Journal of Petrology | Volume 40 | Number 5 | Pages 755-772 | 1999
© Oxford University Press 1999
Fluid Fluxing of Cumulates: the J-M Reef and Associated Rocks of the Stillwater Complex, Montana
Department of Geology, Duke University Box 90227, Durham, NC 27708, USA
Received May 5, 1998; Revised typescript accepted October 28, 1998
 |
ABSTRACT |
|---|
 TOP ABSTRACT IntroductionThe Lower Banded Series...Evidence for Fluid Fluxing...Fluid-Induced Melting of...A Hydromagmatic Model for...ConclusionsReferences |
|---|
Olivine-Bearing zone I (OB I) is host to the principal platinum-group element deposit of the Stillwater complex, the J-M reef. OB I is characterized by a lower Troctolite subzone composed of troctolite, dunite, gabbronorite and anorthosite, and an overlying Anorthosite subzone composed largely of anorthosite and lesser troctolite and norite. A number of petrologic features suggest that the olivine-bearing rocks from the Troctolite subzone are the product of partial melting of gabbronorite fluxed by fluids exsolved during crystallization of underlying intercumulus liquid: (1) Pegmatoids become more abundant immediately beneath OB I. (2) There are discordant, lateral changes from uniform-textured gabbronorite to olivine-bearing rocks in the lower half of OB I. The olivine-rich rocks also define pegmatoidal ‘pothole‘ structures. (3) Rock composition, texture and grain size vary considerably in olivine-bearing units, characterized by segregation between coarser mafic regions and medium-grained felsic regions. (4) Primary plagioclase in olivine-rich rocks has an eroded texture. (5) Modal variations are consistent with progressive incongruent melt reactions, and hydrous minerals are most abundant in olivine-rich rocks. (6) Hydrous melt inclusions (now crystallized to polyphase, hydrous mineral-bearing assemblages) are present in olivine and chromite. (7) Halogen-bearing minerals in the olivine-bearing assemblage have higher Cl contents than those associated with anorthosite and gabbronorite. In addition, the presence of massive sulfide associated with silicate pegmatoids is consistent with theoretical calculations that high-temperature fluids will have a significant component of sulfur. Modeling the thermal effects of fluid-induced partial melting of meter-sized partly solidified layers suggests nearly isothermal melting can occur when fluid is introduced slowly such that the heat of melting is supplied by the surrounding partially molten cumulates. The mineral changes that accompany the volatile-induced partial melting of a gabbronorite protolith are illustrated using the program MELTS and with phase diagrams that show how isotherms vary with water content. It is proposed that volatile fluxing of the lower portion of OB I led to the formation of olivine and chromite by incongruent melting of gabbronorite in the upper part of the crystal pile. The hydrous partial melt produced in this reaction, on mixing with resident pyroxene + plagioclase-saturated liquids in the magma chamber, produced hybrid liquids with crystallization order plagioclase followed by olivine or pyroxene, depending on the proportions mixed. This hybrid liquid crystallized rock of the Anorthosite subzone. The main platinum-group element sulfide concentrations occur at and below the boundary between these subzones, owing to the marked solubility difference of sulfur in volatile fluids compared with silicate liquids.
KEY WORDS: layered intrusions; partial melting; platinum-group elements
|
Introduction |
|---|
|
TOPABSTRACTIntroductionThe Lower Banded Series...Evidence for Fluid Fluxing...Fluid-Induced Melting of...A Hydromagmatic Model for...ConclusionsReferences |
|---|
Basic magmas that crystallized layered igneous intrusions such as the Stillwater complex have traditionally been considered to have been relatively ‘dry’, largely because of the paucity of hydrous minerals. However, investigations of halogen variations suggest that the high-Cl, high-Mg parent liquids of the Stillwater and Bushveld complexes, like their modern-day boninitic counterparts, contained a much higher volatile content than previously recognized—possibly in excess of 1.0 wt % H2O (Boudreau et al., 1997
). Although volatile components can have a marked effect on crystallization behavior of these magmas, most existing models either ignore or downplay their potential importance.
An example of conflicting interpretation involves the petrogenesis of Olivine-Bearing zone I (OB I) of the Lower Banded series of the Stillwater complex (Fig. 1). Although details among the various models vary, a number of studies have attributed crystallization changes to two distinct parent magma types. Below this zone, the rocks are interpreted to have crystallized from a magma of ‘ultramafic’ parentage and record the apparent crystallization order olivine ± chromite 
orthopyroxene orthopyroxene + plagioclase orthopyroxene + plagioclase + augite. This progression is interrupted in OB I where the rocks are predominantly composed of troctolites, anorthosites and dunites. This change is proposed to have resulted from a mixing event that blended the evolved ‘ultramafic’ liquid with an ‘anorthositic’ liquid that had plagioclase as the first mineral to crystallize (e.g. Irvine et al., 1983; Barnes & Naldrett, 1986; Wooden et al., 1991). In these models, the observed stratigraphic modal changes document variations in primary crystallization resulting from these magma mixing events, and volatiles are assumed to have played only a secondary role in modifying original textures (Barnes & Campbell, 1988).
|
In contrast to these ‘orthomagmatic’ models are those that suggest that mineral changes in OB I of the Stillwater complex (and in the Merensky Reef of the Bushveld complex) are the result of incongruent melting of preexisting cumulates induced by fluid fluxing (Boudreau, 1988
; Nicholson & Mathez, 1991). These models involve the interaction of volatile-rich fluids with crystal–liquid assemblages and might be termed ‘hydromagmatic’ to emphasize the interaction of volatile magmatic fluids with partly molten (supersolidus) magmatic assemblages.
A correct petrogenetic interpretation of OB I is of considerable practical importance as this zone hosts the J-M reef, an important platinum-group element (PGE) deposit. This paper summarizes existing work and presents a more complete model describing changes in the rock sequence in OB I as the result of the interaction of volatile fluids as they redissolve and partially melt an original assemblage composed of plagioclase, pyroxene and fluid-undersaturated silicate liquid. The importance of isothermal sections in phase diagrams is emphasized to demonstrate the changing composition of the melt and residual solid assemblage as it undergoes progressive flux-induced melting, as well as the effects of mixing of this hydrous melt with original silicate liquids resident in the chamber. It is shown that changes in crystallization sequence traditionally interpreted to be the result of mixing of parent magmas with different liquid lines of descent can instead arise from processes that occur during crystallization of a single magma. The interactions of fluids, minerals and liquid inferred for OB I may be a common phenomenon in layered intrusions as exsolved fluids interfere with nucleation and alter cotectic mineral–liquid assemblages close to the nucleation front, leading to the development of modal layering as proposed by McBirney (1987).
|
The Lower Banded Series of the Stillwater Complex |
|---|
|
TOPABSTRACTIntroductionThe Lower Banded Series...Evidence for Fluid Fluxing...Fluid-Induced Melting of...A Hydromagmatic Model for...ConclusionsReferences |
|---|
There have been a number of studies of the rock of the Lower Banded series of the Stillwater complex (e.g. McCallum et al., 1980
; Todd et al., 1982; Barnes & Naldrett, 1985, 1986; Page & Moring, 1990). With increasing structural height within the Lower Banded series, the rocks below OB I consist of relatively uniform-textured norite of Norite zone I (N I) (Fig. 2a), overlain by gabbronorite of Gabbronorite zone I (GN I). In the upper part of GN I modal variations become more pronounced to produce modal layering. Pegmatoids, common throughout the Ultramafic zone and along the Ultramafic series–Banded series contact but rare elsewhere in N I and GN I, again become more common in the upper part of GN I and the lower portion of OB I (see also below). The reappearance of these pegmatoids coincides with locally significant layer deformation (Fig. 2b–d). Plagioclase–biotite veins (Fig. 3a) similar to those reported below the Merensky reef by Nicholson & Mathez (1991), are present below and extend upward into OB I.
