Timing of Gold Mineralization Relative to the Peak of Metamorphism at Bronzewing, Western Australia

Transcription

1 2007 Society of Economic Geologists, Inc. Economic Geology, v. 102, pp Timing of Gold Mineralization Relative to the Peak of Metamorphism at Bronzewing, Western Australia F. L. ELMER, CSIRO Exploration and Mining, PO Box 1130, Bentley, Western Australia 6102, Australia R. POWELL, R. W. WHITE, School of Earth Sciences, The University of Melbourne, Victoria 3010, Australia AND G. N. PHILLIPS Range River Gold Ltd, c/- PO Box 3, Central Park, Victoria 3140, Australia Abstract The Bronzewing gold deposit is hosted within the middle to upper greenschist facies, tholeiitic metasalt of the Yandal greenstone belt within the Archean Yilgarn craton. Gold mineralization at Bronzewing is surrounded by an alteration halo hundreds of meters in thickness, composed of distal chlorite, intermediate chlorite-calcite, and proximal carbonate-k-mica zones. Chlorite (in the distal zone) and scovite and otite (in the proximal zone) define a strong foliation that is locally crenulated, indicating that alteration minerals and K 2O addition were produced prior to the crenulation event. Within the proximal alteration zone, otite, ankerite, and calcite cut the crenulated foliation-forming potassic minerals, implying changes in metamorphic conditions subsequent to mineralization. Mineral equilibria modelling indicates that the stle mineral assemblage of otite-scovite-calciteankerite-alte-quartz can only be produced at a temperature of out 440 C (at 2.5 kbars) with a fluid composition of out X CO2 = However, the textural relationships cannot be explained, and the observed alteration zones surrounding mineralization cannot be reproduced by infiltration of fluid (X CO2 = ) into a slightly carbonated actinolite-bearing mineral assemblage at this temperature. A fluid of X CO2 = , between 330 and 375 C (at 2.5 kbars), can reproduce the observed pre-peak metamorphic alteration assemblages at Bronzewing. Calculated internal buffering paths, which model the mineral assemblage and evolution of fluid composition upon further heating, show that the postmineralization assemblage can be explained by further heating to the peak of metamorphism at out 440 C with fluid composition evolving to X CO2 = The implication of the proposed timing relationship is that synmineralization alteration at Bronzewing occurred at temperatures significantly lower than that of peak metamorphism (60 120ºC lower). Introduction THE BRONZEWING gold mine, situated in the Yilgarn craton, Western Australia, is hosted within middle to upper greenschist facies mafic rocks. It has a total endowment (ore reserves, mineral resources, and past production) of 18.4 million tonnes (Mt) of ore at 4.0 g/t gold for 73 t of gold, extending over an area of <2 km 2 and to a depth of 580 m (Kohler et al., 2003b). Bronzewing belongs to a coherent group of gold deposits in Archean greenstone belts in Australia (e.g., Yilgarn craton: Phillips, 1986; Mueller and Groves, 1991; Groves et al., 1992; McCuaig and Kerrich, 1998), Canada (e.g., Superior provinces and Archean Slave province: Corfu and Andrews, 1987; Kerrich and Cassidy, 1994; Robert and Poulsen, 1997) and Africa (e.g., Barberton greenstone belt: Foster and Piper, 1993) which account for a significant part of the world s gold production. When considering the possible sources of ore fluids for greenstone-hosted gold-only deposits, it is critical that the timing of mineralization relative to peak metamorphism be constrained. A broadly synpeak metamorphic timing for gold deposition for higher metamorphic grade deposits has been proposed on the basis of textural relationships, the high variance of mineral assemblages, and the occurrence of fine oscillatory zoning in vein minerals. Corresponding author: , powell@unimelb.edu.au However, this timing has been questioned in studies of other deposits (Phillips and DeNooy, 1988; Tomkins and Mavrogenes, 2001) for which gold mineralization prior to the peak of metamorphism has been suggested. Recent advances in thermodynamic datasets (Holland and Powell, 1990, 1998), computer methods of modelling (Powell and Holland, 1988; Powell et al., 1998), and new activity models for key minerals involved in the alteration of mafic metavolcanic rocks (Dale et al., 2005) have meant that modelling in appropriate chemical systems is now possible. In this paper, the metamorphic evolution of the altered rocks at Bronzewing is investigated via a series of calculated T-X pseudosections testing two hypotheses. The first hypothesis suggests that fluid infiltration and mineralization was synpeak metamorphism, and the second suggests that fluid infiltration and mineralization occurred prior to the peak of metamorphism. T-X pseudosections are phase diagrams drawn for a range of bulk rock compositions and are calculated in the systems Na 2 O-CaO-FeO-MgO-Al 2 O 3 -SiO 2 -CO 2 -H 2 O and Na 2 O-CaO- K 2 O-FeO-MgO-Al 2 O 3 -SiO 2 -CO 2 -H 2 O to predict the conditions of stality of mineral assemblages in terms of temperature, T, X CO2 = CO 2 /(H 2 O + CO 2 ) in the fluid and K 2 O content of the rock. These diagrams are drawn for a fixed pressure, as the evolution of these rocks is dominantly controlled by temperature. The stality of mineral assemblages is predicted /07/3660/

2 380 ELMER ET AL. from these T-X pseudosections and can be comned with key textural observations to deduce the temperature of mineralization and the temperature of the peak of metamorphism. Regional Setting, and Field and Textural Relationships The Bronzewing gold mine is situated within the Yandal greenstone belt, within the extensive Archean Yilgarn craton (Fig. 1). The Yandal greenstone belt is north trending, 300 km long and up to 40 km wide. Three major greenstone sequences (lower, middle, and upper) can be distinguished in the central to northern Yandal belt (Vearncombe et al., 2000; Kohler et al., 2003a). However, ch of the Yandal gold province occurs in the Middle Greenstone sequence of ultramafic intrusions, komatiites, high Mg basalts, tholeiitic basalts, and dolerite (Fig. 1b). These sequences have been subjected to regional metamorphism; the central parts of the belt were metamorphosed to greenschist facies, whereas nearer to granite contacts, amphibolite facies assemblages are found (Vearncombe et al., 2000; Worley et al., 2000). Deformation in the Yandal belt was dominantly the product of eastnortheast