Pesquisar no resumo do documento
The origin of chromitites and related PGE mineralization in the Bushveld Complex: new mineralogical and petrological constraints
A. J. Naldrett & Allan Wilson & Judith Kinnaird & Marina Yudovskaya & Gordon Chunnett
Received: 10 February 2011 /Accepted: 19 May 2011 # Springer-Verlag 2011
Abstract This article reports a study of chromitites from the LG-1 to the UG-2/3 from the Bushveld Complex. Chromite from massive chromitite follows two compositional trends on the basis of cation ratios: trend A—decreasing Mg/(Mg + Fe) with increasing Cr/(Cr + Al); trend B—decreasing Mg/(Mg + Fe) with decreasing Cr/(Cr + Al). The chromitites are divided into five stages on the basis of which trend they follow and the data of Eales et al. (Chemical Geology 88:261–278, 1990) on the behaviour of the Mg/Fe ratio of the pyroxene and whole rock Sr isotope composition of the environment in which they occur. Following Eales et al. (Chemical Geology 88:261–278, 1990), the different characteristics of the stages are attributed to the rate at which new magma entered the chamber and the effect of this on aAl2O3 and, in the case of stage 5, the appearance of cumulus plagioclase buffering the aAl2O3. The similarity of PGE profiles across the MG-3 and MG-4 chromitites that are separated laterally by up to 300 km and the variation in V in the UG-2 argue that the chromitites have largely developed in situ. Modelling using the programme MELTS shows that increase in pressure, mixing of primitive and fractionated magma, felsic contam-
ination of primitive magma or addition of H2O do not promote crystallization of spinel before orthopyroxene (in general they hinder it) and that the Cr2O3 content of the magma was of the order of 0.25 wt.%. Less than 20% of the chromite in the magma is removed before orthopyroxene joins chromite, which implies a >13-km thickness of magma for the Critical Zone. It is suggested that the large excess of magma has escaped up marginal structures such as the Platreef. The PGE profile of chromitites depends on whether sulphide accumulated or not along with chromite. Modelling shows that contamination of Critical Zone magma with a felsic melt will induce sulphide immiscibility, although not chromite precipitation. The LG-1 to LG-4 chromitites developed without sulphide, whilst those from the LG-5 upwards had associated liquid sulphide. Much of the sulphide originally in the LG-5 and above has been destroyed as a result of reaction with the chromite.
The Bushveld Complex (Fig. 1) is the largest known mafic/ ultramafic intrusion extending 450 km east–west and 350 km north–south. The country rocks consist mainly of quartzites, argillites and dolomites of the Transvaal Super- group, in particular the Pretoria Group, although Archean granitoids and greenstones form the footwall to the complex in the extreme north. Although the Bushveld is known primarily for its layered mafic rocks, the Bushveld igneous event comprises four distinctive igneous suites. The first suite consists of early mafic sills (Cawthorn 2002). The classic view is that these were followed by felsites belonging to the Rooiberg Group that now form the roof to much of the layered igneous series and comprise the second suite.The best age for the base of the Rooiberg is 2,057.3±
Editorial handling: C. Li
Electronic supplementary material The online version of this article (doi:10.1007/s00126-011-0366-3) contains supplementary material, which is available to authorized users.
A. J. Naldrett (*) :A. Wilson : J. Kinnaird :M. Yudovskaya : G. Chunnett School of Geosciences, The University of the Witwatersrand, 1 Jan Smuts Avenue, Braamfontein, 2000 Johannesburg, South Africa e-mail: [email protected]
M. Yudovskaya IGEM RAS, Staromonetny 35, Moscow 119017, Russia
Miner Deposita DOI 10.1007/s00126-011-0366-3
3.8 Ma (Harmer and Armstrong 2000), which is within error of the Scoates and Friedman (2008) Merensky Reef age of 2,054±1 Ma. Recently, Armitage et al. (2007) have found evidence for a surprisingly high proportion of mafic lavas and tuffs in the Upper Rooiberg near Mookgophong that might be related to the next major event, the introduction and crystallization of the Rustenburg layered series, along with related dykes and sills (see discussion towards the end of this paper). Repetitive major influxes of magma gave rise to the up to 10-km-thick sequence of mafic/ultramafic rocks constituting the Rustenburg Layered Series. Melting of roof rocks resulted in a series of granophyres, which occur at intervals along the upper margin of the layered series. The final igneous suite that belongs to the Bushveld igneous event is the Lebowa Granite Suite that intrudes into the centre of the mafic/ ultramafic rocks. Seismic and gravity data indicate that this has the form of a sheet-like body spreading out over the mafic rocks (Webb et al. 2004).
The layered series is now present in five major structures. Three of these—the Eastern Bushveld, the Southeastern Bushveld and the Western Bushveld—have
the form of semi-circular, basin-like lobes (Fig. 1). It is not certain whether these lobes have been closely interconnected at one time because the intrusion of the Bushveld granite now obscures any interconnections that may have existed initially, but the gravity and seismic data discussed by Cawthorn and Webb (2001) and Webb et al. (2004) are consistent in suggesting that the Western and Eastern lobes were originally contiguous. These two lobes show a full development of rock types from the Marginal Zone through to the Upper Zone. The Southeastern lobe is poorly exposed, but the gravity data suggest that it lacks the lower two zones. A fourth structure (Far Western Bushveld) lies west of the Pilanesberg. Here, marginal rocks are overlain by the Lower Zone and the Lower part of the Critical Zone (five chromitite horizons are preserved): Another fully developed lobe probably existed in this area, but if so, it has been mostly removed by erosion. The fifth structure, the Northern Limb, may well be part of another lobe, although the rocks exposed along it exhibit some differences to those of the other bodies. Rocks that have been equated (Hulbert 1983; Hulbert and von Gruenewaldt 1982; Maier et al. 2008; van der Merwe 2008) with the
Fig. 1 Map of the Bushveld Complex (modified after Eales and Cawthorn 1996) showing locations of drill holes sampled in this study
Lower and Critical Zones elsewhere are observed in the extreme south of the limb, although this correlation is not universally accepted (McDonald et al. 2005). The Main and Upper Zones are present along much of the limb, and a unit unique to this area, the Platreef, is present along the southern half.
The stratigraphy of the Rustenburg Layered Series is divided into five zones (Fig. 2). The Marginal Zone (0– 800 m thick) is formed of norite along with minor pyroxenite. The Lower Zone (800–1,300 m thick) is composed mainly of orthopyroxenites, harzburgites and dunites (Cameron 1978). Overlying this is the Critical Zone (1,300–1,800 m thick) that forms the principal focus of this paper. The base of the Critical Zone is marked by the incoming of cumulus chromite. The zone is divided into two parts (Cameron 1980, 1982): The Lower Critical Zone consists primarily of orthopyroxenites, chromitites and some harzburgites and the Upper Critical Zone is marked
by the incoming of cumulus plagioclase. The Critical Zone is overlain, in turn, by the norites, gabbros and anorthosites of the Main Zone (3,000–3,400 m thick), which are themselves capped by the ferrogabbros and ferrodiorites of the Upper Zone, 2,000–2,800 m thick (von Gruenewaldt 1973; Molyneaux 1974).
Chromitite horizons and related PGE concentrations
The Bushveld Complex contains 14 major chromitite horizons (21 named horizons in all; Fourie 1959; Cousins and Feringa 1964; Scoon and Teigler 1994; Schurmann et al. 1998; Mondal and Mathez 2007; Maier and Barnes 2008) together with numerous less developed seams that occur intermittently and have not been named. Not all named seams occur in all areas of the complex, but many are restricted to different apparent ‘compart-
Fig. 2 Distribution of chromi- tite seams in the Bushveld Complex. (adapted from Scoon and Teigler 1994)
ments’ or ‘sectors’. All are confined to the Critical Zone, except for seams in the Grasvaly area of the Northern Limb that occur in rocks thought to constitute the Lower Zone.
The chromitite layers are divided into three major groups (Fig. 2): the Lower Group comprising the LG-1 to LG-7; the Middle Group comprising the MG-0, MG-1, MG2a,b,c, MG-3, MG-3a, MG-4, MG-4a, and in some areas the UMG-1 and 2; and the Upper Group comprising the UG-1, UG-2 and, in some areas, the UG-3 and UG-3a (see Schurmann et al. 1998 for a recent overview). Subsidiary seams may occur in the hanging wall of the main seam and are known as leaders (e.g. to the UG-2), or in the footwall (e.g. to the LG-6 and UG-1). Most of the chromitite layers (Hatton and von Gruenewaldt 1987; Schurmann et al. 1998) can be traced for distances in excess of 100 km in both eastern and western limbs. Development of the different chromitites varies around the Bushveld Complex; in some areas, the Lower Group are well developed, and in others the Middle Group show better development. The UG-1 is known for the bifurca- tions of individual chromitite layers, although locally, the UG-2 and LG-5 may also show significant development of bifurcations. For both the UG-1 and UG-2, the combined thickness of the main and subsidiary seams is broadly constant along strike (Nex 2004).
Compositional variation of chromite
Compositional data on Bushveld chromitites have been presented, amongst others, by Scoon and Teigler (1994), Mondal and Mathez (2007) and Naldrett et al. (2009).
Variation in Mg/(Mg + Fe2+) versus Cr/(Cr + Al) ratio
Naldrett et al. (2009) have shown that the Mg/(Mg + Fe2+) versus Cr/(Cr + Al) ratio of Bushveld Complex chromites varies systematically with elevation (Fig. 3). They observed that chromites from the LG-1 to LG-4 chromitites defined a trend of increasing Cr/(Cr + Al) with decreasing Mg/(Mg + Fe2+) (referred to as trend A), that those from the LG-5 to MG-2 defined a trend of decreasing Cr/(Cr + Al) with decreasing Mg/(Mg + Fe2+) (referred to as trend B), and that variation in those from the MG-3 to UMG-2 Cr/Cr + Al) paralleled the LG-1 to LG-4 trend, but at a lower Mg/(Mg + Fe2) ratio (trend A). The area occupied by individual microprobe analyses taken at 3-cm intervals across a sample of the UG-2 chromitite from Waterval shaft is superimposed on the data for the Lower and Middle Group chromitites in Fig. 3. The Waterval UG-2 sample defines a similar trend to all of the chromitites from the MG-3 to UMG-2, albeit at a slightly higher Mg/(Mg + Fe2+) ratio.
Variation in vanadium across the Waterval shaft sample
In the UG-2 sample from Waterval shaft, the V content of the chromite in successive 3-cm slices (determined by microprobe) is constant at about 2,200 ppm in the lower half of the profile, but about halfway up shows a very progressive increase to about 2,800 ppm at the top (Fig. 4). Individual slices were analysed randomly and not in progression with height, therefore confirming that no systematic instrumental bias of the microprobe has oc- curred. This variation, which is regarded as very significant from a genetic point of view, is also discussed below.