|
|
OB I is relatively melanocratic in the lower half [the Troctolite subzone of Todd et al
. (1982)] and leucocratic–anorthositic in the upper half [the Anorthosite subzone of Todd et al. (1982)] (Fig. 1). Olivine-bearing rocks in the Troctolite subzone are medium- to coarse-grained troctolite and subordinate dunite, typically with pegmatoidal texture (Fig. 3b and c). These are interlayered with minor anorthosite, norite, and gabbronorite, with thickness on the order of one to a few meters. Although locally variable, PGE-enriched sulfides of the J-M reef are most commonly concentrated within the lower part of OB I, and are particularly associated with the contact between the Troctolite and Anorthosite subzones (Fig. 1; also see Raedeke & Vian, 1986). The Anorthosite subzone of OB I is composed predominantly of anorthosite and anorthositic troctolite, both with minor interstitial augite and orthopyroxene. In contrast with the Troctolite subzone, pegmatoidal rocks and high-temperature vein assemblages are absent and rock textures are medium grained and more uniform in texture. Plagioclase remains dominant into the lower parts of norite zone II, where orthopyroxene againbecomes a euhedral ‘cumulus’ mineral.
Extensive mining development by the Stillwater Mining Company has revealed a significant region in which OB I cuts down into the underlying cumulates, locally reaching as far down as the lower parts of N I (Turner et al., 1985). Within the unconformable region, the thickness of OB I itself thins, such that at the deepest penetration into the underlying cumulates, there is a very short transition from N I to N II. It is also observed that in the region of this unconformity, the olivine-rich layers of the Troctolite subzone change laterally into melanocratic gabbronorite layers of GN I (Turner et al., 1985). The overlying Anorthosite subzone also thins, giving the regional appearance that the Troctolite and Anorthosite subzones both form an apparent angular unconformity. The surface of the unconformity itself is broadly defined by PGE–sulfide mineralization of the J-M Reef, although mineralization is considerably less abundant and locally absent in the region of the unconformity (Dahy et al., 1995).
Structurally above OB I, the rocks return to the mineral sequence observed in N I and GN I to form the N II and GN II zones. However, mineral compositions in N II and GN II are more evolved than is seen in the lower units (McCallum et al., 1980; Todd et al., 1982; Barnes & Naldrett, 1986). Pegmatoids are rare or not observed in N I. The rocks again become modally diverse in the upper part of GN II as the thick anorthosites of the Middle Banded series are approached.
|
Evidence for Fluid Fluxing in the Formation of Olivine-Bearing Zone I |
|---|
|
TOPABSTRACTIntroductionThe Lower Banded Series...Evidence for Fluid Fluxing...Fluid-Induced Melting of...A Hydromagmatic Model for...ConclusionsReferences |
|---|
Field relations
As noted above, pegmatoids become increasingly common as OB I is approached from below. They are characterized by being modally and compositionally similar to the host rock, the primary difference being their very coarse grain size and, in some instances, a partial replacement of pyroxene with amphibole. These have been interpreted to indicate increasing volatile concentrations and late channeling of fluids exsolved from these and underlying rocks (Braun et al., 1993
). Pegmatoids are not found in the Anorthosite subzone of OB I and in the overlying units. This suggests that the overlying cumulates had not yet crystallized at the time these fluids were migrating upward, and that the Troctolite–Anorthosite transition in OB I marks the approximate floor of the chamber during this fluid introduction event.
Pegmatoidal, olivine-rich rocks locally define potholes developed down into underlying gabbronorite (Turner et al., 1985). Although the origin of potholes is still debated, several studies have suggested these are regions where fluid migration up through the cumulus pile becomes localized at the floor of the magma chamber (e.g. Buntin et al., 1985; Ballhaus, 1988). They have many features that are similar to gas escape ‘pockmark’ structures that develop on the ocean floor (Boudreau, 1992). As with the distribution of pegmatoids, their presence in the lower half of OB I implies this zone was the uppermost part of the crystal pile during fluid introduction.
As noted above, in the region where OB I unconformably cuts down into the underlying cumulates, the texturally and modally heterogeneous olivine-rich layers of the Troctolite subzone change laterally into more uniform-textured melanocratic gabbronorite layers of GN I. This lateral change is consistent with the olivine-rich layers being replacements of initially pyroxene-rich layers.
Locally, pegmatoidal olivine-rich rocks occur as apophyses from the main olivine-bearing unit intruding into overlying units (Fig. 3d). The lack of deformation textures, particularly in the olivine-bearing rock, suggests intrusion of liquid into fractures rather than soft sediment-type deformation. This is consistent with the olivine-bearing rocks having been significantly molten at the time of injection.
Mineralogical, textural and compositional evidence for high volatile concentrations and partial melting in OB I
Biotite and pargasitic amphibole are considerably more common in the coarse, olivine-rich rocks than in associated rocks (Barnes & Naldrett, 1986; Boudreau, 1988). In typical cumulates of the Lower Banded zone, hydrous minerals other than trace apatite are virtually absent. In contrast, in some olivine-rich rocks biotite and amphibole may occur as large grains poikilitically enclosing olivine or other phases, consistent with growth from a volatile-enriched liquid.
The coarse textures in the olivine-rich rocks of the Troctolite subzone are consistent with growth from a hydrous liquid. Although melting would effectively reduce crystal size of original cumulus minerals (i.e. primary cumulus pyroxene and plagioclase), those secondary minerals that grew during incongruent melting (e.g. olivine) or that crystallized during eventual cooling of this hydrous melt (e.g. late pyroxene) would have grown in a wet liquid. Water and possibly the halogens can enhance the transfer of components required for growth of a crystal (Watson, 1994). If a fluid is also present, it would considerable aid transfer of components to growing crystals as well.
Plagioclase in olivine-bearing rocks may be rounded and embayed, in contrast to its typical euhedral habit in gabbronorite and anorthosite. There is also a tendency for the plagioclase to show reverse zoning in the more olivine-rich rocks, particularly where it is preserved as inclusions in larger olivine grains and protected against late reaction with crystallizing interstitial liquids (Barnes & Naldrett, 1986; Boudreau, 1988). These observations are consistent with the incongruent melting of plagioclase.
In addition to being laterally discontinuous, the olivine-bearing layers of the Troctolite subzone are texturally and mineralogically heterogeneous internally. Olivine is commonly concentrated in pegmatoidal, podiform regions surrounded by medium-grained, more leucocratic rock. This internal heterogeneity is present even where olivine-rich rock is regionally stratiform. Barnes & Naldrett (1986) proposed that olivine formed large pegmatoidal glomerocrysts and settled as large boulders. However, the abundance of hydrous minerals and eroded plagioclase in the olivine-rich pods is more consistent with heterogeneous hydration melting. Melting occurred in wet pockets on the local scale and involved mass transport between these hydrated regions and surrounding, partially molten rocks. In this respect, the heterogeneous mineralogy and texture is analogous to the mineral and textural heterogeneity that develops in migmatites. Localized melting will enhance permeability but decreases in the immediate surroundings as the heat of melting is balanced by crystallization in surrounding rocks. This will cause focused fluid flow and localized hydration melting.