west-southwest to east-west shortening, involving the progressive development of folds and shear zones (Vearncombe et al., 2000; Kohler et al., 2003b). Although several large-scale transcurrent ductile shear zones transect the Yilgarn craton, gold-mineralizing fluids are thought to have been focused by a network of middle to upper crustal brittle faults and ductile features which emanated from depths of up to 25 km (Vearncombe, 1998). The Bronzewing gold deposit is hosted within the Middle Greenstone sequence of the central to northern Yandal belt. The stratigraphy in the Bronzewing area has been described in detail by Phillips et al. (1998), Dugdale and Hagemann (2001), Eilu et al. (2001), and Kohler et al. (2003b) and consists predominantly of a north-south striking, steeply dipping Mg- and Fe-tholeiitic basalt sequence bounded to the west by a thick, differentiated, mafic to ultramafic sill (Eilu et al., 2001, Fig. 1c). Deformation at Bronzewing expresses itself as a gently (25 ) south-southeast plunging, west-verging Bronzewing anticline (Fig. 1c). Shear zones at Bronzewing show evidence of sinistral and dextral shearing and well-developed penetrative frics which generally strike north, dipping at moderate to steep angles to the east (Vearncombe et al., 2000; Kohler et al., 2003b). Ore zones within the deposit consist of anastomosing veins with varying mineralogy containing quartz ± carbonate ± iron-sulfide and including scheelite and telluride, breccias with entrained wall-rock fragments, quartz carbonate veins parallel to the foliation and surrounded by pervasive hydrothermal alteration, and later massive quartz reefs. All are major sources of gold. All ore is hosted within heterogeneously deformed tholeiitic basalt (Eilu et al., 2001). It has been noted that massive quartz reefs are folded, and in areas of folding the veins display strong limb attenuation and transposition (Kohler et al., 2003b). Alteration haloes surrounding mineralized zones at Bronzewing range from a few millimeters around veins to hundreds of meters in thickness on the mine scale. These haloes show distinct zonation and, commonly, gradual transitions between zones. Studies of the hydrothermal alteration at Bronzewing (Worley et al., 2000; Dugdale and Hagemann, 2001; Eilu et al., 2001), including this study, have led to a classification of these alteration zones: the carbonate-k-mica zone occurs proximal to veining or shearing; adjacent to this are the chlorite-calcite zone, the chlorite zone, and the actinolite-epidote zone in the country rocks (as shown in Fig. 1d). The mineral assemblage for each zone is outlined in Figure 2 and textural relationships in each zone are discussed in detail in the following section. Petrology Amphibole-epidote zone These rocks are massive, metamorphosed tholeiitic pillow basalt. On a microscopic scale, relict plagioclase laths are almost completely replaced by alte and, in places, by epidote. Actinolite (hornblende), epidote, and chlorite have replaced other igneous minerals. However, in places these minerals still define the characteristic ophitic texture of the protolith. Epidote-rich pillow margins and interpillow matrices are present. Compositions of amphibole from the actinolite-epidote and chlorite zone are listed in Tle 1 (compiled from Worley et al., 2000). Amphibole compositions range from actinolite to hornblende, following the classification of Leake et al. (1997). These assemblages suggest that the rocks were affected by a previous carbonation and hydration event (possibly an early subsea-floor event, presumly occurring shortly after eruption), followed by regional middle to upper greenschist facies metamorphism. Within this zone, rare gold-poor veins consist of carbonate, quartz, epidote, and minor chlorite. Chlorite zone This is the most distal alteration zone directly associated with mineralization. The transition from the unaltered regional greenschist facies rocks is marked by a gradual increase in the undance of chlorite, a decrease in the amount of epidote, and a marked decrease in the modal undance of actinolite. Owing to the change in mineral assemblage, chlorite zone rocks are weakly to moderately foliated in comparison to rocks from the actinolite-epidote zone rocks and consequently are a brighter green. At a microscopic scale, some epidote crystals have retained their shape and appear to overprint chlorite (Fig. 3a), but most fine-grained relict actinolite and rounded epidote crystals are overprinted by chlorite (Fig. 3b). These observations may suggest two stages of epidote growth. Within the matrix, quartz, and alte grains are granoblastic with serrate grain boundaries. Within this zone, there is a slight increase in the frequency of veins compared with the country rocks; veins of calcite + quartz + epidote ± chlorite occur parallel to the foliation, whereas later veins of similar composition cut the foliation. Chlorite-calcite zone A diagnostic feature of this zone is that it has a lighter green color compared to the chlorite zone, caused by the increase in the undance of calcite. The chlorite-calcite zone can extend for hundreds of meters from mineralization. Chlorite defines a foliation that is locally crenulated, and matrix calcite is elongate in the direction of the foliation. Small amounts of rounded, fine-grained epidote can persist through the chlorite-calcite matrix. The foliation within this zone is strongly developed and the modal undance of pyrite and pyrrhotite /98/000/ $

3 TIMING OF GOLD MINERALIZATION AT BRONZEWING, WESTERN AUSTRALIA 381 a NORTH WESTERN GNEISS TERRAIN MURCHISON PROVINCE YANDAL BELT Bronzewing EASTERN GOLDFIELDS PROVINCE b JUNDEE Mylonite Syenite Magnetic granite Granite Felsic Ultramafic & mafic Shear zone Synform Antiform Gold mine/prospect SOUTHERN PROVINCE CROSS MGB NORSEMAN- WILUNA BELT Kalgoorlie BLF Moilers shear zone Gourdis N 0 50 Kilometres Empire Perth KSZ Biddy Well Bills Find DBSZ SOUTH WESTERN GNEISS TERRAIN Celia shear zone Granitoid & gneiss Greenstones Layered bodies Deformation zone Province boundary Movement sense on fault/shear zone 200 km Moongarnoo shear zone Mt McClure Mt Joel Sundowner BRONZEWING Woorana Mandilla Well c m N 16500mE 17000mE LATERITE 17500mE Granodiorite Mafic-ultramafic schist Basalt, dolerite Ultramafic, dolerite Mineralization Inferred Structure d 11000mN m N 10500mN 10000mN CENTRAL H oo k a n ti f o r m 0 N 500 m Bronzewing komatiite 10500mN 10000mN Legend Carbonate-K-mica Chlorite-calcite Chlorite Quartz vein WESTERN DISCOVERY m N 9500mN m E Bapinmarra sill 16500mE Discovery granodiorite 17000mE m E 17500mE 9500mN FIG. 1. (a)-(c) Location maps for the Yilgarn craton, Yandal greenstone belt, and Bronzewing gold mine (compiled and modified after Phillips et al., 1998). a) Location map and simplified geology for the Yilgarn craton. Bronzewing gold mine is marked as the large black square. b) Geology of the Yandal belt, showing main structures and rock types. c) Local geology of the Bronzewing gold mine. d) A schematic diagram showing the spatial relationships of the alteration zones associated with mineralization around veining /98/000/ $

