Formation of the main sulfide zone at Unki Mine, Shurugwi Subchamber of the Great Dyke, Zimbabwe: Constraints from petrography and sulfide compositions

In the Great Dyke of Zimbabwe, the principal platinum group element (PGE)–bearing horizon, termed the main sulfide zone, is hosted in orthopyroxenites (Prendergast and Wilson, 1989). PGE mineralization has been demonstrated to occur bimodally in mafic-ultramafic intrusions, either hosted by sulfides or as discrete platinum group minerals. Pentlandite [(Fe,Ni)₉S₈], for example, is characterized by elevated concentrations of Pd and Rh (Junge et al., 2015). Such investigations of PGE contents of sulfides have routinely been carried out utilizing electron probe microanalysis (EPMA; Godel and Barnes, 2008; Osbahr et al., 2013; Zaccarini et al., 2014). The importance of using the EPMA in analyses of PGEs is that it has a spatial resolution of ∼1 μm, with detection limits of tens of parts per million (ppm).

Several processes may be responsible for the origin of the PGE mineralization in layered intrusions. Naldrett (1989) suggested that PGEs precipitate from magma and accumulate on the top of a crystal pile. Vermaak and Hendriks (1976) and Boudreau (2019), in contradistinction, preferred an upward infiltration process whereby PGEs are precipitated and transported from footwall rocks by ascending, volatile-rich fluids. Although the exact process responsible for the PGE mineralization remains under debate, the close spatial, and possibly genetic, relationship between PGE mineralization and base metal sulfides has been documented by other researchers (Viljoen and Schürmann, 1998).

Base metal sulfides in orthomagmatic Ni-Cu-PGE deposits often carry significant concentrations of PGEs. Several previous studies have helped shed some light on the distribution of the PGEs between various sulfides such as pentlandite, pyrrhotite, chalcopyrite, and pyrite (Dare et al., 2010). From experimental investigations, the distribution of PGEs in base metal sulfides is inferred to be a consequence of sulfide liquid fractionation (Mungall et al., 2005). From such experimental investigations, it has been observed that subsequent to the separation of a Fe-Ni-Cu sulfide liquid from a mafic magma, a monosulfide solid solution (MSS) and a Cu-rich liquid precipitate at ∼1000 °C (Kullerud et al., 1969; Naldrett, 1989). Os, Ir, Ru, and Rh, based on experimental work on partition coefficients, preferentially partition into the Fe-rich MSS, whereas Cu, Pt, Pd, Ag, and Au are enriched in the fractionated liquid rich in Cu. Such partition coefficients depend strongly on the Fe/S ratio of a magmatic sulfide liquid (Barnes et al., 2001). Li et al. (1996) concluded that partition coefficients of all metals increase between MSS and liquid with increasing S content, both in the sulfide liquid as well as in the MSS. The liquid rich in Cu then crystallizes as an intermediate solid solution (ISS) at ∼900 °C. At temperatures below 600 °C, the ISS has been shown to break down to chalcopyrite and cubanite, whereas the MSS breaks down to pyrrhotite and pentlandite. Consequently, pyrrhotite and pentlandite should be enriched in Os, Ir, Ru, and Rh, and chalcopyrite should be enriched in Pd, Pt, Ag, and Au, if no discrete platinum group minerals crystallize or no subsolidus reequilibration occurs upon further cooling.

In the Great Dyke, all three models of orthomagmatic (Irvine, 1983), hydromagmatic (Boudreau and McCallum, 1992; Boudreau, 2016), and micronugget models (Tredoux et al., 1995) have been proposed for the origin of the main sulfide zone (Wilson, 2001). Debate on these models continues (Robb, 2005). Stratigraphic offsets in peak concentrations of PGEs and base metal sulfides, which occur in the main sulfide zone (Oberthür, 2011), have, in part, been attributed by Li et al. (2008) to the interaction between magmatic PGE-bearing base metal sulfide assemblages and hydrothermal fluids. Li et al. (2008) presented mineralogical and textural evidence from the Hartley Platinum Mine, located in the Darwendale Subchamber of the Great Dyke, which they interpreted to indicate that alteration of base metal sulfides and mobilization of metals and S occurred during hydrothermal alteration. Sulfur isotope data of pyrite, pyrrhotite, and chalcopyrite ranging from 0.1‰ to 0.8‰, as well as O isotope data for orthopyroxene ranging from 5.1‰ to 6.5‰ from the main sulfide zone suggest that magmatic fluids were involved in the alteration (Li et al., 2008). Further, actinolite has both O and H isotope data ranging from 5.0‰ to 5.6‰ and from 64‰ to 73‰, respectively, which Li et al. (2008) interpreted to be consistent with magmatic fluids.

Despite numerous studies on mineralized reefs in layered intrusions documenting the association of PGEs with sulfide-enriched horizons, only a tiny fraction of these studies (Brynard et al., 1976; Todd et al., 1982; Kawohl and Frimmel, 2016) have focused on the composition of the sulfide minerals themselves. We had three aims in carrying out this base metal sulfide composition study in the Unki Mine area of the Shurugwi Subchamber of the Great Dyke. The first one was to carry out a petrographic study of the sulfides that occur in the PGE-enriched zone straddling the main sulfide zone from the gabbronorites at the base of the mafic sequence to the footwall of the main sulfide zone. The second aim was to document the concentrations of major and minor elements and their distributions in base metal sulfides occurring in this sequence in order to place some constraints on the origin of the mineralization. Since hydrothermal alteration affected the main sulfide zone samples under study (Chaumba, 2017), our third and final aim was to use MELTS modeling to infer the temperature at which this hydrothermal alteration likely occurred and determine the Great Dyke minerals that likely retained their crystallization temperatures.

The Great Dyke of Zimbabwe is one of the world’s largest and better-studied intrusions. It is a layered mafic-ultramafic intrusion that cuts across the Zimbabwe craton (Fig. 1A), and it also acts as host to the second largest resource of PGEs in the world. The Great Dyke is largely composed of two major chambers, the North and South Chambers (Fig. 1B), which Wilson and Prendergast (1989) further subdivided into subchambers on the basis of continuity of layering, style, and thickness of cyclic units (Fig. 1B). A third and relatively small chamber, the Mvuradonha Chamber, occurs at the extreme north end of the dike (Wilson and Prendergast, 1989).

The structure of the Great Dyke is synclinal, with inwardly dipping layers, where the dips increase from the central axis toward the margins but decrease again near the walls (Worst, 1960; Wilson and Prendergast, 1989). At depth, the Great Dyke has a dike-like feeder that, in places, has been interpreted to be connected to deep-seated magma chambers (Podmore and Wilson, 1987). The transverse section of the Great Dyke (Fig. 2A) has been interpreted to be trumpet-shaped, with individual layers thinning away from the axis and eventually becoming incorporated in the Border Group, which rests against the Great Dyke walls (Wilson and Prendergast, 1989). The longitudinal section of the layers of the Great Dyke plunges gently toward the center of each chamber to form an overall boat-like structure (Worst, 1960). Based on U-Pb dating of zircon and rutile from orthopyroxenites of the P1 pyroxenite layer, the age of the Great Dyke is 2575.4 ± 0.7 Ma (Oberthür et al., 2002). Further Sensitive high-resolution ion microprobe (SHRIMP) U-Pb studies on the Great Dyke and its satellites by Wingate (2000) yielded a comparable Neoarchean emplacement age of 2574 ± 2 Ma for baddeleyite.

