The tungsten-gold veins of Bonnac (French Massif central): new constraints for a Variscan granite-related genesis

The French Massif Central (FMC) represents a prominent metallogenic province for gold (Bouchot et al., 2005) but only a few tungsten deposits have been found, although resources of at least 45,000 t of WO₃ have been estimated (Audion and Labbé, 2012), this is probably the consequence of an under-exploration for this metal in the FMC. The ubiquity of the quartz-wolframite hydrothermal veins makes it the main source of tungsten in the FMC. Some deposits with still a strong economic potential were formerly mined, such as Leucamp, Enguialès and Echassières (Marignac and Cuney, 1999; Fig. 1).

Among the wolframite-bearing veins of the FMC, some display a polymetallic content with bismuth and gold in sub-economic grade (W-As-Bi-Au). They are ascribed either to the perigranitic wolframite vein-type (Marignac and Cuney, 1999) or to orogenic gold deposit (Bouchot et al., 2005). In the FMC, occurrences of W-As-Bi-Au are known in gold mining districts such as the Limousin (Marcoux and Bonnemaison, 1988), Pontgibaud (Marcoux and Picot, 1985), Brioude-Massiac (Périchaud, 1970; Bril, 1982; Marcoux and Bril, 1986; Sandras, 1988; Bril and Beaufort, 1989; Bril et al., 1991), Montredon (Béziat et al., 1980), St-Mélany (Charonnat, 2000; Chauvet et al., 2012), and Leucamp districts (Derré, 1983; Bogdanoff et al., 1987; Demange et al., 1988; Bouchot et al., 2005; Lerouge and Bouchot, 2009). The spatial association of the wolframite-quartz veins with Carboniferous peraluminous two-mica granites argues for an important role of this crustal magmatism in the genesis of the tungsten mineralization (Marignac and Cuney, 1999; Cuney et al., 2002; Bouchot et al., 2005; Harlaux et al., 2018). However, the formation of these peri-granitic deposits is still debated. In particular, the temporal and genetic links between the episodes of magmatism, deformation, and hydrothermal mineralization are not fully clarified.

U-Pb TIMS dating on wolframite from the Limousin, Echassières, Montredon, and St-Mélany districts (Fig. 1) has been recently performed (Harlaux et al., 2018). The obtained U-Pb ages define the existence of three tungsten mineralizing events closely related to magmatic activity:

• Visean to Serpukhovian mineralization (333–327 Ma) coeval with the emplacement ages of large peraluminous two-mica plutons (ca. 335–325 Ma) during the syn-orogenic extension;

• Serpukhovian to Bashkirian mineralization (317–315 Ma) synchronous with the emplacement of syn-tectonic granites (ca. 315–310 Ma) during the late-orogenic extension;

• Gzhelian to Asselian mineralization (around 300–298 Ma), formed during the post-orogenic extension, coeval with the metallogenic peak of the Or300 event during which the main orogenic gold deposits were emplaced in the FMC (Bouchot et al., 2005).

In the central part of the FMC, i.e. in the Pontgibaud, Brioude-Massiac, and Leucamp districts, the timing for the emplacement of the wolframite bearing veins remains poorly constrained. Furthermore, the genetic relationships between the ore deposits and the granitic emplacement is not yet demonstrated. The Bonnac area, in the central part of the Brioude-Massiac district is a good target to address these questions (Fig. 1). Indeed, in this area, a dozen of W-As-Bi-Au bearing quartz veins from the South of the Bonnac village to the Scoufour hamlet (Fig. 3) further north (Périchaud, 1970; Sandras, 1988) are spatially associated with a leucogranitic dyke swarm dated at 322 ± 7 Ma (K/Ar on muscovite, Bril et al., 1991), while the W-As-Bi-Au mineralization yields late Permian age of about 250 Ma (K/Ar on muscovite from Bonnac; Bril et al., 1991). Variable gold grades ranging from 1 to 15 g/t have been reported in ore samples from the region (Périchaud, 1970; Sandras, 1988), but the relationship between gold and tungsten ore is not clearly documented.

The aim of this study is to better understand the W-As-Bi-Au quartz vein deposits in the FMC, and more particularly in the Bonnac area. It focuses on the structural control and the timing of these ore deposits, and their relationships with the Carboniferous tectonics and magmatism, and will be integrated within the hydrothermal framework of the Variscan FMC. Detailed field structural studies, and in situ apatite LA-ICP-MS U/Pb dating were carried out. Mineralogical and geochemical studies were also performed to decipher the tungsten-gold relationships and propose an updated metallogenic model.

Located in the central part of the FMC, the 1000 km² Brioude-Massiac polymetallic district is covered by Cenozoic basaltic flows, and bounded to the northeast by the Oligocene Limagne graben. The district that belongs to the Variscan orogen, is a N120–130°E trending antiform that refolds the stack of metamorphic nappes recognized in the entire Massif Central (e.g.Burg and Matte, 1978; Ledru et al., 1989; Faure et al., 2005, 2009; Figs. 1 and 2). The core of the area is occupied by the Lower Gneiss Unit (LGU) represented by biotite-sillimanite paragneiss, intruded by the Céroux orthogneiss derived from a Cambro-Ordovician alkaline granite (Burg and Matte, 1978; Lasnier et al., 1982; Mathonnat, 1983; Thonat et al., 2014). The LGU is tectonically overlain by the Upper Gneiss Unit (UGU), which is formed by migmatitic paragneiss with rare orthogneiss, and a felsic-mafic association called the “leptynite-amphibolite complex”. The mafic rocks experienced an eclogite-facies or granulite-facies metamorphism during the eo-Variscan evolution (Marchand, 1974; Lasnier, 1977; Bernard-Griffiths et al., 1980).