4 382 ELMER ET AL. Mineral SILICATES alte amphibole otite chlorite epidote scovite quartz titanite CARBONATES calcite ankerite OXIDES ilmenite rutile SULFIDES chalcopyrite magnetite pyrite pyrrhotite GOLD Zone amphiboleepidote chlorite chloritecalcite carbonate- K-mica FIG. 2. Interpreted distribution and paragenesis of the minerals found within each alteration zone. Abbreviations used are as follows: = alte, act = actinolite, ank = ankerite, = otite, cc = calcite, cpy = chalcopyrite, chl = chlorite, dol = dolomite, ep = epidote, Au = gold, hb = hornblende, ilm = ilmenite, mt = magnetite, = scovite, pa = paragonite, py = pyrite, po = pyrrhotite, q = quartz, rut = rutile, and tnt = titanite. increases relative to more distal zones. Minor otite may be present. Rare, randomly oriented actinolite porphyroblasts that overgrow the foliation defined by chlorite-calcite-otitealte have been observed (Kohler et al., 2003b). Domains of granoblastic quartz and alte have replaced most of the original igneous plagioclase laths. Carbonate-K-mica zone Rocks from this zone are light, creamy yellow in color, which reflects high modal undance of carbonate. Primary igneous textures are not observed. As the rocks become richer in calcite and ankerite, the foliation becomes less distinct as the proportion of chlorite decreases. At a microscopic scale, small irregular zones of scovite and otite are observed around veins. The foliation in such zones is better developed and locally crenulated. This first generation of scovite commonly appears in thick mats along the vein boundaries or as bands in areas of high strain. Sulfides, ilmenite, and rutile are concentrated within these highly altered areas. This zone appears to have involved pervasive and intense carbonation and the addition of potassium. Critical textural relationships within the proximal zone are shown in Figure 3. At the edges of the carbonate-k-mica zone, otite partially replaces minor chlorite that defines the foliation (Fig. 3c). Biotite also partially replaces scovite, growing on the edges of scovite laths, most clearly seen in scovite-rich domains. Ankerite and, less commonly, calcite also may overprint crenulated scovite (Fig. 3d). Two generations of otite growth have been observed. Biotite (1) occurs as irregular-shaped crystals with diffuse boundaries throughout the scovite-ankerite alteration and, in the pressure shadows of pyrite, as coarse-grained laths that are parallel to the foliation (Fig. 3e). Biotite (2) occurs in the pressure shadows of euhedral pyrite; both crosscut the foliation (Fig. 3f). A second generation of scovite (not shown in Fig. 3) occurs as random laths that overprint clusters of aligned chlorite. This relationship appears to be associated with late veins and is uncommon. Fine-grained granoblastic quartz and alte are observed throughout the matrix of the proximal alteration zone. Gold is commonly associated with sulfide disseminated in the altered rock but also occurs within veins of quartz + calcite + ankerite + pyrrhotite + alte and minor K-feldspar. Quartz displays serrate grain boundaries, with some crystals exhiting deformation lamellae. Alteration minerals chlorite, scovite, and otite define the foliation and are locally crenulated, with calcite elongated in the foliation. These textural relationships suggest that the alteration occurred prior to or contemporaneously with deformation, possibly as a result of ltistage deformation that crenulated the earlier foliation. A wide range in plagioclase compositions is recorded across the alteration zones with calcic relict plagioclase laths occurring in the unaltered zone and the anorthite content of alte decreases towards the proximal alteration zone (Tle 1). A key textural observation is that otite, ankerite, calcite, and minor actinolite overprint the earlier crenulated foliation. Evolution of alteration zones The mineral assemblages described ove are consistent with the infiltration of an aqueous, CO 2 -bearing fluid, containing K, S, and Au, into a pile of tholeiitic basalt, producing laterally zoned alteration. The proximal zone reflects large fluid/rock ratios, and distal rocks reflect progressively lower fluid/rock ratios. Two hypotheses are considered here to account for the observed textural relationships and evolution of the rocks within the proximal and distal alteration zones at Bronzewing. The first involves infiltration of fluid into the rocks at the peak of metamorphism (Dugdale and Hagemann, 2001). The second involves fluid infiltration into the rocks at greenschist facies conditions during prograde metamorphism with subsequent metamorphism of the alteration zones occurring as temperatures increased to peak conditions. To test these hypotheses, the stle mineral assemblage at each stage in the evolution of the deposit is inferred from textural observations. In the proximal alteration zone, peak metamorphic minerals such as otite overprint the foliation, and potassic minerals are locally crenulated. This suggests that a dynamic environment existed where reactions could proceed and were not inhited by kinetics. However, within distal alteration zones, the weak foliation is not crenulated and there is little mineral growth. Such a static environment may suggest that kinetics inhited new mineral growth during prograde metamorphism distal to the deposits. The mineral assemblages formed within the alteration zones are controlled by the P-T conditions, the composition (i.e. X CO2 and K 2 O content) and volume of the infiltrating fluid, and the composition of the protolith. Metamorphic conditions, /98/000/ $