Association of PGE with chromitite
The spatial association between chromitites and PGE enrichment in the Bushveld Complex, which was first noted by Hall and Humphrey (1908), has been noted subsequently by numerous other authors (e.g. Wagner 1929; Hiemstra 1979; Von Gruenewaldt et al. 1986; Lee and Parry 1988; Teigler and Eales 1993). The data on chromite compositions and PGE contents of most chromitite layers from many different localities in the Bushveld Complex that are referred to here have been drawn from Scoon and Teigler (1994), Mitchell and Scoon (2007), Naldrett et al. (2009) and Naldrett (unpublished data). Scoon and Teigler (1994) recognized that the LG-1 to LG-4 chromitites (their type 1a) had the lowest PGE contents and (Pt + Pd + Rh)/ (Ru + Ir + Os) ratios, the LG-5 to MG-2 (their type 1b) somewhat higher contents and ratios, and chromitites above this (MG-3 to UG-2,3) even higher values of both.
Overall PGE variations from the LG-1 to UG-2
The data of Naldrett et al. (2009), along with unpublished data for the UG-1 and UG-2, are shown in chondrite- normalised format in Fig. 5. The data shown here are only for massive chromitite, not including footwall or hanging wall seams, and the spread for individual locations reflects the zoning across the chromitite. It is seen that in agreement with Scoon and Teigler’s (1994) observation, the total PGE content rises almost progressively from the LG-1 to the UG-2, save for a decrease in the MG-4a and UG-1 chromitites. Of particular interest is the contrast between the low Pt and Pd contents for the LG-1 to LG-4 (no data are available for LG-3) and those for the LG-5 and higher chromitites, as remarked on by both Scoon and Teigler (1994) and Naldrett et al. (2009).
Vertical profiles through the chromitites
Vertical profiles through the LG-6, MG-3, MG-4, UG-1 and UG-2/3 chromitites are summarized in Table 1, and those
most pertinent to this paper are illustrated in Figs. 4 and 6. Very little variation in Pt concentration or key PGE ratios is observed across the LG-6 chromitite (Table 1). In contrast, marked variations are present in the MG-3 and MG-4 in different parts of the Bushveld Complex (Fig. 6). Pt (not shown in the figure) decreases upward across the MG-3, as do the Pt/Pd and Pt/Rh ratios. Pt increases over the lower 20 cm of the MG-4 and then decreases, as do the Pt/Pd and Pt/Rh ratios. The UG-1(Table 1) is a complex chromitite, with many bifurcations, but there is little systematic variation in Pt concentration, which is much lower than the overlying UG-2 or underlying MG-3 and 4 chromitites, or in the PGE ratios. Many workers (see Von Gruenewaldt et al. 1986; Maier and Barnes 2008) have remarked that the highest concentrations of Pt (and of the other PGE) in the UG-2 are found towards the base and top of the profile. This is also shown by our Waterval sample (Fig. 4) that,
despite being sampled at 3-cm intervals, shows no indication of the upward exponential decay described by Hiemstra (1979, 1985). Looking at our UG-2 profiles as a whole, there is no systematic variation in Pt/Pd ratio and only a slight change in the Pt/Rh ratio in the UG-2 of the Northwest and Eastern Bushveld Complex. The Rh/Ir ratio shows no systematic change in any of the chromitite profiles.
Variation in PGE ratios within the chromitites as a whole
The average ratios of Pt/Pd, Pt/Ru, Ru/Ir and Ru/Rh for each of the massive portions of chromitites from the LG-1 to UG-2 are shown in Fig. 7. Each point represents the average from one area of the Bushveld Complex. Ru/Ir and Ru/Rh show much less variation within a given chromitite than do Pt/Ru or Pt/Pd. The limited variation shown by the
Fig. 3 A cation plot showing variations in the Mg/(Mg + Fe2+) versus Cr/(Cr + Al) ratio of chromites from the massive portions of chromitites throughout the Bushveld Complex showing the two trends (A) and (B) that are discussed in the text. The different stages of chromite crystallization (stages 1–5) are compared with the data of Eales et al. (1990) on the Mg/(Mg + Fe2+) ratio of orthopyroxene and 87Sr/86Sr over the same stages of crystallization. In stage 1, chromite compositions follow trend A, during which there was rapid input of fresh magma and the magma in the chamber was becoming more primitive. Trend B, up to the end of stage 4, is one during which orthopyroxene was becoming generally less magnesian and 87Sr/86S more radiogenic, and, it is hypothesized, the crystallization of orthopyroxene was causing the activity of Al2O3 in the magma to increase. During stage 5, orthopyroxene continued to become more
enriched in Fe and the 87Sr/86Sr continued to increase, but plagioclase was or was close to being a liquidus phase, and its crystallization prevented any further increase in the activity of Al2O3; the chromite compositions follow trend A. Note also that the pyroxene En and 87Sr/86Sr data are presented with the lower strata at the top, not the bottom, so that they match the Mg/(Mg + Fe2+) versus (Cr/(Cr + Al) plot. It is seen that the Sr isotope data are consistent with no contamination occurring during stage 1, but that increasing contam- ination is indicated during subsequent stages. Data from the Waterval sample of the UG-2 are overlain on the data for Lower and Middle Group chromitites. See ESM Fig. X1 for further discussion of Mg/ (Mg + Fe2+) versus Cr/(Cr + Al) ratio variation in the Bushveld chromites
Fig. 5 Chondrite-normalised PGE profiles for the massive segments of all chromitites from the LG-1 to UG-2. The data for each location are expressed on the figure by an area that covers all of the profiles for a particular chromitite from a particular area (where the profile is
shown as a single line, the whole chromitite is covered by a single sample.). The chondritic concentrations used for the normalization are those of Naldrett and Duke (1980) (Os=515; Ir=540; Ru=690; Rh= 200; Pt=1,015; Pd=540; Au=150, all in parts per billion)
Fig. 4 Vertical variation in parts per billion Pt and parts per million V in chromite from the Waterval sample
Ru/Ir ratio is to be expected since these elements traditionally track together. The Ru/Rh ratio is perhaps more surprising since Rh is not regarded as a member of the IPGE group, although it certainly follows Ru in the Bushveld Complex chromitites (see discussion concerning the effect of As below). The wide variation in the Pt/Ru ratio is suggestive that these two elements are controlled by different mechanisms, as is discussed below.
Origin of chromitites
Existing models for chromitite formation
There are a number of ideas prevalent in the literature concerning the origin of chromitites. The principal problem is that of explaining how massive layers of chromite can form, undiluted or minimally diluted by other phases such as olivine, orthopyroxene and, in some cases, plagioclase. In our view, these models can be divided into two groups, those calling for the chromite to form outside the portion of the intrusion now preserved and then to be transported into position as a chromite-rich slurry (‘offstage’ models) and those calling on the chromite to crystallise in situ (‘onstage’ models).
Two recent offstage models are illustrated in Fig. 8a, b. Model A was proposed by Eales (2000) and endorsed by Mondal and Mathez (2007) and Voordouw et al. (2009). It calls upon chromite and orthopyroxene to have crystallised in a staging chamber in cotectic proportions and then to have been winnowed out from the orthopyroxene–chro-
mite–magma mixture during subsequent introduction to the Bushveld chamber, with chromite spreading out across the cumulate–magma interface as a slurry. One of the principal arguments that led Eales (2000) to propose this model was his calculation that the volume of magma responsible for the chromite in the Critical Zone must have been 12 times the volume of the Zone itself, an observation that had previously been made by Cawthorn and Walraven (1998). Mondal and Mathez (2007) showed that the Cr content of orthopyroxene above and below the UG-2 chromitite was very similar, which they argued was inconsistent with the onstage crystallization of a significant amount of chromite from the magma. Voordouw et al. (2009) described intrusive relationships between the chromitites (particularly the UG- 1) and enclosing rocks, which favoured introduction of the chromite as a slurry.
Model B, proposed by Maier and Barnes (2008), assumes that initially, the Bushveld Complex was much more extensive than at present and that, as crustal loading caused the centre to subside, partially consolidated chromi- tites became remobilized (they referred specifically to the UG-2) and slid downslope, causing a thickening of the preserved remnants.
An onstage model requires chromite to be the sole liquidus phase, and there have been many suggestions as to how this can be achieved (Fig. 8c). Ulmer (1969) and Cameron and Desborough (1964) suggested that this was caused by an increase in fO2. Lipin (1993), drawing on the experimental studies of Sen and Presnall (1984) and Onuma (1988), suggested that an increase in pressure moved the
Table 1 Summary of variations across profiles across chromitites
Pt (ppb) Pt/Pd Pt/Rh Rh/Ir
LG-6 Uniform Uniform Uniform Uniform
200–400 JGD 1–2, Mill approx. 2, NGD 8–10 All 2 to 3 All 1 to 2
MG-3 Decreases upward Decreases upward Decreases upward Uniform
MAR 2,100 to 1,000, MN 2,200 to 1,000 MAR 2.7 to 1, MN 5.6 to 2.5 MAR 5 to 3, MN 5 to 3 All 2 to 3 Mill 1,600 to 600 Mill 8 to 1.5 Mill 4.5 to 3
MG-4 Increases upward Increases over lower 10–30 cm Increases over lower 80 cm Uniform 1.5 Then decreases over upper 20 cm Then decreases upward Then decreases over upper 80 cm
MAR 500 to 2,000, no decrease MAR 2.5 to 4, then to 1.5 MAR 4 to 7.5, then to 2.5
MN 700 to 2,000, then to 500 MN 4 to 8, then to 3 MN 2.5 to 10, then to 3
Mill 700 to 2,000, then to 700 Mill only decrease 15 to 7 Mill only decrease 12 to 3
UG-1 Uniform Uniform Uniform Uniform
UG-2 Highest at base, then decreases, levelling out in centre with another spike near but not at top
Relatively constant Increases very slightly Uniform
Base Centre Top
SW Bush 6,800 2,400 4,400 SW Bush 2 SW Bush 4 SW Bush 2–3
NW Bush 8,000 2,300 6,500 NW Bush 3 NW Bush 4–5 NW Bush 2–3
E Bush 6,700 1,800 4,400 E Bush 1–2 E Bush 4–5 E Bush 2–3
Key to sectors: JGD northeast, MH east central, MAR east south of Steelpoort fault, MN southwest, east, Mill southwest west, NGD northwest
boundary of the spinel field so that a magma crystallising olivine, orthopyroxene, or lying on the cotectic between either of these minerals and spinel would find itself in the field of spinel alone. Cameron (1980) had previously called upon this mechanism, and subsequently, it was invoked by Cawthorn (2005). Following his work on the Muskox Complex, Irvine (1975) suggested that magma in the
intrusion mixed with a felsic melt that existed at the top of the intrusion and that the hybrid initially had chromite as the sole liquidus phase. Kinnaird et al. (2002), Nex (2004) and Kottke-Levin et al. (2009) have called upon this mechanism. Irvine (1977) subsequently had doubts about the efficacy of his 1975 model and proposed that the mixing of an injection of primitive magma (his proposal
Fig. 6 Vertical variation in Pt/Pd and Pt/Rh for massive chromitite from the MG-3 (a) and MG-4 and MG-4A (b) chromitites in the Rustenburg (BRK SP-04), Mooi Nooi (TF 21) and Mareesburg (8JT7)
areas of the Bushveld Complex. Note: Mareesburg is approximately 270 km east of Rustenburg
referred to primitive magmas with olivine, not orthopyrox- ene as the first liquidus phase) with the fractionated residue of the same magma residing in the magma chamber would cause chromite to appear on the liquidus ahead of olivine. This proposal has been adopted and applied to the Bushveld chromitites (where orthopyroxene is usually the first liquidus phase of the magma responsible for the Critical Zone) by Eales et al. (1986), Scoon and Teigler (1994) and Naldrett et al. (2009).