The nappe stacking in the Haut-Allier area was probably acquired during the early Carboniferous, around 350–340 Ma (monazite U/Pb dating on syntectonic granulite; Pin and Peucat, 1986). Then two extensional events, coeval with granitic intrusions accommodated the crustal thinning widely observed in the FMC (e.g.Malavieille et al., 1990; Faure and Pons, 1991; Faure, 1995):

• the syn-orogenic extension at ca 330–310 Ma, called D4 event (Faure et al., 2009), is characterized by a NW-SE maximum stretching recorded by the syn-tectonic biotite monzogranite and the two-mica granite plutons and their country rocks. In the Brioude-Massiac area, andalusite-leucogranitic dykes and their related two-mica granites (Sandras, 1988; Bril et al., 1991) belong to this D4 event;

• the late-orogenic extension (D5 event, between 310 to 299 Ma; Faure et al., 2009), with a NNE-SSW maximum stretching direction, was contemporaneous with the emplacement of highly differentiated peraluminous intrusions and pegmatites, and rare-metal granites (Cuney et al., 2002). The opening of the Late Carboniferous coal basins belongs to the D5 event (Faure, 1995).

The study area is located in the central part of the Brioude-Massiac polymetallic district which contains more than 200 veins-type deposits and occurrences with variable ore content: W, Sn or W-As-Bi-Au (Bonnac, Vèze-Bosberty veins), Sb (Au), Pb-Zn or F and Ba (Fig. 2). Antimony veins have been the main economic substance mined between 1850 and 1978 with an historical production of 43 000 t Sb metal (Périchaud, 1970; De Gramont et al., 1990). The Bonnac area was only mined for gold between 1880 and 1911 with a production of only 2 kg of gold, and never for tungsten (Périchaud, 1970; De Gramont et al., 1990).

The Bonnac W-As-Bi-Au quartz-veins are spatially associated with Sb (Au) quartz veins (Fig. 2) and are located in the southwestern limb of a regional N120–130°E antiform. All veins seem to crosscut the foliation of the LGU paragneiss (Figs. 2 and 3). Some quartz veins develop immediately at the contact with andalusite leucogranitic dykes (Périchaud, 1970; Sandras, 1988). The ore mineralogy, is mainly represented by arsenopyrite-löllingite with wolframite, minor scheelite, associated with bismuthinite, native bismuth, tetradymite and visible gold in a quartz gangue with muscovite clusters (Périchaud, 1970; Bril, 1982; Bril and Beaufort, 1989).

Detailed studies on fluid inclusions (Bril, 1982), hydrothermal alteration (Bril and Beaufort, 1989), and lead isotopic signature (Marcoux and Bril, 1986) allow to constrain the P-T conditions of vein deposition. The Bonnac veins display two successive hydrothermal stages with an early high temperature (homogenization temperatures between 350–400 °C) CO₂-CH₄-rich ± N₂ aqueous fluid with low-salinity (about 5 equiv wt % NaCl), interpreted as of metamorphic origin, and a late low temperature (homogenization temperatures between 200–300 °C) aqueous fluid, with low-salinity, of meteoric origin, similar to the ore forming fluid from the Sb (Au) deposits. According to the same authors, the formation of the W-As-Bi-Au and Sb (Au) veins could be contemporaneous, the result of a single hydrothermal system coeval with the emplacement of an underlying, albeit hidden, Permian pluton. The ore veins were considered to show a zonal distribution, with W-As-Bi-Au occurring in a limited area, near the intrusion, the Sb (Au) veins being more distal.

Leucogranitic dykes, striking N30°E, N60°E and sometimes N100°E, with a variable thickness (between 50 cm to 20 m), and small extension (less than 3 km), are cropping out in a few clusters in the district, the Bonnac area being one of the most important (Figs. 2 and 3). These dykes display a homogeneous mineralogical and chemical composition in the overall district (Sandras, 1988). Quartz is the most abundant mineral, associated with plagioclase, and sub-euhedral K-feldspar (<5 mm) with biotite aggregates and muscovite. Andalusite is observed as small sized grains (0.6 to 0.8 mm), associated with the K-feldspar and quartz. A temperature emplacement of ca 600–650 °C, and a pressure of 0.4 GPa, was estimated from the andalusite-quartz assemblage (Sandras, 1988). An emplacement age of 322 ± 7 Ma was documented by K/Ar radiometric dating on muscovite (Bril et al., 1991).

The dykes chemical compositions correspond to a peraluminous two-mica granite with 74 to 76% SiO_2, and a Al₂O₃/CaO + Na₂O + K₂O ratio up to 1.6 (see EDM. 1 for analytical details). This bulk chemistry is similar to those found in the Chatelet gold district, and other Visean to Serpukhovian leucogranites from the Limousin district (e.g. St-Sylvestre massif; Scaillet et al., 1996). A local partial melting of the LGU paragneiss could be responsible for the genesis of those intrusive rocks (Sandras, 1988). A spatial and potentially genetic link with the W-As-Bi-Au deposits of the Brioude-Massiac district is assumed. At Bonnac, the dykes are interpreted as the shallow expression of an underlying pluton (Sandras, 1988; Bril and Beaufort, 1989).