5 TIMING OF GOLD MINERALIZATION AT BRONZEWING, WESTERN AUSTRALIA 383 TABLE 1. Representative Mineral Microprobe Analyses from Distal and Proximal Alteration Zones Distal alteration zones 1 Proximal alteration zones BWRCD BWU BWU BWRCD BWRCD BWRCD BWU Sample 1011/ 1070/ 1070/ 1011/ 1011/ 1011/ 1070/ 19378_ 19352_ 19373_ 19380_ 19378_ 19373_ 19378_ 19379_ 17240_ _ _ _ _ _13 K-1 B-1 D-1 C-1 J-1 E-1 C-1 G-1 Chl Chl Chl Ep Amph Amph Fsp Mu Mu Bi Bi Chl Chl Alb Ank No SiO TiO < Al2O <0.01 Cr2O3 <0.01 < FeO MnO < <0.01 < MgO < < CaO < <0.01 < Na2O < < < K2O < < Total Si Ti Al Cr Fe Mn Mg Ca Na K Cations Compiled from Worley et al. (2000) /98/000/ $

6 384 ELMER ET AL. a chl b ep 0.45 mm 2.8 mm c py d chl ank 0.43 mm 0.35 mm e f py py cc mm mm FIG. 3. Petrography of country rock, and distal and proximal alteration zones at Bronzewing gold mine, Western Australia. The scale indicated is the length of the field of view. a) Sample BWRCD1011/250.1 consists largely of chlorite and epidote with sharp boundaries; some epidote crystals overprint chlorite (polarized light). b) Sample BWRCD1011/ consists of fine-grained relict actinolite and epidote (appears brighter than the gray actinolite), which are surrounded by chlorite (polarized light). c) Sample BWRCD 1034/ contains otite pseudomorphs of chlorite, and chlorite shows a preferred orientation. Such a texture is common at transitions from the distal to the proximal zone (polarized light). d) Sample BWRCD 1034/ otite pseudomorphs of scovite, growing on the edges of scovite laths; ankerite overprints crenulated scovite (plain light). e) Sample BWRCD 1099/ contains otite in pressure shadows of pyrite, elongated in the direction of the foliation (plain light). f) Sample BWRCD 1099/ contains otite and calcite in pressure shadow of pyrite; all are interstitial to the foliation (plain light) /98/000/ $

7 TIMING OF GOLD MINERALIZATION AT BRONZEWING, WESTERN AUSTRALIA 385 including the compositions of the infiltrating fluid at the time of mineralization, have been estimated using a variety of methods. Fluid inclusions at Bronzewing have relatively constant homogenization temperatures (Dugdale and Hagemann, 2001), but with highly varile compositions. It has been suggested that fluids were derived from the phase separation of a low salinity parent fluid at 300 ± 50 C and 1.5 kbar (Dugdale and Hagemann, 2001). The fluid inclusion results from gold deposits elsewhere across the Yilgarn suggest a low salinity fluid with X CO2 = fluid that infiltrated at temperatures between 270º and 350ºC (Ho et al., 1990b). Mineral Equilibria Study Mineral equilibria were examined in a series of calculated T-X pseudosections. T- X CO2 pseudosections also can be used to investigate buffering processes (White et al., 2003). External buffering in this context corresponds to a fluid infiltration event. Internal buffering occurs when the fluid composition is controlled by reaction equilibria involving the solid phases, without fluid addition from outside the rock (Baker et al., 1991). In the following sections on the evolution of the distal zone and evolution of the proximal zone, T- X CO2 pseudosections are interpreted in terms of external buffering. An external buffering path on a T- X CO2 is parallel to the X CO2 axis, the change in the mineralogy of the altered rock is caused by the infiltrating fluid and is dependent on the amount and composition of this fluid. An example of such a path is shown on Figure 4 as path I. In a later section that considers heating subsequent to fluid infiltration, T- X CO2 pseudosections are interpreted in terms of internal buffering. Internal buffering paths relate to the reactions taking place in the mineral assemblage with temperature increase. An example of such a path is shown on Figure 4 as path II. These paths track across the T- X CO2 pseudosection and show changes in mineral assemblage and fluid evolution. Mineral equilibria calculations Mineral equilibria calculations used THERMOCALC 3.22 and the internally consistent dataset of Holland and Powell (1998, updated November 22, 2003). A T- X CO2 pseudosection has been calculated in the system Na 2 O-CaO-FeO-MgO- Al 2 O 3 -SiO 2 -CO 2 -H 2 O (NCaFMASCH) for a pressure of 2.5 kbars, temperatures of 300 to 500 C and X CO2 = These conditions were chosen based on estimates from Phillips et al. (1998). Such a diagram has been calculated to show phase relationships in the distal alteration zones at Bronzewing. The calculations involved the phases actinolite (act), alte (), calcite (cc), chlorite (chl), clinozoisite (cz), dolomite (dol), hornblende (hb), plagioclase (pl), with quartz (q) and fluid in excess. The activity model used for amphibole is taken from Dale et al. (2005), for chlorite from Holland et al. (1998), for scovite and paragonite from Coggon and Holland (1992), for otite from Powell and Holland (1999), for plagioclase from Holland and Powell (1992), and for the fluid from an asymmetrical model of Holland and Powell (2003). In the modelling the mineral name of dolomite is used to express all compositions between the dolomite and ankerite end members. The fluid is considered to be dominated by H 2 O and CO 2 ; its composition can then be considered in terms of X CO2 T( C) hb act chl cz pl 400 act chl cz NCaFMASCH + q + fluid hb act cz pl II chl cz cc hb act pl hb act cz chl cz chl chl cc I B chl chl cc dol v = 4 chl cc dol pl chl dol pl chl dol 2.5 Kbar v = 3 hb act hb cc dol pl hb act hb dol pl dol pl hb chl hb act chl hb chl dol pl hb chl dol A v = 5 chl cc dol pl X CO2 Distal alteration zone FIG. 4. T- X CO2 pseudosection in NCaFMASH + q + fluid showing the phase relationships in the distal alteration zone. Additional breviations: cz = clinozoisite, and pl = plagioclase. Box A shows the mineral assemblage that would be stle if infiltration of fluid took place at peak metamorphic temperatures. Box B shows the conditions of stality for fields that best represent the observed fric-forming mineral assemblages at Bronzewing, constrained by fluid inclusion results at Bronzewing and across the Yilgarn. See text for details. = CO 2 /( CO 2 + H 2 O). As potassic alteration is indicative of the proximal alteration zone, several T- X CO2 pseudosections and a T-X K2 O pseudosection were calculated, with different K 2 O contents of the rock, the relative proportions of the remaining oxides being held constant. These pseudosections were calculated in the system Na 2 O-CaO-K 2 O-FeO-MgO-Al 2 O 3 - SiO 2 -CO 2 -H 2 O (NCaKFMASCH) at temperatures of 300º to 500ºC and X CO2 = 0 0.8, involving the additional phases otite () and scovite (). The substitution of Fe 3+ and Ti 4+ cannot, as