In comparison to other layered intrusions such as the Bushveld Complex, isotope data from the interval straddling the contact between the ultramafic and mafic sequences of the Great Dyke indicate a less enriched composition of initial ⁸⁷Sr/⁸⁶Sr ratios (0.7024–0.7028) and ε_Nd (−1 to +1; Maier et al., 2015). Sulfur isotope values (δ³⁴S) of Great Dyke samples from studies carried out by Li et al. (2008) on pyrite, pyrrhotite, and pentlandite ranged from 0.1‰ to 1‰, and Maier et al. (2015) obtained δ³⁴S values on bulk-rock samples that ranged from −0.3‰ to 0.3‰, which fall within the 0 ± 5‰ range of mantle values (Ohmoto and Goldhaber, 1997). Thus, relatively moderate amounts of contamination of the Great Dyke parent magma must have occurred.

The stratigraphic section in the vicinity of Unki Mine in the Shurugwi Subchamber, from just above the mafic sequence–ultramafic sequence contact to the footwall of the mineralized zone, is shown in Figure 2B. Here, plagioclase pyroxenite is overlain by a 6-m-thick layer of plagioclase websterite, which is capped by a very thin chromitite layer (or chromitite stringer; Fig. 2B). Overlying this chromitite stringer, there are gabbronorites of the mafic sequence (Fig. 2B). A base metal–enriched zone, called the base metal sulfide zone, occurs ∼3 m beneath the plagioclase websterite–plagioclase pyroxenite contact (Fig. 2B).

The main sulfide zone at Unki Mine is a 180 cm package of rocks composed of the PGE subzone and a base metal subzone (BM subzone; Fig. 2C). The peaks in Pt, Pd, Rh, Cu, Au, and Ni occur in a sulfide-enriched horizon—the base metal sulfide zone (Fig. 2C). The PGE subzone occurs stratigraphically beneath the BM subzone, with a transitional contact (Fig. 2C). The BM subzone is enriched in Ni and Cu (Fig. 2C). The resource width, the stope width, and the PGE mineralized zone width have thicknesses that vary widely (Fig. 2C). The main sulfide zone, therefore, was encountered toward the lower part of the investigated succession in this work (Figs. 2B and 2C).

Within the Great Dyke succession, there are several PGE-enriched layers, which also include most of the chromitite layers (Oberthür et al., 2002), as well as several silicate horizons in the upper portions of the ultramafic sequence (Prendergast and Keays, 1989; Prendergast and Wilson, 1989; Wilson and Tredoux, 1990; Wilson and Prendergast, 2001; Oberthür et al., 2002; Oberthür, 2011). According to Oberthür (2002, 2011), up to 2 ppm PGEs (Pt/Pd = 0.1) occur in the C1d chromitite and its host rocks.

PGEs in the Shurugwi Subchamber, as in the rest of the Great Dyke, occur mainly in a tabular stratabound layer hosted in pyroxenite termed the main sulfide zone. The main sulfide zone is located in orthopyroxenites close to, and overlapping with, the websterite layer and contains up to 6% sulfide. Having been discovered over a century ago (Wagner, 1914), the main sulfide zone is similar in form in all the chambers of the Great Dyke (Worst, 1960). The width of the main sulfide zone varies from 2 to 8 m, and it can contain up to 5 ppm PGEs and up to 8% sulfides over 2–3 m (Oberthür, 2011). The main sulfide zone is 1.8 m thick at Unki Mine (Fig. 2B) and ∼2.5 m in the Wedza Subchamber (Prendergast, 1991), and it lies close beneath, or overlaps, the boundary between the orthopyroxenites and the overlying websterite (Fig. 2). The vertical distribution of sulfides (pyrrhotite, pentlandite, and chalcopyrite) and PGEs in the main sulfide zone displays a pattern wherein the base metals are dominant in the upper part, termed the BM subzone, and the PGEs are concentrated in the lower part, termed the PGE subzone (Fig. 2C). The PGE subzone itself is further subdivided into a Pt-dominant upper section and a Pd-dominant lower section, with the two sections being generally separated by an intermediate section characterized by moderate Pd/Pt ratios (Wilson and Prendergast, 2001).

A general vertical zonation of minerals occurs in the main sulfide zone (Fig. 2C), whereby Fe-Ni sulfides dominate the lower part, and Cu sulfides dominate the upper part (Coghill and Wilson, 1993). The sulfides occur as interstitial phases to orthopyroxene, and they show a heterogeneous distribution on a scale of millimeters to centimeters; in several cases, they are concentrated at the boundaries of plagioclase oikocrysts (Wilson, 1992). These so-called “offset” metal distribution patterns within the main sulfide zone are, thus, defined by peak Pd levels that occur near the base of the reef, whereas peak Pt and Cu levels occur at progressively higher stratigraphic levels within the main sulfide zone (Oberthür, 2011).

The main sulfide zone has been classified as a sulfide-hosted, magmatic PGE deposit (Wilson et al., 2000). The regular distribution of PGEs in all areas where the main sulfide zone has been investigated has been interpreted to indicate a primary, sulfide-controlled fractionation pattern, whereby the PGEs have been scavenged from the magma due to their strong partitioning into primary sulfide (Naldrett and Wilson, 1990; Prendergast and Keays, 1989). Other analogues of this process have been reported in the Munni Munni intrusion of Western Australia (Barnes et al., 1990; Barnes, 1993).

According to Coghill and Wilson (1993), platinum group minerals in the Shurugwi Subchamber occur in three distinct textural environments: (1) at the boundary of sulfides and silicates/hydrosilicates, (2) entirely enclosed within sulfides, and (3) entirely enclosed within silicate or hydrosilicate minerals. Wilson (2001) concluded, from orthopyroxene composition, PGE, and Cu concentration studies, that PGE-bearing horizons of the Great Dyke are composed of subzones within which constant Pd:Pt ratios occur, although significant variations can occur between the subzones.

Samples utilized in this study were collected from a stratigraphic zone straddling the main sulfide zone at Unki Mine, currently the only PGE mine in the Shurugwi Subchamber (Fig. 2B), and sample descriptions are provided in Table 1. The samples were obtained from a borehole that was drilled toward the axis of the Shurugwi Subchamber at Unki Mine. Samples were collected from the overlying gabbronorites of the mafic sequence (samples MUSGa01/02), stratigraphically above the main sulfide zone (samples MusCr01, MusWebb01a/02a, MusWebb01/02, NusPxhw01, and MusPeg01), and through the main sulfide zone (samples MusBMSZ01 and MusIRUP) to its footwall (samples MUSPxaFwt01 and MUSPxbFwt01; Figs. 2B and 2C; Table 1). Sample MUSCr01 came from the chromitite stringer occurring right at the ultramafic sequence–mafic sequence contact (Fig. 2B; Table 1). Plagioclase pyroxenite is overlain by a 6-m-thick layer of plagioclase websterite, which is capped by a very thin chromitite layer (or chromitite stringer; Fig. 2B). The PGE subzone is located below the base metal sulfide zone (sample BMSZ01, which is part of the main sulfide zone; Figs. 2B and 2C) and the BM subzone; both are distinguished based on their metal profiles, whereby the concentration of the PGEs begins to rise to appreciable amounts (ppm levels) as both base metals and sulfide concentrations increase. Sample MusIRUP01 is the other sample obtained from the main sulfide zone (Figs. 2B and 2C).