The accessory mineral phases were identified using a Merlin compact Zeiss Scanning Electron Microscopy (SEM), co-operated by BRGM-CNRS-Orléans University, and equipped with Energy Dispersive System (EDS) for qualitative analyses. In addition, SEM-cathodoluminescence images have been made using MEB-CL MIRA-3 Tescan (BRGM-CNRS-Orléans University). Chemical analyses of arsenopyrite were obtained with the Cameca SX-Five Electron Probe Micro-Analyzer (EPMA) at the Institut des Sciences de la Terre d’Orléans (ISTO). Analyses were performed using an accelerating voltage of 20 kv, a beam current at 40 nA, a beam diameter of 3 μm, and a counting time of 30 s for Fe, Ni, Co, Sb, Pb, Cu, As, Zn, Ag, Bi, and 60 s for gold. Details on the method, parameters and analytical data are given in ESM2.

U/Pb dating on hydrothermal apatite from the mineralization was conducted in situ by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) at the GeOHeLiS analytical platform (Géosciences Rennes/OSUR, Univ. Rennes) using an ESI NWR193UC Excimer laser coupled to an Agilent 7700 × quadrupole ICP-MS equipped with a dual pumping system to enhance sensitivity (Paquette and Piro, 2014). The analytical procedure followed the method described in Pochon et al. (2016). Details, standards analyses and analytical data are provided in ESM3. All errors are listed at 2σ level. Concordia diagrams and calculated ages have been produced using IsoplotR (Vermeesch, 2018).

The strike of the mineralized quartz veins displays four different orientations similar to those of the leucogranitic dykes. The N30°E trend is represented by the Riolet, Androl, Rode, Borie veins), the N60°E by the Costillon one, while few veins strike N90°E and N160–170°E (Fig. 3). The vein dip varies between 30° to 45° to the SE whatever their strike. Mineralized quartz veins are located in the hangingwall or footwall of the leucogranitic dykes (Figs. 4 and 5). They display similar features all around the area, with a lenticular shape, and a narrow width of 1 to 20 cm. The host-rocks are weakly affected by the hydrothermal alteration. The geometric relationships show that the leucogranitic dykes crosscut the paragneiss foliation, and are cut by the mineralized veins (Figs. 4 and Figs. 4B and 5) with locally developed breccias (Fig. 4B).

Most of the vein wallrocks are fault planes with a down-dip striation, indicating a normal motion (Fig. 5). Some veins display also a striation with a NE 40–50° pitch, in which quartz steps indicate a normal component of displacement (Fig. 5C, D). The observed kinematics of the Costillon vein (Fig. 4A) also indicate a normal motion that complies with the drag folds asymmetry developed around the leucogranitic dykes (Fig. 5B). Late N60–65°E striking sub-vertical barren faults, with gouge, offset the quartz veins (Fig. 4A).

The Scoufour mineralized quartz veins exhibit some differences with the Bonnac ones. Three N10–30°E striking low-angle veins with a dip between 10 and 20° to the SE are observed (Fig. 6). Their thickness is comprised between 30 and 50 cm, but may exceed 1 m. Here, leucogranitic dykes are absent, and the host-rock underwent a hydrothermal alteration related to the mineralization. A greisen-type alteration is thus observed up to 1 or 2 m from the vein, resulting in the bleaching of the paragneiss by transformation of biotite into muscovite with associated silicification. Drag folds observed near the Ferbert vein show a normal motion (Fig. 6B). A late fracturing event, represented by N20–25°E steeply NW dipping barren faults with gouge, is also recognized (Fig. 6A).

All the structural features observed in the Bonnac and Scoufour areas are summarized in Figure 7. In both cases, the emplacement of the mineralized quartz veins occurred during a single deformation phase coeval with a hydrothermal event. The magmatic event responsible for dyke emplacement post-dated the regional metamorphism but preceded the ore veins emplacement. Furthermore, normal faults rework the pre-existing faults that controlled the emplacement of the leucogranitic dykes. Lastly, both dykes and mineralized veins are cut by NE-SW subvertical faults with subhorizontal striation indicating a sinistral motion.

All the mineralized veins of the study area have similar habitus, with a massive infill of coarse white-grey quartz gangue with parallel ribbons of a few centimeters, or aggregates of arsenopyrite and coarse wolframite (Figs. 4A and 8). Secondary scheelite is common in cm-scale pseudomorphic aggregates of early wolframite that may appear as rare relicts and rarely observed as isolated, coarse grains or infilling fractures (Fig. 8).

On the basis of gangue and ore mineral textural observations, 3 stages have been recognized, namely:

• stage 1 with tungsten and arsenic;

• stage 2 with arsenic, and tungsten;

• stage 3 with bismuth, gold, and base metals.

The first opening of the veins corresponds to the deposition of a wolframite and löllingite assemblage in a macrocristalline Qz1 quartz (crystal size >500 μm; Fig. 9A, B). Observed with cathodoluminescence, Qz1 displays a blocky habitus with regular growth bands without significant deformations (Fig. 9B), and numerous fluid inclusions. Coarse euhedral wolframite crystals (up to 1 cm) are associated with cm-size aggregates of löllingite (Fig. 9A). SEM analyses of wolframite exhibits a high iron grade (6–8 wt %) and a lower manganese one suggesting a ferberite. Large apatite crystals (>100 μm) have been observed within Qz1 in close association with wolframite (Fig. 9C). Their euhedral shape and straight grain boundaries suggest it is coeval with Qz1.