yet, be incorporated into the thermodynamic data for the major rockforming minerals such as amphibole. As a consequence of excluding Fe 2 O 3, epidote is modelled as clinozoisite. If the chemical system was extended to include Fe 2 O 3, epidote may have a larger stality field than that of clinozoisite, given its ality to accommodate substantial amounts of Fe 3+. It may also affect the stality of other minerals such as amphibole, chlorite, and otite, all of which contain some of Fe 3+ in lowgrade mafic metavolcanic rocks (Condie and Harrison, 1976). The inclusion of TiO 2 and Fe 2 O 3 would also allow for the incorporation of phases such as magnetite, ilmenite, and rutile; all of which are found in the alteration zones. The addition of sulfur, a major component of the altered rocks, also has not been accounted for within this modelling. However, sulfur does not substitute into the phases included in the modelling here, so results will not be substantially affected /98/000/ $

8 386 ELMER ET AL. by this omission. White et al. (2000) suggest that a small effect may arise from the partitioning of Fe into sulfide minerals, affecting the bulk rock composition used in the calculations (e.g., X Mg trends in chlorite and otite from distal to proximal alteration zones: Tle 1; Fig. 4). In order to model the peristerite solvus the pure end member alte is used comned with the plagioclase model of Holland and Powell (1992) to reflect substitution of Ca. Equilibria in the T- X CO2 pseudosection are dependent on pressure. Here pseudosections are calculated at 2.5 kbars; however, it is noted that Dugdale and Hagemann (2001) estimated pressure conditions of 1.5 kbars at the time of mineralization. If pressure was lower, it would result in the equilibria shifting toward more CO 2 -rich conditions and in the T-X K2 O pseudosection the equilibria would move to lower temperatures with decreasing pressure. Small changes in pressure from 2.5 kbars will have little effect on the geometry of the pseudosections and so do not affect the results presented here. Evolution of the distal zone Mineral equilibria in the transition from the actinolite-epidote zone to chlorite-calcite-bearing rocks in the distal alteration zones in the T- X CO2 pseudosection in NCaMASCH + quartz + fluid can be seen in Figure 4. This pseudosection was calculated using a typical tholeiitic basalt composition from Bronzewing (Phillips et al., 1998; listed in Tle 2) at a pressure of 2.5 kbars, corresponding to P estimates at the time of mineralization (Phillips et al., 1998). The peak metamorphic mineral assemblage from the unaltered zone at Bronzewing is hornblende + actinolite + clinozoisite + calcite + alte + chlorite ± quartz, corresponding to an upper greenschist-lower amphibolite facies assemblage. Thus the peak metamorphic assemblage is represented by the divariant field hornblende + actinolite + clinozoisite + calcite + plagioclase + chlorite + quartz + fluid in Figure 4. This divariant field is stle at temperatures between 410º and 450 C and at X CO2 = , constraining the peak metamorphic temperature at Bronzewing. The mineral assemblage within the chlorite-calcite zone is represented by the quadrivariant field chlorite-dolomite-calcite-plagioclase in Figure 4, between 380º and 460 C and X CO2 = The hypothesis that mineralizing fluids infiltrated at peak metamorphic temperatures is tested first. Figure 4 shows that at peak metamorphic conditions for Bronzewing, indicated by Box A, the observed mineral assemblage (chlorite + ankerite + plagioclase + calcite + quartz + fluid) of the chlorite-calcite TABLE 2. Bulk Rock Compositions Used in the Calculation of Pseudosections to Represent the Rocks at Bronzewing SiO 2 Al 2O 3 CaO MgO FeO Na 2O K 2O Fig Fig Fig. 6 1 a b Fig. 9a Fig. 9b The compositions used for the calculations of X K2O = 0 and X K2O = 0.8 in Figure 6 are also shown as a and b, respectively zone is stalized when in equilibrium with a fluid of X CO2 greater than However, adopting the fluid composition of Archean greenstone gold deposits (X CO2 = ), estimated by Ho et al. (1990b), the expected mineral assemblages are indicated by Box A, solid lines, in Figure. 4. Under these conditions clinozoisite is unstle for the whole peak temperature range. With the addition of progressively more fluid with X CO2 = , first actinolite and finally hornblende is consumed. The predicted succession of mineral assemblages with increasing X CO2, outlined by Box A in Figure 4, fails to reproduce the fric-forming mineral assemblages found in the distal zones at Bronzewing and does not account for the presence of ankerite in the chlorite-calcite zone. The hypothesis that fluid infiltrated and reacted with the rocks at Bronzewing before peak conditions is also tested using the T- X CO2 pseudosection in Figure 4. Box B represents the predicted succession of mineral assemblages if a fluid of X CO2 = infiltrated the rocks at greenschist facies temperatures. Infiltration of such a fluid at lower temperature causes a shift to higher X CO2. Initially actinolite is consumed and the quadrivariant field, chlorite + clinozoisite + calcite + alte, is entered, representing the chlorite zone. Further reaction with the fluid results in the consumption of clinozoisite and an increase in the modal undance of calcite, representing the chlorite-calcite zone. Within the chlorite-calcite zone at Bronzewing, ankerite may also be present; this is represented by the quadrivariant field chlorite + calcite + dolomite + alte, which at low temperature (indicated by Box B) exists over a wide range of X CO2 values. In sence of coexisting alte and plagioclase, this assemblage is stle between 300º and 375ºC (Box B, Fig. 4); i.e., at temperatures lower than peak metamorphic temperatures of between 410º and 450 C. This suggests that if a fluid of X CO2 = infiltrated the rocks at Bronzewing, then the temperature of infiltration was lower than the peak metamorphic temperature. Results in Figure 4 also indicate that mineral assemblages found within the distal chlorite to chlorite-calcite zones can be in equilibrium with fluid having a wide range of possible X CO2 compositions. The zoned nature of alteration at Bronzewing reflects both the composition and volume of fluid that interacted with the metasic rocks. The observed sequence of mineral assemblages is most readily explained by such processes, but this does not explain textural observations of otite, ankerite, calcite, and minor actinolite overprinting the earlier crenulated foliation. Note that modelling here may also represent a process whereby fluid infiltration occurred postpeak metamorphism. Evolution of the proximal zone To model the evolution of the proximal alteration zone at Bronzewing, a T-X CO2 pseudosection was calculated in the system NCaKFMASCH to reflect the addition of potassium in the mafic precursor, as potassic minerals otite and scovite are both observed within this zone. A bulk rock composition representing the unaltered bulk rock with an added component of K 2 O was used (Tle 2). The composition was chosen to represent minor amounts of potassic alteration representative of the outer parts of proximal alteration zones. Although only K 2 O in the rock is varied, potassium is a minor component of the fluid responsible for the large alteration /98/000/ $