The pegmatoidal plagioclase pyroxenite, which ranges from a few centimeters to over 2 m in thickness, was encountered 6–8 m below the websterite–plagioclase websterite contact (0.75–1 m above the base metal sulfide zone) in the Shurugwi Subchamber. The pegmatoidal plagioclase pyroxenite is a very coarse-grained, grayish brown, pegmatoidal plagioclase pyroxenite composed of coarse-grained orthopyroxene, plagioclase, and clinopyroxene crystals with grain sizes up to 2 cm occurring together with coarse net-textured sulfides. The PGE subzone and the BM subzone (from which sample BMSZ01 was obtained) have an overlap of ∼30 cm between them. The BM subzone commences ∼20 cm below the Pt peak. In this work, the base metal sulfide zone refers to the sampled position within the BM subzone. A sample from the BM subzone was obtained ∼3 m below the plagioclase websterite–plagioclase pyroxenite contact (Figs. 2B and 2C).

Thin sections were first examined under a petrographic microscope. Then, sulfides were analyzed under a microprobe on polished thin sections. Mineral compositions of the sulfide minerals were obtained using the JEOL JXA-8530F Hyperprobe housed at Fayetteville State University, Fayetteville, North Carolina (Chaumba et al., 2016). The Hyperprobe was operated with a beam current of 20 nA, an accelerating voltage of 30 kV, and a minimum beam diameter of 1 μm. Images were acquired with both a secondary electron detector and a backscattered electron detector, which provided compositional information to visualize different phases. Compositions were analyzed first by X-ray energy-dispersive spectroscopy (EDS) for qualitative and semiquantitative analysis, and then by X-ray wavelength-dispersive spectroscopy (WDS) for quantitative analysis using Smithsonian standards. At least three spots were collected for each phase. Fifteen second counting times were used on peak and background measurements. For the calculation of the oxides, the ZAF matrix correction system of Armstrong (1988) was used.

Use was made of the rhyolite-MELTS model (v. 1.0.x; Gualda and Ghiorso, 2015) for modeling the crystallization temperatures of Great Dyke minerals by changing the controlling variable of pressure as equilibrium was repeatedly calculated. Isobaric fractional crystallization upon cooling was used to model batches of the parental magma to the Great Dyke at the liquidus at defined pressures. The parental magma composition to the Great Dyke utilized in MELTS calculations was that of the East Dyke (Wilson, 1982).

Under the petrographic microscope, fine- to coarse-grained chromite crystals from the chromitite stringer occur together with either plagioclase or orthopyroxene (Figs. 3A–3F). Where a chromitite stringer occurs at the base of the mafic sequence, such as in the Unki Mine area of the Shurugwi Subchamber, subrounded to idiomorphic chromite crystals are typically enclosed in coarse-grained, polysynthetically twinned poikilitic plagioclase crystals (Fig. 3A). Almost all chromite crystals from the chromitite stringer under investigation enclose very fine-grained plagioclase crystals (Fig. 3). Anhedral plagioclase crystals from the chromitite stringer are typically enclosed in chromite crystals from the chromitite stringer, and they are not in optical continuity with plagioclase crystals that are not included in chromite (Fig. 3C). In some cases, subhedral chromite crystals are oriented across the twin plane of simply twinned plagioclase crystals, in addition to finer-grained chromite crystals from the chromitite stringer, which occur in different twin sets than the simply twinned plagioclase crystals (Fig. 3D).

The boundary between the ultramafic and mafic sequences in the Unki Mine area of the Shurugwi Subchamber is defined by a sharp contact between coarse-grained orthopyroxene and plagioclase crystals (Figs. 3E, 3F, 4A, and 4B). At this boundary, subhedral chromite crystals from the chromitite stringer, which in some places completely enclose very fine-grained plagioclase crystals, are also partially or completely enclosed by fine-grained plagioclase crystals (Figs. 3E, 3F, 4A, and 4B). Minute orthopyroxene crystals also occur as inclusions in chromite from the chromitite stringer (Figs. 4A–4F). Very fine-grained pockets of plagioclase crystals commonly occur as inclusions within the coarse-grained orthopyroxene crystals, which occur near the contact with the underlying ultramafic rocks (Figs. 4A and 4B). In the upper part of the P1 pyroxenite layer in the underlying ultramafic sequence, fine-grained chromite crystals from the chromitite stringer tend to be completely enclosed in very coarse-grained orthopyroxene crystals (Figs. 4C and 4D).

Within the sulfide-enriched lower part in the mafic sequence occurring close to the contact with the underlying ultramafic rocks, some plagioclase crystals that enclose numerous fine-grained chromite crystals from the chromitite stringer show evidence of alteration (Fig. 4E). Here, the plagioclase has minute inclusions of sericite, with even plagioclase crystals that are enclosed within chromite crystals from the chromitite stringer also showing minute inclusions of sericite (Fig. 4E). Approximately 1 cm away from the ultramafic-mafic sequence contact in the lower mafic succession, coarse-grained clinopyroxene and plagioclase, which display cloudy appearances, comprise the gabbroic rocks (Fig. 4F).

At the contact with the overlying lower mafic succession, very fine-grained inclusions of orthopyroxene are also common in chromites from the chromitite stringer (Figs. 5A–5F), and they are included in poikilitic orthopyroxenes (Fig. 5A). Sulfides rarely occur, partially rimmed by chromite crystals (Fig. 5A). Sulfides in the investigated succession commonly occur as minute grains, which also occur as inclusions in silicate phases such as orthopyroxenes (Fig. 5B). Coarse-grained orthopyroxenes wrap around both chromites and sulfides (Fig. 5B).

In the lowermost part of the lower mafic succession, relatively coarser-grained sulfide crystals tend to occur in interstices between chromite crystals from the chromitite stringer and plagioclase crystals, and they also appear to partially surround chromite crystals from the chromitite stringer (Fig. 5C). Numerous very fine-grained sulfide crystals occur scattered within the coarse-grained orthopyroxene (Fig. 5B) and plagioclase (Fig. 5C), as well as at the boundary of chromite crystals from the chromitite stringer and orthopyroxene crystals (Fig. 5C). One sulfide crystal and two chromite crystals displayed a triple junction (Fig. 5D); the sulfide crystals typically occur either partially rimming chromite crystals or as inclusions in orthopyroxene crystals in the P1 pyroxenite layer.

Some fine-grained sulfide crystals occur as inclusions in relatively coarser-grained chromite crystals from the chromitite stringer, whereas other fine-grained sulfide crystals occur at the boundary between chromite crystals (Fig. 6A). The base metal sulfide pentlandite typically occurs as inclusions within both chromite crystals from the chromitite stringer and orthopyroxene crystals (Figs. 6A and 6B). Pyrrhotite, another common base metal sulfide encountered in the chromitite stringer, is typically very fine-grained (Figs. 6A and 6B). Chlorite tends to occur together with pentlandite inclusions in chromite, forming a rim that partially surrounds the pentlandite (Fig. 6B).

Both pyrrhotite and pentlandite occupy interstices between silicate and chromite crystals from the chromitite stringer, with other finer-grained pyrrhotite crystals occurring within silicate minerals and at the boundaries between chromite and orthopyroxene crystals (Fig. 7A). Chalcopyrite crystals occur wholly enclosed within plagioclase crystals, with the latter also containing a cluster of both pentlandite and pyrrhotite crystals (Fig. 7B). Pyrrhotite, pentlandite, and chalcopyrite can all occur as inclusions in some silicate phases such as plagioclase (Fig. 7B).

In some patches within the chromitite stringer, intergranular and fine-grained crystals (∼60 μm) of pyrrhotite, chalcopyrite, and pyrite (Fig. 8A) have a tendency to form sulfide clusters that are relatively coarser grained (can exceed 1000 μm in size) than sulfides that occur as inclusions or between silicate minerals. Besides the pyrite crystals occurring within the chromite stringer, the only other lithologies that contain pyrite crystals are the gabbronorite from the mafic sequence (sample MusGa02) and the pegmatoids occurring just above the BM subzone (sample MusPeg01; Tables 1 and 2; Supplemental Table¹). All three base metal sulfides of pyrrhotite, chalcopyrite, and (subhedral) pyrite were observed occurring together only in the contact zone chromitite sample MusCr01 (Fig. 8B).