The beginning of stage 2 is marked by an intense and heterogenous episode of deformation (Fig. 9A–D), which has affected the previous stage 1 assemblage. Qz1 grains experienced a dynamic recrystallization (Stipp et al., 2002) shown by undulose extinction, subgrain rotation and recrystallization with golfed or jagged shape of the Qz1 boundary, bulging and quartz neograin (Qz2) formation at the expense of the early Qz1 (Figs. 9B, D and 10A and B). Qz2 displays undulose extinction and grain size reduction and is located along 10 to more than 500 μm shear bands that surround the edges of Qz1, or crosscut it.

The crystallization of arsenopyrite (Fig. 9B) and scheelite are coeval with the deformation episode. Arsenopyrite forms small rhombic crystals (up to 100 μm) and fills the fracture network of the stage 1 löllingite or wolframite and sometimes totally pseudomorphs after löllingite. Xenomorphic scheelite forms at this stage by partial or total pseudomorphosis of previous wolframite (Fig. 9A, C).The scheelites, show rare inclusions of niobium oxides rich in niobium as on the Ferbert vein (65 wt % Nb₂0₅), iron (25 wt % FeO), manganese (4.5 wt % MnO), tungsten (2.4 wt % WO₃), titanium (2.2 wt % TiO₂) and scandium (0.9 wt % Sc₂O₃). Those uncommon oxides have been observed in the Puy-les-Vignes tungsten deposit (Harlaux et al., 2015).

The 56 EPMA analyses of arsenopyrite (see Tab. 1 and ESM 2 for detail) have been compared with previous study of Bril (1983) from the Bonnac area and from other W-As-Au regional veins (Vèze and Bosberty, see Fig. 2 for localization). Arsenopyrite has a very similar composition in all deposits with As % average values of 33.4%, 32.8%, and 33.0% for Bonnac, Vèze, and Bosberty veins, respectively. Minor elements, e.g. Co, Ni, Sb, Au are under the detection limit of the method. According to Kretschmar and Scott (1976), and Sharp et al. (1985), the use of the arsenopyrite chemical compositions to assess ore formation temperatures may be considered with a good confidence level, because of the absence of minor element. Assuming the arsenopyrite-pyrrhotite equilibrium, the calculated formation temperature was around 480 °C. These values are in the same range than those deduced from the H₂O-CO₂ L1-type fluid-inclusions (homogenization temperature between 300 and 420 °C; Bril, 1982).

This stage is marked by the deposition of a new discrete quartz generation (Qz3) in the fractures of arsenopyrite or löllingite (Fig. 10C, D). The medium grain-size (around 100 μm) Qz3 displays a subeuhedral shape or elongate-fibrous habitus. Its hyaline aspect is due to the scarcity of fluid and solid inclusions. This Qz3 never experienced dynamic recrystallization (Fig. 10C, D) and is coeval with a bismuth-telluride-gold-base metals assemblage (Fig.10C–E). Rhombic arsenopyrite crystals are commonly crosscut by large mm-size, subeuhedral pyrite, associated with bismuth-telluride and base-metals minerals (Fig. 11D, E). The EPMA chemical analysis of the pyrite from Bonnac veins does not show minor element.

Bismuthinite and native bismuth are the dominant minerals species of this stage 3. They occur as 2 to 80 μm inclusions aligned along the cleavage plane of löllingite (Fig. 11A), also within arsenopyrite, or wolframite (Figs. 10C, D and 11A–C), and as small-size aggregates (<500 μm) or in fractures crosscutting löllingite or arsenopyrite (Fig. 10E, F). Native bismuth is frequently observed as ovoidal inclusions within bismuthinite (Figs. 10F and 11C). Xenomorphic hedleyite (Bi₇Te₃) is the main tellurium mineral, always closely associated with the bismuthinite and native bismuth. Tetrahedrite [(Cu,Fe)₁₂Sb₄S₁₃] and a bismuth-antimony-sulfide, close to the chemical composition of the horobetsuite [(Bi,Sb)₂S₃, have been identified as inclusions in a hedleyite aggregate (Fig. 10F). In the Ferbert vein, a late deposition of Cu-Zn-Pb minerals overprints and brecciating all the previous minerals. Chalcopyrite is the main mineral, with small tetrahedrite inclusions (50 μm). Sphalerite and minor galena complete this assemblage. This late deposition stage is associated with a Qz4 quartz generation displaying undeformed pyramidal grains (Fig. 11E).

Visible gold has been commonly observed in thin-sections from all veins of the Bonnac area. Most of the gold intergrowths with the Bi-Te minerals (Figs. 10B and 11A–C), or more rarely appears within sulfides (mainly chalcopyrite, Fig. 11D, E), or sometimes pyrite or isolated between Qz2 grains. Gold patches range between 1 μm to 25 μm in size but can reach 50 μm within sulfides. All display a xenomorphous rounded shape (Fig. 11). Rare maldonite (Au₂Bi) has been observed by SEM analysis as a few microns inclusions intergrowths with native bismuth.