Polyphase inclusions consisting of biotite, pargasitic amphibole and, locally, pyroxene and a now serpentinized mineral (presumably originally olivine) occur in chromite, apatite and possibly olivine (original relationships in olivine are commonly obscured by a later partial serpentinization that has affected primarily this mineral). These inclusions commonly define negative crystal boundaries with the host and have been interpreted to have crystallized from hydrous melt inclusions (Barnes, 1983; Boudreau et al., 1986; Boudreau, 1988). These inclusions are not observed in the eroded plagioclase noted above in these same rocks, nor in plagioclase or pyroxene elsewhere in the Lower Banded series. Both observations are consistent with the growth of olivine, chromite and apatite, and dissolution of plagioclase during a hydration event.
If the rocks of the Troctolite subzone were the result of a partial melting event caused by the influx of fluid into this zone, then the progress of this melting event should be evident in regular modal variations. In other words, mineral modes should record an increase in the fluid/rock mass ratio. In this respect, Boudreau (1988) suggested that the modal amounts of olivine could be used as a measure of the fluid/rock mass ratio, or progress of reaction, as fluid interacted with a fluid-undersaturated, partially molten mineral assemblage.
Modes from the olivine-bearing rocks of the Troctolite subzone have the following general characteristics (Barnes & Naldrett, 1986; Boudreau, 1988): (1) Biotite and amphibole are most abundant in dunites rich in olivine and are rare or absent in troctolitic rocks with lesser amounts of olivine. (2) Chromite is most abundant in rock with intermediate amounts of olivine. It is absent or rare in plagioclase-rich troctolites and in dunitic rocks with abundant biotite. (3) Plagioclase decreases in abundance, and changes from a ‘cumulus’ mineral to one with a late-crystallizing interstitial habit in the olivine-rich rocks. (4) Pyroxenes are typically interstitial or oikocrystic, invariably in reaction-relationships with earlier-formed olivine. (5) Pyrite is observed as a part of the exsolved sulfide assemblage (developed during cooling of immiscible sulfide liquids) that is hosted by gabbronorite, but not by olivine-bearing rocks.
Using the abundance of olivine as a monitor of the fluid/rock mass ratio, Boudreau (1988) argued that the partial melting event had the following characteristics. The early stage of the reaction was characterized by the formation of olivine by the incongruent dissolution of cumulus pyroxene. Cumulus plagioclase also melted incongruently to form locally rounded, embayed grains with reverse zoning. Chromite was saturated during the middle stages of the reaction, only to be melted as the reaction proceeded toward completion, as evidenced by its locally rounded and embayed habit (Barnes & Naldrett, 1986).
Eventual cooling and crystallization of the partial melt produced during the reaction and not lost from the crystal pile partially reversed the mineral changes caused by the fluxing event. This later crystallization produced pyroxene (especially orthopyroxene as reaction rims after olivine), interstitial plagioclase, and, in the wettest rocks, abundant interstitial biotite and amphibole. Sulfur, added during the early stage of reaction, was remobilized and lost in part during late degassing of the volatile-rich melts produced by the reaction and not lost from the crystal pile.
The presence of sulfide well in excess of expected cotectic proportions of 
0.1 wt % is consistent with the presence of high-temperature fluids. Calculations of Shi (1992) for Fe–Si–O–H–S fluids for the fayalite + quartz + magnetite + pyrrhotite + water f(O2)–f(S2) buffering assemblage are broadly applicable to sulfur speciation and relative gas species abundance in high-temperature fluids for mafic magmatic systems. In this system, the fugacities of H2S and SO2 both increase with temperature. At 1000°C, the fluid is composed of about 10 mol % H2S and lesser SO2, or about 20% sulfur by weight (assuming ideal gas). Degassing of even a small amount of intercumulus fluid will lead to remobilization of pre-existing sulfide. This can lead to sulfur loss from regions where fluid is exsolving and precipitation of sulfides where fluid interacts with fluid-undersaturated liquids as sulfur redissolves in the liquid (Boudreau & McCallum, 1992). The most striking evidence of this sulfide redistribution is the presence of massive sulfide that locally occurs as cores to silicate pegmatoids (Fig. 4).
|
Dunn (1986)
has shown that unaltered minerals of the Stillwater complex have retained their magmatic oxygen isotopic composition. He also suggested that modest 18O depletions observed in the olivine-bearing zones of the Banded series are consistent with the influence of an igneous late-stage fluid.
The addition of magmatic volatiles to the lower parts of OB I is also supported by the Cl/F ratios of associated apatite. In the Stillwater complex, apatite is a common interstitial phase and has been used as a monitor of relative halogen variations of intercumulus liquids (Boudreau et al., 1986; Boudreau & McCallum, 1989). Apatites from the immediate underlying rocks to OB I are among the most Cl rich observed in any intrusion, with compositions approaching endmember chlorapatite. However, changes in Cl/F ratio begin to occur within OB I, with apatite from olivine-bearing rocks of the Troctolite subzone being more Cl rich than those of anorthosite and gabbronorite of this and the overlying anorthosite subzone (Fig. 5). The halogen compositions of apatite from gabbronorite and anorthosite of OB I are otherwise similar to those from the sills and dikes that intrude the base of the complex. Because chlorine preferentially partitions into a separating fluid, the addition of this fluid to fluid-undersaturated interstitial liquids would increase the Cl/F ratio of that liquid and, by inference, of other volatiles as well. The marked change in Cl/F in OB I also implies that the fluxing agent was not simply a hydrous silicate liquid, as crystallization of anhydrous cumulates would not fractionate the halogens.
|
Because of the later serpentinization of the olivine-bearing rocks, it is not possible to unambiguously deduce the extent to which carbon is inherited from magmatic or high-temperature processes. However, based on graphite in unaltered Bushveld samples and other similarities between the Bushveld and Stillwater systems, Mathez et al
. (1989) argued that in the C–O–H–Cl system the principal species are CO2, CO, and HCl. Little water would be present if the Cl/H molar ratio exceeds unity. However, high Cl/F ratios, as are characteristic of the Stillwater and Bushveld parent magmas, are typically found in modern mafic magmas that have relatively high water concentrations (Boudreau et al., 1997). This, and the presence of hydrous minerals in the more olivine-rich rocks, implies that any fluids did have some non-trivial amount of water.
In addition, degassing of relatively insoluble CO2 would result in saturation in a gas phase earlier than would be expected if water were the primary volatile (e.g. Tait et al., 1989). Indeed, as suggested by Anderson et al. (1989), separation of a CO2-rich fluid deep in a degassing system can dehydrate overlying liquids as it rises upward. This dehydration would tend to suppress hydrous mineral crystallization from the rocks undergoing degassing beneath OB I while delivering water to OB I. This can explain why amphibole and phlogopite are a common late interstitial component in the Ultramafic series (e.g. Page & Zientek, 1987) but are then absent until they again become common in OB I.
