Journal of Petrology Advance Access originally published online on December 6, 2006
Journal of Petrology 2007 48(3):511-535; doi:10.1093/petrology/egl070
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
On the Initiation of Metamorphic Sulfide Anatexis
1Department of Geology and Geophysics, University of Calgary, Calgary, AB, Canada T2N 1N4
2School of Geosciences, Monash University, PO Box 28E, VIC. 3800, Australia
3Department of Geology and Geophysics, University Of Wyoming, Laramie, WY 82071, USA
RECEIVED FEBRUARY 26, 2006; ACCEPTED OCTOBER 18, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONOVERVIEW OF CONDITIONS REQUIRED...EARLIEST MELTING IN SULFIDE...MOBILIZATION OF POLYMETALLIC...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Mineral assemblages in common sulfide ore deposits are examined together with phase relations to (1) investigate the pressure–temperature conditions required for the onset of metamorphically induced partial melting involving economic minerals, and (2) place constraints on the amount of melt produced. Deposits that contain sulfosalt or telluride minerals may start to melt at conditions ranging from lowest greenschist facies to amphibolite facies. Deposits lacking sulfosalt and/or telluride minerals may begin to melt once P–T conditions reach the upper amphibolite facies, if galena is present, or well into the granulite facies if galena is absent. The result is two broad melting domains: a low- to medium-temperature, low melt volume domain involving melting of volumetrically minor sulfosalt and/or telluride minerals; and a high-temperature, potentially higher melt volume domain involving partial melting of the major sulfide minerals. Epithermal gold deposits, which are especially rich in sulfosalt minerals, are predicted to commence melting at the lowest temperatures of all sulfide deposit types. Massive Pb–Zn (–Cu) deposits may start to melt in the lower to middle amphibolite facies if pyrite and arsenopyrite coexist at these conditions, and in the upper amphibolite facies if they do not. Excepting sulfosalt-bearing occurrences, massive Ni–Cu–PGE (platinum group element) deposits will show little to no melting under common crustal metamorphic conditions, whereas disseminated Cu deposits are typically incapable of generating melt until the granulite facies is reached, when partial melting commences in bornite-bearing rocks. The volume of polymetallic melt that can be generated in most deposit types is therefore largely a function of the abundance of sulfosalt minerals. Even at granulite-facies conditions, this volume is usually less than 0·5%. The exception is massive Pb–Zn deposits, where melt volumes significantly exceeding 0·5 vol. % may be segregated into sulfide magma dykes, allowing mobilization over large distances.
KEY WORDS: sulfide melt; ore deposits; melt migration; metamorphism
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONOVERVIEW OF CONDITIONS REQUIRED...EARLIEST MELTING IN SULFIDE...MOBILIZATION OF POLYMETALLIC...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Petrologists have a good understanding of the metamorphic conditions under which common crustal rocks start to melt and the chemical factors that influence the onset of partial melting, such as composition of fluid, minerals and bulk rock, yet there remains a class of minerals for which anatectic relationships are relatively poorly understood: the sulfides and sulfosalts that are concentrated in ore deposits.
Interpretations prior to 2001 involved two widely accepted mechanisms by which ore minerals could be remobilized during deformation and metamorphism of a sulfide ore deposit. These were solid state mechanical mobilization (where strain is preferentially partitioned into easily deformable sulfide minerals) and hydrothermal dissolution and reprecipitation (Marshall et al., 2000
). A third mechanism, partial melting and consequent remobilization of this melt, was proposed by Lawrence (1967) and Vokes (1971) for base metal deposits at Broken Hill, Australia, and Bleikvassli, Norway, respectively. However, some doubt was expressed over the validity of these early interpretations (Plimer, 1987; Skinner & Johnson, 1987). Metamorphic sulfide anatexis was discussed briefly by Marshall et al. (2000), but at the time only one deposit, Legenbach in Switzerland, had been shown to have partially melted (Hofmann, 1994; Hofmann & Knill, 1996; Knill, 1996).
Since 2001, it is being recognized, at an increasing number of metamorphosed sulfide ore deposits around the world, that part of the ore mineral assemblage underwent partial anatexis during metamorphism (Mavrogenes et al., 2001; Tomkins & Mavrogenes, 2002; Tomkins et al., 2004; Baile & Reid, 2005; Sparks & Mavrogenes, 2005; Ciobanu et al., 2006). The ore minerals in these deposits that melted include sulfides, sulfosalts (containing As or Sb and S), tellurides and native minerals (metallic minerals lacking S and Te). Thus, an extensive range of minerals is referred to in this paper and abbreviations are presented in Table 1.
|
The metamorphic conditions inferred for these previously described deposits varied from
520°C, at the Lengenbach deposit in Switzerland (Hofmann, 1994), to as high as 850°C at the Challenger deposit in Australia (Tomkins & Mavrogenes, 2002). The lower end of this range shows that the onset of partial sulfide anatexis can occur at significantly lower temperatures than silicate partial melting. However, because large accumulations of sulfide minerals are rare, metamorphic partial sulfide anatexis is unlikely to have ever generated very large volumes of sulfide magma compared with silicate magmas. Nevertheless, understanding the processes behind sulfide partial melting during metamorphism is important for a range of reasons. During high-temperature metamorphism and deformation of sulfide ore bodies, even small proportions of sulfide melt are easily mobilized because their viscosities are very low (e.g. Dobson et al., 2000). Further, the high viscosity contrast between solid sulfide minerals and sulfide melts implies that generally only melt will be mobilized and not a mixed liquid–solid mobilizate. This sulfide melt mobilization leads to development of complex mineralogical–structural associations that are impossible to interpret correctly if partial sulfide melting is not recognized (Tomkins et al., 2004). Therefore, awareness that partial sulfide melting can occur facilitates a more accurate interpretation of ore body evolution, leading to better genetic models to be used in finding similar deposits. Understanding sulfide melt mobilization is also important to mine planning and near-mine exploration because segregation of even small amounts of sulfide magma can significantly redistribute ore metals, especially precious metals, in and around pre-existing mineral deposits (Tomkins & Mavrogenes, 2002; Tomkins et al., 2004). A number of important questions remain regarding metamorphically induced partial sulfide melting. (1) Under what metamorphic conditions do typical sulfide ore deposits start to melt? (2) Is this process capable of generating enough melt to be mobilized through sulfide magma veins and dykes over hundreds of meters or kilometers? (3) Can this process actually form new ore deposits? (4) Can our understanding of this process help us find undiscovered ore deposits?