The vertical profiles through the MG-3, MG-4 and UG-2 chromitites described above place constraints on the mode of emplacement of these units. Referring first to the progressive increase in V content in the upper half of the Waterval sample (Fig. 4), this is strongly supportive of the chromite having crystallised and settled in situ. If the chromite had been ‘winnowed out’ of a suspension of orthoproxene and chromite containing only 3% chromite and then introduced as a slurry composed of chromite alone, no such progressive variation would have developed. The same would be true if the chromite had been deposited over a wider area of the Bushveld Complex, an area that has now been removed by erosion, and had slumped down dip towards the centre of the intrusion.
The similar variations in Pt/Pd and Pt/Rh ratio (Fig. 6) and Pt distribution across profiles through the MG-3 and MG-4, which are separated by as much as 300 km, also argue for a progressive in situ deposition of PGE, not emplacement in the jumble of a slurry. On the basis of these considerations, our conclusion is that the chromitites formed primarily ‘on stage’, i.e. in their present position.
In the following section, we apply the programme MELTS (Ghiorso and Sack 1995) to place constraints on mechanisms for the development of monomineralic chro- mite cumulates.
Results of modelling using MELTS
Eales (2000) argued that 0.15 wt.% Cr2O3 was the maximum that is likely to be contained in U-type magma of the composition proposed by Barnes and Maier (2002) (Table 2). This estimate is close to the Cr2O3 content (0.142 wt.%) of the parent magma to the Bushveld Complex proposed by Davies et al. (1980). However, Harmer and Sharpe (1985) found the average of three samples of their B1 magma (equivalent to the initial Critical Zone magma) to be 0.20±0.01 wt.% Cr2O3, Curl (2001) found his three samples of analogous magma to average 0.19±0.06 wt.%, and Allan Wilson (2011, personal com- munication) found that four samples of a direct chill to the Lower Zone averaged 0.24±0.04 wt.% Cr2O3.
Application to Bushveld compositions
We have used MELTS to evaluate the effect of variable proportions of Cr2O3 in the Barnes and Maier (2002) magma and also to investigate which of the proposed mechanisms that are discussed above is relevant to the formation of chromite cumulates. Our approach has been to determine the temperature of appearance of spinel (the composition of the spinel is not always that of chromite) and orthopyroxene on the liquidus as a function of fO2. If a spinel cumulate is to form, spinel must appear before orthopyroxene. Figure 9a, b shows hypothetical diagrams to aid in the interpretation of our results that are shown in Fig. 10a–d. It is assumed, in line with many authors (Reynolds 1985; Murck and Campbell 1986; Ballhaus and Sylvester 2000; Tegner et al. 2003) that the Bushveld crystallised at an fO2 of QFM or less. In Fig. 9a, the magma is at an fO2 and the temperatures of the appearance of spinel
Fig. 7 Average Pt/Ru, Pt/Pd, Ru/Ir and Ru/Rh ratios for massive chromitites from the LG-1 to UG-2/3. The spread for a given chromitite horizon reflects the variation from area to area
and orthopyroxene are such that, on cooling, the magma will first crystallise orthopyroxene and then orthopyroxene and spinel together. No chromite cumulate will form. Figure 9b shows what must happen if chromite is to crystallise alone, before orthopyroxene. The chromite curve must move to the left, or the temperature of the orthopyr- oxene curve lowered. The plots comprising Fig. 10a–d are the results of our runs with MELTS (magma compositions used are shown in the Electronic supplementary material
(ESM) available online). Figure 10a shows that for spinel to precede opx, the Cr2O3 content of the magma must be >0.2 wt.% and that increasing pressure has little effect on the order of crystallization. Figure 10b shows that the addition of H2O adversely affects the appearance of spinel before orthopyroxene. Figure 10c shows that mixing primitive magma with fractionated magma also has an adverse effect. Figure 10d shows that mixing Critical Zone magma with felsic melt has a slight adverse effect. Our
Fig. 8 Summary of suggestions in the literature for the formation of massive chromitites, i.e. ‘off- stage’ models (a, b) and ‘on- stage’ models (c)
conclusion on the basis of MELTS is that none of the proposed mechanisms promote the appearance of spinel before orthopyroxene. We conclude that the Critical Zone magma must have contained more than 0.2 wt.% Cr2O3 and that the widely assumed maximum Cr2O3 content of 0.15 wt.% (that is close to that of the peripheral sill proposed by Davies et al. 1980, as the primary magma for the Bushveld) is lower than that responsible for the Critical Zone. This is not unexpected since removal of all the early crystallising spinel in our MELTS runs only involves about 1 wt.% crystallization, and it is quite likely that a magma penetrating cool country rocks as a thin sill would fractionate to this extent, leaving a scattering of chromite cumulate in its wake and ending up on the chromite– orthopyroxene cotectic.
Figure 11a is a plot of the decrease in the Cr2O3 content of the Critical Zone magma whilst crystallising spinel alone, before orthopyroxene joins it on the liquidus. It is seen that at fO2 < QFM, Cr2O3 is only removed from the
magma containing 0.25 wt.%, not the other two, and that this amounts to about 0.04 wt.%, i.e. about 20% of the Cr2O3 contained in the magma The amount of Cr2O3 removed in the situation involving the hybrid magma composed of Critical Zone magma contaminated by felsic melt is only slightly lower (Fig. 11b). The fact that only a fraction of the contained Cr2O3 is removed whilst a chromite mono-cumulate forms has implications as to the amount of magma required to form it. A 70-cm-thick chromitite containing 80% chromite and 35% wt.% Cr2O3 will require 150 vertical metres of magma containing 0.25 wt.% Cr2O3, provided that all of the Cr2O3 is removed. If 0.04 wt.% or less is removed, as our calculations with MELTS indicate, the column of magma must have been at least 938 m. Given that there are about 9.5 m of massive chromitite (Eales 2000) in the Bushveld, this implies 13 km of magma. BUT there are only about 1.2–1.7 km of Critical Zone cumulates, as Maier and Teigler (1995) and Eales (2000) have pointed out. There-
U-type 10% fract U 20% Fract U Av upper crusta
SiO2 55.47 55.42 55.49 67.36
TiO2 0.37 0.41 0.45 0.54
Al2O3 12.46 13.61 15.05 14.20
Cr2O3 b 0.15, 0.20, 0.25 0.15 0.09
FeOc 9.08 9.15 9.05 4.12
MnO 0.21 0.23 0.26
MgO 12.56 10.31 7.71 2.32
Cao 7.24 7.94 8.80 4.28
Na2O 1.51 1.68 1.89 3.59
K2O 0.76 0.85 0.95 3.58
Table 2 Magma compositions used in MELTS modelling
aWedepohl (1995) b U magma = 0.15, U2 magma = 0.20, U3 = 0.25 wt.% Cr2O3 c Total Fe as FeO
Fig. 9 Figure to explain Fig. 10. In a, the magma is at a temperature and fO2 such that, on cooling, orthopyroxene will crystallise first, followed by orthopyroxene + chromite together in cotectic propor-
tions. As is shown in b, in order for chromite to crystallise alone, either the spinel ‘in’ curve must be moved to the left (lower fO2) or the orthopyroxene ‘in’ curve must move to a lower temperature
Fig. 10 Temperature at which spinel and orthopyroxene crystallise from Barnes and Maier’s (2002) estimated composition of the Critical Zone magma as a function of fO2 and Cr2O3 content. U magma contains 0.15 wt.% Cr2O3, U2 magma 0.2 wt.% and U3 magma 0.25 wt.%. All modelling has been undertaken using the programme MELTS (Ghiorso and Sack 1995). a Effect of Cr2O3 content of the magma and pressure—note that pressure has only a very slight effect on the fO2 at which spinel crystallises after orthopyroxene. In the case of the U and U2 magmas, spinel does not crystallise before orthopyroxene when the fO2 is below QFM. b The addition of water also increases the fO2 at which spinel crystallises after orthopyroxene. Modelling undertaken at 1 kb. c Results of mixing the primary magma with a fractionate of the same magma—mixing of U3 magma with an
increasing proportion of fractionate increases the fO2 at which spinel crystallises after orthopyroxene, i.e. mixing with a fractionate is not a mechanism for promoting spinel as the sole liquidus phase, although such mechanism still permits spinel to remain on the liquidus alone at QFM or below, so long as the proportion of fractionate is relatively low. All modelling was undertaken at 1 kb. d Results of mixing U3 magma with a felsic contaminant (average upper continental crust; Wedepohl 1995). Increasing proportions of the contaminant increase the fO2 at which spinel crystallises after orthopyroxene, although not to the extent of negating spinel crystallising as the sole liquidus phase below fO2 = QFM. It does show that felsic contamination is not a mechanism for promoting spinel as the sole liquidus phase. Modelling undertaken at 1 kb
fore, how and where has the missing magma gone? (see “Exit channels for the Critical Zone magma” below).
Figure 12 compares the trivalent cations, Fe3+, Cr and Al, that were derived from the MELTS runs with the field occupied by the actual observations of Naldrett et al. (2009). The high oxygen fugacities required for spinel to be present on the liquidus before orthopyroxene when the Cr2O3 content is ≤0.20 wt.% gives rise to spinels very rich in Fe3+. It is only when the Cr2O3 content is >0.2 wt.% and fO2 less than or equal to QFM that the MELTS-generated spinel compositions approach the observed compositions.
The question of how much Cr2O3 a basaltic magma can dissolve was addressed in the experimental study of Poustovetov and Roeder (2000). They showed that at an fO2 = QFM, their ‘401’ basalt could dissolve 0.275 wt.% Cr2O3 when saturated with chrome spinel. The concentra- tions proposed above for Critical Zone magma are therefore consistent with experimental observations.