In the following, ‘water’ is considered the principal fluxing agent, as its effects on crystallization are relatively well known. However, it is emphasized that other components of the fluid can have similar effect.
|
Fluid-Induced Melting of Cumulates |
|---|
|
TOPABSTRACTIntroductionThe Lower Banded Series...Evidence for Fluid Fluxing...Fluid-Induced Melting of...A Hydromagmatic Model for...ConclusionsReferences |
|---|
Thermal considerations
The effects of volatile migration and its effect on inducing melting in a partially molten cumulus section have been discussed by McBirney (1987)
, Boudreau (1988), Boudreau & McCallum (1989) and Nicholson & Mathez (1991). Let us consider the situation discussed by Boudreau & McCallum (1989) and Nicholson & Mathez (1991) involving a sizeable thickness of partially molten cumulates. At some depth within the crystal pile, solidification of intercumulus liquid leads to the eventual saturation and separation of a volatile-rich fluid phase. On migrating upward, this fluid will encounter liquid–solid assemblages that have not yet reached fluid saturation. The effect of this introduction of fluid will be to induce partial melting of the solid assemblage (Fig. 6a). This non-equilibrium situation may be even more extreme if fluid moves upward rapidly into rocks that are still far from fluid saturation, perhaps via the transgressive pegmatoids and plagioclase–biotite veins seen below OB I.
|
For flux melting, the heat of melting must be supplied by the rocks themselves. One can make a simple quantitative model of the actual thermal changes involved in flux melting of a thin layer. The evolution of temperature in the fluxed zone with time is governed by two main factors: (a) thermal changes involving the heat of fusion; (b) thermal diffusion between the zone and its surroundings. The heat of melting can be supplied by both the excess heat capacity present in the hydrated assemblage (i.e. the heat content difference between the assemblage’s original temperature and the lower equilibrium temperature defined by the addition of the volatiles) and by heat diffusing in from the surroundings. Flux melting with heat supplied by the melting rocks themselves will lower the temperature of the zone, whereas diffusion of heat from the surroundings will counter this. A new equilibrium temperature is thus never reached (except on a long time scale after fluxing has ended or the rock has reached fluid saturation); rather it changes as fluid is continually added and these two thermal processes compete.
If we make the simplifying assumption that the heat capacities of the liquid and solid phases are equal, these effects can be modeled by a numerical analog to the following one-dimensional heat transport–reaction equation:
 |
where t is time, x is the one (vertical) dimension, 
is the bulk thermal diffusivity, cp is the bulk heat capacity, 
Hcryst is the latent heat of crystallization, T is the temperature and 
is the rate of change of the fraction solid with time.
As a first approximation, we can assume that the addition of volatiles will result in a lowering of the solidus–liquidus relationships to such an extent that all the heat of melting can be supplied by the excess heat capacity of the rocks undergoing flux melting. Hence the change in the amount of crystalline material can be expressed as a function of the gain or loss of heat, q, with time:
 |
where 
/msoild is the rate of change in the fraction solid with the change in the mass of the solid (= 1/
, where is the density of solid), msolid/mfluid is the mass solid melted per unit mass volatiles added, and mfluid/t is the fluid influx rate into the zone.
In the numerical experiments, the temperature is assumed to be initially uniform at 1200°C throughout a pile of cumulates containing 30% intercumulus liquid. Fluid is added to a zone 1 m thick, and causes melting in a ratio of 10 g rock melted for every 1 g of fluid added. Space steps used in the simulation are 10 cm long and time step is taken as
 |
The results of three numerical models for three different fluid influx rates are shown in Fig. 6b, in which is plotted the initial thermal profile and the profile after half of the initial solid has melted. In all cases it is the influx rate of fluid that controls the rate at which melt is produced and hence is the dominant control on the thermal evolution over time.
For the case of a fast fluid influx rate of 1.0 g/cm2 per day (equivalent to 1.0 kg of fluid/m2 per day), 50% of the original solid is melted in just over 10 days. For this very high flux rate, the heat of melting is supplied largely from within the melting interval itself and thermal diffusion is not particularly effective at equalizing temperature. None the less, the influx of heat from the surroundings does cause additional melting and the maximum temperature change is a modest 60°C.
For slower infiltration rates, where the time required to melt 50% of the original solid may be months or years, the thermal effects are much more modest and most of the heat of melting can be supplied by the surrounding rocks. In short, flux melting of a thin layer over sufficiently long time scales may be considered, to a first approximation, as an isothermal process. This considerably simplifies phase diagram considerations, as discussed below.
In the numerical models, no solidification of interstitial liquid in the rocks surrounding the melting interval was considered. However, heat lost from the surroundings to the melting layer will result in crystallization of the surrounding partially molten rocks, and the heat of crystallization would be an additional thermal source to partially balance the heat of melting within the melting zone. In other words, the thermal profiles as shown in Fig. 6b would be the maximum thermal variations expected.
In a real system, the increasingly molten layer would be surrounded by increasingly solidified rocks as they supply the heat of melting. This could lead to the situation shown in Fig. 3d where a partially molten, olivine-saturated liquid intruded the overlying rigid, solidified anorthosite. Further fluid migration from the partially molten layer into overlying cumulates must first pass through this increasingly impermeable cap rock, and this would probably occur by fracture, perhaps as represented by the plagioclase–biotite veins noted previously. Mathez & Marcantonio (1995) have suggested that infiltration metasomatism in the Merensky unit proceeded by just such local fracture transport of fluid to produce observed stratigraphic variations in initial Sr isotopic characteristics. Channeled fluid migration could give rise to a succession of partial melt layers with distinct chemistry from intermediate layers.
Hydration of solid + liquid assemblages—MELTS calculations
The effect of bulk water content on mineral stability in basaltic and intermediate liquid compositions has been discussed by Ford et al. (1972), Eggler & Burnham (1973), Cawthorn (1976), and Nicholson & Mathez (1991), among others. The addition of water to a typical basaltic system has four major effects: (1) olivine is stabilized at the expense of pyroxene; (2) plagioclase stability is depressed relative to the mafic silicate minerals (e.g. Holloway & Burnham, 1972; Helz, 1976; Rutherford et al., 1985); (3) spinel stability is enhanced relative to the silicate phases (Ford et al., 1972); (4) isotherms will migrate in such a way that, at any given temperature, the amount of liquid will increase with the addition of water.
Using the thermodynamic modeling program MELTS (Ghiorso et al., 1994; Ghiorso & Sack, 1995) one can model the effect of adding water to a partially molten gabbroic rock. Shown in Table 1 are the equilibrium phase relationships as a function of water added to a bulk rock analysis of a gabbronorite from GN I of the Stillwater complex (Table 1, composition 1). With the exception of chrome spinel, the silicate phases in the dry assemblage all begin to crystallize within 15 degrees of each other, the narrow crystallization interval being expected if the bulk composition represents a binary mixture of cotectic cumulus minerals and a multiply saturated liquid. (A chrome spinel remains saturated throughout the crystallization of the dry bulk composition as MELTS does not make provisions for Cr incorporation into pyroxenes. In the natural situation, the Cr component is incorporated in the pyroxene, especially augite, once these minerals begin to crystallize. The location of the chrome spinel saturation line is otherwise a strong function of the bulk Cr content; rocks with more modal augite will crystallize chromite earlier.) With increasing bulk water content, olivine becomes increasingly stable over a larger temperature range.
|
Let us consider a partially molten dry assemblage at 1250°C represented by point A in Fig. 7. This dry assemblage is composed of the cotectic assemblage plagioclase + orthopyroxene + clinopyroxene (+ chrome spinel in the model calculation), in equilibrium with a multiply saturated liquid whose composition is listed in Table 1, composition 2. The rock at this point is 60% crystallized.
|
Addition of water to the bulk assemblage at constant temperature leads to increasing amounts of liquid at the expense of orthopyroxene followed by clinopyroxene and finally plagioclase. ‘Secondary’ olivine becomes stable after a small amount of water is added but with increasing water content it also begins to melt such that at point B only a small amount of olivine remains as the only phase on the liquidus.
