| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology Volume 42 Number 10 Pages 1845-1867 2001
© Oxford University Press 2001
Compositional and Lithological Controls on the PGE-Bearing Sulphide Zones in the Selukwe Subchamber, Great Dyke: a Combined Equilibrium–Rayleigh Fractionation Model
SCHOOL OF GEOLOGICAL AND COMPUTER SCIENCES, UNIVERSITY OF NATAL, DURBAN, 4041, SOUTH AFRICA
Received November 30, 1999; Revised typescript accepted March 14, 2001
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONThe platinum group element (PGE)-bearing Main and Lower Sulphide Zones of the Selukwe Subchamber of the Great Dyke are made up of a series of subzones within which the ratios Pd:Pt are effectively constant, whereas these ratios vary significantly between the subzones. Fractionation of Pd with respect to Pt varies by a factor of 10 and cannot be modelled using a sulphide collector phase and constant partition coefficients. The link with sulphide is indisputable and the control is likely to have been the degree of oversaturation of PGE micro-nuggets in the magma. The apparent partition coefficients for Pt and Pd between silicate and sulphide liquid are dependent on the degree of oversaturation and thereby exhibit spurious correlation with the PGE content of the sulphide. Modelling replicates the Pt and Pd distribution and ratios only by dramatically changing the effective partition coefficients. Pyroxene compositions (including TiO2) are shown to be strongly dependent on the incompatible element content of the whole rock, and specific linear arrays relating these variables can be related to the PGE subzones. The overall control is Rayleigh fractionation, but constancy of the ratio Pd:Pt and the initial pyroxene composition (before re-equilibration with trapped liquid) within the subzones is indicative of equilibrium crystallization. This layered structure may have been derived from liquid layers in the magma chamber.
KEY WORDS: platinum group elements; sulphide; Great Dyke; pyroxene compositions
|
INTRODUCTION |
|---|
An increasing number of mafic-layered intrusions are reported to have distributions of platinum group elements (PGE), Au and base metals characterized by vertically offset profiles. Offset profiles involve the vertical separation of metal peaks and were first reported in the Great Dyke (Prendergast & Keays, 1989
; Wilson et al., 1989; Naldrett & Wilson, 1990; Wilson & Tredoux, 1990). The phenomenon has since been recognized in the Munni Munni intrusion of Western Australia (Barnes et al., 1990), the Skaergaard intrusion (Bird et al., 1991) and the Rincón del Tigre Complex of eastern Bolivia (Prendergast, 2000). PGE are also postulated to be transported in fluids emanating from expulsion of intercumulus liquid from the crystal pile, and there is strong evidence to support such processes in several gabbroic sections of layered intrusions such as the Skaergaard intrusion (Andersen et al., 1998) and in the Stillwater Complex (Boudreau & McCullum, 1986, 1992). In contrast, the mineralized zones in the Great Dyke and Munni Munni intrusion are located in pyroxenite, and primary magmatic processes have been invoked to explain the mineralization in these intrusions (Prendergast & Wilson, 1989; Barnes et al., 1990; Naldrett et al., 1990). A fundamental question is whether processes postulated for a particular magmatic environment could be applied to other dissimilar environments. A thorough understanding of the mechanisms involved is hampered, in many cases, by a lack of detailed information on the silicate framework in the mineralized zones. This paper attempts to link compositional variations in silicates and whole rocks to distinct patterns of PGE mineralization in the Main and Lower Sulphide Zones in the Selukwe Subchamber of the Great Dyke, and thereby to establish if the same processes operated in these two mineralized zones of the intrusion.
The nature of the offset profiles in the Great Dyke was initially represented as contiguous variations of metal values giving systematic variations through the sequence (Prendergast, 1988; Prendergast & Keays, 1989; Wilson & Tredoux, 1990). Proposed models were largely based on fractional segregation of sulphide with strong partitioning of the PGE into sulphide (Naldrett & Wilson, 1990). Precise replication of the trends by numerical modelling using established partition coefficients for PGE between sulphide and silicate melts proved elusive (Barnes, 1993), indicating that other mechanisms were wholly or partly responsible for the observed profiles (Peach & Mathez, 1996). Transport of PGE by fluid migration, largely considered to result from compaction of underlying cumulates, has been proposed in the form of a chromatographic model (Boudreau & Meurer, 1999). The association of sulphides with PGE enrichment has been recognized in most mineralized environments in layered intrusions, and therefore the role of late-stage fluids is of major significance because of their potential interaction with sulphide. Late-stage fluids would also affect silicate minerals either by their infiltration and expulsion, or by reaction when contained.