Ten EPMA analyses of gold grains intergrowths with native bismuth and within chalcopyrite from the Ferbert vein are provided in Table 1 (see EDM 2 for detailed analyses). The small size of the gold grains explains why the analyses do not reach 100%. Silver is the only element present with gold (average of 16.7 wt % Ag). “Gold” of Bonnac area is thus an electrum with a gold fineness of 830 (Morisson et al., 1991).

The three economic-stages described above are summarized in Figure 12, however, this evolution represents increments of a single progressive crystallization-deformation event.

It should be noted that, a poorly developed and uneconomic post-ore stage, macroscopically highlighted by a dark-gray quartz facies, has been only observed in the Ferbert vein. All the previous stages are brecciated and cemented by a chalcedonic quartz gangue with minor carbonates. The associated metallic-minerals are pyrite-marcasite aggregates with pyrrhotite relics, with a new generation of chalcopyrite, and sphalerite. This does not greatly affect the previous assemblages.

Hydrothermal apatite from the early tungsten-arsenic stage 1 (Figs. 9C and 12) of the Androl (Bonnac area) and Ferbert veins (Scoufour area) have been carefully characterized and 15 grains have been selected for dating. Namely, 9 from the Ferbert vein (mineralized quartz vein sample BM46D) and 6 from Androl vein (mineralized quartz vein sample BM39A; ESM 3 for analytical details). The cathodoluminescence image shows growth zones without evidence complex zoning (Fig. 13A, B). Between 2 and 6 U/Pb analysis have been performed on each grain (Fig. 13C, D). Apatite grains from the Ferbert vein have variable ²⁰⁷Pb/²⁰⁶Pb ratios, ranging between 0.1849 and 0.7010. The discordant points plot along a discordia line that yields a lower intercept date of 315.6 ± 2.8 Ma (MSWD = 3.7; Fig. 13C). The data plot along a discordia line yielding a lower intercept date of 323.4 ± 3.2 Ma (MSWD = 2.4, Fig. 13D). The intersect point of the discordia give the isotopic composition of the initial lead of the dated apatite (Fig. 13C, D). Their composition shows some difference between the Ferbert vein and the Androl vein with respectively 0.823 ± 0.005 and 0.853 ± 0.013. Those results are consistent with initial ²⁰⁷Pb/²⁰⁶Pb ratios measured in different nearby mineralized veins, and main lithological formations of the Brioude-Massiac area (Marcoux and Bril, 1986). Thus, the Fournial (Pb-Ag-Sn-As veins) have ²⁰⁷Pb/²⁰⁶Pb ratios different 0.86234 ± 0.00027 (n = 14), but that of the Sb-(Au) vein of the Céroux is closer 0.8598 ± 0.0014 (n = 1). On the other Sb-(Au) veins, the isotopic compositions of the lead remain variable within de district. The aplitic dykes close to a leucogranitic composition from the la Margeride pluton, in the southern part of the district, show the strongest correlation with the Androl vein with a ²⁰⁷Pb/²⁰⁶Pb ratio of 0.8569 (n = 1, composition recalculated at 320 Ma).

During stage 1, the initial vein opening, coeval with the massive crystallization of the Qz1 under low-strain, was contemporaneous with a high temperature (up to 400 °C) paragenesis with wolframite and löllingite from an As and W enriched fluid with low-salinity (about 5 equiv wt % NaCl) CO₂-CH₄-rich ± N₂ aqueous fluid (V-type fluid inclusions in the classification of Bril, 1982).

A ductile/brittle deformation episode marks the beginning of stage 2 (Fig. 12). The first Qz1 experienced a dynamic recrystallization shown by undulose extinction, subgrain rotation recrystallization, bulging and quartz neograin (Qz2) formed at the expense of the early Qz1. Ductile/brittle deformation features argue for a deep-seated formation between 12 to 9 km (Scholz, 1988). This is supported by the range of temperature for this stage, estimated around 480 °C, based on the arsenopyrite thermometry and, in agreement with the estimated temperature range (400 to 500 °C) deduced from the quartz textural observations (e.g.Stipp et al., 2002). Those depth and temperature estimation are in good agreement with the estimated emplacement conditions of the leucogranitic dykes of Bonnac (see Sect. 2.3). The transformations of wolframite into scheelite, and löllingite into arsenopyrite suggest a change in fluid chemistry, still rich in As and W, and marked by an increase in sulfur fugacity and in the calcium activity. The pseudomorphic textures argue for a remobilization of the tungsten from the previous ferberite. The increase of the Ca/Fe ratio in the fluid, associated with a slightly decrease of the temperature, as predicted by Wood and Samson (2000) could possibly correspond to higher fluid interaction with the surrounding metamorphic formations. This fluid could correspond to the L1-type fluid inclusions defined by Bril (1982), with low-salinity (3.5–0 equiv wt % NaCl), lower CO₂ aqueous fluid, and trapping temperature estimated around 300 to 400 °C (Fig. 14).