The aim of this paper is therefore to explore the initial partial melting relationships in the common types of sulfide mineral deposits, and, in the process, gain an understanding of how much melt can be generated in various deposit types. A preliminary review of partial sulfide melting during metamorphism was conducted by Frost et al. (2002). Based primarily on an examination of textures at the Broken Hill Pb–Zn deposit in Australia (metamorphosed at granulite-facies conditions), those workers reviewed some of the published phase relations before developing a model for the formation of sulfide melts in massive Pb–Zn deposits. Five characteristic features were identified that may be used as evidence of partial sulfide melting, and then some other deposit types were briefly discussed as part of a section aimed at identifying additional partially melted ore deposits. The present study takes a more detailed look at the natural constraints on initial melting in all sulfide-bearing ore deposit types and attempts to constrain the proportions of melt generated as a function of natural conditions. This is achieved through bringing together previous experimental studies and observations of several ore deposits metamorphosed at conditions ranging from mid-amphibolite facies to granulite facies.
|
OVERVIEW OF CONDITIONS REQUIRED FOR SULFIDE MELTING |
|---|
|
TOPABSTRACTINTRODUCTIONOVERVIEW OF CONDITIONS REQUIRED...EARLIEST MELTING IN SULFIDE...MOBILIZATION OF POLYMETALLIC...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
The temperature of sulfide melting
The most fundamental constraints on sulfide melting are provided by experimentalists, who have investigated phase relations in numerous systems relevant to sulfide ore deposits (almost all experiments were conducted at atmospheric pressures). These studies defined the melting temperatures for many individual ore minerals and more importantly, for many naturally occurring ore mineral assemblages. Table 2 lists the melting temperatures (Tm) for a range of naturally occurring ore minerals and mineral assemblages (the list is not complete because experimental data are lacking for many possible systems). This table summarizes the most basic melting relationships, and more detailed information is contained in numerous phase diagrams spread throughout the geological literature. Many of the relevant phase diagrams have been compiled by Tomkins et al. (2004
).
|
Table 2 is broken into two parts to highlight the difference between melting of individual ore minerals in isolation and partial melting of coexisting ore minerals. In multiple mineral systems, the more minerals that coexist (i.e. are able to communicate chemically), the lower the initial partial melting temperature of the combination. For example, the melting temperature of gold is 1064°C, but if a small amount of bismuth is added, the combination partially melts at 375°C (at 1 bar; if >33 at. % Bi is added the melting point is even lower, at 241°C). Because Table 2 lists the melting temperatures of mostly individual minerals and binary systems, the actual initial partial melting temperatures of many natural ore assemblages, which have more coexisting minerals, are lower. The exception to this rule occurs where there is solid solution between two or more components in individual minerals. A good example is the system Sb2S3–PbS (stibnite–galena). Where a large ratio of Sb2S3 to PbS occurs, PbS is saturated in Sb2S3 and partial melting occurs at moderate temperatures (<550°C). In contrast, a low ratio of Sb2S3 to PbS (<
3 mol % Sb2S3) results in a solid solution that does not melt until much higher temperatures are reached (far higher than the melting point of Sb2S3 alone), although the solid solution nevertheless melts at lower temperature than pure PbS (Salanci, 1979). The information in Table 2 shows that numerous ore minerals melt at low temperatures, many of them within the greenschist facies. However, the minerals that melt at low temperatures are mostly rare sulfosalts as well as tellurides and some native minerals. In contrast, the sulfides melt at significantly higher temperatures.
Factors affecting the temperature of sulfide melting
Pressure
In the few studies that have investigated the effects of pressure on partial sulfide melting reactions, it has been found that melting temperatures are only slightly affected by pressure. The eutectic temperature of the system pyrrhotite–galena–sphalerite increases by 6°C/kbar (Mavrogenes et al., 2001), and the temperature of the reaction arsenopyrite+pyrite=As–Smelt+pyrrhotite increases by 14°C/kbar (Sharp et al., 1985), whereas the temperature of the reaction jordanite=galena+Pb–As–Smelt does not vary with pressure (Roland, 1968). The influence of pressure on these reactions is shown in Fig. 1. The temperature of the eutectic in the system Fe–FeS has been studied at pressures ranging up to 60 kbar and it has been found to vary only slightly with changing pressure (see Brett & Bell, 1969; Ryzhenco & Kennedy, 1973; Usselman, 1975).
|
In S-absent native metal systems most elements melt at slightly higher temperatures with increased pressure. Bismuth and antimony are noteworthy for the fact that they melt at progressively lower temperatures with increased pressure (–4·5°C/kbar and –0·2°C/kbar respectively; Liu & Bassett, 1986
). The effect of pressure on the melting temperature of telluride minerals remains unknown.