There are two main mineralized zones in the Great Dyke—the Main Sulphide Zone and Lower Sulphide Zone—and although an internal structure has been previously recognized (Prendergast & Keays, 1989; Wilson & Tredoux, 1990) no systematic dependence has been demonstrated relating these to the silicate host rocks. In addition, these two mineralized layers were seen as separate entities with no common features relating to their structure. This paper establishes common aspects between the two mineralized zones and shows a strong relationship to the silicate framework.
|
GENERAL ASPECTS OF THE GREAT DYKE |
|---|
The Great Dyke is a linear intrusion of mafic and ultramafic rocks that cuts across the Zimbabwe Craton (Fig. 1). The Craton comprises dominantly Archaean rocks and is bounded by the Zambezi metamorphic belt in the north, the Mozambique belt to the east and the Limpopo belt to the south. The Great Dyke, aligned approximately NNE, is 550 km in length and 4–11 km wide (Worst, 1960
). Gabbro and quartz gabbro satellite dykes are located parallel to the intrusion. These satellite dykes are closely associated with a major fracture pattern that is postulated to be the result of a craton-wide tectonic control that gave rise to the form and alignment of the major Great Dyke intrusion (Wilson, 1996).
|
Previous age determinations of the Great Dyke using the Rb–Sr method yielded ages of the order of 2·4 Ga (Allsopp, 1965; Davies et al., 1970) with the most precise being 2455 ± 16 Ma (Hamilton, 1977). Recent studies (Mukasa et al., 1998; Armstrong & Wilson, 2000) indicate a significantly older age of emplacement of 2579 ± 7 Ma.
The Great Dyke is longitudinally subdivided into a series of narrow contiguous magma chambers and subchambers (Fig. 1). Two main magma chambers (the North and South Chambers) have been recognized, with further subdivision into a series of subchambers on the basis of structure, style of layering and continuity of layers (Podmore & Wilson, 1987; Prendergast, 1987; Wilson & Prendergast, 1989). The North Chamber is subdivided into the Musengezi, Darwendale and Sebakwe Subchambers. The South Chamber comprises the northern Selukwe Subchamber, the subject of the current study, and the southern Wedza Subchamber. Similar to the other subchambers, the Selukwe Subchamber has extensive development of PGE resources in the Main Sulphide Zone (MSZ) and Lower Sulphide Zone (LSZ).
|
THE SELUKWE SUBCHAMBER |
|---|
Structure and location of the studied section
The Selukwe Subchamber comprises an upper Mafic Sequence (280 m thick) overlying the Ultramafic Sequence, which has an exposed thickness of 1600 m. In transverse section, the Subchamber is synclinal in shape with essentially the same lithological sequence being exposed on both sides of the longitudinal axis. Asymmetry in the layering pattern close to the walls is attributed to the physical shape of the chamber and the contrasting nature of the wall rocks, which are greenstones on the west flank and granite on the east side (Wilson et al., 2000
). The layering has a shallow plunge from north to south in the northern segment and vice versa for the southern segment, giving an overall boat-like shape to the layering configuration. Repetition of rock units of the Mafic Sequence on the longitudinal axis arises from a number of major transverse, steeply dipping faults that interrupt the overall continuity of the layering. This has important implications for the exposed strike length of the economic MSZ (Fig. 2), which is located entirely within pyroxenite beneath the boundary of the Ultramafic–Mafic Sequences, and consists of a zone of base metal sulphides, the lower part of which is enriched in PGE.
|
The current study focuses on two drill core intersections (MR87 and MR92) of the MSZ in an area east of the axis of the Subchamber in close proximity to the Unki prospect shaft (Fig. 1). Both drill cores intersected the MSZ, but only MR92 intersected the LSZ (Fig. 2). This is the first intersection of the LSZ in the Selukwe Subchamber and so permits the most detailed investigation to date on the PGE distribution in the LSZ of the Great Dyke.
Stratigraphy and rock types
The Mafic Sequence is exposed in the central region of the Selukwe Subchamber and is underlain by a series of cyclic units of the Ultramafic Sequence (Wilson et al., 2000). The uppermost lithology of the Ultramafic Sequence of Cyclic Unit 1 (Fig. 2) is the P1 Pyroxenite layer, comprising websterite in the upper 4–8 m, which in turn overlies an orthopyroxenite unit 220 m in thickness. Cyclic units below Cyclic Unit 1 are dominated by dunite and granular harzburgite grading into olivine pyroxenite. The pyroxenites have 1–15% interstitial plagioclase, together with other late-stage minerals, and range from adcumulates to orthocumulates (Irvine, 1987), as compared with dominantly adcumulate pyroxenites in the North Chamber. Except for a narrow zone that separates the Websterite Layer from the gabbroic rocks of the Mafic Sequence, pegmatoids are almost completely absent from the succession.
Textural relations in the P1 Pyroxenite and their interpretation
The principal development of sulphide in the Selukwe Subchamber is the MSZ located mainly within orthopyroxenite, close to the websterite–orthopyroxenite boundary (Fig. 2). The LSZ is located in the same orthopyroxenite layer, some 35 m below the MSZ and with no obvious petrological break between the two sulphide-bearing zones. The pyroxenite consists of medium-grained, prismatic to tabular cumulus orthopyroxene crystals enclosed within a matrix of postcumulus minerals comprising plagioclase and clinopyroxene, together with pockets of late-stage minerals such as magnetite, phlogopite, K-feldspar, quartz, apatite, rare rutile and zircon. The clinopyroxene and plagioclase occur mainly as large oikocrysts, the former up to 20 mm in length, and the plagioclase as zoned crystals up to 10 cm in diameter. Preferential weathering on the margins of the plagioclase oikocrysts gives rise to the characteristic nodular texture commonly associated with exposures of the MSZ in the field.
Sulphide is interstitial to the cumulus pyroxene, and its distribution is closely associated with the late-stage minerals, giving rise to local development of a net texture where it is most abundant. The sulphide also has a tendency to concentrate around the margins of the plagioclase oikocrysts, further accentuating the weathering feature of the nodules (Wilson, 1992). These textures indicate the close association of trapped silicate liquid with early-formed sulphide liquid.