The onset of stage 3 (Fig. 12) was marked by brittle fracturing. The previous assemblage was micro-brecciated, and invaded by a hyaline, low-strain Quartz (Qz3). This new quartz generation was coeval with the deposition of native bismuth, bismuthinite, hedleyite, electrum, pyrite, and base-metals minerals (chalcopyrite, sphalerite, galena). The microcrack infill as well as fluid and solid-inclusions bands (Fig.10A, B) suggest fluid-assisted fracturation under a hydrostatic regime at a depth comprised between 7 and <5 km (Scholz, 1988). This stage 3 might be interpreted as a consequence of a crack-and-seal trigged by a fluid overpressure mechanism (Ramsay, 1980). The emplacement of the previous massive quartz generations could have sealed the porosity in the fault zone and act like an impermeable barrier to fluid flow allowing an increase in the fluid pressure. The crack-and-seal process has triggered pressure drop from lithostatic to hydrostatic regime, and might be interpreted as a consequence of a fault-valve mechanism (Sibson et al., 1988; Cox et al., 1991; Robert et al., 1995; Gaboury and Daigneault, 2000). The textural evolution reflects the transition from a lithostatic pressure context during stages 1 and 2 with a limited permeability to pervasive fluid circulation in connected fractures, therefore implying a transition from, an aseismic to, a seismic environment (Cox, 1987; Cox et al., 1991).

The deposition temperature of gold during stage 3 can be deduced from the paragenetic association of native bismuth and bismuthinite. According to Barton and Skinner (1979), a range of temperature between 320 to 240 °C can be proposed. However, the association of native bismuth and hedleyite can only be stable below a temperature of 266 °C (Dimitrova and Kerestedjian, 2006). Therefore, we propose a temperature range between 266 to 240 °C for the deposition of stage 3, which is in good agreement with a ca 200 °C homogenization temperature deduced from the L2-type secondary aqueous fluid inclusions (Bril, 1982). This relatively low temperature accounts for the deposition of antimony-rich Bi-Te minerals like the horobetsuite or the tetrahedrite.

The common occurrence of electrum within Bi-Te minerals might emphasize the role of a polymetallic melt/liquid for the concentration of precious metal, as already reported for similar mineralization (Ciobanu et al., 2006; Cook et al., 2007; Ciobanu et al., 2010; Zachariáš et al., 2014). The polyphase character of the ore-inclusions (electrum intergrowth in native bismuth, or electrum, native bismuth intergrowths in hedleyite), the ovoidal shape of the electrum inside native bismuth, and the presence of minerals such as maldonite or horobetsuite might support this mechanism (Ciobanu et al., 2006; Zachariáš et al., 2014). But our deposition temperature estimate, between 266 to 240 °C, is too close to, or even below, the bismuth solidus temperature, and the eutectic temperature of the Au-Bi system (241 °C). This leads us to suggest that a large part of the gold and Bi-Te phases probably precipitated directly form a hydrothermal fluid enriched in Bi-Te-Au and were not scavenged by a bismuth-rich liquid/melt.

Between stages 1 and 3, a temperature decrease is observed (Fig. 14). In addition, the stage 3 underwent a significant pressure drop associated with a change in tectonic regime, which might have triggered the input of a meteoritic fluid (Bril, 1982; Fig. 14). Furthermore, the decrease in the sulfide activity in the ore-fluid could be responsible for the gold deposition by destabilization of gold-sulfide complexes associated with the ongoing sulfide precipitation (Williams-Jones et al., 2009).

The vein geometry observed in the field is controlled by the inherited structures. The medium dip (30–50°SE) of the Bonnac veins reused the previous fault system coeval with dyke emplacement, whereas the low-angle dip Scoufour veins are controlled by the host-rock foliation. These anisotropy surfaces can be easily reactivated during faulting, and channeled the ore-bearing fluids.

Kinematic data on the striated fault planes coeval with W-As-Bi-Au veins allowed us to derive the stress tensor by an optimized inversion method (Delvaux and Sperner, 2003). The result indicates an extensional tectonic regime with the maximum (σ₁), intermediate (σ₂), and minimum (σ₃) principal stress axes oriented 39°E/86°, 37°E/04° and 127°E/0°, respectively (Fig. 15). The W-As-Bi-Au quartz veins deposition was controlled by a NW-SE horizontal extension direction, consistent with the NW-SE stretching direction documented for the D4 tectono-magmatic event in the FMC (Faure, 1995). This widespread event, related to the syn-orogenic Variscan extension in the FMC, dated between 325 and 310 Ma, is contemporaneous with the emplacement of the leucogranitic and monzogranitic syn-tectonic plutons (Figs. 1 and 16; Faure and Pons, 1991; Faure, 1995; Faure et al., 2009; Talbot et al., 2005a, 2005b; Joly et al., 2007, 2009).

The hydrothermal apatite grains from two localities in the Bonnac area yield LA-ICP-MS U/Pb dates of 315.6 ± 2.8 for the Ferbert vein (Scoufour area) and 323.4 ± 3.2 Ma for the Androl vein (Bonnac area), respectively interpreted as the emplacement age of the W-As-Bi-Au veins. This age of ca. 320 Ma is consistent with the D4 tectonic, which is also coeval with the Serpukhovian-Bashkirian generation of tungsten ore deposits with minor gold in the FMC as the Puy-les-Vignes or St-Mélany (Fig. 16; Harlaux et al., 2018). Closer to the Brioude-Massiac district, the Leucamp and Enguialès ore deposits in the Châtaigneraie district yielded the most important tungsten resources of the FMC (Derré, 1983; Bogdanoff et al., 1987; Demange et al., 1988; Lerouge and Bouchot, 2009) are other examples of this tungsten ore deposition event.