Within the base metal sulfide zone, some sulfide grains are slightly coarser grained, often >200 μm across and 400 μm in length, with pyrrhotite, which tends to occupy the center of the sulfide cluster, occurring together with chalcopyrite crystals occurring near the margins of pyrrhotite crystals (Fig. 9A). Also, in the base metal sulfide zone, crystals of pyrite exceeding 100 μm in size occur at the center, with pentlandite again occurring at the edges of the pyrite crystals (Fig. 9B). Chlorite occurs in association with these base metal sulfides (Figs. 9A and 9B).

In the plagioclase websterite, which occurs stratigraphically below the chromite stringer, both pyrrhotite and chalcopyrite commonly form elongated crystals (Fig. 10A). Anhedral crystals of pyrrhotite and pentlandite typically occur together in the plagioclase websterite, often surrounded by chlorite crystals (Fig. 10B). In summary, both sulfides and chromite crystals occur as inclusions in one or the other, and both sulfides and chromite occur as inclusions in both orthopyroxene and plagioclase crystals. Further, some sulfides partially wrap around both chromite and some silicate crystals.

Representative pyrrhotite, pentlandite, chalcopyrite, and pyrite analyses from the main sulfide zone at Unki Mine are shown in Table 2, with the rest of the analyses shown in the Supplemental Table (see ^{footnote 1}). With the exception of Ir concentrations in pyrrhotite and pentlandite, concentrations for metals such as Pt, Pd, Rh, Au, Ag, and Zn were almost always at, or just above, detection limits in all three sulfides observed during this study (Table 2; Supplemental Table). Consequently, no useful plots of these elements could be obtained.

Pyrrhotite analyses (240) constituted ∼39% of all sulfide analyses, totaling 614. Out of the 240 pyrrhotite analyses, 163 of these analyses fell in the 38.1–40 wt% S range, representing the bulk of pyrrhotite analyses (Fig. 11A). The bulk of pyrrhotite S analyses fell within the 36.1–42 wt% range, with very few analyses falling below 36.1 wt% S and greater than 42 wt% S (Fig. 11A). Average Co contents in pyrrhotite, which were low, occurred at even lower concentrations in the sample from gabbroic rocks at the top of the main sulfide zone (Fig. 12A). Samples from within the main sulfide zone footwall were characterized by the lowest Co in pyrrhotite values compared to those from the main sulfide zone. Samples from areas stratigraphically above the main sulfide zone, such as those from the chromitite stringer (sample MusCR01) as well as in plagioclase pyroxenite (sample MusPxhw01), were also characterized by low concentrations of Co in pyrrhotite (Fig. 12A).

Sulfur concentrations in pentlandite were concentrated in the 28.1–38 wt% range, with the 32.1–34 wt% range constituting the bulk of the analyses with a frequency of 125 (Fig. 11B). Pentlandite S concentrations typically ranged between 30 and 40 wt% for the lower portions of the sampled stratigraphic interval, and only in the upper part did these S concentrations show a wider range from 28 to 57 wt% (Fig. 12B). Average values of S concentrations in pentlandite across the sampled stratigraphic interval were ∼33 wt%. Cobalt concentration in pentlandite across the investigated interval showed a spiked pattern, ranging from a low of 16.2 wt% just above the footwall of the main sulfide zone (sample MusIRUP01) to a high of 33.4 wt% in the base metal sulfide zone of the main sulfide zone (sample MusBMSZ; Fig. 12B). The chromitite stringer near the top of the main sulfide zone had the second highest Co concentration in pentlandite (33 wt%), with the sample above the main sulfide zone having the second lowest Co concentration of 24 wt% (Fig. 12B).

Average Co concentrations in pentlandite also showed a spiked pattern with stratigraphic height across the investigated section, with the lowest Co concentration occurring in the base metal sulfide zone, followed by those in the mafic rocks above the top of the studied interval (Fig. 12C). The concentration of Ni in pentlandite (Fig. 12D) broadly mimicked that of Co in pentlandite (Fig. 12C; sample MusPxaFwt01).

Chalcopyrite analyses with the highest frequency (62) fell in the 32.1–34 wt% S range (Fig. 11C). Next in frequency (38) were chalcopyrite analyses that plotted in the 34.1–36 wt% S range (Fig. 11C). With minor deviations, average Cu concentrations in chalcopyrite across the main sulfide zone (Fig. 12D) broadly mimicked the variation displayed by both Co and Ni in pentlandite (Figs. 12C and 12D, respectively), although average Co contents in chalcopyrite occurred in much lower concentrations than those in the latter.

Only 11 pyrite analyses were obtained, representing a mere 0.02% of all sulfide analyses. S concentrations in pyrite fell within the 52.1–58 wt% range, with seven of these falling in the 54.1–56 wt% range (Fig. 11D).

The base metal sulfides that were observed within the main sulfide zone at Unki Mine, as well as stratigraphically below and above it, were pyrrhotite, pentlandite, chalcopyrite, and pyrite, in order of decreasing abundance. Similar base metal sulfides have been reported from elsewhere in the main sulfide zone in other subchambers of the Great Dyke (Prendergast and Wilson, 1989; Wilson and Prendergast, 2001). These sulfides occur both as inclusions in primary silicate phases such as orthopyroxene, clinopyroxene, and plagioclase, as well as in association with secondary silicate phases such as amphibole and chlorite. This suggests the occurrence of both primary magmatic sulfide mineralization as well as remobilization of some of that primary mineralization. It must be pointed out, though, that coexistence does not imply cogenesis, as sulfide liquid is denser than silicate liquid (Murase and McBirney, 1973). However, if sulfide segregation and silicate crystallization occurred concurrently, they are likely to both have accumulated together and formed disseminated sulfide ores, as is the case in the main sulfide zone. It is also possible that late-stage mobilization of high-density, low-viscosity sulfide-rich melts, which may range from sulfide-rich silicate magma pulses to coherent pulses of pure sulfide liquid (Saumur et al., 2016; Saumur and Cruden, 2017), may have been responsible for the concentration of the PGEs at Unki Mine. Evidence for this comes from the iron-rich ultramafic pegmatites in the Unki Mine area (Chaumba, 2017). The altered nature of the silicate minerals in the Unki Mine area suggests that at least some of the PGEs may have been concentrated by hydrothermal fluids.

Coghill and Wilson (1993), Evans and Buchanan (1991), Evans et al. (1996), Oberthür et al. (1997), Prendergast (1990), and Wilson and Tredoux (1990) all favored a primary magmatic origin for the PGE mineralization in the Great Dyke but allowed for local-scale remobilization of the PGEs. These authors based their interpretations on the close correlation of PGEs and primary silicate geochemistry, as well as on the textures and distribution of interstitial sulfides in the mineralized horizons of the Great Dyke. These authors argued that the main sulfide zone, for example, provides evidence that the sulfides were segregated from the magma and accumulated, with minor redistribution, within the partly consolidated silicate crystals. Wilson and Prendergast (2001) observed that the heterogeneous small-scale distribution of sulfides in the main sulfide zone formed due to the growth of plagioclase oikocrysts within the pyroxene crystal framework, which forced the droplets to concentrate around their margins. The distribution of platinum group minerals and the varying PGE content of sulfide mineralization were inferred by Wilson and Prendergast (2001) to be the result of multiple processes, which involved the scavenging of PGEs from the magma by early-formed sulfides, subsolidus annealing of sulfide resulting in both remobilization and concentration of PGEs, and later remobilization of PGEs by hydrothermal fluids at elevated f_S2/f_O2 (pyrite), leading to the formation of discrete platinum group minerals.