A detailed assessment of the textures of the pyroxenite is important in understanding the distribution of trace elements and is illustrated by a study on two samples located within the MSZ and LSZ (Fig. 2). The textural forms are largely controlled by the fabric of the orthopyroxene and crystal associations, summarized as follows:
- Grains with preserved crystal faces. These make up a high proportion of the primocryst orthopyroxene assemblage, and are most commonly (but not totally) in contact with interstitial plagioclase. In some cases (in the sections observed), all faces of individual crystals are preserved and in other cases several faces are preserved and others are obscured. In general, prism faces [110] show the strongest retention of facial development, whereas domal faces [101] have a greater tendency to be obscured by mutual interaction. Identification of the status of crystal boundaries as facial, or planar but non-facial, is readily determined by universal stage measurements. In this study approximately 400 such crystal boundaries were measured and related to the textural associations discussed below.
- Grains of orthopyroxene with mutual planar boundaries. Although few crystal faces are preserved it is not unusual for two crystals to be joined along faces. Commonly six to 20 grains are in close contact with little or no interstitial feldspar. Generally, linear boundaries of the crystals meet at triple points and there is a relatively constant grain size within individual clusters.
- Grains of orthopyroxene enclosed within clinopyroxene oikocrysts. These show resorption textures with marked rounding and grain size reduction. Similar textures have been reported for the Stillwater Complex (Jackson, 1961), and are the result of down-temperature stability relations between ortho- and clinopyroxene.
Textural types (1)–(3) are consistent with variable degrees of densification occurring on a local scale by which porosity has been reduced by a combination of processes, including re-equilibration, overgrowth, enlargement of crystals, enclosure by late-stage minerals and compaction (Hunter, 1996). The preservation of crystal faces of orthopyroxene grains bounded by plagioclase indicates that cementation of cumulus orthopyroxene took place at an early stage in domains within the cumulate pile. In adjacent domains, annealing of grains caused mutual interaction of orthopyroxene, with recrystallization overcoming face boundary locations. Triple junctions at 120° indicate that textural equilibrium has been achieved in these zones (Hunter, 1987, 1996). There are no instances of such mutually interacting clusters of orthopyroxene crystals being enclosed by a single grain of optically continuous interstitial plagioclase. This is consistent with the description of oikocrysts for this rock unit (Wilson, 1992).
The range of textural types and domains within individual samples is illustrated in Fig. 3a and b. Annealed orthopyroxenes have few crystal faces, and are dominated by mutually interacting crystal boundaries. Framework orthopyroxene crystals are defined here as largely isolated crystals commonly having several preserved crystal faces and tending to be variable in grain size. They also usually touch other crystals at point contact or along one or two crystal faces. The stratigraphically lower sample (Fig. 3a), located within the LSZ, has a higher degree of annealed zones (and less interstitial plagioclase) compared with that from the stratigraphically higher position within the MSZ (Fig. 3b) (insets to these figures show textural type-zones). On a thin-section scale, both samples have locally developed strongly annealed domains, which are not specifically located along layering planes. Therefore, although grain boundary readjustment has taken place in the annealed domains, it is not uniformly developed. The framework domains are considered to represent the earliest stage by which the crystal network was frozen in place by the oikocrystic and interstitial plagioclase, and which would have played an important or even a critical role in the entrapment of interstitial liquid. The P1 Pyroxenite exhibits a range in textures, which reflect various stages in the sequence of events that led to complete solidification of the cumulate. Original packing of crystals in the immediate subliquidus stage could not have been greater than 30–50% (Wager & Brown, 1968), although flow or slumping of the crystal mush could have produced a higher packing efficiency (Higgins, 1991). Magmatic shearing (Benn & Allard, 1989) may have caused rotation and interaction of grains, and flow partitioning could have reduced the melt fraction to <20% (Nicolas, 1992).
|
Individual domains within the two examples have areas where pyroxene crystals are annealed to attain textural equilibrium and although planar boundaries are observed, few of these represent crystal faces. This is in contrast to adjacent areas where groups of crystals possess several faces unhindered by growth. Other sides of the crystals are rounded by dissolution or by interaction with other crystals. Sulphide tends to be mainly concentrated on pyroxene–pyroxene boundaries and to a lesser extent on pyroxene–plagioclase boundaries but is seldom enclosed entirely within plagioclase. Sulphide was confined to melt-rich zones by the early crystallization of the plagioclase cement, which rapidly formed zoned oikocrysts (Wilson, 1992
). Local alteration of pyroxene is developed in close association with small pockets of late-stage minerals (quartz, K-feldspar, phlogopite and primary amphibole), and this also tends to be most prevalent in zones bounded by framework-textured orthopyroxene crystals.
