower grades and smaller scale (subeconomic) of gold mineralization.

In conclusion, this study indicates that the Buffalo gold deposit shows both similarities and differences with the world-class Campbell-Red Lake deposit in fluid composition and pressure-temperature conditions. The predominance of carbonic inclusions in the Buffalo deposit suggests that the CO2-dominated fluid was not limited in the eastern part of the Red Lake greenstone belt, but was possibly present beneath the whole Red Lake greenstone belt. This CO2-dominated fluid reservoir was tapped by major deformation zones to form major gold deposits during the main phase of D2 deformation, and continued to flow during late D2 or D3. The flux of this deep fluid system, however, might have decreased over time. The Campbell-Red Lake deposit was dominated by the deep CO2-rich fluid throughout the mineralization process, whereas in the Buffalo deposit, aqueous fluid from the country rocks intermittently invaded the vein system during vein formation and mineralization. Furthermore, the aqueous fluids in the Buffalo country rocks appear to have had higher salinities than those in the Campbell-Red Lake deposit, and the ambient temperatures at Buffalo were slightly lower. It appears that one of the key factors controlling the formation of large gold deposits in the Red Lake greenstone belt was prolonged focusing of deeply sourced CO2-dominated fluids along a deformation zone; smaller deposits were formed when the connection with the deep fluid reservoir was relatively weak or short lived.

Acknowledgments

This study is supported by an NSERC Discovery grant (to Chi). We would like to thank Claude Resources for permitting access to the Buffalo mine property, and A. Litchblau, C. Storey, K. Williamson, and Y. Liu for assistance in the field work and sampling. F. Longstaffe and K. Law of the University of Western Ontario are thanked for oxygen isotope analysis. The manuscript has benefited from critical reviews by Drs. C. Hart, E. Marsh, and J. Richards.