On cooling of this hydrated liquid, plagioclase joins olivine at point C (the liquid composition at point C is shown in Table 1, analysis 3), followed in order by spinel, clinopyroxene and finally orthopyroxene. It should be noted that the solid assemblage at point C, even though at lower temperature than that of the dry assemblage at point A, is in equilibrium with a modestly more calcic plagioclase. In addition, the system at this point has about a 1% lower enthalpy than the assemblage at point A, reflecting added heat of melting. Finally, if no liquid is lost, then crystallization of the hydrated assemblage should lead to eventual resorption of olivine. That is, the preservation of secondary olivine requires loss of the partial melt generated during hydration.
If melting is occurring at or near the top of the crystal pile, then the loss of hydrous partial melt from the cumulus pile leads to a secondary effect: the mixing of hydrous partial melts with the drier melts resident at the base of the magma chamber. As an example, we consider the mixture of liquid at point A (a dry liquid multiply saturated in plagioclase, orthopyroxene, clinopyroxene and nominally chromite) with that of point C (a hydrous liquid saturated in olivine and plagioclase). A 50:50 mixture of these two liquid compositions is shown in Table 1, composition 4, and has the crystallization order plagioclase followed by olivine, pyroxenes and finally spinel. What this implies is that the zone of secondary olivine produced by flux melting (i.e. the rocks of the lower half of OB I) should be overlain by a liquid that had its crystallization order altered from a multiply saturated liquid producing a gabbronorite cumulate to one producing more plagioclase-rich compositions (anorthosite, troctolite or norite, depending on proportions added). These are the types of rock observed in the upper part of OB I. Further, the amount of the plagioclase-rich rocks should be a function of the amount of secondary olivine produced. That is, the thickness of the lower and upper parts of OB I should broadly complement each other.
Finally, it is noted that only about 0.5 wt % water is needed to cause almost complete melting; partial melting requires less. How much total water would be required to make the several olivine-rich units of OB I? The very olivine-rich units do not amount to more than about 10 m total stratigraphic thickness, which would have required an initial thickness of gabbronorite of perhaps 20 m. If the underlying cumulates originally contained a bulk water content of only 0.01 wt % and all of this water were lost to degassing and available for fluxing, then the thickness of cumulates beneath OB I would need to be on the order of 1 km. This thickness is readily present below the reef.
Phase diagram interpretations
A phase diagram view of hydration melting was suggested by Boudreau (1988) and is illustrated by the isothermal phase relations in the system forsterite–silica–water (Fig. 8). Let us consider an assemblage of enstatite and a liquid of composition L1 that has not yet solidified to the point where it would separate a fluid. The initial bulk composition of the assemblage will lie on a tieline connecting enstatite and the original liquid composition, shown here as point B. Addition of water to the fluid-undersaturated liquid will cause the bulk composition to change along the dotted ‘fluid-mixing line’ toward the H2O apex. Because the bulk solid composition initially remains fixed (i.e. pyroxene is the only solid present), a change in bulk composition must result in a change in liquid composition. Thus, the isothermal addition of water to bulk composition B causes the liquid composition to change (by melting of enstatite) along the isotherm on the liquidus surface, as shown by the open path. Once the liquid composition reaches the olivine boundary at L2, however, the liquid composition is held constant while the enstatite melts incongruently to olivine, the reaction again being driven by the bulk increase in water. Once all the enstatite is converted to olivine, the liquid is free to move into the olivine phase field. With continued influx of fluid, the olivine itself will melt. At the point the liquid composition reaches the dotted fluid-mixing line at L3, all the solid will have melted and the liquid can move into the liquid-only field until it reaches fluid saturation at point L4. Proportions of liquid, enstatite and olivine at any point can be determined by use of the lever rule for the changing bulk composition. At low pressures, only a few weight percent H2O is required to completely melt the original liquid + enstatite assemblage and saturate the liquid. In other words, a small amount of fluid can produce many times its weight in partial melt.
|
The extension of the isothermal reaction in the forsterite–silica–water system shown in Fig. 8 can be extended into the anorthite–forsterite–silica–H2O system, as shown schematically in Fig. 9. The surface labeled ‘isothermal surface’ is a portion of the isothermal, fluid-undersaturated surface that is defined by the migration of an isotherm as a function of bulk water concentration. This surface is equivalent to the lower, fluid-undersaturated boundary on the isothermal liquid-only field shown in Fig. 8 (i.e. the heavy dark line portion of the isotherm of Fig. 8), and generally slopes downward toward the SiO2 apex. The phase boundaries are defined by where the migrating isotherm intersects the various cotectic surfaces, which are themselves migrating as a function of bulk water concentration as noted above. The isothermal phase relations in hydrous systems are not well constrained, and may change markedly as a function of bulk composition, particularly for changes in chromium content.
|
Let us consider a liquid, L1, originally with low water concentration, that is in equilibrium with anorthite + enstatite with a bulk solid composition S1. The bulk composition, denoted here B1, lies on the dashed tieline that connects the liquid composition with the bulk solid composition. The addition of water will cause the bulk composition to migrate toward the H2O apex. To maintain equilibrium, the solid assemblage will melt during the isothermal addition of water such that the liquid will change along the open path L1 to L2 to L3. All of the original solid is melted once the liquid composition reaches the bulk composition at L3.
As the original assemblage is hydrated, the solids undergo the following progressive changes: (1) Initially, plagioclase and enstatite melt congruently, leading to an increase in the amount of melt. (2) Enstatite begins to melt incongruently once the olivine field is reached, and is the first mineral to be used up. (3) Plagioclase is present during early and middle stages of melting, persisting even when olivine becomes stable, but its abundance decreases as the reaction proceeds; its disappearance is associated with the incoming of spinel. (4) Spinel (i.e. chromite) is stable only in rocks with intermediate water contents; it eventually disappears during the end stages of melting. (5) Olivine will be the last mineral to melt. The associated liquids will not change composition but will increase in amount at peritectic points until reactant minerals are used up.
This sequence is consistent with observed modal variations in OB I noted by Boudreau (1988) and described previously. However, the sequence of residual mineral assemblages can change depending on the original bulk composition. For example, a more anorthositic protolith would shift S1 and lead to spinel as the final mineral rather than olivine.