All previous works (Périchaud, 1970; Bril and Beaufort, 1989) in the Brioude-Massiac district, proposed a single hydrothermal event, responsible for the formation of both antimony and tungsten-gold from a perigranitic metal zonation. This model seems to be challenged by our results. Despite their spatial link with the antimony veins, the W-As-Bi-Au ones yield a distinct age, since the antimony veins of the Brioude-Massiac district belong to a younger hydrothermal event referred to as the “Or 300” event (Bouchot et al., 2005; Cheval-Garabédian, 2019). We have shown that the Bonnac W-As-Bi-Au mineralizations have a polyphased history in a continuum of time (Fig. 12), which fits with a unique deformation event, decorrelated with the Or300 event. The three mineralized stage are all trapped in quartz-veins with same and distinct structural control (D4 tectonic event), same kinematics, a same infilling chronology with a continuity from a ductile/brittle to only-brittle regime, associated with a progressive cooling of the ore-fluid (Figs. 12 and 14), and by the absence of late remobilization textures during the gold stage 3. This, is also attested by the obvious age difference.

Our results, together with the emplacement ages obtained on the gold-bearing quartz veins from the Cévennes district (Chauvet et al., 2012) argue for an earlier Bashkirian-Moscovian gold-hydrothermal event, sometimes associated with tungsten, during the D4 event. Our data allow us to propose a polyphase history in the FMC characterized by two distinct hydrothermal events carrying gold, namely: (i) a first Serpukhovian to Bashkirian, and (ii) a second, Gzhelian-Asselian, as the Or300 event, that might have re-used the structural discontinuities that controlled the early event. Recent new data obtain on the orogenic-gold deposit of the la Bellière district (Armorican Massif) argue also for a tectonic-hydrothermal gold event earlier than the Or300 event around 340–325 Ma (Cheval-Garabédian et al., 2020).

Based on our results, the genetic relationships between the mineralization and the leucogranitic dykes can also be re-assessed. The leucogranitic dykes from Brioude-Massiac, dated at 322 ± 7 Ma (Bril et al., 1991) were emplaced contemporaneously with the D4 event and are spatially related to the W-As-Bi-Au deposits. The mineralizing event that followed the intrusive rocks emplacement could be interpreted as a magmatic-to-hydrothermal evolution (Fig. 17; Linnen and Cuney, 2005; Gloaguen, 2006) with a relatively long-time duration, around 329–316 Ma (dating of the leucogranitic magmatic event). Several lines of evidence support this interpretation as:

• the spatial and temporal link between dykes and mineralized veins;

• their formation during the D4 tectonic event;

• the high temperature paragenesis of stage 1 at 400–500 °C, close to the 600–650 °C emplacement temperature estimated for the andalusite-leucogranitic dykes;

• their close emplacement depths estimated around 10 km (Sandras, 1988);

• the reworking of previous structural discontinuities.

Throughout the Variscan Orogen, there are examples of W-Au deposits that were emplaced within a time gap of ca. 10 Ma after the intrusions of peraluminous magmas (Romer and Kroner, 2016). In France, the Puy-les-Vignes deposit (Harlaux et al., 2018), and the Salau deposit in the French Pyrenees (Poitrenaud et al., 2019), or in the Bohemian Massif, the Mokrsko deposit (Zachariáš et al., 2014), are good examples of this relation.

A two-stage evolutive model for the magmatic-hydrothermal system in the Bonnac area is proposed in a continuum of time (Fig. 17). Between 329 and 315 Ma, the emplacement of leucogranitic dykes, related to an underlying pluton, was controlled by normal faulting in the regional NW-SE extensional regime. Slightly younger from the dyke crystallizations, between 321 to 316 Ma, the deformation became more localized and contemporaneous with quartz veins emplacement that recorded a magmatic-to-hydrothermal transition (Fig. 17, stage 2). These hydrothermal fluids have reused the previous discontinuities as feeder channels.

The metamorphic or magmatic origin of the ore-forming fluid for peri-granitic ore deposits is still a matter of debate (Lang and Baker, 2001; Boiron et al., 2001; Hart et al., 2002; Bouchot et al., 2005; Hart, 2007; Zachariáš et al., 2014; Marcoux et al., 2015). In the Bonnac area, the magmatic origin of the fluid is not fully demonstrated by the existing fluid inclusion study (Bril, 1982) as they rather display a metamorphic origin (low salinity, complex chemistry enriched in CO₂). However, a magmatic fluid affinity is pointed out by: (i) the presence of magmatic affinity element such as the W and the Bi-Te-Au association (Lang and Backer, 2001; Blevin, 2004) and by (ii) the spatial and temporal link with leucogranitic dykes. In addition, the lead isotopic compositions measured for different mineralized veins and main lithological formations of the Brioude-Massiac area support such magmatic origin of the metals (Marcoux and Bril, 1986). The ²⁰⁷Pb/²⁰⁶Pb ratios estimated on our dating are close to those obtained on the aplitic dykes of leucogranitic composition of the la Margeride granite (see Sect. 6 for the data).

To take into account all those features, we propose a genetic link between those mineralized veins and the leucogranitic dykes, where the initial magmatic fluids were derived from the crystallization of an underlying granitic pluton. The heat source supplied by the pluton was responsible for fluid migration toward the high permeability zones. The high temperature (400–500 °C) interaction between the fluid and the host-rocks (greisen type alteration in the vein hanging walls) could have re-equilibrated the fluid, overprinting its initial magmatic signature. Then fault-valve mechanisms channeled the ore-fluids into structural discontinuities (Sibson et al., 1988), and fluid overpressure associated with potential fault-valve mechanism could be responsible for vein opening and the gold-stage 3 deposition, in a shallow crustal environment.