Unlike in other areas of the Great Dyke, the concentration of sulfides in the Unki Mine area increases gradually from the base of the PGE subzone upwards, whereas Pt and Pd (and other PGEs) rise sharply (Wilson and Prendergast, 2001). An inverse relationship was reported by Wilson and Prendergast (2001) between Pt and Pd, with Pd being concentrated at the base and Pt at the top of the PGE subzone. Based on metal profiles, Pt and Pd appear to be decoupled from each other, but this relationship is not supported by interelement plots (Wilson and Prendergast, 2001). The main sulfide zone in the Unki Mine area has been shown to be unusually uniform due to its location, which is entirely in the axis of the Great Dyke. Results from the present study, however, tend to suggest a more pronounced effect of hydrothermal alteration on the mineralization in the main sulfide zone at Unki Mine than previously envisaged, as discussed below.

Barnes and Liu (2012) interpreted the correlation between Pt and Pd in sulfide-poor Australian komatiites to be controlled by original crystallization and accumulation of olivine, whereas in disseminated ores, the correlation was controlled by the abundance of sulfide. In massive ores, Barnes and Liu (2012) attributed the Pt and Pd concentrations in komatiite-hosted massive ores to fractional crystallization of the sulfide liquid (Barnes and Naldrett, 1987; Barnes, 2004). Such fractionation, however, involves the high-temperature MSS phase, which is characterized by low partition coefficients for both Pt and Pd (Barnes et al., 1997; Ebel and Naldrett, 1996), resulting in an inability to produce order-of-magnitude variations in Pt and Pd. Results by Barnes (2004) from modeling the fractionation of the MSS of the Silver Swan massive sulfide orebody at Black Swan in Western Australia showed that the concentrations of Pt and Pd would not be accounted for by accumulation of MSS orthocumulates, which they ascribed to hydrothermal alteration and remobilization. PGEs in the Great Dyke were probably transported via chloride and bisulfide, which are the most important ligands for Pd and Pt transport in hydrothermal fluids (Wood et al., 1989; Ridley, 2013), although hydroxyl and mixed species (van Middlesworth and Wood, 1999) are also thought to have contributed to the mobility PGEs in hydrothermal fluids in a minor way. The PGE-enriched main sulfide zone, consequently, is likely to have been concentrated by hydrothermal alteration, as evidenced by the alteration of the associated silicate phases (Chaumba, 2017).

Some sulfides, such as those occurring in association with chromite, plagioclase, and orthopyroxene (Fig. 7), appear to be primary, whereas others, which occur in the presence of either chlorite or both chlorite and chromite (Figs. 6, 9, and 10), appear to be secondary. Chlorite occurring in association with chromite crystals (Fig. 6) is interpreted to have been formed as a reaction product of the reaction MgAl₂O₄+ 4·MgO + 3SiO₂ + 4·H₂O = chlorite (Mg₅Al₂Si₃O₁₀[OH]₈) (Evans and Frost, 1975; Kimball, 1990), where MgAl₂O₄ occurs in the solid solution state in the spinel phase. The timing of this reaction is interpreted to have been during hydrothermal alteration that accompanied sulfide mineralization, due to the presence of sulfides such as pentlandite, which occur together with chlorite (Figs. 9A and 9B).

The observations in this study—that concentrations of PGEs such as Pt, Pd, Rh, and (to some extent) Ir in the main sulfide zone occur at or just above detection limits in the base metal sulfides encountered during this study—lend support to an earlier study by Evans et al. (1996), who concluded that in the Darwendale Subchamber of the Great Dyke, the PGEs are not hosted in the single chromitite layer they investigated (the sulfide-enriched C1d chromitite layer), which hosts anomalous but sporadic PGEs. This is unlike in the Bushveld Complex, where Merkle (1992) observed that some platinum group minerals containing Pt, Pd, and Rh are concentrated in the intercumulus silicates and are also frequently associated with base metal sulfides. It must be noted that even in the Merensky reef, most of the PGEs are hosted in silicates (Merkle, 1992). Backscattered electron images from within, as well as stratigraphically below and above, the main sulfide zone (see Figs. 2C and 5–10) show some close association of hydrosilicate phases and base metal sulfides, suggesting that scavenging of the PGEs by hydrothermal fluids likely helped concentrate the mineralization in the main sulfide zone. The original source of the mineralization, however, is likely to have been magmatic (Prendergast and Keays, 1989; Wilson and Prendergast, 2001). Prendergast and Wilson (1989), for example, observed that, regardless of the extent of alteration in the main sulfide zone, the uniform vertical profiles of both Cu + Ni and Pt + Pd throughout the Great Dyke are in agreement with an “entirely” magmatic origin. However, results from this study, as well as those of Li et al. (2008), support an interaction between primary magmatic PGE-bearing base metal sulfide assemblages and late-stage hydrothermal fluids of magmatic origin, based on both petrographic observations and sulfide compositions. Oberthür et al. (2003), from a study of primary distribution patterns in the main sulfide zone in the Darwendale Subchamber of the Great Dyke, also concluded that hydrothermal alteration helped concentrate mineralization.

Somewhat mimicking the trends displayed by S concentrations in pyrrhotite, S concentrations in pentlandite across the main sulfide zone tend to vary widely only in the uppermost part, as compared to the middle and lower parts (Fig. 12A). Average concentrations in pentlandite across the main sulfide zone are ∼34 wt% S (Fig. 12A). Across the main sulfide zone, average concentrations of Co in pentlandite display a spiked pattern (Fig. 12B). With the exception of samples from the top and near the bottom of the investigated section, Fe in pyrrhotite tends to not always mimic the variations of both Co and Ni in pentlandite and Cu in chalcopyrite (Figs. 12B and 12C). Although occurring in lower concentrations, the variation of Co in both chalcopyrite and pyrrhotite shows similar patterns across the main sulfide zone except for the topmost and lowermost samples (Fig. 12D). Co and Ni in pentlandite vary in a similar manner across the main sulfide zone (Figs. 12B and 12C, respectively). Cu in chalcopyrite behaves in a similar manner to both Co and Ni in pentlandite, except in the lower part of the investigated succession (Figs. 12B and 12C). Variations in Fe, Co, and Ni in pentlandite have been shown to display linear trends (Hem et al., 2001). In other layered intrusions, Ni contents in pentlandite have been shown to correlate negatively with Fe and Co (Cogulu, 1993). In the Crystal Lake intrusion, Thunder Bay, Ontario, Canada, Cogulu (1993) observed that Co contents increase with depth due to contamination of the magma by assimilation of sedimentary xenoliths. Co and Ni in the Crystal Lake intrusion have been shown to be antipathetic, correlating negatively with each other. Unlike in the Crystal Lake intrusion, where Co and Ni in pentlandite are ascribed to contamination by country rocks, Co and Ni in pentlandite in the investigated succession are interpreted to reflect magmatic processes, because they vary sympathetically and were likely to have been subsequently affected by the same hydrothermal processes. The depletion in both these elements in sample MusIRUP01 near the base of the main sulfide zone, as well as in sample MusGa01/02 at the top (Figs. 12B and 12C), may be due to hydrothermal processes.