Sulfur and oxygen fugacity
Sulfur fugacity has an important influence on partial melting reactions that proceed with pyrite as a reactant. Some examples include
|
| (1) |
|
| (2) |
|
| (3) |
|
| (4) |
; Craig & Vokes, 1993). Oxygen fugacity can exert a major influence on pyrite stability in deposits characterized by disseminated sulfide assemblages (see Tomkins et al., 2006), such that low fO2 conditions promote pyrite consumption during metamorphism (e.g. in graphite-bearing rocks), thus indirectly affecting sulfide melting. Although sulfur fugacities high enough for reactions (1) and (2) may be possible in a variety of natural rocks, the very high fugacities required for reactions (3) and (4) (Fig. 2) may be rare (Tomkins et al., 2006).
|
Water and other hydrothermal phases
Water affects partial sulfide melting both directly and indirectly. Wykes & Mavrogenes (2005
) showed that water induces a melting point depression in the system FeS–PbS–ZnS of 35°C (from 900°C to 865°C at 15 kbar), although they suggested that this effect may be muted at lower pressures. The effect of water on other sulfide systems remains unknown. Indirectly, at amphibolite-facies conditions and above, the presence of water requires consumption of pyrite to maintain equilibrium proportions of H2S, SO2 and S2 in the fluid (Connolly & Cesare, 1993). As grade increases, increasing amounts of pyrite are consumed.
Little is known about whether other common components of the fluid phase such as CO2, CH4, H2S, SO2 and halogens have any direct effect on the melting temperatures of sulfide minerals. In a study on the chalcophile behaviour of halogens, Mungall & Brenan (2003) found that the melting point of monosulfide solid solution (mss) in the system Fe–Cu–Ni–S is lowered by the presence of Cl, and that Cl is progressively enriched in the melt during crystallization. Br and I, but not F, were also found to dissolve in small amounts in this system.
Comparison with silicate metamorphism
Figure 3 illustrates the conditions required for partial sulfide anatexis in relation to the classic metamorphic facies. Only a few key reactions out of the many listed in Table 2 are shown. The metamorphic facies boundaries come from Spear (1993), with the exception of the amphibolite–granulite facies boundary, which comes from Pattison et al. (2003).
|
Deposits that contain sulfosalts or tellurides may start to melt at conditions ranging from lowest greenschist facies to amphibolite facies, well below conditions required for partial melting of common silicate rocks. An important and potentially widespread reaction involves partial melting of pyrite+arsenopyrite at lower amphibolite conditions.
Deposits lacking sulfosalts and/or tellurides may begin to melt once P–T conditions reach the upper amphibolite facies, if galena and chalcopyrite are present, or well into the granulite facies, if galena is absent. Partial melting of another widespread and common sulfide association, chalcopyrite+pyrrhotite, does not occur within the range of normal metamorphism. However, partial melting of bornite+pyrrhotite or bornite+chalcopyrite commences at lower temperature ( 800°C, at 1 bar), near the amphibolite–granulite facies boundary. Thus, there are two broad sulfide melting domains: a low- to medium-temperature, low melt volume domain involving the melting of minor sulfosalts and/or tellurides (this may involve complete melting of some phases and partial melting of others); and a high-temperature, potentially higher melt volume domain involving the partial melting of major sulfide minerals. The boundary between minor and major melting is depicted as a shaded band to reflect the effect of trace metals, as well as the possible effect of water, on lowering the eutectic melting temperature between the major sulfide minerals, galena+chalcopyrite+pyrrhotite+sphalerite. The lower temperature side of the band represents partial melting in the presence of water (35°C melting point depression; Wykes & Mavrogenes, 2005) and trace metals, whereas the upper temperature side represents partial melting in the absence of water and trace metals.
Mineral communication and mobilization-assisted melting
In any typical sulfide deposit each sulfide mineral will coexist with a variety of other minerals. The degree of chemical communication between the different minerals will affect the extent of melting. As an example, we may consider partial melting between galena, sphalerite and pyrrhotite (Fig. 4a). Within a given deposit, some galena grains might coexist with sphalerite only, others with pyrrhotite only, and still others with both sphalerite and pyrrhotite (Fig. 4b). As in most systems, partial melting occurs first where all three sulfide minerals coexist at a triple point, whereas there is no melting where only two coexist (Fig. 4c). Partial melting continues until all of one of the phases is consumed (in our example it is galena; Fig. 4d and e), at which point the melt is saturated in all three components and has a eutectic composition, which in this system is dominated by galena. As temperature rises above the eutectic temperature, sphalerite and pyrrhotite continue to partially melt, thereby remaining saturated in these phases. Because all of the galena at this locality has already been consumed, the melt is now no longer saturated in this mineral and the melt composition is now on the ZnS–FeS cotectic. If this galena-undersaturated melt is now mobilized it will melt enough of any galena that it comes in contact with to regain equilibrium (Fig. 4f). We refer to this last stage as mobilization-assisted melting.
|
Through mobilization-assisted melting a more diverse array of elements can be incorporated in the melt. For example, if the galena grain isolated within sphalerite in Fig. 4e was instead chalcopyrite, it too would be at least partly incorporated in the melt in Fig. 4f. Because adding more elements typically lowers the melting temperature, mobilization-assisted melting also allows more of the melt's saturated components (ZnS and FeS in the example) to be incorporated. It is clear that considerably more melt can be generated via this process.
A further important point to make is that sulfide ore deposits are not mineralogically homogeneous. Therefore, although mobilization-assisted melting acts to generate a more homogeneous melt composition, in reality the bulk compositions of local aliquots of sulfide melt across an entire deposit will vary considerably. Deposit-wide mineralogical variability also affects the volumes of melt that can be produced from different parts of a given deposit. Locally higher concentrations of the low melting point minerals and elements, a natural consequence of the original hydrothermal mineralizing process, will produce locally higher melt volumes. Thus, although on average only 1% of the deposit may melt (for example), the local melt proportion at small scales might exceed 10%. This applies to all sulfide deposits that melt.