Finally, as discussed for the MELTS model calculation, any hydrated liquids moving out of the crystal pile and mixing with the resident liquid in the chamber can produce a series of blended liquids. A ‘typical mixing line’, in which the hydrous partial melt L2 is mixed with the original liquid L1, will produce a hybrid liquid that lies in the anorthosite field. This is true for any intermediate melts between L1 and L2 that mix with L1 as well. Depending on the proportions of the endmembers, this hybrid liquid can have the crystallization sequence anorthite anorthite + olivine or anorthite anorthite + enstatite. Hence, above the hydration melting zone one might expect anorthosite, troctolite and norite to represent crystallization products of this hybrid liquid, as is observed. This mixing model is similar to those proposed by Irvine et al. (1983) and Barnes & Naldrett (1986), but in this case the liquids are not from different parental liquids but are generated from events involving the crystallization of a single magma.
|
A Hydromagmatic Model for the Formation of Olivine-Bearing Zone I |
|---|
|
TOPABSTRACTIntroductionThe Lower Banded Series...Evidence for Fluid Fluxing...Fluid-Induced Melting of...A Hydromagmatic Model for...ConclusionsReferences |
|---|
Volatile zones in a crystal pile
One can define a series of volatile enrichment and fluid evolution zones within a thick crystal–liquid pile as one moves from fully solidified cumulates upward into the magma chamber that is not yet fluid saturated. These zones are suggested by the numerical modeling of solidification and degassing in a growing and compacting crystal pile by Boudreau & Meurer (1999)
and shown in Fig. 10.
|
Solidified zone
At some depth within the crystal pile, the rocks have fully solidified. All original volatile content present in interstitial liquids has degassed, is contained in trace hydrous minerals, or remains as pore fluids. All degassed material is added to the volatile budget of the overlying cumulate–magma system.
Fluid-saturated zone
Above the solidified zone are partially molten cumulates with fluid-saturated interstitial liquids. As this zone degasses, the fluid migrates upward until it dissolves in undersaturated interstitial liquids, thereby enriching the overlying intercumulus liquids in volatile components. This continues until the overlying interstitial liquids themselves reach fluid saturation.
As the crystal pile grows and the lower portions solidify, the net effect of degassing from below is that an increasingly thicker section is enriched in volatiles and is saturated in fluid. Because sulfur is a significant component of the fluid, loss of sulfur and resorption of sulfide minerals from the fluid-saturated zone is an expected consequence of fluid separation and loss.
Fluid-undersaturated zone
Above the saturated zone the crystal + liquid assemblage is not yet fluid saturated. This zone continues into the magma chamber proper and may extend to near the top of the chamber where lower pressures may again allow the resident magma to become fluid saturated. This undersaturated zone can be subdivided into the following two additional zones.
- Fluid–liquid–solid mixing zone. As exsolved fluid from cooler underlying cumulates rises and redissolves in structurally higher, hotter and fluid-undersaturated intercumulus liquids, melting is induced. In addition, because of the low solubility of sulfur in a silicate liquid, dissolution of S-bearing fluid into the fluid-undersaturated liquid can lead to sulfide precipitation. Early in the growth of the crystal pile, the fluid-saturated zone will be relatively thin, the amount of fluid degassed is small and hence the amount of melting will be modest. However, bulk volatile contents are increased. In addition, the volatile-enriched interstitial liquid may move upward by compaction or convection.
- Liquid–liquid mixing zone. As the process of volatile enrichment of intercumulus liquids advances, compaction or interstitial liquid convection can allow these liquids to escape the crystal pile and mix with the resident magma in the uppermost part of the crystal pile or at the floor of the chamber. This will most probably occur when the fluid–liquid–solid mixing zone is near the top of the crystal pile as shown in Fig. 10. Although some proportion of solid may be present, it is the interaction of these two liquids that is the predominant control on bulk composition and the minerals that crystallize from the hybrid liquid. The effect of this is to alter the crystallization of the resident magma in the chamber.
Finally, it should be noted that if the fluid-saturation front reaches the top of the crystal pile, then volatiles will be continuously added to the main body of (fluid-undersaturated) magma. In this case, crystallization of the magma will temporarily cease and melting and resorption of the floor will result. Under such conditions, the crystal pile may degas significantly before the magma resumes crystallizing, at which point the buildup of the various volatile fronts will begin again.
Evolution of OB I
With this in mind, one can make the following model for the evolution of OB I. The original cumulates to crystallize in the Troctolite subzone of OB I continued the crystallization of GN I and were composed of plagioclase and pyroxene. Contemporaneous fluid separation and migration deeper in the crystal pile led to a progressively thicker zone of volatile-rich intercumulus liquids that eventually affected nucleation and crystal growth at the top of the crystal pile.
The initial stages of hydration affected the nucleation and growth kinetics of the magma crystallizing at the floor of the chamber, as seen in the increasing development of modally segregated pyroxene and plagioclase-rich layers in the upper part of GN I and the lower half of OB I. The increase in volatiles in the crystallizing magma affected cotectic crystallization such that pyroxene was favored over plagioclase, leading to the formation of pyroxene-rich layers. The modestly hydrated residual liquid, enriched in the plagioclase component excluded during crystallization of the pyroxene-rich portions, mixed with overlying, drier liquids and led to the formation of associated thin anorthosites within the Troctolite subzone. The pyroxene-rich layers became the protoliths that, on later partial melting, formed the olivine-rich layers.
As more fluid is exsolved from deep in the crystal pile, melting became more extensive and eventually generated olivine-saturated liquid. Although broadly stratiform on a regional scale and mainly affecting melanocratic gabbronorite layers, melting was not uniform on the local scale, producing the texturally and modally mixed ‘mafic migmatite’ on an outcrop scale. At high degrees of partial melting, olivine-saturated liquids were locally mobilized and intruded overlying cumulates.
As fluid continued to exsolve from deep in the crystal pile, melting and the fluid-saturation front became progressively closer to the top of the crystal pile. Because the mass of magma in the chamber is large relative to the mass of separating fluid, the magma in the chamber does not become fluid saturated (except perhaps near the roof where pressure is lowest). When the fluid-saturation front reached the top of the crystal pile, the fluid dissolved directly into the magma and any additional crystallization from the resident magma (or hybrid liquids) is suppressed and cumulates at the floor undergo melting. Where upward fluid migration was channeled, melting was more localized and olivine-bearing pegmatoidal potholes developed. Sulfur and PGE, soluble in the high-Cl fluid but insoluble in magma, precipitate as an immiscible sulfide liquid in the fluid–liquid–solid mixing zone, the top of the zone eventually defined by the top of the crystal pile (e.g. Boudreau & McCallum, 1992). After the fluid-saturation front reached the top of the crystal pile there is fluid loss directly to the magma. The volatiles returned to the magma in the chamber are subsequently lost to degassing at the roof and through surface eruptions.
Mixing of the increasing volumes of olivine ± plagioclase-saturated liquid that escape the crystal pile with resident magma in the chamber produced larger volumes of hybrid liquid lying within the plagioclase field. Depending on the curvature of the isothermal surface in the plagioclase-saturated surface shown in Fig. 9, the hybrid liquid may have been under- or over-saturated in plagioclase at the time of formation. However, the lack of pegmatoids in the upper part of OB I suggests that it is more likely that extensive crystallization from this hybrid liquid did not occur until much of the fluid was lost from the crystal pile and ambient temperature began to drop. A schematic representation of the development of OB I at this stage is shown in Fig. 11.
|
Eventually, fluid separation from underlying cumulates ended and the hybrid liquid produced as the partial melt mixed with the resident magma crystallized. The hybrid was generally saturated in plagioclase initially, followed by olivine or pyroxene depending on the proportions mixed. At this point, the anorthosite, troctolite and norite in the upper half of OB I crystallized. On further cooling any interstitial melt remaining in the Troctolite subzone and not expelled and mixed with overlying magma itself separates a late fluid that leads to a modest late sulfide and PGE redistribution, largely within the lower half of OB I.