Figure 13A shows plots of the variation of both Cu and Ni in pyrrhotite versus stratigraphy from the investigated succession. Both Cu and Ni in pyrrhotite have high concentrations in the main sulfide zone, although a spiked pattern is noticeable over the section. Since Cu in particular has low abundances in silicates due to its chalcophile nature, its abundance is correlated with the abundance of sulfides. The spiked pattern of both Cu and Ni across the investigated section (Fig. 13A) suggests that some of these metals, as well as sulfur, were remobilized by hydrothermal fluids. Figure 13B shows the percent relative abundances of the sulfides, and, in general, pyrrhotite is the most dominant sulfide, followed by pentlandite, chalcopyrite, and pyrite.

A good positive correlation is displayed between Cu and Fe in chalcopyrite from the main sulfide zone at Unki Mine (Fig. 14A). This positive correlation is interpreted to indicate magmatic processes that are preserved in the main sulfide zone samples at Unki Mine. For pyrrhotite from the main sulfide zone at Unki Mine, however, Ni and Co (which, along with Cu, are the most common elements that substitute for Fe in pyrrhotite; Vaughan, 2011) show a positive correlation (Fig. 14B). This suggests that Ni and Co in pyrrhotite substituted for Fe in roughly equal proportions, implying that the original magmatic signatures are still retained in some sulfides despite later hydrothermal remobilization. This observation is supported by the behavior of elements such as Ni and Co in pentlandite, which behave sympathetically (Figs. 12B and 12C).

Pentlandites are typically characterized by two different trends of Co substitutions (Riley, 1977). One trend is restricted to serpentinites and hydrothermal veins, where Co tends to substitute for Fe, whereas the other trend is observed in metamorphosed massive sulfide ores in Scandinavia, where Co tends to replace Ni (Riley, 1977). In the main sulfide zone at Unki Mine, a plot of atomic % Co versus atomic ratio Ni/(Ni + Fe) shows no clear trend of Co replacing Ni (Fig. 14C). This is unlike the upper zone of the Bushveld Complex, where Co-rich pentlandite follows a trend of preferential replacement of Ni (Merkle and von Gruenewaldt, 1986), suggesting different mineralizing processes in the two intrusions.

The Unki Mine main sulfide zone pentlandites are characterized by very low Co contents (Table 2; Supplemental Table [^{footnote 1}]), an indication that Fe substitution by this element was very limited. On a ternary plot of Co-Ni-Fe, Unki Mine main sulfide zone samples plot in the low-Co part of the field of natural pentlandites (Fig. 14D). According to Merkle and von Gruenewaldt (1986), pentlandite in the upper zone of the Bushveld Complex is consistently Co-rich, with the lowest observed value being 8.2 atomic %, a highest value of 46.4 atomic %, and a mean of 28.7 atomic %, values that are much higher than those from the main sulfide zone at Unki Mine. In the upper zone of the Bushveld Complex, very high Fe/Ni ratios ensure that the samples plot toward higher Fe/Ni values (Fig. 14D) than typical pentlandites (Merkle and von Gruenewaldt, 1986). However, it must be noted that Bushveld Complex pentlandites from the Merensky Reef are comparable to those from sulfide-bearing zones from both the Great Dyke and Stillwater Complex, probably indicating comparable processes (likely magmatic) in early stages that were responsible for the formation of the mineralized horizons in these two large layered mafic intrusions.

Pentlandite compositions from the main sulfide zone at Unki Mine are plotted on an S-Fe-Ni ternary diagram in Figure 15A, and these samples plot above and below the pentlandite “field” (Kullerud et al., 1969). Most Unki Mine main sulfide zone pentlandites do not define a trend of pentlandites that were exsolved from pyrrhotite, but rather they define pentlandite compositions that are consistent with an origin between MSS nickel-poor (monosulfide-poor, MSS1) and MSS nickel-rich (MSS2) pentlandites (e.g., Guo et al., 1999). Bushveld Complex pentlandites, for example, are nickel-rich and are MSS2 pentlandites (Fig. 15A). At temperatures of 900–1100 °C (Fig. 15A), main sulfide zone pentlandites plot in the field of MSS plus liquid, likely indicating that the pentlandites crystallized from magma (Kullerud et al., 1969; Guo et al., 1999). This is consistent with the sulfides that formed early in the crystallization history of the main sulfide zone. Later pentlandite crystals likely were formed during hydrothermal alteration, likely at temperatures of ∼300 °C, since no pyrrhotite was observed in the samples investigated, which would be consistent with pentlandite being exsolved from pyrrhotite (Kullerud et al., 1969; Guo et al., 1999). The occurrence of pentlandite in the main sulfide zone at Unki Mine indicates that the final stage of mineralization in this part of the Great Dyke occurred at temperature below 610 °C, as pentlandite is only stable below this temperature (Vaughan and Craig, 1978).

Phase relations in the Fe-Cu-S ternary system at 700 °C and 300 °C (Barton and Skinner, 1979; Vaughan and Craig, 1997) are shown in Figure 15B to deduce information from crystallization of chalcopyrite from magma. Melts were entirely crystallized at 700 °C with two extensive solid solution fields (Vaughan and Corkhill, 2017). The first of these, the intermediate solid solution (ISS), includes chalcopyrite and other phases that are close to CuFeS₂ in composition, whereas the second solid solution field occurs around bornite (Fig. 15B). Unki Mine main sulfide zone chalcopyrite compositions plot close to the stoichiometric composition of chalcopyrite but define a trend from high (31%) to low S (28%), i.e., between ∼29 and ∼34 wt% Cu (Fig. 15B). A few samples (MusIRUP, MusPeg01, MusBMSZ) have somewhat lower Cu concentrations below 29 wt% (Fig. 15B). On further cooling to 300 °C, these fields shrink, with the associated separation by solid-state diffusion that produces exsolution textures (Vaughan and Craig, 1997). These samples were probably affected more by later processes than those that plot closer to the stoichiometric composition of chalcopyrite. Since no exsolution textures were observed in the samples under study, we conclude that sulfide precipitation occurred at temperatures higher than 300 °C.

As can also be observed in Figure 15B, samples from the Stillwater Complex are comparable to a number of Unki Mine main sulfide zone samples, probably indicating a similar origin in the mineralized zones in these two intrusions. Compared to the Unki Mine main sulfide zone samples, the J-M reef of the Stillwater Complex is characterized by more samples that have compositions close to the stoichiometric composition of chalcopyrite, probably indicating less alteration in this intrusion during mineralization in its J-M Reef than in the main sulfide zone of the Great Dyke.

At Unki Mine, the main sulfide zone samples may be explained by precipitation of sulfide after an introduced S-rich fluid has stripped Fe, Ni, and Cu from the silicate assemblages, as described in the equations above. Several sulfides from the main sulfide zone at Unki Mine occur in close association with hydrous silicates such as amphibole and chlorite (Figs. 6, 9, and 10B). Ballhaus and Stumpfl (1986) also documented sulfides that occur in association with hydrous silicate phases like biotite and phlogopite in the Merensky Reef of the Bushveld Complex, which they interpreted to have formed from the liberation of H₂O during the sulfidation reaction in Equation 2 above. A comparable scenario is likely to also have occurred in the Great Dyke, but with amphibole and chlorite (Chaumba, 2017) rather than biotite being the phases that were formed from the liberation of H₂O during the sulfidation reaction in Equation 2. Further, Lorand and Gregoire (2006) also attributed the sulfides in some phlogopite-rich peridotite xenoliths from the Bultfontein kimberlite pipe from the Kaapvaal craton in South Africa to the reaction in Equation 2 above. Some of the chlorite in main sulfide zone samples probably originally crystallized as biotite and/phlogopite that was later altered. Biotite and phlogopite, which occur together with sulfides in the Bushveld Complex, were similarly thought have crystallized in the vicinity of sulfides from a fractionated intercumulus melt that was highly enriched in volatiles (Ballhaus and Stumpfl, 1986). In the Bushveld Complex, Ballhaus and Stumpfl (1986) further observed that amphibole, chlorite, and talc occur as later alteration phases of cumulus minerals such as pyroxene, as well as intercumulus phases like plagioclase. It is also possible that this may also have been the case for some hydrous phases observed in samples from the main sulfide zone at Unki Mine (Figs. 5, 8, and 10B).