|
EARLIEST MELTING IN SULFIDE ORE DEPOSITS |
|---|
|
TOPABSTRACTINTRODUCTIONOVERVIEW OF CONDITIONS REQUIRED...EARLIEST MELTING IN SULFIDE...MOBILIZATION OF POLYMETALLIC...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
We have categorized the various types of sulfide ore deposits into four general groups: massive Pb–Zn±Cu deposits; gold deposits; magmatic Ni–Cu–PGE (platinum group element) deposits; disseminated Cu deposits. These groups are split into the more well-known ore deposit types and the typical sulfide mineral assemblages found in each type are listed in Table 3. This organization provides a geochemical framework for our analysis of initial melting within each of the four general groups. Several highly metamorphosed ore deposits of different types have been investigated to provide a basis for theoretical discussions on the conditions of initial melting in sulfide mineral deposits, and the volumes of melt expected to be generated. The location of these deposits is shown in Fig. 5.
|
|
Massive Pb–Zn±Cu deposits
Observations
The Lengenbach deposit (Switzerland) is a Pb–Zn-rich massive sulfide deposit that has been shown to have partly melted during moderate-temperature metamorphism at approximately 520°C (Hofmann, 1994
; Hofmann & Knill, 1996). It has an unusually Tl- and As-rich assemblage that is not typical of this group of deposits. We have investigated two more intensely metamorphosed deposits that contain massive Pb–Zn±Cu sulfide assemblages, the Broken Hill and Montauban deposits. The Broken Hill deposit (New South Wales, Australia) was metamorphosed at granulite-facies conditions of approximately 800–850°C and 5 kbar (see Phillips, 1980; Stevens et al., 1988; Cartwright, 1999; Frost et al., 2005) and has been previously demonstrated to have partly melted (Mavrogenes et al., 2001; Frost et al., 2005; Sparks & Mavrogenes, 2005). The Montauban deposit (Quebec, Canada; Stamatelopoulou-Seymour & MacLean, 1984) was metamorphosed at middle to upper amphibolite-facies conditions (650°C, 5 kbar; Bernier et al., 1987). Possible arsenopyrite melting at Montauban and at two additional deposits (Geco and Osborne Lake mines, both in Canada) was discussed by Tomkins et al. (2006), and several other deposits where partial sulfide melting may have taken place were identified by Frost et al. (2002). In Fig. 6 back-scattered electron (BSE) images of mineral associations from Broken Hill and Montauban are presented, which may be interpreted as crystallized accumulations of melt when the assemblage and metamorphic conditions are considered together. Experimental verification of this interpretation at Montauban is described below. These examples all come from small fractures in the wall rock adjacent to the massive sulfide units. An important characteristic of each is the abundance of sulfosalts and, in some, native minerals, relative to the volumetrically dominant massive sulfide minerals. The examples in Fig. 6 are highly enriched in elements such as Sb, As, Bi, Ag, Au and Pb, but depleted in Zn and Fe when compared with the bulk composition of the massive sulfide unit.
|
Several samples from each deposit contain numerous inclusions, fractures and intergranular aggregations filled with these sulfosalts, sulfides and native minerals. In several of these samples we have attempted to measure and calculate local bulk compositions of a statistically significant number of accumulations. This was done by taking a BSE image of each accumulation and obtaining a wavelength-dispersive spectromety (WDS) analysis of each mineral in the accumulation using the JEOL JXA8200 electron microprobe at the University of Calgary. Approximate bulk compositions of each accumulation were obtained using image analysis techniques within Adobe Photoshop to calculate the area of each analysed mineral in each BSE image. This technique cannot provide an exact local bulk composition because we can only ever view a two-dimensional section through each sulfosalt accumulation. Figure 7 shows a variation diagram produced from these data for a sample from the Montauban deposit (the same sample as used in the experiment described in the next section). There is an inverse correlation between the amount of Pb (as PbS) in each accumulation and the amount of Sb, As, Ag, and Au.
|
An experimental test of sulfide melting at the Montauban deposit
To test if native mineral–sulfosalt–sulfide accumulations at the Montauban deposit were once melts, a natural sample was held at temperature and pressure approximating peak metamorphic conditions in a Boyd & England (1960
) type piston cylinder apparatus (the experiment was conducted at the Research School of Earth Sciences at the Australian National University). One half the sample shown in Fig. 7 was polished and examined by reflected light microscopy to confirm that the sample used in the experiment contained numerous native mineral–sulfosalt–sulfide accumulations. The other half of the sample was shaped to fit inside a graphite sleeve. The sample and graphite sleeve were then encased in a talc sleeve. A corundum disc separated the sample from a thermocouple and boron nitride (BN) plugs fitted into both ends of the sleeves. The thermocouple, inserted through one of the BN plugs, allowed temperature readings from within 0·5 mm of the sample. This assembly was held in the piston cylinder apparatus at 650°C and 5 kbar, the estimated metamorphic conditions of the deposit, for 24 h and then quenched. The experiment product was analysed using reflected light and electron microscopy.
Metallic and sulfide melts do not quench to glasses, but crystallize on quenching to a microcrystalline intergrowth of phases (e.g. Sparks & Mavrogenes, 2005). In contrast, unmelted material remains as coarse individual grains that are not mixed with other material. Gravity separation of melt from residuum in experiments by Stevens et al. (2005) showed that these quench textures indicate the presence of a sulphide melt. Observation of these quench textures in the experiment product is therefore taken as evidence that the accumulations in question were once at least partly molten. Analyses by reflected light and electron microscopy of the experiment product show quench textures (Figs. 6d), which indicate that the native mineral–sulfosalt–sulfide accumulations from the Montauban deposit were at least partly molten at the estimated metamorphic conditions. This therefore suggests that a small proportion of melt containing Ag, As, Au, Pb, Sb and S was present during peak metamorphism.