A note on more olivine-rich compositions than those of the Bushveld
Figure 13 is schematic and has been adapted from the figure of Irvine (1977) in which he explained his mechanism of magma mixing to promote the precipitation of chromite alone. We have modelled the scenario in this figure using MELTS. To do this, we have taken the Barnes and Maier (2002) composition and added 20% olivine and
raised the Cr2O3 content to 0.3 wt.% (magma A). We have fractionated this magma using MELTS at fO2 values of QFM, QFM-1 and QFM-2 until chromite joins it on the liquidus (magma B—note that the composition of magma B changes with the fO2 at which the fractionation proceeds). The lower the fO2, the greater was the degree of fractionation required to reach the olivine–chromite cotec- tic). We have then fractionated magma B along the olivine– chromite cotectic until orthopyroxene replaced olivine (magma C). The two end members (magmas B and C) were then mixed in different proportions and fractionated. In all cases, chromite was the first phase to appear on the liquidus, followed in most cases by olivine (the exception is that of the mixture richest in the C composition, 95% C and 5%B, from which at fO2’s of QFM-1 and QFM-2, orthopyroxene rather than olivine followed spinel). The temperatures of the appearance of the phases for the three fO2 investigated are shown in Fig. 14. This diagram supports the conclusion of Irvine (1977) that mixing of primitive and fractionated liquids, both on the olivine/ orthopyroxene–chromite cotectic, will promote the crystal- lization of chromite alone. However, the MELTS modelling shows that the amount of chromite removed before it is joined by olivine/orthopyroxene is very small, i.e. that the curvature of the olivine–chromite cotectic boundary is slight. This is illustrated in Fig. 15 and in ESM Table 1 in which the decrease in the Cr2O3 content of the liquid whilst chromite is crystallising alone is very small (<5.5% of the
Fig. 11 Results of modelling using the programme MELTS in which the amount of Cr2O3 removed from the magma before orthopyroxene joins chromite on the liquidus is plotted against fO2. a Effect of the Cr2O3 content of the magma and increasing confining pressure (note that increasing confining pressure decreases the amount of Cr2O3
removed at a given fO2, i.e. it reduces the stability field of chromite, which is contrary to the proposal of Lipin 1993). b Mixing U3 magma with average upper continental crust (Wedepohl 1995) has only a slight adverse effect on the amount of Cr2O3 removed in comparison with the unhybridised magma
total in the magma B–C mixture), so small as to require volumes of magma that would appear to be unrealistically large (of the order of 4 km) to produce a 70-cm chromitite layer.
The question of the association of PGE with chromitites
Before addressing this question directly, it is necessary to consider how the composition of chromite constrains its association with PGE.
Mg/(Mg ± Fe2±) versus Cr/(Cr ± Al) variation in chromite
As indicated above, the data of Naldrett et al. (2009) for massive chromitites from the Bushveld Complex show two trends on an Mg/(Mg + Fe2+) versus Cr/(Cr + Al) variation diagram, identified as A and B in Fig. 3. Hatton and von Gruenewaldt (1985) recognized trend A in their data for chromitites from the Bushveld Complex and equated it to
re-equilibration with adjacent silicates; however, it is only their analyses of disseminated chromite that show the trend, and their data for massive chromite are clustered together. Barnes and Roeder (2001) recognized trend A in their comprehensive data set for massive chromitites from continental layered intrusions and termed it the ‘CrAl’ trend; they noted that it involved essentially no change in Fe3+/(Fe3+ + Cr + Al) ratio and an increase in weight per cent TiO2 with decreasing Mg/(Mg + Fe
2+). Naldrett et al. (2009) showed that the reciprocal exchange reaction of Allan et al. (1988) involving the substitution of Cr and Fe2+
for Al and Mg between spinel and liquid will have a major effect on the Mg–Fe exchange partition coefficient between spinel and liquid (see ESM Fig. X1).
The explanation of Naldrett et al. (2009) for trend B is illustrated in Fig. 3. Here, the variations in chromite composition are compared with the data of Eales et al. (1990) for variation in the Mg/(Mg + Fe2+) ratio of orthopyroxene over the same intervals as those over which
Fig. 12 Relative proportions of trivalent cations, Fe3+, Cr and Al resulting from the MELTS modelling compared with the field occupied by the Bushveld chromites from massive chromitites observed by Naldrett et al. (2009). The two identical symbols for any set of conditions reflect the composition of spinel when it appears
the liquidus and its composition when orthopyroxene joins it on the liquidus. The latter is always less rich in Fe3+ than the former. It is seen that the only modelled compositions that overlap the observed field are those for a magma containing 0.25 wt.% Cr2O3 at fO2<QFM
different groups of chromitites were forming and with variations in the 87Sr/86Sr over the same interval. Stage 1, involving formation of the LG-1 to LG-4, is one of a steady upward increase in Mg/(Mg + Fe2+) of the orthopyroxene, that Eales et al. (1990) interpret as the consequence of an input of primitive magma that overrode the effect of
fractional crystallization. The Sr isotopic ratio also remains constant throughout this stage. Stage 2 (LG-4 to LG-6) is one involving a decrease in Mg/(Mg + Fe2+) of orthopyr- oxene, over which, in the view of Eales et al. (1990), fractional crystallization overrode the effect of any input of magma. Stage 3 (LG-6 to LG-7) is a short interval over which the input of new magma overrode the effect of fractional crystallization, whilst stages 4 and 5 are charac- terized by fractional crystallization overriding the input of fresh magma. The systematic increase in 87Sr/86Sr over stages 2–5 is explicable if the new magma influxes became increasingly crustally contaminated before or during ascent.
Naldrett et al. (2009) considered how the Al2O3 content of the magma could have varied during the formation of the Lower Critical Zone and the lowermost part of the Upper Critical Zone. They argued that in stage 1, during which input of magma was exceeding the effect of crystallization of olivine and orthopyroxene, Al2O3 content in the magma was not changing substantially, perhaps even decreasing, and that during this stage trends in chromite composition were largely controlled by the reciprocal exchange reaction referred to above, resulting in trend A. Apart from the short interval represented by stage 3, stages 2–4 are dominated by a decrease in Mg/ (Mg + Fe2+), which is attributed primarily to fractional crystallization of orthopyroxene. This would have resulted in a substantial increase in Al2O3 in the magma, giving rise to the decrease in the Cr/(Cr + Al) ratio of the chromitites with decreasing Mg/(Mg + Fe2+). This trend did not continue into stage 5 because the boundary between stages 4 and 5 (the boundary between the MG-2 seams and the MG-3, and also the boundary between the Lower and Upper Critical Zones) is marked by the first appearance of cumulus plagioclase in the Critical Zone. As a result, the activity of Al2O3 in the magma was
Fig. 14 Temperature of appearance of chromite and olivine/orthopyr- oxene over the fractionation path a–b (see Fig. 13) as a function of the fO2 and proportions of magmas B and C in the mixture
Fig. 13 Schematic representation of Irvine’s (1977) olivine–chro- mite–silica diagram, modified to show the stages of crystallization and mixing modelled here using MELTS. The Barnes and Maier (2002) Critical Zone magma was combined with 20% olivine (Fo90) and additional Cr2O3 to bring this to 0.3 wt.% to produce magma A. This was fractionated under controlled fO2 (QFM, QFM-1, QFM-2) until chromite joined olivine on the liquidus to produce magma B. Magma B was then fractionated at controlled fO2 until orthopyroxene replaced olivine (magma C). Magmas C and B were then mixed in different proportions (mixing line is blue dotted line) and fractionated at controlled fO2 along the path a–b (i.e. until olivine, or, in rare cases (see text) orthopyroxene joined spinel). Temperatures of the appear- ance of phases on the liquidus and the decrease in the Cr2O3 content of the magma along the path a–b are shown in Figs. 14 and 15
Fig. 15 Amount of Cr2O3 removed from the magma during fractionation from a to b (see Fig. 13) as a function of the fO2 and proportions of magmas B and C in the mixture
essentially buffered by plagioclase crystallization so that trend B terminates at this level, and any trends in chromite composition above this are dominated by the reciprocal exchange reaction (trend A).
Detailed compositional variation across the Waterval sample
The progressive upward increase in V concentration across this sample (Fig. 4) is unrelated to any variation in the Pt concentration of the sample. V is clearly compatible in chromite (Barnes and Maier 2002 estimated that the Critical Zone magma contained 179 ppm V, giving a Dspinel–magma= 12–16 for the Waterval sample) so that fractionation of chromite would lead to an upward decrease in V, not an increase. It is concluded that the upper part of the UG-2 was the consequence of chromite crystallising from a magma that was either becoming progressively richer in V or whose composition was changing in a manner that increased the partition coefficient of V into chromite from about 12 to about 16. In general, the average V content of chromites in Bushveld chromitites increases progressively upward (as does Ti), which is attributed to the increasing degree of fractionation of the magma(s) responsible for them Naldrett (unpublished data), but there is no such correlation between Ti and V in the Waterval sample. Canil (1999) has shown that the partitioning of V between spinel and basalt/komatiite magma at f02 = QFM is strongly dependent on the composition of the melt, although his experiments pertain to olivine-rich melts and are not directly applicable to Bushveld compositions. Thus, we prefer the concept of a changing partition coefficient over change in the concentration of V in the magma. The change in partition coefficient may be due to a collapse in the boundary between two convecting magma layers, as is discussed below and also illustrated in Fig. 17.
Provenance of PGE in chromitites
It has long been remarked (Hiemstra 1979; McCandless et al. 1999; Naldrett et al. 2009) that the PGE/sulphur ratio of most Bushveld Complex chromitites is much higher than is found in most magmatic sulphide deposits. Naldrett et al. (2009) showed that the Pt/S ratios for the chromitites are an order of magnitude higher than those of the Merensky samples, despite the latter having much higher ratios than most magmatic sulphides. The reason for the very high PGE/S ratios of Bushveld Complex chomitites has pro- voked considerable debate. Many have suggested that the PGE were contributed by sulphide, which has subsequently been lost (Maier and Barnes 2008; Merkle 1992; Naldrett and Lehmann 1988), whilst others have suggested that the PGE are concentrated by other phases such as Fe–PGE
alloys that may piggyback on settling chromite grains (Hiemstra 1979) or are the consequence of PGE concen- tration in sulphide by ascending fluids (Boudreau and McCallum 1992). As remarked above and illustrated in Fig. 7, the available data suggest that Pt and Pd were controlled by a different process from that responsible for much of the Rh, Ru, Ir and Os concentration. Naldrett et al. (2009) contended that the Pt and Pd, along with some Rh, Ru, Ir and Os, were contributed by the segregation of small amounts of sulphide liquid along with chromite, whilst chromite grains themselves concentrated a significant proportion of the IPGE, largely by encapsulating laurite and alloys that were present in the magma. The high observed PGE/S ratios are the consequence of the bulk of the sulphide having been lost subsequently. They suggested that the difference between PGE profiles of the LG-1 to LG-4 and overlying chromitites is because sulphide liquid did not segregate in the case of the LG-1 to LG-4, but did so in the higher chromitites.