At a temperature of 600 °C, equilibrium phase relations in the central portion of the ternary Fe-Ni-S system are dominated by MSS and pentlandite (Fig. 16). A wide range in composition is exhibited by the pentlandite, which coexists with the MSS (Fig. 16), with pyrite and/or vaesite being stable with the MSS along the S-rich boundary of the MSS (Hill, 1984). A narrowing of the MSS field occurs with falling temperature (Fig. 16). Naldrett and Kullerud (1967) investigated the isotherms that limit the solvus between the S-rich and S-poor limits of the solvus shown in Figure 16. The temperature of the breakdown of the MSS is interpreted to occur above 260–300 °C, and this temperature determines the stable coexistence of pyrite and pentlandite (Fig. 8B) in the Fe-Ni-S system (Naldrett and Kullerud, 1967). This 260–300 °C breakdown temperature of the MSS is thought to constrain the upper limits on the formation of the coexistence of pyrite and pentlandite (Hill, 1984), probably during hydrothermal alteration in the case of the main sulfide zone at Unki Mine. In the main sulfide zone at Unki Mine, only a couple of samples were observed with both pyrite and pentlandite, a likely indication that the bulk of the samples must have been formed at temperatures above 300 °C, resulting in the lack of the coexistence of pyrite and pentlandite. The paucity of pyrite in the main sulfide zone Unki Mine samples lends support to the formation of the main sulfide zone at temperatures above 300 °C.

Pentlandite compositions from the main sulfide zone at Unki Mine are plotted on an S-Fe-Ni ternary diagram in Figure 15A, and these samples plot above and below the pentlandite “field” (Kullerud et al., 1969). The pentlandites under study range in S from 39.5% to 46.7% and have near-constant Fe and Ni proportions (Fig. 16), with no major differences in the concentrations of primary and secondary pentlandites. Most Unki Mine main sulfide zone pentlandites (both primary and secondary) do not define a trend of pentlandites that were exsolved from pyrrhotite, but rather they define pentlandite compositions that are consistent with an origin between Ni-poor MSS (monosulfide-poor, MSS1) and nickel-rich (MSS2) pentlandites (Guo et al., 1999). Bushveld Complex pentlandites, for example, are nickel-rich and are MSS2 pentlandites (Fig. 15A).

It has already been noted from experimental investigations that the distribution of PGEs in base metal sulfides is inferred to be a consequence of sulfide liquid fractionation (Peregoedova et al., 2004; Naldrett, 2004; Mungall et al., 2005). The occurrence at or below detection limits of PGEs in the main sulfide zone samples investigated here probably indicates that PGEs that may have crystallized together with base metal sulfides during the sulfide liquid fractionation stage in the main sulfide zone were probably redistributed/remobilized during a later hydrothermal alteration event(s).

The relative shrinkage rates of the S-poor and S-rich limits of the MSS with temperature, as well as the position of the bulk compositions between the S-rich and S-poor stability limits (Fig. 16), determine the order of appearance of pentlandite and pyrite (Naldrett and Kullerud, 1967). For the S-poor and Cu-rich majority of samples from the main sulfide zone at Unki Mine, which plot below or within the pentlandite solvus (Fig. 16), pentlandite would have crystallized from temperatures as high 610 °C (Kullerud, 1963) to temperatures as low as the breakdown of MSS at 260–300 °C. In this case, pentlandite would have exsolved from the melt, making pyrite an unstable phase until the establishment of pyrite-pentlandite tie lines at temperatures below the 260–300 °C breakdown of the MSS (Fig. 16). This scenario is likely to have been the more widespread one given the abundance of pentlandite and the paucity of pyrite in the main sulfide zone samples at Unki Mine. In certain localized portions, however, the primary bulk composition would have fallen in the S-rich side of the MSS, which would have enabled pyrite to exsolve from the melt, resulting in pentlandite being an unstable phase until the establishment of pyrite-pentlandite tie lines at temperatures below the 260–300 °C breakdown of the MSS (Fig. 16).

Unlike in the PGE-enriched UG-2 chromitite layer of the Bushveld Complex, which is characterized by higher concentrations of both Pd and Rd in pentlandite and where the PGEs are interpreted to have formed due to magmatic differentiation (orthomagmatic processes; Junge et al., 2015), the low concentrations of PGEs in pentlandite in the main sulfide zone at Unki Mine (Table 2; Supplemental Table [^{footnote 1}]) may lend support to remobilization of the PGEs by hydrothermal fluids. Consequently, the reported occurrences of micrometer-sized discrete platinum group minerals in ores of the Bushveld Complex (Junge et al., 2014) are consistent with a dominantly orthomagmatic origin for the PGEs in the Bushveld Complex, which is not what we observed in the main sulfide zone at Unki Mine. Chaumba (2017) also presented evidence of hydrothermal alteration in the main sulfide zone at Unki Mine in the form of the close association of base metal sulfides with alteration phases such as amphibole and chlorite. The plagioclase websterite, which occurs at the top of the P1 pyroxenite layer of the ultramafic sequence, and which is overlain by gabbroic rocks of the mafic sequence, may have been formed due to compaction (Meurer and Boudreau, 1996). From studies of concentrations of trace elements such as P and Zr that are weakly compatible in cumulate minerals, Meurer and Boudreau (1996) observed that when a sharp drop in the density of cumulus minerals occurs, such as that encountered at the ultramafic sequence–mafic sequence boundary of the Great Dyke, the interstitial liquid distributions closely mimic those predicted by compaction models. The position of the feldspathic websterite is comparable to that of the mineralized zone of the Munni Munni Complex (with a stratigraphic section that is broadly comparable to the Great Dyke’s) that was investigated by Meurer and Boudreau (1996).

From their study of the origin of mineralization in several important layered intrusions, Peach and Mathez (1996) made broadly comparable observations, i.e., that mineralization in the main sulfide zone of the Great Dyke was not a single event. Peach and Mathez (1996) argued that the metal profiles in the main sulfide zone are not consistent with a model of continuous, single-process sulfide fractionation. They suggested that the mineralization in the main sulfide zone originated from processes that postdated the initial sulfide liquid or that concentration of the PGEs took place by an entirely different mechanism (Peach and Mathez, 1996). Observations from this work lend support to that hypothesis.