9 TIMING OF GOLD MINERALIZATION AT BRONZEWING, WESTERN AUSTRALIA 387 haloes associated with gold mineralization, and potassic alteration is restricted to the narrower proximal zones. Using the peak metamorphic temperature and the temperature of infiltration constrained from the distal zone, box A in Figure 5 predicts the mineral assemblage that should be observed if fluid infiltration took place at the peak of metamorphism. Box B in Figure 5 represents the conditions of fluid infiltration within the proximal alteration zone if temperatures were lower than peak metamorphism. The observed mineral assemblage within the outer part of the proximal alteration zone at Bronzewing, where potassic alteration is minor, is best represented by the trivariant field chlorite + otite + scovite + calcite + dolomite + plagioclase + quartz + fluid in Figure 5, from 435º to 445ºC, in equilibrium with a fluid having X CO2 = Figure 6 shows that the observed mineral assemblage within the outer parts of the proximal alteration zone can only be reproduced upon fluid infiltration at peak temperatures with a fluid with X CO2 = The predicted mineral assemblages that would be stle over the temperature range of peak metamorphic conditions, with a fluid having X CO2 = , if infiltrated at the peak metamorphic temperature is indicated by Box A in Figure 5. The assemblage predicted by Figure 6 at X K2 O = 0.05 is chlorite + calcite + plagioclase, including scovite and or otite, depending on exact peak temperature. None of the predicted mineral assemblages account for the presence of ankerite, a major fric-forming mineral within the proximal zone at Bronzewing. Nor can fluid infiltration of any composition at peak temperatures account for overprinting by otite (2), calcite and ankerite of the crenulated foliation. At lower temperatures (Box B in Fig. 5), the mineral assemblage chlorite + scovite + dolomite + alte + paragonite is stle. This predicted mineral assemblage accounts for the presence of dolomite within the fric-forming mineral assemblages of the proximal zone. However, paragonite has not been observed within the proximal or distal alteration zones at Bronzewing but could have been present prior to peak metamorphism. The pseudosections presented here do not account for observed depletions of Na coincident with K addition (Eilu et al., 2001). Such depletion of Na may account for the instality of paragonite at Bronzewing. Variations in K 2 O Due to the heterogeneous permeality of the tholeiitic basalt at Bronzewing, the rocks are varily altered and contain varile amounts of K 2 O. A T-X K2 O pseudosection (Fig. 6) was constructed for a fluid composition at a constant X CO2 = 0.25 and P = 2.5 kbars, representing estimated conditions at the time of alteration in the proximal zones. The range in rock compositions used in construction of the diagram is listed in Tle 2 and corresponds to X K2 O = 0 to X K2 O = 0.8. Vertical lines A, B, and C on Figure 6 correspond to T-X CO2 pseudosections presented in this study; the bulk rock compositions for these pseudosections are given in Tle 2. The temperature range lelled (i) in Figure 6 highlights the mineral assemblages predicted for varying degrees of potassic alteration if fluid of X CO2 = 0.25 reacted with the rock T( C) 500 hb act cz pl 450 hb act chl cz pl v = 3 hb chl cz 400 hb act chl cz NCaKFMASCH + q + fluid hb act cz hb act pl chl cz chl cz cc chl cz cc hb act chl hb act hb cc dol 2.5 Kbar hb act dol pl hb dol pl hb chl dol pl hb chl hb chl cc dol pl chl dol pl A chl v = 5 chl chl dol pl chl cc pl chl dol pl pa chl cc chl cc pa dol pl B 350 chl dol pa v = X CO2 Proximal alteration zone - minor potassium addition FIG. 5. T- X CO2 pseudosection in NCaKFMASH + q + fluid showing phase relationships in the proximal alteration zone. Box A shows the mineral assemblage that would be stle within the proximal alteration zone if fluid infiltration took place at peak metamorphic conditions. Box B, with constraints from fluid inclusion data on the composition of the infiltrating fluid and constraints from Figure 4, indicates the expected mineral assemblage within the proximal alteration zone if fluid infiltration took place at a temperature that was lower than that of the peak of metamorphism /98/000/ $

10 388 ELMER ET AL. 500 NCaKFMASCH + q + fluid act pl hb act pl act hb 450 v=3 chl chl cc pl pa hb chl chl cc pl v=5 X CO2 = 0.25 act pl cc chl hb chl chl cc cc 2.5 Kbar cc pl cc i T( C) 400 chl cc pa 350 A chl dol cc chl dol pa chl dol chl dol cc B chl dol v=4 dol cc dol C ii 300 dol dol cc X K2 O FIG. 6. T-X K2 O pseudosection in NCaKFMASH + q + fluid showing the effect of addition of varying amounts of potassium to the metasic rock. The vertical lines correspond to compositions used for T- X CO2 pseudosections and show the assemblages seen at varying T but at a constant X CO2 of Line A corresponds to Figure 4, B to Figure 7a and C to Figure 7b. Temperature interval indicated by (i) gives the mineral assemblages that are stle at peak metamorphic conditions with varying X K2 O. Temperature interval indicated by (ii) gives the mineral assemblages that are stle at temperatures constrained by Figures 4 and 5 and fluid inclusion studies. at peak metamorphic conditions. None of the assemblages predicted by Figure 6 at peak metamorphic temperature for any range of X K2 O can account for the presence of ankerite within the fric-forming mineral assemblage at Bronzewing, in agreement with results from Figure 5. The upper limit of the stality of dolomite in Figure 6 suggests that if the infiltrating fluid had X CO2 = 0.25, then mineralization