Mineralogical constraints on initial melting
The ore mineral assemblage in volcanogenic massive sulfide (VMS), Mississippi Valley type (MVT), and sedimentary exhalative (SEDEX) type deposits is typically dominated by some combination of pyrite, pyrrhotite, sphalerite, galena and chalcopyrite (in this section we use chemical formulae when discussing experiments and mineral names when discussing natural minerals). Based on an experimental study in the system CuFeS2–PbS–FeS–ZnS–S conducted at 2 kbar, it has been suggested that where these minerals coexist they may start to melt at temperatures between 700 and 730°C (Stevens et al., 2005). In an early study conducted at atmospheric pressures, Craig & Kullerud (1967) suggested that CuFeS2+PbS start to melt at 630°C. Our observations of textures in ore deposits suggest that the Stevens et al. (2005) study is more accurate; coexisting galena and chalcopyrite appear stable at Montauban, whereas chalcopyrite is a common component of the Broken Hill melts. Consequently, the Stevens et al. (2005) data are listed in Table 2 and shown in Fig. 3.
However, in many deposits galena and chalcopyrite do not occur together. In the absence of chalcopyrite and any minor phases, partial melting may not occur until temperatures of 800°C (at 5 kbar) are reached, when galena+sphalerite+pyrrhotite+H2O (+trace Ag) start to melt (Mavrogenes et al., 2001; Wykes & Mavrogenes, 2005). In galena-free massive sulfide assemblages containing pyrrhotite, chalcopyrite and sphalerite, partial melting may not occur within the limits of typical crustal metamorphism. Although the eutectic temperature in this system has never been determined by experimentalists, Kojima & Sugaki (1984) indicated extensive solid solution fields at 800°C (at 1 bar). However, some minor sulfide and sulfosalt phases do occur in many of these deposits and they may be critical to the low-temperature generation of small quantities of melt.
In many occurrences, minor Sb-, Bi- and Ag-bearing sulfosalt minerals appear to have formed during cooling, as these elements exsolved from galena, with which they are commonly associated (Fig. 8a and b). Figure 9 shows the large extent of coupled substitution of Ag+–Sb3+ and Ag+–Bi3+ for 2Pb2+ in PbS (e.g. Blackburn & Schwendeman, 1977). Given the typically low concentrations of Sb, Bi and Ag in these deposits relative to Pb (ores typically contain <0·1% Ag, Sb and Bi, but percent levels of Pb as galena), it is likely that all would substitute into galena during metamorphism without causing any partial melting. However, Sb-, Bi- and Ag-doped PbS starts to melt at lower temperatures than pure PbS, when combined with other sulfide minerals (e.g. Mavrogenes et al., 2001).
|
|
One mineral that typically occurs in minor amounts in these deposits and that cannot form as an exsolved mineral during cooling is arsenopyrite. In contrast to Sb, Bi and Ag, there is no known solid solution of As in PbS (Fig. 10). Arsenopyrite can partly melt during metamorphism through reaction (1), when high fS2 conditions are generated through decomposition of pyrite. Reaction (1) occurs at 491°C at 1 bar and
560°C at 5 kbar (Clark, 1960; Sharp et al., 1985). This and several other metamorphic processes that can also lead to formation of As–S melt from arsenopyrite breakdown have been discussed extensively by Tomkins et al. (2006). Because there is no substitution of As in PbS, the presence of small quantities of As–S melt causes partial melting of galena at temperatures as low as 549°C (at 1 bar), the melting point of Pb14(As,Sb)6S23 (jordanite; Fig. 10). Experiments by Roland (1968), conducted at pressures of up to 2 kbar, found that the melting point of Pb14(As,Sb)6S23 (jordanite) was not affected by pressure within the uncertainty of measurement. It should be noted that Fig. 10 does not represent the phase relations exactly because the As–S melt produced through arsenopyrite melting does not have the composition As2S3; it has a composition that has slightly less sulfur than arsenic (Tomkins et al., 2006). The ternary phase relations in the system As–Pb–S (Roland, 1968) suggest that this should not greatly affect the topology of the phase diagram above 550°C (see Tomkins et al., 2006, fig. 4). The amount of PbS that melts when combined with an As–S melt of appropriate composition should be slightly greater than as represented by Fig. 10 (as suggested by the dashed liquidus surface on the right of the diagram).
|
Other minor As-bearing sulfosalts may also melt at mid-amphibolite-facies temperatures (e.g. Maske & Skinner, 1971
), but in these deposits these minerals are typically rare compared with arsenopyrite. At middle amphibolite-facies conditions (c. 550–700°C), interaction between As–S melt and PbS results in extensive partial melting of the latter (Fig. 10). For example, at 650°C (and 1 bar) 1 mole of As–S melt will cause at least 3·8 moles of PbS to melt. However, typical bulk compositions in massive sulfide deposits have a very large ratio of galena to arsenopyrite, so the As–S melt would become saturated in galena, and most of the galena would not melt. Nevertheless, it is of interest to know how much melt might be generated in these deposits as a result of the presence of minor arsenopyrite.
Volume of melt generated during initial melting
We have derived an equation to estimate the amount of As–Pb–S melt present in any given galena-bearing rock (above 549°C) using the arsenic content from bulk-rock analysis (based on Fig. 10). First, the amount of As–S melt from arsenopyrite breakdown is calculated; the amount of sulfur required is calculated from the As–S phase diagram (Massalski et al., 1990), assuming that the As–S melt composition is on the liquidus in this system (see Tomkins et al., 2006):
|
| (5) |
Then, the amount of PbS incorporated by the As–S melt is calculated by fitting a curve to a graph of moles of PbS melted vs temperature, according to the phase diagram (Fig. 10) of Kutolglu (1969):
|
| (6) |
Finally, these are incorporated into one formula and converted to volume percent:
|
|
|
| (7) |
), the effect of pressure is likely to be minor. The density of the As–S component of the melt is assumed to be 3560 kg/m3 and that of the Pb–S component to be 7580 kg/m3, based on the density of realgar and galena, respectively. The result represents the maximum amount of As–Pb–S melt that can exist, assuming that all of the As in the rock forms a melt and that all As–S melt can communicate chemically with excess galena. In many rocks these are not valid assumptions (see Tomkins et al., 2006) and care should be taken when applying this formula. Table 4 shows the results of this calculation performed on a range of ore deposits. It can be seen that although the overall As content of massive Pb–Zn deposits is typically low, a significant melt fraction can be generated locally in As-rich portions of the deposit if all of the As in the rock forms a melt. It should be noted that this calculation considers only As-bearing sulfosalts and galena, and the elements As, Pb and S. If other sulfide and sulfosalt minerals contribute to the melt, or other elements partition into the melt from unmelted sulfide minerals, then the melt ratio may actually be higher.