These textures indicate that parts of the rock formed a rigid framework with cementation arising from the development of plagioclase oikocrysts. Hunter (1996) pointed out that such poikilitic cementation would restrict growth of cumulus grains (hence preserving their original crystal form) and would also restrict compaction. At the same time, mutually bounding pyroxene crystals may undergo grain boundary readjustment and annealing. The proportions of cumulus orthopyroxene to oikocrystic and late-stage phases indicates that densification must have occurred with expulsion of liquid at an early stage. Melt (together with liquid sulphide if present) is therefore likely to have been concentrated in localized zones where orthopyroxene formed a rigid framework and ultimately this would have formed a closed environment. Sulphide tends to concentrate at the boundary of the annealed and framework zones in both samples (Fig. 3a and b). These observations imply that consolidation of the crystal mush took place before the cementation stage, dependent on the temperature interval between the liquidus and solidification temperatures, defined as 
Tcem (Hunter, 1996). In the Great Dyke P1 Pyroxenite, Tcem is indicated to be relatively small and therefore solidification took place soon after the compaction stage. The combined, and generally contrasting effects of early removal of melt by compaction and annealing of primocrysts, and its retention within the crystal framework, would profoundly affect the major and trace element geochemistry of the rock.
|
VERTICAL DISTRIBUTION OF PGE AND BASE METALS IN THE MAIN SULPHIDE ZONE |
|---|
General characteristics of the MSZ
Vertical metal distributions in the MSZ are reputed to be remarkably consistent throughout the Great Dyke, although variations are observed in the magnitude and vertical intervals over which they occur (Prendergast & Wilson, 1989
; Wilson et al., 1989). The variations are known to exist on both regional and small scales (Wilson & Tredoux, 1990; Wilson, 1996). The regional variation is one of mainly systematic changes of peak values for the metals and the widths of the mineralized zones in relation to the position between the axis and margin of the Great Dyke. No link has previously been established between the metal distributions and silicate mineral compositions except for an increase in iron content in the pyroxenes in the region of the mineralized zone (Prendergast & Keays, 1989; Evans & Buchanan, 1991; Wilson, 1992). The general pattern of metal and sulphide distribution in the MSZ gives rise to well-established subdivisions (Fig. 4). The MSZ encompasses an upper sulphide-rich Base Metal Subzone (BMsz) and a lower Platinum Group Element Subzone (PGEsz) generally 1·5–2·5 m thick. The base of the PGEsz is coincident with the visual appearance of sulphide in small amounts (0·3–0·5% by volume). The unit stratigraphically overlying the PGEsz is called the Hanging Wall section. The Footwall section is the interval between the PGEsz and the top of the LSZ. The vertical patterns for Pt and Pd do not show identical form; Pd is concentrated at the base of the PGEsz and Pt towards the top. Concentrations of Pt and Pd in the PGEsz range from 500 to 5000 ppb.
|
Distribution of base metals, S, Pt and Pd
Distribution of base metals and PGE is represented by two drill sections (MR87 and MR92), which were studied in detail using continuous 15 cm samples (representative analyses are given in Table 1 and the complete dataset is available at Website http://www.duck.cs.und.ac.za/geology/greatdyke). As Cu, Ni and S contents start to rise, so the concentrations of PGE rise to appreciable amounts (ppm levels) (shown for MR87 in Fig. 5). (Analyses for Pt and Pd were carried out by fire assay using Ni sulphide collection and inductively coupled plasma mass spectrometry using an Elan6000 instrument. Blanks were pure quartz, and the international standard SARM-7 returned values of 3640 ppb Pd and 1500 ppb Pt. Five samples from the Footwall Zone at sub-ppb level were analysed by radiochemical neutron activation techniques at the Department of Earth Sciences, University of Melbourne.) In many sections, the BMsz commences one sample position (15–20 cm) below the Pt peak position. The peak positions of S, Cu (Fig. 5a) and Ni occur one or two sample positions (15–30 cm) above the Pt peak, where the sulphide content is 4·5–6%. The amount of sulphide decreases upwards in the section by way of several minor peaks, but at overall elevated levels compared with that in the PGEsz.
|
|
The PGEsz is 1·6–2·2 m in width in this region where Pt and Pd (as well as the other PGE and Au) rise to markedly high values. The patterns of Pt and Pd (Fig. 5b) are different, with Pt increasing in concentration upwards in the profile to its peak position near the top of the PGEsz whereas Pd tends to decrease in concentration upwards in the succession through a series of minor peaks. The position of the highest Pd peak is 60–90 cm below the highest Pt peak, with a clear crossover point of the metal profiles. The general form of the peaks is a gradual rise on the lower limbs with a sharp fall-off on the upper limbs to give a pattern of asymmetric peaks. The displacement of the Pd peaks to slightly below those of Pt, as well as the PGE to the base metal peaks in general gives the so-called offset profiles that characterize the MSZ.
Ratios of Pd to Pt and PGE to base metals in the PGE Subzone
The ratio Pd/(Pd + Pt) emphasizes the contrasting profile patterns of Pt and Pd in the PGEsz and the Footwall section (Fig. 5c). Values for this ratio are relatively low (0·5) in the Footwall to the PGEsz, below the Pd peak. Upwards in the section in the PGEsz, values for the ratio rise to a maximum close to the Pd peak and then decrease, with lowest values coinciding with the Pt peak. Ratios of Pt/Cu and Pd/Cu (Fig. 5c) have relatively constant values below the Pd peak position, rise sharply immediately above the Pd peak and then decrease again at a position below the Pt peak but to a much lower level than in the Footwall. A clear crossover point exists for the Pd/Cu and Pt/Cu curves.