The chemical composition of the Brioude-Massiac andalusite-leucogranitic S-type magma dykes (EDM 1) is characterized by an oxidizing redox state (Fig. 18), although a reduce redox state is more commonly observed. These characteristics are considered as more favorable than a reduce redox state to form a tungsten ore deposit. Indeed the early magnetite precipitation allows the preservation of Au in the melt fraction of granitic magmas (Blevin, 2004).

Because of the metamorphic origin of fluids, the Bonnac deposits was previously classified as an orogenic gold deposit (Bouchot et al., 2005), although they do not fulfill all the necessary criteria (Groves et al., 1998).

The Bonnac deposits share some of the characteristics of the IRGD type (Thompson et al., 1999; Lang and Baker, 2001; Hart et al., 2002; Gloaguen, 2006; Hart, 2007; Zachariáš et al., 2014; Marcoux et al., 2015) such as:

• the spatial and temporal relationships between ore-deposition and peraluminous granite emplacement;

• the peculiar relation between magmatic-source-related ore phases such as bismuth, tellurides, wolframite, löllingite or molybdenite;

• the contemporaneous bismuth-telluride-gold deposition;

• the vein deposition during the syn-orogenic extensional period.

The redox state of the leucogranitic dykes that spans the boundary of the ilmenite series (Fig. 18) is also a favorable context for IRGD deposits (Lang and Baker, 2001; Blevin, 2004).

On the other hand, the structural evolution from a ductile/brittle deformation to a brittle one, fault-valve mechanism features, and the presence of a late meteoric fluid input, are classical features found for the orogenic gold type deposits (Boiron et al., 1990, 2003; Groves et al., 1998; Bouchot et al., 2005).

Therefore, the Bonnac ore-zone represents a mineralization type intermediate between the orogenic-gold and IRGD models as already demonstrated for the Mokrsko gold deposit in the Bohemian-Massif for example (Zachariáš et al., 2014).

The Bonnac veins show a W-As-Bi-Au paragenesis that results from polyphase hydrothermal and magmatic events. Three stages of ore deposition have been identified starting with wolframite and löllingite in a deep-seated setting, between 12 to 9 km and up to 400 °C. The second ductile/brittle deformation stage was responsible for scheelite and arsenopyrite deposition at 480–300 °C. The late stage was marked by brittle micro-fracturing infilled with native bismuth, bismuthinite, hedleyite, electrum, pyrite and Cu-Zn-Pb minerals at 266 to 240 °C and 7 to <5 km depth. This could be the result of fault-valve mechanism. The decrease of the temperature-pressure and the sulfide activity caused by ongoing sulfide precipitation seems to be the major factors responsible for the gold deposition during the late stage of deposition.

The mineralized veins are spatially and temporally linked to the emplacement of the andalusite-leucogranitic dykes of the Brioude-Massiac district around 329–315 Ma. Our new LA-ICP-MS U/Pb dating on hydrothermal apatite from the deposit documents an emplacement age around 320 Ma. This genetic link suggests a magmatic-to-hydrothermal transition for the Bonnac ore-forming process. In the newly proposed model, the cooling of a peraluminous pluton might have supplied magmatic fluids, and heat source for mineralized fluid flows toward high permeability zones such as previous discontinuities. The magmatic origin of the fluid has been partly overprinted by fluid-rocks interaction.

The emplacement of both the andalusite-leucogranitic dykes and quartz veins was controlled at the regional scale by a NW-SE extensional direction, consistent with the tectono-magmatic D4 event corresponding to the syn-orogenic extension in the FMC Variscan orogen. This geodynamic event is older than the “Or 300” one, suggesting therefore an earlier fertile metallogenic period at ca. 320 Ma for tungsten and gold deposition in the FMC.

The genetic classification of the Bonnac mineralization is equivocal. The W-As-Bi-Au-quartz veins exhibit the features of both orogenic gold (relatively deep level of emplacement) and IRGD model (spatial-temporal link between the mineralization and the peraluminous intrusion). We propose that it represents an intermediate type between the typical orogenic-gold and IRGD model.

The Serpukhovian to Bashkirian peraluminous andalusite-leucogranitic dyke swarms appear as a favorable metallotect for future mining exploration in the FMC. The widespread occurrences related to this type of magmatism in the FMC suggest that other Bonnac-like, or IRGD-like, deposits probably exist. IRGD deposits, and the economic high-grade gold content in tungsten vein-type, might have been underestimated in the FMC.

This work was funded by e-Mines Company through its R&D activity, and the ISTO with contribution from ANRT. The authors thank the following persons for their technical support and valuable advices: T. Poitrenaud, B. Cochelin, A. Beaudoin, R. Augier, C. Montmartin, J. Gouin, M. Picault. We particularly thank I. Di-Carlo (ISTO) for the SEM, cathodoluminescence and microprobe analyses and for valuable advices. S. Janiec, G. Badin, and P. Benoist-Julliot (ISTO) for thin sections preparation, and technical assistance, and the GeOHeLiS platform for the LA-ICP-MS dating. We also gratefully acknowledge Pr. L. Jolivet and Dr. O. Vanderhaeghe, for their efficient editorial handling, and Pr. D. Béziat and one anonymous reviewer for their constructive comments that improved this manuscript significantly.