|
Once partial melting between the major sulfide minerals commences in the upper amphibolite or granulite facies (750–800°C, Table 2 and Fig. 3), the volume of melt in massive Pb–Zn sulfide deposits can substantially exceed 0·5 vol. %. The volume of melt produced at this grade of metamorphism again depends on the bulk composition. Deposits with a high proportion of galena and chalcopyrite together with some sphalerite and pyrrhotite will generate greater melt volumes than deposits containing only minor galena. Granulite-facies metamorphism of such a galena- and chalcopyrite-rich deposit could generate melt volumes in the 5–30% range, volumes that would in all probability allow segregation of sulfide magma dykes.
Our observations of crystallized sulfide melt accumulations at the Broken Hill and Montauban mines indicate significant enrichment in Sb, Ag and Bi relative to Pb, when compared with galena in the massive sulfide units (Fig. 6). This observation is matched by that of Sparks & Mavrogenes (2005), who described similarly enriched sulfide melt inclusions at Broken Hill. Therefore, these elements, which initially would be in solid solution in galena, are interpreted to diffusively partition into As–Pb–S melts where melt communicates chemically with unmelted galena. This addition of Sb, Ag and Bi to the melt should enhance its capacity to cause further partial melting, thus causing a positive feedback effect during initial melting, although the overall effect is likely to be small because of the rarity of the elements.
Mobilization-assisted melting in massive Pb–Zn±Cu deposits
Because the bulk compositions of massive Pb–Zn±Cu deposits are dominated at high metamorphic grades by sphalerite, pyrrhotite, galena and in some cases chalcopyrite, the small volumes of initial melt will be saturated in most or all of these phases. Thus, extraction of a partial melt will leave some or all of these minerals in the restite. In cases where PbS is a saturated component of the partial melt, interaction between the melt and unmelted galena through mobilization-assisted melting is thought to induce preferential partitioning of Ag, Sb and Bi from unmelted galena into the melt. If this galena-saturated melt interacts with a large proportion of unmelted galena during mobilization, the melt may become extremely enriched in these elements relative to the restite. This is analogous to using R-factors to explain PGE enrichment in magmatic sulfide melts. During formation of some massive Pb–Zn±Cu deposits, Au and Ag are preferentially precipitated together with low melting point minerals and elements (see Tormanen & Koski, 2005), increasing the likelihood that they would be incorporated into any melt produced during metamorphism, and therefore mobilized.
Evidence for preferential metal accumulation in melts
Support for the idea that initial partial melts in massive Pb–Zn deposits should be rich in As, Sb, Ag, Bi and Au (if present) and saturated in PbS, but low in Fe and Zn, is provided by the data in Fig. 7. These diagrams are a form of Harker diagram, as used by igneous petrologists to show fractionation in silicate melts, but here used to show fractionation of a sulfide melt. Figure 7 shows the approximate local bulk compositions of a number of sulfide–sulfosalt–native mineral aggregations in one small (3 cmx2 cm x1 cm) sample from the gold-rich part of the Montauban deposit adjacent to the massive sulfide unit (e.g. Figs. 6b and c). The abundant dyscrasite (Ag3Sb; Tm=585°C) in many of these, and pyrargyrite (Ag3SbS3; Tm=485°C) along micro-veinlets connecting the aggregations, leaves little doubt that there was a significant proportion of sulfide melt in this sample. This is confirmed by the experiment conducted on this sample, which produced melt when subjected to P–T conditions approximating peak metamorphism (Fig. 6d).
As suggested by Fig. 10, the initial partial melt, which probably formed within the massive sulfide unit before migrating into the wall rock, was probably Pb-rich and contained significant As (+Sb, Ag and Au in this case). Importantly, the initial partial melt would have been saturated in galena, so when it started to cool, galena was the first mineral to crystallize. Assuming continual migration during cooling, through either gravitational or deformational processes, fractionation of the melt from the residual galena would make the melt increasingly rich in As, Sb, Ag and Au, producing the fractionation trend of Fig. 7a. The melt should be thought of as a complex many-element system, rather than as the simple binary system of Fig. 10. Thus, although there is a thermal divide in Fig. 10 that a cooling melt could encounter, the elemental complexity of the melt means that there is likely to be a series of diverging cotectics that the melt composition follows as it cools towards an ultimate eutectic. Therefore, as other sulfosalts started to crystallize with further cooling, local melt patches as well as local solid residue packets deviated in composition from the ideal fractionation line produced by galena-only crystallization. This may explain the scatter seen in Fig. 7b and c. In contrast to the bulk composition of the massive sulfide unit as a whole, which is essentially a restite, these melt accumulations contain little Fe or Zn. The Fe that is present in the accumulations may represent interaction during cooling between the Fe-poor melt and the pre-existing disseminated pyrrhotite that occurs throughout the sample. It can be seen that the initial partial melt at Montauban was enriched in Pb, As, Sb, Ag and Au, relative to the massive sulfide unit and became increasingly enriched in all but Pb as it cooled and crystallized.
Textural evidence for the existence of a sulfide melt is difficult to recognize within the massive sulfide restite at Montauban. There are no cumulate textures at this locality because only a very small proportion of the deposit melted. In any example of partial melting and melt migration within a massive Pb–Zn±Cu deposit, the restite is likely to be relatively depleted in Pb, As, Sb, Ag, Au and possibly Cu, and enriched in Fe and Zn (Fig. 7).