Formation of the regional unconformity
It was noted previously that OB I cuts down into the underlying cumulates by as much as several hundred meters. The unconformity between the J-M reef and both the Troctolite and Anorthosite subzones is interpreted to be the result of loss of the volatile-enriched source rocks and incomplete development of OB I as a consequence of these lost volatiles. It is not considered to be the result of thermal erosion, for three reasons: (1) Thermal erosion should occur everywhere, and not particularly limited to one, albeit large, region. (2) Pyroxene-bearing rocks do not normally melt congruently, they should melt to produce olivine (see Fig. 7). As the amount of olivine is less common where the unconformity is greatest, this is inconsistent with a thermal erosion model. (3) If the unconformity actually represented a hole filled by OB I assemblages, then OB I should be thicker through the unconformity, when in fact the rocks of both the Troctolite and Anorthosite subzones actually thin substantially.
It is suggested that the lost section represents a region where cumulates were deposited on a locally steeper slope than the surrounding cumulates (Fig. 12). The section may have become oversteepened by contemporaneous deformation in this part of the magma chamber. Because of this local oversteepening, a large amount of volatile-enriched cumulates were lost to slumping. Two points support this interpretation: (1) There is abundant evidence in the underlying cumulates of incipient layer disruption and slumping within GN I (see Todd et al., 1982; Page & Moring, 1990) (Fig. 2). (2) The buildup of volatiles and the early stages of hydration melting would enhance the tendency of the crystal pile to slump, much as a heavy rain would cause a saturated hillside to fail. The slumped region could move a long way, perhaps all the way back to the mantle feeder to the entire complex. Assuming these slumped rocks are not in a now-eroded part of the complex, one may find them in future mine development (perhaps when mine development along the reef reaches beneath Billings, MT!)
|
Because of the loss of a considerable thickness of volatile-enriched cumulates in the slumped region, there will be less fluid generated to induce melting in overlying rocks. Consequently, there will be less secondary olivine produced (perhaps none, as some partial melting must occur before the onset of olivine saturation). Also, there will be less partial melt to mix with the resident melt in the chamber to produce the plagioclase-saturated liquids of the Anorthosite subzone. Consequently, the fluid–liquid–solid mixing zone and the liquid–liquid–solid mixing zones will both be thinner where they rest above N I as compared with over GN I. Because of this, both the underlying and overlying rocks will appear to form angular unconformities against the original surface of the unconformity. Finally, because there is less fluid in the underlying cumulates in the region of the unconformity, one would expect less PGE–sulfide in this region as well. All of these are consistent with the features of the regional downcutting (Dahy et al., 1995
). |
Conclusions |
|---|
|
TOPABSTRACTIntroductionThe Lower Banded Series...Evidence for Fluid Fluxing...Fluid-Induced Melting of...A Hydromagmatic Model for...ConclusionsReferences |
|---|
The model presented here for the origin of Olivine-Bearing zone I of the Stillwater complex implies that the features of this zone and the surrounding rocks can be explained by processes acting during the solidification of a single magma. As such, it is an extension of the general process of the constitutional zone refining process proposed by McBirney (1987)
. Hence one can envision a gradational series offeatures that could develop in layered intrusionsin response to volatile fluxing that range frommodest variations in mineral mode to produce modal layering to the more extensive, incongruent partial melting reactions seen in OB I of the Stillwater complex.
|
Acknowledgements |
|---|
Reviews by W. P. Meurer, J. Longhi, P. Candela, and E. Sonnenthal are very much appreciated. Comments by the participants of the Second Bostok Conference are also very much appreciated. This work was supported through NSF Grants EAR 94–17144 and EAR 97–05507.
|
FOOTNOTES |
|---|
* Fax: (919) 684-5833. e-mail: boudreau{at}eos.duke.edu
|
References |
|---|
|
TOPABSTRACTIntroductionThe Lower Banded Series...Evidence for Fluid Fluxing...Fluid-Induced Melting of...A Hydromagmatic Model for...ConclusionsReferences |
|---|
Anderson A. T., Newman S., Williams S. N., Druitt T. H., Skirius C., Stolper E. H2O, CO2, Cl and gas in plinian and ash-flow Bishop rhyolite. Geology (1989) 17:221–225.
Ballhaus C. G. Potholes of the Merensky Reef at Brakspruit Shaft, Rustenburg Platinum Mines: primary disturbances in the magmatic stratigraphy. Economic Geology (1988) 83:1140–1158.
Barnes S. J. Petrology and geochemistry of a portion of the Howland (J.-M.) Reef of the Stillwater Complex, Montana. (1983) Toronto: University of Toronto. 215. Ph.D. Dissertation. Ont.
Barnes S. J., Campbell I. H. Role of late magmatic fluids in Merensky-type platinum deposits: a discussion. Geology (1988) 16:488–491.
Barnes S. J., Naldrett A. J. Geochemistry of the J-M (Howland) reef of the Stillwater complex, Minneapolis Adit area I. Sulfide chemistry and sulfide–olivine equilibrium. Economic Geology (1985) 80:627–645.
Barnes S. J., Naldrett A. J. Geochemistry of the J-M reef of the Stillwater Complex, Minneapolis Adit area II. Silicate mineral chemistry and petrogenesis. Journal of Petrology (1986) 27:791–825.
Boudreau A. E. Investigations of the Stillwater Complex, IV. The role of volatiles in the petrogenesis of the J-M reef, Minneapolis adit section. Canadian Mineralogist (1988) 26:193–208.[Web of Science]
Boudreau A. E. Volatile fluid overpressure in layered intrusions and the formation of potholes. Australian Journal of Earth Sciences (1992) 39:277–287.[Web of Science]
Boudreau A. E., McCallum I. S. Investigations of the Stillwater Complex: Part V. Apatite as indicators of evolving fluid composition. Contributions to Mineralogy and Petrology (1989) 102:138–153.[Web of Science]
Boudreau A. E., McCallum I. S. Concentration of platinum-group elements by magmatic fluids in layered intrusions. Economic Geology (1992) 87:1830–1848.
Boudreau A. E., Meurer W. P. Chromatographic separation of the platinum-group elements, base metals, gold and sulfur during degassing of a compacting and solidifying igneous crystal pile. In: Contributions to Mineralogy and Petrology (1999) in press.
Boudreau A. E., Mathez E. A., McCallum I. S. Halogen geochemistry of the Stillwater and Bushveld Complexes: evidence for transport of the platinum-group elements by Cl-rich fluids. Journal of Petrology (1986) 27:967–986.
Boudreau A. E., Stewart M. A., Spivack A. J. Stable Cl isotopes and origin of high Cl magmas of the Stillwater Complex, Montana. Geology (1997) 25:791–794.