In supporting their case for sulphide loss, Naldrett et al. (2009) recalled Naldrett and Lehmann’s (1988) observation that at temperatures between 1,597°C (melting point) and 900°C, magnetite is non-stoichiometric, showing a signif- icant range of solid solution towards hematite, i.e. it contains a considerable number of vacant Fe2+ (depending on the fO2 at which it crystallised). On cooling from a typical mafic magma liquidus temperature (about 1,250°C) to 900°C, magnetite would have to gain additional Fe to fill the Fe2+ vacancies. Naldrett and Lehmann (1988) demon- strated thermodynamically that the magnetite component of chromite would behave in the same way and showed that the source of this Fe could be magmatic sulphide liquid occurring interstitial to the grains of chromite in chromitite. Such a reaction would raise the S/metal ratio of any remaining sulphide, eventually reaching values of fS2 that were unsustainable, with the consequent loss of S to the surroundings.
Naldrett et al. (2009) proposed that there are two provenances for PGE associated with chromitites: (1) those (predominantly Ru, Ir and Os as laurite and alloys) trapped within growing chromite grains and (2) those (Pt, Pd and some IPGE) associated with initial interstitial sulphide liquid. With respect to Ru, Ir and Os, both Brenan and Andrews (2001) and Bockrath et al. (2004) have shown that laurite can crystallise from mafic magma at sulphur fugacities below that consistent with sulphur content at sulphide saturation (SCSS; Shima and Naldrett 1975). In fact, both Brenan and Andrews (2001) and Bockrath et al. (2004) emphasize that if sulphide saturation occurs, Ru, Ir and Os will partition into the sulphide liquid rather than form laurite. A continuing problem is that it has not been shown that laurite will form when these metals are at the relatively low concentrations found in natural melts.
McLaren and de Villiers (1982) found that laurite in the UG-2 contained substantial amounts of Ir and Os. Consid- eration of data on the association of laurite with chromitites from the MG-1 to UG-1 (Merkle 1992) indicates that 80% is present as inclusions in chromite, 15% in silicate or at silicate–chromite interfaces, and 4% associated with base metal sulphides. Maier et al. (1999) also found a very strong association between laurite and chromite. It is logical to conclude that growing chromite grains attracted and enclosed laurite that was present in the magma, possibly catalyzing its growth at the margins of such grains, and that the PGE incorporated in this way were protected and not susceptible to loss as interstitial magmatic sulphides were destroyed. In addition to laurite, very fine particles of IPGE alloy may be present. Finnigan et al. (2008) found that in their experiments involving chromite and natural basaltic liquids, the localised reduction front around growing or, in other cases, re-equilibrating chromite crystals caused the solubility of PGE to fall below their concentration in the magma, with the result that PGE alloys crystallised and were subsequently encapsulated within enlarging chromite grains. They equated their results with an earlier prediction by Mungall (2002) that this would occur.
Rh presents a special problem. Most (but not all) podiform chromitites in ophiolite complexes exhibit high (Ru + Ir + Os)cn/(Pt + Pd)cn
1 ratios (see the summary in Naldrett et al. 2009), as do the LG-1 to LG-4 chromitites of the Bushveld Complex. However, ophiolite complexes typically show (Ru/Rh)cn ratios >>1, whilst in the Bushveld Complex, ratios are significantly lower. Rh is generally regarded as a PPGE (i.e. it follows Pt and Pd in its geochemical behaviour), but clearly it behaves differ- ently in the chromitites of the Bushveld Complex. In his study of the Middle Group and UG-1 chromitites, Merkle (1992) found that one of the most common associations of Rh was with Pt in arsenosulphide minerals. McDonald (2008) has described PtAs2 (sperrylite), IrAsS (irarsite) and RhAsS (hollingworthite) inclusions in olivine in mafic/ultramafic intrusive complexes near Lavatrafo, Madagascar (Ohnenstetter et al. 1999). Whilst he con- cluded that they crystallised early from a sulphide liquid with a high As/S ratio and then became detached, they could have formed before sulphide immiscibility occurred as direct precipitates from a silicate magma with a higher than normal As content. In either case, direct crystalliza- tion from silicate magma or early crystallization from a sulphide liquid, it is possible that a higher concentration of As in the Bushveld Complex magma in comparison with that in magmas responsible for chromitites in ophiolite complexes can explain the difference in the mineralogical setting and thus differing behaviour of Rh. The higher As
content of the Bushveld magma as compared with magmas responsible for ophiolitic chromitites may be due to the crustal contamination that has undoubtedly affected all Bushveld magmas.
Formation of immiscible sulphide
Naldrett et al. (2009) modelled the effect of the mixing of an influx of primitive magma with resident magma in the chamber and showed that this can induce sulphide saturation in some cases. In order to cause this to happen, they had to assume that the ascending magma differenti- ated about 10% on ascent. This maintained the S content of the magma at about the saturation level during the ascent since a decrease in total pressure results in an increase in the SCSS of a mafic magma, and this was sufficient to cause sulphide saturation when the fresh magma mixed with that in the chamber. They found that if the ascending primitive magma rose rapidly without fractionating, the increase in the SCSS with decreasing pressure (see Li and Ripley 2009) meant that the mixing of the two magmas did not give rise to sulphide saturation, and they presented this as the reason for the low Pt and Pd content of the LG-1 to LG-4 chromitites. However, fractionation of 10% would cause the magmas discussed above to be well along the spinel–orthopyroxene cotectic and therefore not capable of forming chromite monocu- mulates; the amount of fractionation that we found using MELTS before orthopyroxene joined spinel on the liquidus varied between 0.2% and 1.2%. This has led us in this article to investigate the effect of mixing unfractionated U (or U3) magma with average upper continental crust as a possible cause of sulphide immiscibility.
The investigation has been undertaken using the spread- sheet developed by Li and Ripley (2009). Obviously, the sulphur content of the assimilated crust is necessary for the calculation. Wedepohl’s (1995) estimate of this is 953 ppm, whilst Rudnick and Gao’s (2004) estimate is 600 ppm. The SCSS of Wedepohl’s (1995) average upper crustal compo- sition is 389 ppm at fO2 = QFM and 397 at QFM-1. The δ34S values of sulphides within the Bushveld (i.e. excluding those in the Platreef and other marginal rocks) are close to mantle values (Liebenberg 1970) so that it is likely that relatively little crustal sulphur has been added along with the contaminant. Our modelling therefore has been under- taken assuming two alternatives: (1) that the crustal contaminant is saturated in sulphur and (2) that it contains no sulphur. Results are shown in Fig. 16 as plots of the SCSS minus the actual sulphur content of these mixtures. Where values are negative, the mixture contains more sulphur that it can dissolve, and this sulphur will be present as immiscible sulphide liquid; where they are positive, the mixture will be unsaturated in sulphide. Figure 16a shows1 cn = chondrite normalised
that when sulphide-saturated U3 is mixed with sulphide- saturated average upper crust, sulphide immiscibility will develop in all proportions of the two; however (Fig. 16b), when the average crust contains no sulphide, no immiscible sulphide will form when more than about 40% contaminant is present (solid red and blue lines). The scenario when the U3 magma is unsaturated, containing 30 ppm less sulphur than its saturation value (SCSS) of about 1,500 ppm, is also
shown in Fig. 16b (dashed red and blue lines). If the U3 magma is somewhat more undersaturated than this, immis- cible sulphides will not develop, unless the crustal melt with which it mixes contains some sulphur.
It is suggested that the rapid ascent of magma during stage 1 (see Fig. 3) precluded much additional contamina- tion and that an immiscible sulphide liquid consequently did not form in these situations. The slower ascent of magma in stages 2, 4 and 5 resulted in contamination and the formation of sulphide. Figure 3 also shows the data of Eales et al. (1988) for the variation in 87Sr/86Sr ratio throughout the Critical Zone up to the top of the UG-2 (taken from their study of the Northwestern Bushveld). This supports our proposal that little additional contamination over that that had already affected the Lower Zone magma occurred in our stage 1, but that the magma became progressively more contaminated thereafter (except for stage 3). However, we are conscious that our model does not address the precise timing of sulphide saturation with respect to chromite deposition. If the sulphides had formed prior to ‘onstage’ chromite crystallization, one might expect sulphide globules to be trapped in chromite (Holwell et al. 2011), which they are not. In this situation, one would also have to explain how chromite could trap laurite and PGE alloys, in light of the data of Bockrath et al. (2004) and Brenan and Andrews (2001) that the development of a sulphide liquid should so deplete the magma in PGE that these minerals would not form. It is possible that the contamination might have brought the composition of the magma so close to SCSS that towards the end of chromite crystallization the formation of the chromite tipped the balance and a sulphide liquid segregated. Further work on this aspect is clearly warranted.
Proposed model for the formation of chromitites and contained PGE
A simplified version of our model is illustrated in Fig. 17a, b. As has been understood for a long time (McBirney and Noyes 1979; Irvine et al. 1983; Campbell et al. 1983), magma chambers are stratified into convecting layers of upwardly decreasing density, separated by thin transitional zones across which heat and chemical components diffuse. It is suggested that periodically, injections of fresh, relatively dense magma with chromite as the sole liquidus phase intruded across the cumulate pile, depositing, as they cooled, chromite followed by cotectic proportions of orthopyroxene and chromite. Chromitites thus define the initiation of cyclic units. Whilst some chromite may have crystallised in the magma conduit and have been introduced as a bottom load in the intruding magma, most crystallised in situ, probably as a bottom growth (see Jackson 1961). Very high 87Sr/86Sr ratios observed by Kinnaird et al. (2002) in silicate layers
Fig. 16 Results of mixing U3 magma with average upper continental crust shown as the sulphur content at sulphide saturation (SCSS) of the mixture minus its sulphur content; thus, when the value is negative, the SCSS is less than the sulphur content of the magma and immiscible sulphide will form. a Mixing S-saturated U3 magma with a S-saturated melt with the composition of average upper continental crust—it is to be noted that the SCSS value for the continental crust is significantly lower than estimates of its sulphur content (Wedepohl 1995; Rudnick and Gao 2004). b Because S isotope data indicate that relatively little country rock sulphur has been added to Bushveld magma, except in marginal zones such as the Platreef, this figure shows the results of mixing when the continental crust contains no sulphur. Under these constraints, sulphide immiscibility still occurs when the proportion of contaminant is less than about 40 wt.%. The dashed lines indicate results when the U3 magma contains about 30 ppm less than its saturation value. If the U3 magma contains much less S than this, sulphide immiscibility will not develop unless some S is present in the contaminant
within the UG-2, UG-1, MG-4 and LG-6, and attributed by them to a fountain of magma rising to the roof and interacting with melted roof rocks (Rooiberg felsite), are an indication that hybridization of the magma responsible for this chromitite did occur. Our MELTS calculations suggest that it was not responsible for initiating chromite crystalli- zation. It did, however, initiate sulphide immiscibility, with the consequent concentration of PGE.