The variation of the ratio Pd/(Pd + Pt) indicates an apparently progressive change through the section. However, the combined data for MR87 and MR92 show clear groups for this ratio in the PGEsz (Fig. 6a). The pattern is the same in both intersections, with the stratigraphically lowest zone in each group (CPSZ zone) having the highest values for Pd/(Pd + Pt) and the stratigraphically highest zone (APSZ zone) the lowest values. The intermediate BPSZ zone is a narrow zone 
35 cm wide, which corresponds to the crossover region of the Pd and Pt profiles and is therefore intermediate to the wider APSZ and CPSZ zones.
|
Therefore, superimposed on the apparently continuous trends of Pd and Pt variation in the vertical profiles, there exists a certain consistency of the ratios within these zones. Data from 16 drill hole intersections (Fig. 6b) in the vicinity of the Unki shaft exhibit a high degree of correlation for the APSZ zone (Pt-rich) and CPSZ zone (Pd-rich). Using the larger dataset, the narrow intermediate BPSZ zone is shown to comprise two distinct subsets, which relate to one sample position above and generally two sample positions below the crossover point for the Pt and Pd profiles, and is effectively a transitional zone between the dominant zones. The percentage population for the various ratio groups is shown in Fig. 6b.
The results of this analysis show that: (1) irrespective of the variable absolute values for Pt and Pd in different drill core sections, the Pd/Pt ratio is remarkably constant over certain intervals; (2) for any one section, Pd/(Pd + Pt) values lie between well-defined limits for each zone, with boundary samples ‘linking’ the dominant zones to give the apparent continuous variation in metal values observed in the vertical profiles.
Footwall and Hanging Wall sections to the PGE Subzone
The 30–35 m thick section of poorly mineralized pyroxenite that lies between the PGEsz and the LSZ is called the Footwall to the PGEsz (Fig. 4). It is a homogeneous orthopyroxenite containing very finely disseminated sulphide of essentially constant composition (100 ppm Cu and 0·3% sulphide). Apart from the basal 10 m where Pt and Pd fall close to or below detection limits (<4 ppb) these elements are present in the Footwall to the PGEsz in the concentration range 5–150 ppb.
Two zones of essentially constant Pd/(Pd + Pt) are identified in the Footwall to the PGEsz (Fig. 7). Higher values for the ratio Pd/(Pd + Pt) occur in the stratigraphically lowest zone (BFWZ) and lower values in the overlying AFWZ zone. The Hanging Wall section to the PGEsz (Fig. 7) also shows a remarkably constant ratio of Pd/(Pd + Pt), with a value slightly higher than that of the underlying APSZ zone.
|
|
THE LOWER SULPHIDE ZONE |
|---|
The LSZ (Fig. 8) is a sequence of continuous mineralization which in this area is >85 m thick. The sulphide content does not exceed 0·9 wt % and on average is 0·3 wt %. The PGE are associated with sulphide but the distribution is highly variable and falls to detection limits at the base of the Footwall to the PGEsz. The PGE distribution is cyclical on a scale of 10–15 m, giving a series of rhythmic units. As will be seen in the following sections, the LSZ can be broadly subdivided into a Lower Section (83–142 m below the mafic contact) and Upper Section (47–83 m) on the basis of a sharp, geochemically defined break. In this sense there is a direct parallel to the overlying and more strongly mineralized PGEsz and the Footwall to the PGEsz.
|
PGE and base metal distributions
A series of well-defined units is superimposed on general trends of metal variations in the LSZ. Sharp discontinuities are observed within the Lower Section. Although the patterns are broadly similar for Pt and Pd (Fig. 8a and b), in detail they are different. Pt has a tendency to increase upwards in the Lower Section whereas Pd decreases by way of well-defined steps. In the Upper Section, the fall-off of Pd is at a greater rate than for Pt. These characteristics are similar to those observed in the MSZ, but on a much larger scale, in that a series of peaks is present with gradually rising lower limbs and sharp fall-off on the upper limbs giving an asymmetric saw-tooth pattern. The Upper Section to the LSZ differs from the Lower Section by a smoother variation, with values for Pt and Pd falling to detection limits at the base of the Footwall of the PGEsz.
Variation of Cu (Fig. 8c) shows very low values (<10 ppm) in the Lower Section of the LSZ, rising sharply at the top of the zone before decreasing at the base of the Upper Section, where it remains at levels of 200–350 ppm through the zone. The concentration of Cu falls to low values in the Footwall to the PGEsz but remains essentially constant over 30 m in this section, rising sharply at the base of the PGEsz.
Variations in metal ratios
Ratios of Pt to Pd through the LSZ (Fig. 9) suggest apparently variable values, but in reality most of the samples in discrete vertical intervals exhibit ratios for these elements within clear limits. Samples located at the boundaries to these intervals have intermediate values. In both the Upper and Lower Sections of the LSZ the stratigraphically lowest zones have the highest Pd:Pt values, which then decrease systematically upwards in each section. The relative values of the ratios are extreme, with Pd/(Pd + Pt) values ranging from 0·06 to 0·25 in the Upper Section and from 0·07 to 0·77 in the Lower Section but the same pattern of variation occurring in both sections. This pattern shows a striking similarity with the PGEsz and Footwall to the PGEsz. The sequence of steps is particularly well defined in the Lower Section. Therefore, although the overall stratigraphic distribution of the elements indicates a continuous variation (Fig. 8), in detail the dominant characteristic is one of ratios lying within clearly defined limits on a step-wise basis. Four zones are recognized in the Upper Section (labelled ALZU–DLZU in Fig. 10a and b) and six zones in the Lower Section (labelled ALZL–FLZL in Fig. 10a and b). The BLZL zone should possibly be regarded as the intermediate zone between ALZL and CLZL. The pattern also shows a strong dependence on the variation of sulphide, with a pronounced break between the Upper and Lower Sections of the LSZ (Fig. 10a).
|
|
The pattern of relative variation of Pd and Pt through the LSZ is therefore similar to that of the MSZ in that Pd:Pt is highest in the lowest units and decreases systematically by way of a series of steps upwards in the sequence. The apparently continuous variation over the profile is the result of intermediate samples linking groups of samples with similar Pd:Pt ratios. This may indicate that (1) some migration of metals has taken place across the boundaries of the well-defined zones and must have occurred after the general pattern was well established, or (2) the primary processes that produced the change may not have been instantaneous, but gradual, and the intermediate samples reflect the rate of change.