Gold deposits—the influence of sulfosalts
Observations
Two gold deposits, the Challenger deposit (South Australia) and the Hemlo deposit (Ontario, Canada), have been shown to have undergone metamorphic sulfide anatexis (Tomkins & Mavrogenes, 2002; Tomkins et al., 2004). The Challenger deposit, metamorphosed at granulite-facies conditions, is relatively sulfosalt-poor with a restricted ore mineral assemblage, whereas Hemlo, metamorphosed at mid-amphibolite-facies conditions (Powell et al., 1999), is a sulfosalt-rich deposit with a broad array of ore minerals. Challenger probably had an assemblage typical of Archean orogenic gold mineralization (Tomkins & Mavrogenes, 2002), whereas Hemlo is thought to have had a pre-metamorphic ore mineral assemblage analogous to both epithermal and porphyry mineralization (Tomkins et al., 2004).
Figure 11 shows some examples of mineral associations from each deposit that, when considered together with experimental studies, metamorphic grade and structural association, are indicative of a sulfide melt (Tomkins & Mavrogenes, 2002; Tomkins et al., 2004). As in the massive Pb–Zn deposits, the examples shown in Fig. 11 are enriched in elements such as As, Sb, Bi, Au, Ag, Hg, Tl and Pb (the LMCE of Frost et al., 2002), but depleted in Fe (as well as Zn and Mo at Hemlo) relative to the bulk of the sulfide material in each deposit. An additional example, from the Soimus Ilii prospect (Romania), shows that metal-rich melts can influence the distribution of gold even at comparatively low temperatures during greenschist-facies metamorphism and deformation (> 400°C, no pressure estimate; Ciobanu et al., 2006). The sulfide melt at this locality is thought to have contained Au, Bi, Pb, Te and S.
|
Mineralogical constraints on initial melting
Gold deposits are extremely diverse in their ore mineral associations, reflecting a wide range of formation mechanisms (Table 3). Five gold deposit types are listed in Table 3. Of these, only the alluvial gold deposits are unlikely to melt during metamorphism, simply because, in these deposits, gold is not associated with any sulfide or sulfosalt minerals that might depress the melting point of gold to lower temperatures.
The gold deposits that may start to melt during very low temperature metamorphism are those that contain sulfosalt minerals, telluride minerals, and/or native bismuth or mercury (typically epithermal deposits as well as some orogenic gold deposits). There is a diverse range of minerals and mineral systems in these deposits that can melt at low temperatures (Table 2), and many, if not all, of the resulting melts have the capacity to incorporate and mobilize gold and silver. In the example at Soimus Ilii, remobilization of Au-bearing melt commenced when native Bi melted (completely) and migrated during deformation, incorporating some Au, Pb, Te and S as it went (Ciobanu et al., 2006). Most of the common sulfide minerals, including pyrite, pyrrhotite, sphalerite and chalcopyrite (as well as molybdenite), do not contribute greatly to these sulfosalt-rich low-temperature melts (Tomkins et al., 2004), and therefore remain behind as residual phases during melt mobilization. This explains the enrichment of Au and LMCE and depletion of elements such as Fe, Zn and Mo in the examples in Fig. 11. This is also essentially the same fractionation tendency as that seen in the massive Pb–Zn deposits.
Gold deposits that contain only pyrite, pyrrhotite and arsenopyrite (i.e. many orogenic Au deposits) may not start to melt during metamorphism until considerably higher temperatures are reached. Arsenopyrite+pyrite may melt at lower amphibolite-facies conditions to form As–S melt+pyrrhotite [reaction (1)], but only under high fS2 conditions (see Tomkins et al., 2006). Generation of high fS2 conditions during metamorphism is favoured by pyrite-rich rocks that maintain moderate fO2 (rocks that are graphite-free, for example), and that are low in Fe silicates and oxides. This is not to say that high fS2 conditions are impossible in less favourable rocks, just that more pyrite must be consumed to generate enough sulfur to maintain equilibrium. Thus, if a deposit hosted by less favourable rocks is initially low in pyrite, all pyrite may be consumed before high fS2 conditions are reached, precluding reaction (1) from taking place. Consequently, if the gold deposit is hosted in graphitic schist, or Fe-rich rocks such as banded iron formation or dolerite, arsenopyrite may not melt until the following reaction occurs:
|
| (8) |
770°C (at 5 kbar, dP/dT
14°C/kbar; see Clark, 1960; Sharp et al., 1985). Even these high-temperature partial melts contain little or no Fe. This reaction may also not proceed at low fS2 conditions, as it is a desulfidation reaction and thus also governed by fS2. It may be, therefore, that many orogenic gold deposits would not melt even at granulite-facies conditions.
Volume of melt generated
As in the massive Pb–Zn deposits, the extent of chemical communication between ore minerals in gold deposits affects melting. The disseminated nature of gold deposits means that there is little physical contact between ore minerals. However, as shown in Table 2, the many sulfosalts found in some gold deposits melt in isolation at low temperatures. Furthermore, arsenopyrite melting via reaction (1) probably does not require physical contact between arsenopyrite and pyrite. The only role of pyrite in reaction (1) is to liberate sulfur, thereby causing an increase in sulfur fugacity (Tomkins et al., 2006). Therefore, if enough pyrite is breaking down in the vicinity of an arsenopyrite crystal, a pervasive state of elevated fS2 may be achieved, causing the arsenopyrite to melt. Once some sulfosalts have melted, mobilization caused by deformation will lead to physical interaction between melt and unmelted ore minerals, including gold, which can then be incorporated into the melt. This mobilization-assisted melting thus leads to generation of a larger proportion of melt than would otherwise be possible. However, mobilization-assisted melting may be less effective in disseminated deposits because the low melt volumes can only migrate as far as the nearest structurally dilatant domain.