could not have occurred ove 400 C. Within the most altered zone at Bronzewing, the foliationforming minerals are otite, scovite, ankerite, calcite, and alte, which were stle prior to the metamorphic peak. The mineral equilibria in Figure 6 predicts that this assemblage is stle at X K2 O < 0.46 and temperatures as low as 330ºC, constraining the temperature of fluid infiltration to 330º to 400ºC. The temperature range lelled (ii) in Figure 6 indicates the predicted mineral assemblages for a range of X K2 O within the transition from proximal to distal alteration zones, with chlorite stalized at lower X K2 O. The textural observations of otite, calcite, ankerite and minor actinolite overprinting the earlier crenulated foliation in both the distal and proximal zones at Bronzewing preclude a model involving a synpeak metamorphic timing for infiltration of fluid with composition of X CO2 = Infiltration of such a fluid at temperatures between 330º and 375ºC is consistent with the fric-forming mineral assemblages within the distal and proximal alteration zones. Infiltration at peak metamorphic conditions cannot explain otite replacing chlorite and scovite, ankerite overprinting crenulated scovite, and calcite and otite within pressure shadows not aligned with the foliation. Although porphyroblasts truncating a fric do not necessarily indicate that the porphyroblasts are later than the matrix minerals (Ferguson and Harte, 1975), the textural relationships between overprinting minerals and crenulated foliation in proximal and distal alteration zones demonstrate that mineral growth continued subsequent to fluid infiltration. Therefore, it is necessary to consider the metamorphic changes involved in heating subsequent to mineralization. Internal buffering paths calculated across T-X CO2 pseudosections can predict prograde development of mineral assemblages subsequent to fluid infiltration. Heating subsequent to fluid infiltration In this study it is assumed that internal buffering is the dominant process occurring within the rocks as temperatures continued to increase following large-scale fluid infiltration. In modelling internal buffering, paths are calculated using a constant porosity model, whereby it is assumed that no fluid infiltrates from an external source after gold deposition. Metamorphic rocks have a limited porosity, the exact magnitude of which is unknown; Lotka et al. (2002) suggests that the porosity of metamorphic rocks at these conditions is likely to be <1 percent. Fluid is produced by devolatilization (prograde metamorphism) of the rocks over the temperature range considered here. Fluid st therefore be lost from the rocks along the buffering path. In applying a constant porosity model to calculate paths, a stepwise procedure is used in /98/000/ $

11 TIMING OF GOLD MINERALIZATION AT BRONZEWING, WESTERN AUSTRALIA 389 which a threshold is set for the amount of fluid allowed to acculate within the rock. Once this threshold is reached, fluid is removed from the system, involving a change in the H 2 O and CO 2 in the bulk rock composition at each step. Internal buffering paths have an initial fluid mole proportion of 0.01, and the threshold is set at a mole proportion of Paths A-C in Figure 7a are calculated in the system NCaF- MASCH to represent internal buffering in the distal zone during heating. Paths D-F (Fig. 7b) are calculated in the system NCaFKMASCH to represent internal buffering in the proximal zone. a) T( C) T( C) hb act chl cz pl chl cz cc b) NCaFMASCH + q + fluid hb act cz pl A hb act pl hb act cz chl cc B hb act chl chl cz chl chl cc dol X CO2 NCaKFMASCH + q + fluid hb act pl hb act hb act chl hb chl chl chl chl cc pl chl cc pa D E F chl hb act hb chl C hb chl cc dol pl chl dol pa hb cc dol pl hb act dol pl hb chl dol pl hb chl dol hb cc dol pl chl cc dol pl chl dol pl hb dol pl hb dol pl hb chl dol pl chl dol pl chl dol 2.5 Kbar chl cc dol pl chl dol chl dol 2.5 Kbar hb act dol pl chl dol pl pa X CO2 FIG. 7. T-X CO2 pseudosection in NCaFMASCH and NCaKFMASH + q + fluid showing phase relationships in the distal and proximal alteration zones. Internal buffering paths are calculated across the T-X CO2 pseudosection to depict the evolution of mineral assemblages as a result of heating subsequent to the fluid infiltration event responsible for gold deposition. a) Buffering paths representing prograde evolution subsequent to fluid infiltration for rocks within the distal alteration zone at Bronzewing. b) Buffering paths representing rocks within the proximal zone at Bronzewing. The three paths (A-C in, Fig. 7a), representing the different distal alteration zones, evolve very differently during prograde metamorphism, owing to large variations in the initial starting X CO2 of the fluid phase and the corresponding mineral assemblage. Path A in Figure 7a is representative of the chlorite zone. The buffering path is initially almost a vertical line within the quadrivariant chlorite + clinozoisite + calcite + alte field as temperature increases. The near vertical nature of this segment of the buffering path reflects minimal reaction. Hornblende is stalized at T > 410 C, and the path is entirely within the trivariant hornblende + chlorite + clinozoisite + calcite + plagioclase field until clinozoisite is consumed. A rock following path A would only develop minor amounts of hornblende during prograde metamorphism. Overprinting by hornblende has not been observed within the chlorite zone. Such results possibly reflect simplifications involved in the modelling or incomplete equilibration. Path B represents the evolution of a rock within the chlorite-calcite zone, (Fig. 7a). As the rock, undergoes prograde metamorphism there is little variation in the stle mineral assemblage, although the mineral compositions may change. During internal buffering, the fluid composition that is in equilibrium with the rock becomes progressively richer in CO 2 due to reactions within the quadrivariant chlorite-calcite-plagioclase-alte field. Finally, at T > 430 C, alte is consumed. Path C is representative of rocks within the chlorite-calcite zone adjacent to highly altered rocks within the proximal zone (Figure 7a). It is assumed that such a rock would be in equilibrium with a fluid composition close to X CO2 = 0.25 prior to prograde metamorphism. Path C is buffered across the quadrivariant field chlorite + calcite + dolomite + alte until plagioclase is stalized, the slope of the path is then very