Interdependence of S and base metals
The variation of S in the LSZ (Fig. 10a) indicates distinct patterns with very low values (<0·02% S) in the Lower Section rising to 0·2% immediately below the base of the Upper Section. In the Upper Section S rises gradually and then remains effectively constant. The ratio of Pd to Cu (expressed as 1000 x Pd/Cu) in Fig. 10b highlights significant differences in both overall distribution pattern and trends for the Upper and Lower Section of the LSZ. A major discontinuity is also observed between the Footwall to the PGEsz and the Upper Section of the LSZ.
The concentration of Cu is strongly dependent on S (Fig. 11a) over the range 10–7000 ppm Cu. Ni variation with S (Fig. 11b) highlights the dominant control of Ni by sulphide where it is abundant, but orthopyroxene dominates the distribution where sulphide is in low abundance, containing 650 ppm Ni. Discontinuities in the Ni sulphide distribution are observed particularly in the Footwall to the PGEsz and the Upper Section of the LSZ.
|
Significance of the PGE variations
Previous studies of the Great Dyke have interpreted the PGE trends as resulting from selective partitioning of the PGE into primary sulphides by a continuous process. The offset trends reflect the effective efficiency in uptake of the PGE by sulphide (Prendergast & Keays, 1989
; Naldrett & Wilson, 1990; Wilson & Tredoux, 1990; Barnes, 1993). However, the apparent systematic and smooth variations mask a more pronounced relationship by which zones are characterized by ratios that lie within clearly defined limits. These zones form a stepped distribution separated by transitional zones. To maintain these ratios, a continuous sulphide fractionation model cannot be applicable to either the strongly mineralized or the poorly mineralized sections. |
PYROXENE COMPOSITIONS |
|---|
Compositional variations in the vertical profile of the P1 Pyroxenite
Previous models of PGE mineralization in the Great Dyke have not considered the silicate framework in detail. Such information is important in assessing the interdependence of the mineralization and primary controls of the silicate compositions. Compositions of orthopyroxene have been determined on the basis of close sampling of drill core in the same sections as analysed for PGE.
Precise compositional determinations were carried out on orthopyroxene using mineral separates and analysed by X-ray fluorescence spectrometry (XRFS). Electron microprobe studies on the same samples indicate that, within error of the determinations, no zoning was detected in the pyroxenes. Advantages of using XRFS are higher analytical precision for large numbers of samples, particularly for the minor components TiO2, Cr2O3 and NiO, and that it overcomes small-scale heterogeneities arising from exsolution. Separates were determined on carefully sized fractions using the magnetic barrier separator, heavy liquids and final hand picking. Each separation was carried out in duplicate and then each analysed in duplicate. Analytical errors are indicated in the diagrams on the basis of standard deviations of replicate analyses.
Main Sulphide Zone and Footwall to the PGEsz
The variation of mg-number [as Mg/(Mg + Fe2+)] in orthopyroxene in section MR87 (Fig. 12a), commencing with the Footwall to the PGEsz, is a series of troughs and peaks. There is a tendency for the troughs in mg-number to be located in the mid-points of these zones. The variation of Cr2O3 in orthopyroxene (Fig. 12b) is similar to that of mg-number, except that Cr2O3 attains a maximum in the Hanging Wall to the PGEsz or in the upper part of the PGEsz in both drill core sections. The cyclicity is even more pronounced for TiO2 in orthopyroxene in MR87 (Fig. 12c), with sharp changes observed at the boundaries of the zones APSZ, BPSZ and CPSZ. The variation for TiO2 is strongly antipathetic to that of mg-number.
|
The distribution of incompatible elements is not dependent on the abundance of the liquidus silicate and sulphide phases and therefore has the potential to monitor changes in the amount or composition of the liquid. The distribution of Zr (Fig. 12d) in the whole rock indicates variations that are essentially identical to that of TiO2 in orthopyroxene. This observation suggests a strong link between minor element compositions in orthopyroxene and a fundamental parameter of the whole rock that will have been imposed by the primary crystallization process, namely the incompatible element content.