Because the ore minerals in gold deposits are disseminated, the total bulk sulfide content is low, so any melt that is generated during metamorphism is likely to comprise only a small proportion of the rock volume. In this case the proportional volume of melt can be appraised by conducting bulk-rock analysis for As, Sb, Te, Tl, Pb, Cu, Hg, Bi, Au and Ag, the likely melt components, and estimating how much each element contributed to the melt (sulfur must also be taken into account). Clearly, these chemical systems can be very complex, so the result can only give a maximum (by assuming that 100% of each element contributed to the melt). In Table 5 some estimates of the melt percentage for metal-rich parts of the Hemlo gold deposit are presented. Observations of small sulfosalt accumulations that formed during mobilization of a sulfosalt melt (Tomkins et al. 2004) indicate that locally and at small scales the melt volume may have exceeded 25%. However, the amount of melt in the deposit as a whole would have been low (<0·5 vol. %).
|
Initial metamorphic melting in magmatic Ni–Cu–PGE deposits
Highly metamorphosed deposits
There have been no previous studies of Ni–Cu–PGE deposits that show that partial melting occurred during high-temperature metamorphism. However, there are at least two localities where such a possibility exists. The deposits at both localities have been metamorphosed at upper amphibolite-facies conditions and contain minerals that would melt under such conditions. These are the various deposits near Thompson in Manitoba, Canada (e.g. Peredery et al., 1982
), and the Las Aguilas Ni–Cu–PGE deposit in Argentina (Gervilla et al., 1997). At both localities a range of rare minerals including arsenides, sulfarsenides, Pd-bismuthotellurides and antimonides have been previously interpreted to be part of a late-mobilized component of ore. At Thompson, Chen et al. (1993) interpreted this mobilization to be associated with late hydrothermal alteration involving conditions of 250–300°C and 3–4 kbar, whereas at Las Aguilas, Gervilla et al. (1997) offered no interpretation to explain the mobilization. In neither case was partial melting considered, although the mineralogy of the remobilized ore strongly suggests this as a possibility.
Constraints on initial melting
Magmatic Ni–Cu–PGE deposits form by fractionation of sulfide melt from mafic to ultramafic silicate magma (e.g. Naldrett, 1989). When this fractionation process is protracted during cooling, the sulfide melt itself evolves small amounts of lower temperature melts, which are variably enriched in As, Cu, Sb, Pb, Sn, Bi, Te, Pd, Pt, Ag and Au (Li et al., 1992; Gervilla et al., 1997; Pritchard et al., 2004). Although the bulk of the sulfide material in these deposits typically crystallizes at temperatures beyond the realm of normal crustal metamorphism, the highly fractionated low-temperature melts may be subject to remelting, and thus mobilization, during amphibolite- to granulite-facies metamorphism.
The late fractionated component of the sulfide magma in these deposits dominantly crystallizes minerals such as arsenopyrite (FeAsS), gersdorffite (NiAsS), cobaltite (CoAsS), nickeline (NiAs), maucherite (Ni11As8), löllingite (FeAs2), westerveldite [(Fe,Ni)As], safflorite [(Co,Fe)As2], chalcopyrite (CuFeS2) and cubanite (CuFe2S3) (Li et al., 1992; Gervilla et al., 1997; Pritchard et al., 2004). At individual localities, this melt can be Cu-rich (e.g. Sudbury, Canada, Li et al., 1992; Noril'sk, Russia, Naldrett et al., 1997; Uruguayan dyke swarm, Pritchard et al., 2004) or As-rich [e.g. Thompson, Canada, see Chen et al., 1993; Carratraca (Spain) and Beni Bousera (Morocco) massifs, Gervilla et al., 1996; Las Aguilas, Argentina, Gervilla et al., 1997]. A range of rare tellurides and sulfosalts occur as minor components that crystallize from these late-stage fractionates. Of the main minerals in As-rich and Cu-rich fractionates, only the sulfarsenides (minerals with As and S) are likely to melt under normal metamorphic conditions (Clark, 1960; Craig & Kullerud, 1967; Sharp et al., 1985; Singleton & Nash, 1986; Makovicky et al., 1992), although the others may melt during ultrahigh-temperature (UHT) metamorphism.
Because many of these deposits contain little or no pyrite and are dominated by pyrrhotite, pyrite decomposition during metamorphism may not buffer fS2 to the levels required to destabilize the sulfarsenides [as in reaction (1)]. With a very high ratio of pyrrhotite to pyrite, all of the pyrite will be consumed in maintaining pyrrhotite stoichiometry, which becomes more S-rich with increasing temperature along the Py–Po buffer (Fig. 2), as well as in maintaining fluid equilibrium. In such a situation, the sulfarsenides are unlikely to melt during metamorphism through reactions that occur on the Py–Po buffer, such as reaction (1). However, mass balance calculations based on Fig. 2 suggest that as long as Po:Py <99:1 and the system is closed to external fluids, pyrite will be preserved to high enough temperatures for arsenopyrite melting to take place through reaction (1).
It is also possible that sulfarsenides in a pyrrhotite-rich massive Ni–Cu deposit could melt with increasing temperature and fS2 through reactions within the pyrrhotite stability field, such as reaction (8) (without the Au). During metamorphism sulfur is liberated from pyrrhotite as temperature rises, leading to an increase in sulfur fugacity (Fig. 2). This desulfidation produces a steep log fS2–T trajectory that is essentially parallel to the NFeS isopleths in Fig. 2 (i.e. pyrrhotite stoichiometry stays almost the same); a trajectory that in some cases may lead to arsenopyrite melting (see Tomkins et al., 2006). Cobaltite and gersdorffite may also melt by the same process, although the fS2–T stability limits of these minerals is not known. Reaction (8) may not be applicable to other deposit types containing disseminated rather than massive pyrr