Mineral compositions of the Great Dyke can be modeled by fractional crystallization processes through the MELTS software (Gualda and Ghiorso, 2015) by using the parental magma composition of the Great Dyke (Wilson, 1982). Isobaric crystallization was modeled at pressure intervals of 500 bars from 1 to 10 kbar as temperature was lowered at 50 °C intervals starting at ∼1200 °C. All initial models were performed as equilibrium crystallization simulations at anhydrous water contents, a pressure of 2 kbar, a temperature of 1300 °C, and oxidation states of QFM +0.6 and QFM +0 (QFM is quartz-fayalite-magnetite). Mineral composition results were then compared with those obtained from the present investigation to determine if Unki Mine mineral compositions were affected by hydrothermal alteration. Calculations for temperature by assuming isobaric fractional crystallization upon cooling were conducted from 5.5 to 1 kbar. No mineral composition results of MELTS modeling were obtained at pressures of ≤1 kbar and ≥6 kbar, suggesting that crystallization of the Great Dyke likely occurred at pressures ranging from 1 to 5.5 kbar. Results of MELTS modeling that were obtained for pressures ranging from 1 to 6 kbar for orthopyroxene (Fig. 17A) showed that the average orthopyroxene Mg# (100 × Mg/[Mg + Fe²⁺]) composition of Unki Mine main sulfide zone samples (=81.8) was likely to have been attained at temperatures of at least 1150 °C and pressures on the order 2–2.5 kbar (Fig. 17A; Chaumba, 2017). Orthopyroxene compositions for the ultramafic sequence of the Great Dyke (Wilson, 1982) would have been crystallized at temperatures higher than 1150 °C (Fig. 17A).

For clinopyroxene, the average Mg# obtained for Unki Mine main sulfide zone samples of 87.4, which is higher than the modeled Mg#_CPX (Fig. 18B; Chaumba, 2017), may imply that this mineral was likely to have been hydrothermally altered (Fig. 17A). Clinopyroxene compositions reported by Wilson (1982) are likely to have been crystallized at temperatures ranging from 800 °C to almost 950 °C at pressures of 2–4 kbar (Fig. 17B). In addition to the orthopyroxene having crystallized earlier at higher temperatures than those at which clinopyroxenes crystallized, the Great Dyke clinopyroxenes likely crystallized over a wider temperature range (Figs. 17A and 17B). This observation is supported by the first appearance of cumulus orthopyroxene rather than clinopyroxene in the Great Dyke (Worst, 1960; Wilson, 1982; Wilson and Prendergast, 1989). Further, the lower crystallization temperatures shown in Figure 17B are also consistent with either the postcumulus nature of this mineral (clinopyroxene is a postcumulus phase in the P1 layer) in the ultramafic sequence or the temperature at which hydrothermal alteration occurred (Chaumba, 2017).

For plagioclase, the anorthite content (An = 100 × Ca)/[Na + Ca]) of plagioclase in the main sulfide zone at Unki Mine was 68.5, which is higher than the modeled values (Fig. 18A; Chaumba, 2017). Both Unki Mine measured An contents and modeled An contents are consistent with crystallization at high temperatures in excess of 1150 °C at relatively low pressures of 1–2 kbar (Fig. 18A). Some of the Unki Mine measured An contents are higher than those obtained from MELTS modeling pressures greater than 2 kbar, which we interpret to indicate hydrothermal alteration of the plagioclase. Although no olivine crystals occur in the P1 layer, as this mineral crystallized stratigraphically lower than the main sulfide zone, calculated olivine compositions are consistent with formation at temperatures of 1150 °C and higher, and at pressures ranging from 1 to 5.5 kbar (Fig. 18B). However, it is also interesting to note that some Great Dyke olivines may have been crystallized at temperatures of 800–930 °C and pressures of 2–4 kbar (Fig. 18B). These olivines, which crystallized at these relatively lower temperatures, are also interpreted to have been affected by the hydrothermal fluids that resulted in the concentration of PGEs. The pressures obtained from MELTS modeling of 1–5.5 kbar are comparable with pressures of 1–4 kbar that were obtained by Chaumba (2017) from thermobarometry calculations, lending support to the conclusion that the Great Dyke must have been emplaced at depths of 6.1–12.6 km (Chaumba, 2017).

The lack of anomalous enrichment of PGEs in base metal sulfides from the Shurugwi Subchamber of the Great Dyke is consistent with observations from other “layered intrusions” such as the Noril’sk (Barnes et al., 2006), the Bushveld Complex (Godel et al., 2007), and the Sudbury Complex (Dare et al., 2010), for example, which also lack such PGE enrichments in their base metal sulfides. These findings lend support to the involvement of postmagmatic processes in the redistribution of magmatic sulfide ores (the hydromagmatic model). However, in previous studies on different Ni-Cu-PGE deposits (Godel et al., 2007), distinct enrichments of Pd and Pt in chalcopyrite were not observed. Thus, postmagmatic processes are thought to play an important role in the redistribution of PGE in magmatic sulfide ores (Dare et al., 2010). For example, Barnes et al. (2006) and Dare et al. (2010) suggested that the enrichment of Pd in pentlandite is due to diffusion during its exsolution from the MSS. Small quantities of Pd, however, may originate from Pd in the original MSS structure, whereas most Pd can be derived from the nearby ISS by diffusion (Dare et al., 2010).

Unlike in the Platinova Reef of the Skaergaard intrusion, where zones of PGE enrichment and precious metal minerals are intimately associated with Cu sulfide globules mostly located at, or close to, silicate and oxide boundaries (Holwell et al., 2016), in the main sulfide zone of the Great Dyke at Unki Mine, this is not the case, possibly pointing to different origins of mineralization in the two layered intrusions. Although both the Great Dyke and the Skaergaard intrusion fall under the “offset reefs” classification, the mineralized horizons are hosted in different lithologies (ultramafic rocks in the case of the main sulfide zone of Great Dyke; mafic rocks in the Platinova Reef of the Skaergaard intrusion). In the Platinova Reef, precious metal minerals have been shown to be intimately associated with Cu sulfide globules, mostly located at, or close to, silicate and oxide boundaries (Holwell et al., 2016), whereas no elevated PGE concentrations in base metal sulfides were observed in this study.

Pyrrhotite, pentlandite, chalcopyrite, and pyrite were the sulfides encountered during the present study in the main sulfide zone at Unki Mine. Some of these base metal sulfides, which occur as inclusions in silicates and chromite, were most likely formed early in the crystallization history of the main sulfide zone. Other sulfides, which occur in association with hydrosilicates, were likely formed later during a late-stage event associated with hydrothermal alteration.

As is the case of the C1d chromitite layer in the Darwendale Subchamber of the Great Dyke, low concentrations of PGEs such as Pt, Pd, and Rh in base metal sulfides may imply that the PGEs in the main sulfide zone at Unki Mine are hosted in silicates and/or platinum group minerals. The broadly comparable trends, with minor variations across the main sulfide zone, of Fe in pyrrhotite, of Co and Ni in pentlandite, and of Cu in chalcopyrite, for example, are interpreted to reflect magmatic processes. Very low Co contents in Unki Mine main sulfide zone pentlandites indicate very limited Fe substitution by Co, and substitution of Ni by Co. The concentrations of metals such as Co, Ni, and Cu in sulfides tend to vary sympathetically, implying that they either are primary magmatic signatures, or they were likely to have been subsequently affected by the same hydrothermal processes. Depletions in some elements, which occur near the base and at the top of the main sulfide zone, however, suggest that these metals (and mineralization in the main sulfide zone) were likely to have been subsequently affected by hydrothermal processes. Positive correlations of elements like Cu and Fe in chalcopyrite from the main sulfide zone at Unki Mine, for example, indicate preservation of magmatic processes in the main sulfide zone samples at Unki Mine.

Chaumba gratefully acknowledges funding for this work from the University of North Carolina–Pembroke Office of Graduate Studies and Research that covered the cost of microprobe analyses. Thin sections costs were covered by a Summer Research Fellowship at University of North Carolina–Pembroke, which is also gratefully acknowledged. Anglo American Corporation Zimbabwe supported this research and contributed financially in several ways. We are extremely grateful to two anonymous reviewers for their careful and helpful reviews of this article.