Lower Sulphide Zone
Variation in mg-number in orthopyroxene in the LSZ (Fig. 13a) illustrates an overall systematic decrease with a very well-defined small-scale structure of humps and troughs on a scale of 10–15 m. This variation highlights the same zones as defined on the basis of Pd/(Pd + Pt) (ALSU, BLSU, CLSU, etc. for the Upper Section and ALSL, BLSL, CLSL, etc. for the Lower Section). Cr2O3 in orthopyroxene (Fig. 13b) exhibits small-scale fluctuations superimposed on major oscillations, which correspond to the observed significant changes in the distribution of PGE in the Lower Section (see Fig. 8). In contrast, Cr2O3 in the Upper Section follows mg-number closely in both the overall and smaller-scale variations. Small amounts of interstitial magnetite are unlikely to have influenced the distribution of Cr2O3 and chromite is not present in this section. The TiO2 content of orthopyroxene (Fig. 13c) increases systematically through the entire LSZ but the small-scale oscillations are particularly well developed. In many cases, the boundaries of the zones, defined on the basis of Pd/(Pd + Pt) (see Fig. 10), correspond closely to sharp changes in TiO2 concentrations. Also in a similar pattern to that observed for the MSZ, the boundaries of these delineated zones in many cases tend to correspond to low TiO2 concentrations in orthopyroxene, with gradually rising and falling peaks forming well-defined rhythmic units. The boundaries are less well defined in the Lower Section, but the systematic variations are still present. A break in the overall trend occurs in the Footwall to the PGEsz.
|
Zr in the whole rock (Fig. 13d) has an almost identical small-scale variation to that of TiO2 in orthopyroxene in the Upper and Lower Sections of the LSZ. The overall increase in Zr content upwards through the LSZ is rather more subdued compared with that of TiO2 in the orthopyroxene. Again, these two elements provide evidence of a substantial link between a compositional parameter for the cumulus orthopyroxene and an incompatible element that would mainly reflect the trapped liquid in the cumulate.
|
SILICATE INTER-ELEMENT VARIATIONS |
|---|
P and Zr in the whole rock
P and Zr are incompatible elements in mafic cumulates and therefore they should display similar variation in chemical profiles for the whole rock. Such corresponding variation is demonstrated in Fig. 14, with three distinct groupings of data conforming to: (1) the LSZ and Footwall to the PGEsz; (2) the PGE Subzone itself; (3) the Hanging Wall to the PGEsz, which includes the Websterite Layer. Zr is not completely incompatible in orthopyroxene, and a small amount of Zr will have been partitioned into the mineral as is indicated for the Zr intercept of 3–5 ppm (Fig. 14) (confirmed by ion probe studies in progress). The slightly greater intercept for the pyroxenes higher in the succession reflects both the evolving composition of the liquid and the increasing amount of oikocrystic clinopyroxene. The displaced trends correspond to the more abundant development (up to 5% of the whole rock) of clinopyroxene oikocrysts, particularly in the region of the PGEsz, and to the development of up to 50% cumulus clinopyroxene in the Websterite Layer. Clinopyroxene has a higher partition coefficient for Zr, which therefore results in the progressive displacement of the trends. This relationship also indicates that the clinopyroxene did not form from the interstitial liquid but as an early primary crystallization phase.
|
Interdependence of orthopyroxene compositional components and whole-rock incompatible elements
The comprehensive and close sampling of this study shows the existence of a strong inverse relationship for mg-number of orthopyroxene with Zr content of the whole rock (Fig. 15) for all sections of the P1 Pyroxenite. The section that includes the Websterite Layer, the Hanging Wall to the PGEsz, PGEsz itself and Footwall to the PGEsz (Fig. 15a) has a range of mg-number of 0·81–0·83 in a series of well-defined arrays, or groupings of data points, which correspond exactly to the zonal distribution of PGE (see Figs 6 and 7). The same relationship is observed for the Upper Section of the LSZ (Fig. 15b and c) (with a range of 0·828–0·848). The arrays for zones BLSU and DLSU overlap and lie within one field indicating a compositional reversal within the sequence. The Lower Section of the LSZ (Fig. 15c) shows very well-defined separation of zones in the Zr vs mg-number diagram. These zones exactly correspond to the zonal distribution of PGE in the LSZ (see Figs 9 and 10). P could have been used in place of Zr and gives the same result. Points that lie on the boundaries between the various arrays, and are therefore intermediate to the arrays, are also identified. In a similar fashion to the PGE variation, zone BLSL can be regarded as intermediate to ALSL and CLSL.
|
There is a general shift of mg-number of orthopyroxene to progressively lower values for each of the arrays or groupings of data points upwards through the succession. On the basis of their average slope, these arrays become steeper progressing upwards in the succession (Fig. 16a). This may be explained by a slight overall enrichment of Zr in the magma, which then is also reflected in the higher Zr content of the trapped liquid. The compositionally defined small-scale rhythmic units are characteristic features of the P1 Pyroxenite and indicate a primary control on the crystallization process. Variation of TiO2 in orthopyroxene, plotted against mg-number (Fig. 16b), also exhibits well-defined linear arrays with an indication that the arrays become steeper higher in the succession. These arrays highlight the separation of the Upper and Lower Section of the LSZ and the separation of the LSZ, as a whole, from the overlying rock units.
|
Concomitantly, Zr in whole rock exhibits strong interdependence with TiO2 in orthopyroxene (Fig. 17a). All data for the Hanging Wall to the PGEsz, the PGEsz and the LSZ sections fall on a well-defined trend. The Footwall section to the PGEsz displays a well-constrained linear trend but is displaced from all other data. This is consistent with this zone having an apparent break in the overall evolutionary pattern for the P1 Pyroxenite. This relationship suggests that compositional equilibrium was attained between the cumulus pyroxene and the interstitial liquid.