Braun K., Meurer W., Boudreau A. E., McCallum I. S. Compositions of pegmatoids beneath the J-M Reef of the Stillwater Complex, Montana. Chemical Geology (1993) 113:245–257.[Web of Science]
Buntin T. J., Grandstaff D. E., Ulmer G. C., Gold D. P. A pilot study of geochemical and redox relationships between potholes and adjacent normal Merensky Reef of the Bushveld Complex. Economic Geology (1985) 80:975–987.
Cawthorn R. G. Melting relations in part of the system CaO–MgO–Al2O3–SiO2–Na2O–H2O under 5 kb pressure. Journal of Petrology (1976) 17:44–72.
Dahy J. P., Corson S. R., Geraghty E. P., Koski M. S., Langston R. B., Leroy L. W. Stratigraphic relationships of the J-M Reef package within the troctolite–anorthosite zone I, Stillwater Complex, Stillwater Mine, Montana. Geological Society of America, Abstracts with Program (1995) 27:7–8.
Dunn T. An investigation of the oxygen isotope geochemistry of the Stillwater Complex. Journal of Petrology (1986) 27:987–997.
Eggler D. H., Burnham C. W. Crystallization and fractionation trends in the system andesite–H2O–CO2–O2 at pressures to 10 kilobars. Geological Society of America Bulletin (1973) 84:2517–2532.
Ford C. E., Biggar G. M., Humphries D. J., Wilson G., Dixon D., O'Hara M. J. Role of water in the evolution of the lunar crust; an experimental study of sample 14310; an indication of lunar calc-alkaline volcanism. Proceedings of the Third Lunar Science Conference. Geochimica et Cosmochimica Acta, Suppl. 3 (1972) 1:207–229.
Ghiorso M. S., Sack R. O. Chemical mass transfer in magmatic processes. IV. A revised and internally consistent thermodynamic model for the interpretation and extrapolation of liquid–solid equilibria in magmatic systems at elevated temperatures and pressures. Contributions to Mineralogy and Petrology (1995) 119:197–212.[Web of Science]
Ghiorso M. S., Hirschmann M., Sack R. O. MELTS: software for thermodynamic modeling of magmatic systems. EOS Transactions, American Geophysical Union (1994) 75:571.
Helz R. T. Phase relations of basalts in their melting ranges at PH2O = 5 kbar, part II. Melt compositions. Journal of Petrology (1976) 17:139–193.
Holloway J. R., Burnham C. W. Melting relations of basalt with equilibrium water pressure less than total pressure. Journal of Petrology (1972) 13:1–29.
Irvine T. N., Keith D. W., Todd S. G. The J-M platinum–palladium reef of the Stillwater Complex, Montana; origin by double diffusive convective magma mixing and implications for the Bushveld Complex. Economic Geology (1983) 78:1287–1334.
Kushiro I. On the nature of silicate melt and its significance in magma genesis: regularities in the shift of the liquidus boundaries involving olivine, pyroxene, and silica minerals. American Journal of Science (1975) 275:411–431.
Kushiro I., Yoder H. S., Nishikawa M. Effect of water on the melting of enstatite. Geological Society of America Bulletin (1968) 79:1685–1692.
Mathez E. A., Marcantonio F. Sr isotopes and magma mixing in the Bushveld complex. EOS Transactions of the American Geophysical Union (1995) 76:F641. Abstract.
Mathez E. A., Dietrich V. J., Holloway J. R., Boudreau A. E. Carbon distribution in the Stillwater complex and evolution of vapor during crystallization of Stillwater and Bushveld magmas. Journal of Petrology (1989) 30:153–173.
McBirney A. R. Constitutional zone refining of layered intrusions. In: Origins of Igneous Layering—Parsons I., ed. (1987) Boston: MA: D. Reidel. 437–452.
McCallum I. S., Raedeke L. D., Mathez E. A. Investigations of the Stillwater Complex. Part I. Stratigraphy and structure of the Banded zone. American Journal of Science (1980) 280A:59–87.[Web of Science]
Nicholson D. M., Mathez E. A. Petrogenesis of the Merensky Reef in the Rustenburg section of the Bushveld Complex. Contributions to Mineralogy and Petrology (1991) 107:293–309.[Web of Science]
Page N. J., Moring B. C. Petrology of the noritic and gabbronoritic rocks below the J-M reef in the Mountain View area, Stillwater Complex, Montana. US Geological Survey Bulletin (1990) 1674-C:C1–C47.
Page N. J., Zientek M. L. Composition of primary postcumulus amphibole and phlogopite within an olivine cumulate in the Stillwater Complex, Montana. US Geological Survey Bulletin (1987) 1674-A:A1–A35.
Raedeke L. D., McCallum I. S. Investigations of the Stillwater Complex: Part II, Petrology and petrogenesis of the Ultramafic series. Journal of Petrology (1984) 25:395–420.
Raedeke L. D., Vian R. W. A three dimensional view of mineralization in the Stillwater Complex, Montana. Economic Geology (1986) 81:1187–1195.
Rutherford M. J., Sigurdsson H., Carey S., Davis A. The May 18, 1980, eruption of Mt. St. Helens, I. Melt composition and experimental phase equilibria. Journal of Geophysical Research (1985) 90:2929–2947.
Shi P. Fluid fugacities and phase equilibria in the Fe–Si–O–H–S system. American Mineralogist (1992) 77:1050–1066.[Abstract]
Tait S., Jaupart C., Vergniolle S. Pressure, gas content and eruption periodicity of a shallow, crystallising magma chamber. Earth and Planetary Science Letters (1989) 92:107–123.[Web of Science]
Todd S. G., Keith D. W., LeRoy L. W., Schissel D. J., Mann E. L., Irvine T. N. The J-M platinum–palladium reef of the Stillwater Complex, Montana: I. Stratigraphy and petrology. Economic Geology (1982) 77:1454–1480.
Turner A. R., Wolfgram D., Barnes S. J. Geology of the Stillwater county sector of the J-M Reef, including the Minneapolis adit. In: The Stillwater Complex, Montana: Geology and Guide. Montana Bureau of Mines and Geology Special Publication—Czamanske G. K., Zientek M. L., eds. (1985) 92:210–231.
Watson E. B. Diffusion in volatile-bearing magmas. In: Volatiles in Magmas. Mineralogical Society of America, Reviews in Mineralogy—Carroll M. R., Holloway J. R., eds. (1994) 30:371–412.
Wooden J. L., Czamanske G. K., Zientek M. L. A lead isotopic study of the Stillwater complex, Montana: constraints on crustal contamination and source regions. Contributions to Mineralogy and Petrology (1991) 107:80–93.[Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. J. Naldrett, A. Wilson, J. Kinnaird, and G. Chunnett PGE Tenor and Metal Ratios within and below the Merensky Reef, Bushveld Complex: Implications for its Genesis J. Petrology, April 20, 2009; (2009) egp015v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. J. Spandler, S. M. Eggins, R. J. Arculus, and J. A. Mavrogenes Using melt inclusions to determine parent-magma compositions of layered intrusions: Application to the Greenhills Complex (New Zealand), a platinum group minerals-bearing, island-arc intrusion Geology, November 1, 2000; 28(11): 991 - 994. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. E. Brandriss and D. K. Bird Effects of H2O on Phase Relations during Crystallization of Gabbros in the Kap Edvard Holm Complex, East Greenland J. Petrology, June 1, 1999; 40(6): 1037 - 1064. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||