Convective overturn within the basal layer kept refresh- ing the magma from which the chromite was crystallising so that a continuous chromite cumulate developed. Any temporary retardation in the convection could result in orthopyroxene joining chromite as a crystallising phase, only to be eliminated as convection became re-invigorated; this would account for the orthopyroxene-rich ‘partings’ observed in a given chromitite in some areas and not in others. In many cases, it is likely that crystallization in the magma conduit, particularly after a relatively quiet period when the conduit had cooled down, could result in the composition entering the field of joint orthopyroxene and chromite crystallization before entering the Bushveld Complex chamber, resulting in the deposition of a cumulate composed of the two minerals. Some of the chromite deposited in the conduit whilst chromite was still the sole liquidus phase could be remobilized and dragged into the chamber to produce chromitite layers (i.e. a mixed ‘offstage–onstage’ origin). The LG-6, which shows rela- tively little zoning of PGE or chromite composition, could have originated in this way. An intruding magma carrying some chromite as a bottom load could rip up inclusions of underlying rocks, giving rise to some of the features described by Voordouw et al. (2009).
Crystallization of chromite would also reduce the density of the magma crystallising the chromite (albeit very slightly because the proportion of chromite crystallising to magma from which it crystallises is so low) so that it could initiate mixing with the immediately overlying magma layer (see
Fig. 17b), and this could have affected the chromite/magma partition coefficient of V, accounting for the upward increase in the V content observed in the sample from Waterval. In our view, this explanation is unconstrained and is again a subject for further work.
Exit channels for the Critical Zone magma
Given the extremely large volumes of magma that are required to explain the chromite budget (and to a lesser degree the PGE budget) for the Critical Zone and the relatively small volume of cumulates, the question must be answered as to where the excess magma has gone. Eales (2000) has shown that the composition of the Main Zone precludes it representing this excess. Naldrett et al. (2008) discussed this problem and suggested (their ‘pudding basin’ model) that the characteristics of the different lobes of the Bushveld Complex were the result of different levels of erosion exposing (1) a level close to the base of the central part of a lobe (Far Western Bushveld Complex), (2) a middle level of the Bushveld Complex exposing the marginal zone and layered series (Western and Eastern Bushveld Complex) and (3) an upper level where magma that has escaped up was frozen to the sides as a series of layers (Northern Limb; see ESM Fig. X2). The Northern Limb has been described recently by Kinnaird et al.(2005), McDonald et al. (2005, 2009), Holwell and McDonald (2007), Holwell et al. (2007), Maier et al. (2008) and van der Merwe (2008). The majority of these authors are of the opinion that the Platreef is the result of many injections of magma, not all of them laterally extensive, that form a series of closely associated lenses that interacted with and cooled against the country rocks. In general, the lowermost of these interacted more strongly and became more contaminated than those farther from the contact (see the papers on sulphur isotope ratios in the Platreef by Sharman- Harris et al. 2005; Holwell et al. 2007; Penniston-Dorland
Fig. 17 Our model for the deposition of chromite and PGE (see text for discussion)
et al. 2008), probably because the latter were insulated by rocks deposited by previous waves of magma. In the Sandsloot area, the variation in PGE tenor in sulphides within individual waves generally increases upward (ESM Fig. X3), with the tenor in some parts of the last wave overlapping with that characteristic of the Merensky Reef. It is proposed here that this represents the escape of much of the magma responsible for the Reef. Subsequently, emplacement of large volumes of Main Zone magma filled the chamber further, coming into contact with, eroding and in part eliminating the Platreef. Other holes were sampled (not shown in ESM Fig. X3) where the high tenor sulphides were missing (holes SS 211, SS 215); this is attributed either to the high tenor zone having been eroded subse- quently by intrusion of the Main Zone or to it never having been deposited. The sulphur isotope data indicating that country rock sulphide has significantly contaminated the lower (outer) injections of magma account in part for the lower tenor of these sulphides. Alternatively, it is possible that the magma responsible for the upper, high tenor parts of the Platreef represent the high tenor magma responsible for the Merensky Reef and that earlier waves of magma contained a lower concentration of PGE. Another possible escape route for Bushveld magma is represented by the Sheba’s Ridge mineralized zone near Groblersdal in the area between the Eastern and Southeastern chambers of the Bushveld (Stevens and Kinnaird 2010).
The above suggestion, which is favoured by the senior author, is by no means accepted by all of those currently working on the Bushveld Complex. The magma lenses closest to the Main Zone contact with the Platreef do not always contain sulphides with the highest tenors (McDo- nald et al. 2005; Holwell and McDonald 2006). Hutchinson and McDonald (2008) found that the Pt tenor of sulphides in the Upper Platreef on the farm Turfspruit are <50 ppm, and Holwell et al. (2007) found the highest tenors on the farm Witrivier to be near the base of the Platreef. McDonald and Holwell (2007) have pointed to the variable and, in some areas, low Ni contents of olivine in the Lower Zone. They suggested that the Ni depletion was the result of Lower Zone magma interacting with sulphide, that Lower Zone conduits were storage chambers for these sulphides, and that the sulphides were subsequently entrained by later Platreef magmas passing along the same conduits.
A question prompted by this admittedly speculative solution to the missing magma is where is the escaping magma now? The Bushveld Complex is overlain uncon- formably by the Waterberg sediments that are indistinguish- able from it on the basis of Shrimp ages; Dorland et al. (2006) dated a lava at the base of the Waterberg at 2.054 Ga. As mentioned above, Armitage et al. (2007) have described mafic tuffs and lavas in the upper horizons
of the Rooiberg rocks, and it is possible that these represent some of the missing magma. However, it is unlikely that these are volumetrically sufficient to account for all that is missing, and it is unreasonable to argue that in this geologically brief interval most of the lava representing the missing magma has been eroded. Our tentative suggestion is that widespread intrusions of Bushveld age, such as the Malopo Farms Complex, have formed from this magma (Meixner and Peart 1984; Gould et al. 1986, 1987; von Gruenewaldt et al. 1989; Reichhardt 1994; Walker et al. 2010).
1. Bushveld chromitites can be divided on the basis of their composition into five stages which correlate with the studies of Eales et al. (1990) of the variation in the Mg/(Mg + Fe) ratio of orthopyroxene. These are: stage 1—when magma additions caused the magma chamber to become steadily more primitive and the aAl2O3 to decrease or remain constant; stage 2—when fraction- ation of orthopyroxene in the chamber caused the resident magma there to become less primitive despite addition of new magma, leading to an increase in aAl2O3; stage 3—a brief interval when addition of magma again overrode the effect of fractionation; and stages 4 and 5—when fractionation again overrode the effect of magma addition. Stage 4 was marked by an increase in aAl2O3 due to orthopyroxene crystallization, but stage 5 differed in that the aAl2O3 was ‘buffered’ due to the crystallization of cumulus plagioclase.
2. The similarity in the vertical profiles of PGE variation through the MG-3 and MG-4 chromitites over distances of nearly 300 km, coupled with variation in the V concentration in chromite at Waterval shaft, led to the conclusion that much of the chromitite was precipitated in situ and is not the consequence of emplacement as a dense slurry.
3. Modelling using the programme MELTS indicates that mixing of primitive magma of Critical Zone composi- tion (i.e. with orthopyroxene on the liquidus) with a fractionate of the same magma will not promote the crystallization of chromite before orthopyroxene, nor will contamination with a felsic melt. Increase in pressure in the range 1–3 kb has no effect on the relative order in which spinel and orthopyroxene appear on the liquidus, nor does the addition of H2O. Given the constraint that the Bushveld crystallised at an fO2 of QFM or less, the only way in which chromite will precede orthopyroxene (and therefore give rise to a chromite mono-cumulate) is if the Cr2O3 concentration of the magma is >0.20 wt.%.
4. The amount of Cr2O3 removed from the magma in the above case is less than 20% of that present in the magma, which leads to the conclusion that a 70-cm layer of chromitite must have accumulated from about 1 km of magma.
5. Similar modelling using as an initial composition the Critical Zone composition plus 20% olivine (Fo90) and 0.3 wt.% Cr2O3 shows that once this magma has fractionated to reach the chromite–olivine cotectic (magma B) and then fractionated along the cotectic until orthopyroxene replaces olivine on the liquidus (magma C), mixtures of magmas B and C will precipitate spinel as the first liquidus phase, in agreement with Irvine (1977) conclusion. However, the amount of chromite precipitating before olivine rejoins it on the liquidus is very small, requiring several kilometres of the mixture to produce a 70-cm chromi- tite layer.
6. It is proposed that PGE in chromitites are contributed from two sources: (1) the chromite itself (principally the IPGE) and (2) sulphide that has largely been destroyed through reaction with non-stoichiometric chromite (principally the PPGE plus some IPGE). The LG-1 to LG-4 chromitites of the Bushveld, which are characterized by low (Pt + Pd)/(Ru + Ir + Os) ratios, are thought to have developed without sulphide, whilst from the LG-5 upward to the UG-2/3, sulphide has always accumulated along with the chromite.
7. Modelling using Li and Ripley’s (2009) equation for sulphur solubility shows that mixing of Critical Zone magma with a felsic contaminant will induce sulphide immiscibility. Data of Eales et al. (1990) on Sr isotope ratios and orthopyroxene compositions in the Critical Zone are consistent with little additional contamina- tion (the initial Critical Zone magma was crustally contaminated) occurring during formation of the LG-1 to LG-4 chromitites (we suggest due to the rapid ascent of the magma). Thereafter, except for a brief interval around the LG-7, contamination occurred to an increasing extent (we suggest because the ascent of magma and its introduction into the chamber was slower). This contamination is responsible for sul- phides accumulating along with chromite from the LG-5 upward.
Acknowledgements The data obtained during the study reported here were on samples kindly provided by Anglo American Platinum Ltd, Lonmin Plc., Impala Platinum Holdings Ltd., and Samancor Chrome. We are most grateful for this assistance and for the information that was shared with us by many of the geologists working for these companies. The senior author is grateful to Judith Kinnaird who has made it possible for him to visit South Africa over the past 8 years, and to the School of Geosciences, The University of the Witwatersrand, for making him feel so welcome.