|
Breaks in the trends of mg-number with TiO2 and Cr2O3 contents in orthopyroxene (Figs 16b and 17b) occur at stratigraphic levels that immediately precede zones where PGE undergoes a sharp increase. A dramatic break occurs within the lower section of the LSZ. Zones of enrichment of Ni in orthopyroxene (Fig. 17c) follow those of Cr2O3. Orthopyroxene in the PGEsz is markedly enriched in Ni and this may partly result from re-equilibration between the cumulus orthopyroxene and Ni-rich sulphide liquid in the base metal portion of the MSZ (Barnes & Naldrett, 1985
; Li & Naldrett, 1999). |
DISCUSSION |
|---|
The precise determinations of orthopyroxene compositions over narrow sampling intervals reveal variations on a scale of 2–10 m, which cannot, in detail, be explained by simple fractionation or by repeated influx of magma, although the overall progressive change in composition through the section is most easily attributed to bulk fractionation. Injection of small amounts of primary liquid, combined with continued crystallization, is likely to have occurred at several points within the section. However, the small-scale patterns of variation, some of which have a symmetrical disposition within the layers, are not consistent with magma influx. The role of trapped, and possibly migrating liquids could also have been important in both the final textures observed and the mineral and whole-rock compositions. In addition, there exists a striking correlation between PGE characteristics within specific rock units and compositional variation of orthopyroxene, which has not previously been demonstrated in the Great Dyke or in other layered intrusions.
Characteristics of the more highly mineralized MSZ are remarkably similar to those of the LSZ, indicating that similar processes took place. In addition, the processes that controlled the PGE enrichment are indicated to have been strongly linked to those that controlled the compositions of the silicate minerals. The observed cyclicity in the pattern of mineralization is inconsistent with previous models of progressive fractional segregation of sulphide. It is also not consistent with models that rely on fluid transport of PGE as the dominant control on mineralization. The interdependence of several geochemical factors requires examination of accepted processes for magma chambers, particularly where these relate to the concentration of PGE.
Liquidus and post-liquidus controls on pyroxene composition
Textural and compositional equilibrium is dependent on many processes subsequent to the first stage of crystallization. These include reaction with trapped liquid (Barnes, 1986), migration of interstitial liquid through the crystal pile (Irvine, 1980), solid-state crystal readjustment, and annealing processes (Hunter, 1996). Observed textures cannot be attributed to any single control and it is likely that all processes contributed by variable degrees to the final textural form. It is shown that in the P1 Pyroxenite, zones of strongly annealed pyroxenes are interspersed within a framework structure of cumulus crystals that have not all undergone annealing to the same extent. The degree to which annealing has taken place is variable and is related to the stratigraphic position in the sequence, with rocks lower in the sequence showing greater degrees of annealing. The extent to which annealing took place is dependent on the development of cementing phases, which then also controlled the redistribution and retention of trapped liquid. A combination of processes will have contributed to the final mineral composition and it is unlikely that the preserved composition would be that of the initial liquidus composition of the early stage mineral phases. The degree to which liquidus cumulus compositions have been modified would depend very largely on the relative influence of the trapped liquid, the passage of evolved liquids expelled by compaction and the stage at which the system became sealed, preventing expulsion or migration of liquid.
The linear arrays linking pyroxene compositions to incompatible element content in the whole rocks indicate two important controls in the crystallization and solidification process: (1) equilibrium crystallization gave rise to the essentially constant pyroxene compositions that existed initially within each of the layers; (2) the trapped liquid played an important role in establishing the final compositions, thereby giving rise to the observed range of compositions within each of the layers. For each of the small-scale rhythmic units, the initial (liquidus) composition of the orthopyroxene must have been constant, or very nearly constant, and reaction with trapped liquid would have driven the pyroxenes to more evolved compositions. The degree by which the pyroxenes changed composition would have been dependent on the local amount (for a given sample) of trapped liquid.
It may therefore be deduced that there are essentially no rock types in this succession that contain pyroxenes of liquidus compositions. An estimate of the liquidus pyroxene composition for each of the units is given by the intercept of the array on the mg-number axis where the concentration of the incompatible element in the interstitial liquid is zero. For the whole-rock composition, this equates to the amount of Zr contained within the pyroxene (2–5 ppm from Fig. 14). Using the approach of Barnes (1986), the amount of trapped liquid required to cause the displacement of the compositions is calculated to range between 2 and 12%. This small amount of trapped liquid would not have caused the amount of Zr incorporated within the pyroxene to vary greatly, and the Zr content of the whole rock remains strongly dependent on that incorporated within late-stage mineral phases, mainly zircon and baddeleyite. These results also indicate that the initial porosity of the cumulate was reduced rapidly at an early stage such that the final trapped liquid strongly influenced the re-equilibrated pyroxene compositions.
The role of fluids and re-equilibration processes
If crystallization accompanies liquid migration then an assumption of a closed system binary mixture of cumulus crystals and crystallized liquid is not valid. In the Banded Sequence of the Stillwater Complex, there is little correlation between the estimates of trapped liquid using the mg-number shift for the whole rock and the trace element content (Meurer & Boudreau, 1998). Therefore using these approaches would not give reliable estimates of the amount of trapped liquid that crystallized. Meurer & Boudreau (1998) concluded that in the Banded Sequence the observed signatures result from the effects of trapped interstitial liquid combined with a continuous flux of fractionated liquid driven by compaction.
There are several fundamental differences between the rocks studied in the Stillwater Complex and the P1 pyroxenite of the Great Dyke: (1) in the Banded Sequence evidence points to extensive