steep as it evolves up temperature through the quadrivariant chlorite + calcite + dolomite + plagioclase field. For a rock in the chlorite-calcite zone near the transition of the proximal alteration zone, represented by path C, little change is apparent within the fric-forming mineral assemblage. Internal buffering paths A-C reproduce the textural features observed within the distal zones at Bronzewing. Path A involves small modal increases in clinozoisite, consistent with textural observations of large epidote crystals with defined edges. If the chemical system was extended to include Fe 2 O 3, then epidote may exhit a greater modal increase with increasing temperature. Path A involves growth of actinolite during heating, following fluid infiltration. However, such actinolite growth has not been observed within the distal zones. This may reflect simplifications in modelling or incomplete equilibration. Path D (Fig. 7b) represents rocks in the chlorite-calcite zone, and paths E-F represent the rocks in the outer parts of proximal alteration zone (paths with scovite present involve rocks that have been affected by addition of K 2 O). Path D evolves at almost constant X CO 2, through the quadrivariant chlorite + scovite + calcite + alte + paragonite field, until paragonite is consumed at T > 405 C. Path D then crosses the quadrivariant chlorite + scovite + calcite + alte + plagioclase field whereby otite is stalized at higher temperature. Path D results in a rock that at peak metamorphic temperature is in equilibrium with a CO 2 -rich fluid. At temperatures /98/000/ $

12 390 ELMER ET AL. > 465 C, hornblende is stalized. As hornblende is not observed within rocks that show evidence of potassic alteration, it can be assumed that the rocks at Bronzewing did not exceed temperatures of 465 C. This is consistent with the maxim temperature constraints for the distal zones. For rocks that are represented by compositions starting within the same field but at lower X CO2, hornblende will be stalized at lower temperatures consistent with the observations of Kohler et al. (2003b) of rare randomly oriented actinolite porphyroblasts that overgrow the foliation defined by chlorite + calcite + otite + alte (an assemblage that represents the gradual transition from distal to proximal alteration zones). Any rock represented by path D will involve the consumption of paragonite. Therefore, the process of internal buffering may account for the sence of paragonite from the rocks at Bronzewing. Paths E and F evolve in a similar fashion (although at different X CO2 values), as nearly vertical lines as temperature increases until calcite is stalized (Figure 7b). Both paths are then buffered along the trivariant chlorite + scovite + calcite + dolomite + alte + paragonite field until paragonite is consumed, consistent with the sence of paragonite and with textural observations. Along these paths, modes of calcite and otite increase, providing an explanation for overprinting growth of calcite and otite (2), observed in pressure shadows not aligned with the foliation within the rocks at Bronzewing. There may have been only a small proportion of K 2 O in the infiltrating fluid but, as K 2 O is extremely reactive with the protolith rock, potassium was lost from the fluid close to the point of infiltration. This leads to a nonlinear relationship between X K2 O and X CO2. It is evident from Figure 7 that the geometry of T-X CO2 pseudosections will vary with increasing K 2 O. Therefore, to more accurately constrain the evolution of the rocks subsequent to fluid infiltration within the proximal alteration zone at Bronzewing, two T-X CO2 pseudosections with internal buffering paths are calculated for rock compositions of X K2 O = 0.3 (Fig. 8a) and X K2 O = 0.62 (Fig. 8b). Paths G I in Figure 8 are calculated to represent rocks in the proximal alteration zones. Paths G and H cross the quadrivariant chlorite + otite + scovite + dolomite + alte field to more H 2 O-rich conditions as dehydration reactions occur. These paths cross modal otite contours and show that the process of internal buffering will result in the growth of otite. It is unlikely that rocks in these zones were in equilibrium with a CO 2 -poor fluid as represented by path G. However, path G shows that any path with starting compositions between G or H will evolve in a similar fashion. Path I (Fig. 8b) represents rocks with the most added K 2 O and shows that the mineral assemblage of such rocks varies little during heating. Discussion Gold-only deposits in Archean greenstone belts, including Bronzewing, have distinctive ore mineral associations and element ratios (Groves and Phillips, 1987), and distinctive wallrock alteration, depending on the host-rock composition. These distinctive features are compatible with a uniform low salinity, CO 2 -rich, potassium-, sulfur-, and gold-bearing fluid (Phillips and Brown, 1987; Mueller and Groves, 1991; Kerrich, 1993; Mikucki and Ridley, 1993). The deposits also show a) 450 hb act cz act cz cc 400 T( C) b) 500 act cz pl cc T( C) NCaKFMASCH + q + fluid chl cz cc G act pl cc act hb chl chl cc act chl cz chl dol act hb chl chl hb chl 2.5 Kbar chl dol pl chl dol pl chl dol hb dol pl X CO2 NCaKFMASCH + q + fluid chl cc 0.25 H I dol dol 2.5 Kbar X CO2 Proximal alteration zone - moderate and intense potassium addition FIG. 8. T-X CO2 pseudosection in NCaKFMASH + q + fluid to show phase relationships in the proximal zone. Dashed lines indicate modal contours for otite. a) Internal buffering paths for rocks within the proximal alteration zone at Bronzewing. b) Internal buffering paths for rocks within the most intensely altered proximal zone at Bronzewing. strong structural control of gold mineralization in and around vein networks and shear zones (Robert and Poulsen, 1997). The current main competing genetic models for the formation of Archean greenstone gold-only deposits are the crustal continuum model and the metamorphic model. The crustal continuum model postulates the formation of subgreenschist facies to granulite facies gold deposits at the peak of metamorphism and rules out the possility that in higher metamorphic grade deposits the ore fluid was derived solely from the greenstone pile itself (Colvine, 1989; Groves et al., 1992; Groves, 1993). The metamorphic model suggests that an ore fluid is generated by devolatilization within the greenstone belt with deposits forming at greenschist facies conditions, /98/000/ $