Allan JF, Sack RO, Batiza R (1988) Cr-rich spinels aspetrogenetic indicators: MORB-type lavas from the Lamontseamount chain, Eastern Pacific. Am Mineral 73:741–753
Armitage PEB, McDonald I, Tredoux M (2007) A geological investigation of the Waterberg hydrothermal platinum deposit, Mookgophong, Limpopo Province, South Africa. Applied Earth Science (Transactions of the Institution of Mining and Metallur- gy, Series B) 116:113–129
Ballhaus C, Sylvester P (2000) Noble metal enrichment processes in the Merensky Reef, Bushveld Complex. J Petrol 41:545–561
Barnes S-J, Maier WD (2002) Platinum-group element distributions in the Rustenburg Layered Suite of the Bushveld Complex, South Africa. In: Cabri LJ (ed) The geology, geochemistry, mineralogy and mineral beneficiation of platinum-group elements. Canadian Insti- tute of Mining and Metallurgy Special volume 54, pp 431–458
Barnes SJ, Roeder PL (2001) The range of spinel compositions in terrestrial mafic and ultramafic rocks. J Petrol 42:2279–2302
Bockrath C, Ballhaus C, Holzheid A (2004) Stabilities of laurite RuS2 and monosulphide solid solution at magmatic temperature. Chem Geol 208:265–271
Boudreau AE, McCallum IS (1992) Concentration of platinum-group elements by magmatic fluids in layered intrusions. Econ Geol 87:1830–1848
Brenan JM, Andrews D (2001) High-temperature stability of laurite and Ru–Os–Ir alloy and their role in PGE fractionation in mafic magmas. Can Mineral 39:341–360
Cameron EN (1978) The lower zone of the eastern Bushveld Complex in the Olifants Tiver trough. J Petrol 19:437–462
Cameron EN (1980) Evolution of the lower Critical Zone, central sector, Eastern Bushveld Complex, and its chromite deposits. Econ Geol 75:845–871
Cameron EN (1982) The Upper Critical Zone of the Eastern Bushveld Complex—precursor to the Merensky Reef. Econ Geol 77:1307– 1327
Cameron EN, Desborough GA (1964) Origin of certain magnetite- bearing pegmatites in the eastern part of the Bushveld Complex, South Africa. Econ Geol 59:197–225
Campbell IH, Naldrett AJ, Barnes SJ (1983) A model for the origin of the platinum-rich sulfide horizons in the Bushveld and Stillwater Complexes. J Petrol 24:133–165
Canil D (1999) Vanadium partitioning between orthopyroxene, spinel and silicate melt and the redox states of mantle source regions for primary magmas. Geochimica et Cosmochimica Acta 63:557– 572
Cawthorn RG (2002) The role of magma mixing in the genesis of PGE mineralisation in the Bushveld Complex. Thermodynamic calculations and new interpretations—a discussion. Econ Geol 97:663–666
Cawthorn RG (2005) Stratiform platinum-group element deposits in layered intrusions. In: Mungall JE (ed) Exploration for platinum- group element deposits. Mineralogical Association of Canada Short Course Series 35, Chapter 3, pp 57–73
Cawthorn GR, Walraven F (1998) Emplacement and crystallization time for the Bushveld Complex. J Petrol 39:1669–1687
Cawthorn RG, Webb SJ (2001) Connectivity between the Eastern and Western limbs of the Bushveld Complex. Tectonophysics 330:195–209
Cousins CA, Feringa G (1964) The chromite deposits of the Western belt of the Bushveld Complex. In: Haughton SH (ed) The geology of some ore deposits in Southern Africa, Johannesburg. Geological Society of South Africa 2, pp 183–202
Curl EA (2001) Parental magmas of the Bushveld Complex, South Africa. PhD thesis, Monash University, Australia, p 140
Davies G, Cawthorn RC, Barton JM Jr, Morton M (1980) Parental magma to the Bushveld Complex. Nature 287:33–35
Dorland HC, Beukes NJ, Gutzmer J (2006) Precise SHRIMP U Pb zircon age constraints on the lower Waterberg and Soutpansberg Groups, South Africa. S Afr J Geol 109:139–156
Eales HV (2000) Implications of the chromium budget of the Western Limb of the Bushveld Complex. S Afr J Geol 103:141–150
Eales HV, Cawthorn RG (1996) The Bushveld Complex. In: Cawthorn RG (ed) Layered intrusions. Elsevier, Amsterdam, pp 181–230
Eales HV, Marsh JS, Mitchell AA, de Klerk WJ, KrugerFJ FM (1986) Some geochemical constraints upon models for the crystallization of the upper critical zone—main zone interval, Northwestern Bushveld Complex. Mineral Mag 50:567–582
Eales HV, Field M, de Klerk WJ, Scoon RN (1988) Regional trends of chemical variation and thermal erosion in the upper critical zone, Western Bushveld Complex. Mineral Mag 52:63–79
Eales HV, de Klerk WJ, Teigler B (1990) Evidence for magma mixing processes within the critical and lower zones of the northwestern Bushveld Complex, South Africa. Chem Geol 88:261–278
Finnigan CS, Brenan JM, Mungall JE, McDonough WF (2008) Experiments and models bearing on the role of chromite as a collector of platinum group minerals by local reduction. J Petrol 49:1647–1665
Fourie GP (1959) The chromite deposits in the Rustenburg area. South Africa Geological Survey Bulletin 27:45
Ghiorso MS, Sack RO (1995) Chemical mass transfer in magmatic processes IV. A revised and internally consistent thermodynamic model for the interpolation and extrapolation of liquid–solid equilibria in magmatic systems at elevated temperatures and pressures. Contrib Mineral Petrol 119:197–212
Gould D, Rathbone PA, Kimbell GS, Burley AJ (1986) The Molopo Farms Complex, Botswana—a possible target for Bushveld-type mineralisation? In: Gallagher MJ, Ixer RA, Neary CR, Prichard HM (eds) Metallogeny of basic and ultrabasic rocks. The Institution of Mining and Metallurgy, London, pp 319–332
Gould D, Rathbone PA, Kimbell GS (1987) The geology of the Molopo farms complex, southern Botswana. Geological Survey of Botswana, Bulletin 23:178
Hall AL, Humphrey WA (1908) On the occurrence of chromite along the southern and eastern margins of the Bushveld plutonic complex. Transactions of the Geological Society of South Africa 11:69–77
Harmer RE, Armstrong RA (2000) Duration of Bushveld Complex (sensu lato) magmatism: constraints from new SHRIMP zircon chronology. Workshop on the Bushveld Complex, Gethane Lodge, Burgersfort, University of the Witwatersrand, Johannes- burg
Harmer RE, Sharpe MR (1985) Field relations and Sr isotope systematics of the marginal rock of the Eastern Bushveld Complex. Econ Geol 80:813–837
Hatton CJ, von Gruenewaldt G (1985) Chromite from the Swartkop chrome mine—an estimate of the effects of sub-solidus equili- bration. Econ Geol 80:911–924
Hatton CJ, von Gruenewaldt G (1987) The geological setting and petrogenesis of the Bushveld chromitite layers. In: Stowe CW (ed) Evolution of chromium ore fields. Van Nostrand Reinhold, Stroudsburg, pp 109–143
Hiemstra SA (1979) The role of collectors in the formation of platinum deposits in the Bushveld. Can Mineral 17:469–482
Hiemstra SA (1985) The distribution of some platinum group elements in the UG-2 chromitite layer of the Bushveld Complex. Econ Geol 80:944–957
Holwell DA, McDonald I (2006) Petrology, geochemistry and the mechanisms determining the distribution of platinum-group element and base metal sulfide mineralization in the Platreef at
Overysel, Northern Bushveld Complex, South Africa. Miner Depos 41:575–598
Holwell DA, McDonald I (2007) Distribution of platinum-group elements in the Platreef at Overysel, northern Bushveld Complex: a combined PGM and LA-ICP-MS study. Contrib Mineral Petrol 54:171–190
Holwell DA, Boyce AJ, McDonald I (2007) Sulfur isotope variations within the Platreef Ni–Cu–PGE deposit: genetic implications for the origin of sulfide mineralization. Econ Geol 102:1091–1110
Holwell DA, McDonald I, Butler IB (2011) Precious metal enrichment in the Platreef, Bushveld Complex, South Africa: evidence from homogenized magmatic sulfide melt inclusions. Contrib Mineral Petrol 161:1011–1026. doi:10.1007/s00410-010-0577-0
Hulbert LJ (1983) A petrological investigation of the Rustenburg Layered Suite and associated mineralization south of Potgieters- rus. DSc thesis (unpublished), University of Pretoria, p 511
Hulbert LJ, von Gruenewaldt G (1982) Nickel, copper and platinum mineralization in the Lower Zone of the Bushveld Complex, South of Potgietersrus. Econ Geol 77:1296–1306
Hutchinson D, McDonald I (2008) Laser ablation ICP-MS study of platinum-group elements in sulfides from the Platreef at Turfspruit, Northern Limb of the Bushveld Complex, South Africa. Miner Depos 43:695–711
Irvine TN (1975) Crystallization sequences in the Muskox intrusion and other layered intrusions, II. Origin of chromitite layers and similar deposits of other magmatic ores. Geochim Cosmochim Acta 39:991–1020
Irvine TN (1977) Origin of chromite layers in the Muskox intrusion and other stratiform intrusions: a new interpretation. Geology 5:273–277
Irvine TN, Keith DW, Todd SG (1983) The J-M platinum–palladium Reef of the Stillwater Complex, Montana: II. Origin by double- diffusive convective magma mixing and implications for the Bushveld Complex. Econ Geol 78:1287–1334
Jackson ED (1961) Primary textures and mineral associations in the ultramafic zone of the Stillwater Complex, Montana. U.S. Geol Surv Prof Pap 358:1–106
Kinnaird JA, Kruger FJ, Nex PAM, Cawthorn RG (2002) The chromitite formation—a key to understand processes of formation and platinum enrichment. Trans Inst Min Metall 111:23–35
Kinnaird JA, Hutchinson D, Schurmann L, Nex PAM, deLange R (2005) Petrology and mineralization of the southern Platreef, northern limb of the Bushveld Complex, South Africa. Miner Depos 40:576–597
Kottke-Levin J, Tredoux M, Grab P-J (2009) An investigation of the geochemistry of the Middle Group of the eastern Bushveld Complex, South Africa Part 1—the chromitite layers. Applied Earth Science (Transactions of the Institution of Mining and Metallurgy, B) 118:111–130
Lee CA, Parry SJ (1988) Platinum group element geochemistry of the Lower and Middle Group chromitites of the Eastern Bushveld Complex. Econ Geol 85:877–883
Li C, Ripley EM (2009) Sulfur contents at sulfide-liquid or anhydrite saturation in silicate melts: empirical equations and example applications. Econ Geol 104:405–412
Liebenberg L (1970) The sulfides in the layered sequence of the Bushveld Igneous Complex. Geological Society of South Africa, Special Publication No. 1, pp 108–207
Lipin BR (1993) Pressure increases, the formation of chromite seams, and the development of the ultramafic series in the Stillwater Complex, Montana. J Petrol 34:955–976
Maier WD, Barnes S-J (2008) Platinum-group elements in the UG1 and UG2 chromitites, and the Bastard Reef, at Impala platinum mine, Western Bushveld Complex, South Africa: evidence for
late magmatic cumulate instability and reef constitution. S Afr J Geol 111:159–176
Maier WD, Teigler B (1995) A facies model for the Western Bushveld Complex. Econ Geol 90:2343–2349
Maier WD, Prichard HM, Fisher PC, Barnes SJ (1999) Compositional variation of laurite at Union Section in the western Bushveld Complex. S Afr J Geol 102:286–292
Maier WD, de Klerk L, Blaine J, Manyeruke T, Barnes S-J, Stevens MVA, Mavrogenes JA (2008) Petrogeneis of contact style mineralization in the northern lobe of the Bushveld Complex: comparison of the farms Rooiport, Townlands, Drenthe and Nonnenworth. Miner Depos 43:255–280
McBirney AR, Noyes RM (1979) Crystallization and layering of the Skaergaard intrusion. J Petrol 20:487–554
McCandless TE, Ruiz J, Adair BI, Freydier C (1999) Re–Os isotope and Pd/Ru variations in chromitites from the Critical Zone, Bushveld Complex, South Africa. Geochim Cosmochim Acta 63:911–923
McDonald I (2008) Platinum-group element and sulfide mineralogy in ultramafic complexes at Western Andriamena, Madagascar. Applied Earth Science (Trans Inst Min Metall B) 117:1–10
McDonald I, Holwell DA (2007) Did Lower Zone magma conduits store PGE-rich sulfides that were later supplied to the Platreef? S Afr J Geol 110:611–616
McDonald I, Holwell DA, Armitage PEB (2005) Geochemistry and mineralogy of the Platreef and “Critical Zone” cumulates of the Northern limb of the Bushveld Complex, South Africa: implica- tions for Bushveld stratigraphy and the development of PGE mineralization. Miner Depos 40:526–549
McDonald I, Holwell DA, Wesley B (2009) Assessing the potential involvement of an early magma staging chamber in the generation of the Platreef Ni–Cu–PGE Deposit in the Northern Limb of the Bushveld Complex: a pilot study of the Lower Zone Complex at Zwartfontein. Applied Earth Science (Transactions of the Institution of Mininhg and Metallurgy B) 118:B5–B20
McLaren CH, de Villiers JPR (1982) The platinum group chemistry and mineralogy of the UG-2 chromitite layer of the Bushveld Complex. Econ Geol 77:1348–1366
Meixner HM, Peart RJ (1984) The Kalahari drilling project. Geol Surv Botswana, Bull 27:224
Merkle RKW (1992) Platinum-group element minerals in the Middle Group of chromitite layers, Western Bushveld Complex: indica- tions for collection mechanisms and postmagmatic modification. Can J Earth Sci 29:209–221
Mitchell AA, Scoon RN (2007) The Merensky Reef at Winnaarshoek, Eastern Bushveld Complex: a primary magmatic hypothesis based on a wide reef facies. Econ Geol 102:971–1009
Molyneaux TG (1974) A geological investigation of the Bushveld Complex in Sekhukhuneland and a part of the Steelpoort Valley. Geological Society of South Africa Transactions 77:329–338
Mondal SK, Mathez EA (2007) Origin of the UG2 chromitite layer, Bushveld Complex. J Petrol 48:495–510
Mungall J (2002) A model for co-precipitation of platinum group minerals with chromite from silicate melts. Abstracts, 9th Interna- tional Platinum Symposium, July, Stillwater, USA. http://www. nicholas.duke.edu/people/faculty/boudreau/IPS__Abstracts.htm
Murck BW, Campbell IH (1986) The effects of temperature, oxygen fugacity and melt composition on the behaviour of chromium in basic and ultrabasic melts. Geochim Cosmochim Acta 50:1871– 1887
Naldrett AJ, Duke JM (1980) Platinum metals in magmatic sulfide ores. Science 208:1417–1428
Naldrett AJ, Lehmann J (1988) Spinel non-stoichiometry as the explanation for Ni-, Cu- and PGE-enriched sulfides in chromi- tites. In: Prichard H, Potts P, Bowles J, Cribb S (eds) Geo- platinum’87. Elsevier, London, pp 93–110
Naldrett AJ, Kinnaird JA, Wilson A, Chunnett G (2008) The concentration of PGE in the earth’s crust with special reference to the Bushveld Complex. Earth Science Frontiers 15(5):264–297
Naldrett AJ, Kinnaird JA, Wilson A, Yudovskaya M, McQuade S, Chunnett G, Stanley C (2009) Chromite composition and PGE content of Bushveld chromitites: Part 1—the Lower and Middle Groups. Applied Earth Science (Transactions of the Institution of Mining and Metallurgy (B) 118:131–161
Nex PAM (2004) Formation of bifurcating chromitite layers of the UG1 in the Bushveld Igneous Complex, an analogy with sand volcanoes. J Geol Soc London 161:903–909
Ohnenstetter M, Johan Z, Cocherie A, Fouillac AMC, Guerrot C, Ohnenstetter D, Chaudisson O, Rouer MO, Makovicky E, Mackovicky M, Rose-Hansen J, Karup-Moller S, Vaughan DJ, Turner G, Pattrick RAD, Gize AP, Lyon I, McDonald I (1999) New exploration methods for platinum andrhodium deposits poor in base-metal sulfides—NEXTPRIM’. Applied Earth Science (Transactions of the Institution of Mining and Metallurgy Section B) 108:B119–B150
Onuma K (1988) Effect of Cr on the liquidus phase relations in the system Mg2SiO4–CaAl2Si2O8–CaMg2Si2O6–SiO2 in air at 1 atm. J Mineral, Petrol Econ Geol 83:405–417 (in Japanese)
Penniston-Dorland SC, Wing BA, Nex PAM, Kinnaird JA, Farqhar J, Brown M, Sharman ER (2008) Multiple sulfur isotopes reveal a magmatic origin for the Platreef platinum group element deposit, Bushveld Complex, South Africa. Geology 36:979–982
Poustovetov AA, Roeder PL (2000) The distribution of Cr between basaltic melt and chromian spinel as an oxygen geobarometer. Can Mineral 39:309–317
Reichhardt FJ (1994) The Molopo Farms Complex, Botswana: history, stratigraphy, petrography, petrochemistry and Ni–Cu–PGE min- eralization. Explor Min Geol 3:263–284
Reynolds IV (1985) Contrasted mineralogy and textural relationship- sin the uppermosttitaniferous magnetite layers of the Bushveld Complex in the Bierkraal area north of Rustenburg. Econ Geol 80:1027–1048
Rudnick RL, Gao S (2004) Composition of the continental crust. In: Holland HD, Turekian KK (eds) Treatise on geochemistry, vol 3. Elsevier, Amsterdam, pp 1–64
Schurmann LW, Grabe P-J, Steenkamp CJ (1998) Chromium. In: Wilson MGC, Anhaeusser CR (eds) The mineral resources of South Africa: Handbook 16. Council for Geosciences, CTP Book Printers, Cape Town, pp 90–105
Scoates JS, Friedman RM (2008) Precise age of the platiniferous Merensky Reef, Bushveld Complex, South Africa, by the U–Pb zircon chemical abrasion ID-TIMS technique. Econ Geol 103:465–471
Scoon RN, Teigler B (1994) Platinum-group element mineralization in the Critical Zone of the Western Bushveld Complex: I. Sulfide- poor chromitites below the UG-2. Econ Geol 89:1094–1121
Sen G, Presnall DC (1984) Liquidus phase relations on the join anorthite–forsterite–quartz at 10 kb with applications to basalt petrogenesis. Contrib Mineral Petrol 85:404–408
Sharman-Harris ER, Kinnaird JA, Harris C, Horstmann UE (2005) A new look at sulfide mineralisation of the northern limb, Bushveld Complex: a stable isotope study. Applied Earth Science: Trans- actions of the Institute of Mining Metallurgy B 114:B252–B263
Shima H, Naldrett AJ (1975) Solubility of sulfur in an ultramafic melt and the relevance of the system Fe–S–O. Econ Geol 70:960–967
Stevens FJ, Kinnaird JA (2010) Stratigraphic correlations of the Sheba’s Ridge Area, Southeastern Bushveld Complex, South Africa. Volume of Abstracts, 11th International Platinum Sympo- sium, Sudbury, Canada
Tegner C, Delaney JS, Dyar MD, Lundgaard KL (2003) Iron in plagioclase as a monitor of oxygen fugacity in Skaergaard, Bushveld, and Bjerkreim-Sokndal layered intrusions, and anor-
thosite of the Rogaland Igneous Province. EGS–AGU–EUG Joint Assembly, Abstracts from the meeting, Nice, France, 6–11 April 2003, abstract no. 8789
Teigler B, Eales HV (1993) Correlation between chromite composition and PGE mineralization in the Critical Zone of the Western Bushveld Complex. Miner Depos 28:291–302
Ulmer GC (1969) Experimental investigations of chromite spinels. Economic Geology, Monograph 4:114–181
Van der Merwe MJ (2008) The geology and structure of the Rustenburg Layered suite in the Potgietersrus/Mokapane area of the Bushveld Complex. S Afr Miner Depos 43:405–419
Von Gruenewaldt G, Hatton CJ, Merkle RKW, Gain SB (1986) Platinum group element–chromitite associations in the Bushveld Complex. Econ Geol 81:1067–1079
Von Gruenewaldt G (1973) The main and upper zones of the Bushveld Complex in the Roossenekal area Eastern Transvaal. Geological Society of South Africa Transactions 76:207–227
Von Gruenewaldt G, Behr SH, Wilhelm HJ (1989) Some prelim- inary petrological investigations of the Molopo Farms Com-
plex, Botswana, and its Ni–Cu sulfide mineralization. In: Prendergast MD, Jones MJ (eds) Magmatic sulfides—the Zimbabwe volume. The Institution of Mining and Metallurgy, London, pp 95–105
Voordouw R, Gutzmer J, Beukes NJ (2009) Intrusive origin for Upper Group (UG1, UG2) stratiform chromitite seams in the Dwars River area, Bushveld Complex, South Africa. Mineral Petrol 97:75–94
Wagner PA (1929) The platinum deposits and mines of South Africa. Oliver and Boyd, Edinburgh, 326 pp
Walker ASD, Key RM, Pouliquen G, Gunn G, Sharrock J, McGeorge I, Koketso M, Farr J (2010) Geophysical modelling of the Molopo Farms Complex; Implications for its emplacement within the 2 Ga large igneous provinces of southern and central Africa. S Afr J Geol 113:381–400
Webb SJ, Cawthorn RG, Nguuri T, James D (2004) Gravity modeling of Bushveld Complex connectivity supported by South African seismic results. S Afr J Geol 107:207–218
Wedepohl KH (1995) The composition of the continental crust. Geochim Cosmochim Acta 59:217–239