| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 10 | Pages 1787-1804 | 2003
© Oxford University Press 2003; all rights reserved
The Concentration of the Platinum-Group Elements in South African Komatiites: Implications for Mantle Sources, Melting Regime and PGE Fractionation during Crystallization
1 CENTRE FOR RESEARCH ON MAGMATIC ORE DEPOSITS, DEPARTMENT OF GEOLOGY, UNIVERSITY OF PRETORIA, PRETORIA 0002, SOUTH AFRICA
2 SCIENCES DE LA TERRE, UNIVERSITÉ DU QUÉBEC, CHICOUTIMI, QUE. G7H 2B1, CANADA
* Corresponding author. E-mail: wdmaier{at}scientia.up.ac.za
RECEIVED JULY 3, 2002; ACCEPTED MARCH 26, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
We have analysed 18 samples of komatiite from five consecutive lava flows of the Komati Formation at Spinifex Creek, Barberton Mountain Land. Our samples include massive komatiite, various types of spinifex-textured komatiite, and flow-top breccias. The rocks have low platinum-group element (PGE) contents and Pd/Ir ratios relative to komatiites from elsewhere, at 0·45–2 ppb Os, 1–1·4 ppb Ir, <1–5 ppb Ru, 0·33–0·79 ppb Rh, 1·7–6 ppb Pt, 1·6–6·1 ppb Pd, and Pd/Ir 3·3. Pt/Pd ratios are c. 1·1. Platinum-group elements are depleted relative to Cu (Cu/Pd = 15 300). They display a tendency to increase in the less magnesian samples, suggesting that the magmas were S-undersaturated upon eruption and that all PGE were incompatible with respect to crystallizing olivine. Komatiites from the Westonaria Formation of the Ventersdorp Supergroup and the Roodekrans Complex near Johannesburg have broadly similar PGE patterns and concentrations to the Komati rocks, suggesting that the PGE contents of South African ultrabasic magmas are controlled by similar processes during partial mantle melting and low-P magmatic crystallization. Most workers believe that the Barberton komatiites formed by relatively moderate-degree batch melting of the mantle at high pressure. Based on the concentration of Zr in the Komati samples, we estimate that the degree of partial melting was between 26 and 33%. We suggest that the low PGE contents and Pd/Ir ratios of all analysed South African komatiites are the result of sulphides having been retained in the mantle source during partial melting. The difference in Pd/Ir between our samples and Al-undepleted komatiites from elsewhere further suggests that the PGE are fractionated during progressive partial melting of the mantle. Thus, our data are in agreement with other recent studies showing that the PGE are hosted by different phases in the mantle, with Pd being concentrated by interstitial Cu-rich sulphide, and the IPGE (Os, Ir, Ru) and Rh resting in monosulphide solid solution included within silicates. Pt is possibly controlled by a discrete refractory phase, as Pt/Pd ratios of most komatiites worldwide are sub-chondritic.
KEY WORDS: platinum-group elements; komatiites; Barberton; mantle melting; South Africa
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The study of platinum-group element (PGE) concentrations in komatiites may provide important information relating to a range of petrological questions. First, there has been a wealth of recent data indicating that much of the PGE budget of the post-Archaean mantle may be hosted by sulphides, with monosulphide solid solution concentrating the IPGE (Os, Ir, Ru) and Rh, and Cu-rich sulphides hosting most of the Pd and Au (Bulanova et al., 1996; Alard et al., 2000; Lorand & Alard, 2001; Luguet et al., 2001). Pt may be hosted by discrete phases, possibly metal alloys (Alard et al., 2000; Luguet et al., 2001). However, PGE concentration patterns in the Archaean mantle, particularly its deep portions, remain much less constrained. This is because there are few examples of Archaean mantle rocks available for study (largely from the Kaapvaal Craton and Udachnaya in Siberia). Komatiites that remain S-undersaturated during ascent and emplacement may preserve most of their original PGE concentration patterns and thereby provide the only insights into the PGE systematics of the Archaean deep mantle. Second, PGE concentration patterns in different types of komatiites may constrain the melting processes in the Archaean mantle. Early Archaean komatiites tend to have different compositions from late Archaean komatiites, e.g. they are relatively depleted in Al2O3 and heavy rare earth elements (HREE) (Nesbitt et al., 1979; Herzberg, 1995). It has been proposed that these compositional characteristics are the result of relatively low degrees (<30%) of batch partial melting of the mantle, at a high pressure (10–15 GPa, Herzberg, 1995), possibly as a result of dictinct mantle dynamics in the early Archaean mantle (Richter, 1988). The different melting regimes should be reflected in the PGE concentration patterns of the different types of komatiites. Third, the behaviour of the PGE during magmatic fractionation remains controversial. The effect of sulphide precipitation on PGE fractionation is relatively well established, in that the PGE are believed to partition broadly evenly into segregating sulphides (Fleet et al., 1996), but the relative roles of chromite, olivine, and platinum-group minerals (PGM) are less certain [see Barnes & Maier (1999) for a discussion]. Most komatiite flows are S-undersaturated during emplacement and much of their crystallization history, and thus offer ideal opportunities to study the effect of the non-sulphidic phases on PGE fractionation (Keays, 1995).
Despite these important applications, PGE studies of komatiite sequences remain relatively rare. Much of the pioneering work was done by D. Crocket and R. Keays (e.g. Crocket & Skippen, 1966; Keays & Scott, 1976; Keays, 1983), but until the mid-1980s, only Ir, Pd and Pt could be routinely analysed at the low levels found in sulphide-free rocks. More complete PGE analyses of komatiites appeared in the late 1980s. Crocket & McRae (1986), Brügmann et al. (1987) and Dowling & Hill (1992) studied PGE in Archaean komatiites in the Abitibi greenstone belt of Canada and Kambalda, Australia, and at Gorgona Island, to constrain the effects of fractionation on the PGE concentrations. Both groups noted a compatible behaviour of Ir during solidification, but incompatible behaviour of Pt and Pd. Crocket & McRae (1986) and Dowling & Hill (1992) proposed that Ir is concentrated by a metal phase, whereas Brügmann et al. (1987) favoured partitioning of Ir (as well as Os and Ru) into olivine.
More recently, Rehkämper et al. (1999) analysed a limited number of komatiite samples from Munro, Alexo and Belingwe, and Puchtel & Humayun (2000) have added data on komatiites in Russian Karelia. Both studies constrained the behaviour of the PGE during partial melting of the mantle and proposed some implications for core–mantle interaction.
The only published PGE data on the Barberton komatiite type locality are the Pd and Ir determinations of Keays (1983, 1995), and an analysis of a bulk sample by Tredoux & McDonald (1996). No PGE information is available from other South African komatiites, including the Westonaria Formation of the Ventersdorp Supergroup and the Roodekrans greenstone remnant near Johannesburg. To address some of the questions highlighted above, and to determine whether South African komatiites share certain compositional characteristics, we decided to study, in detail, five consecutive komatiite lava flows at Spinifex Creek, six samples of komatiite from the Westonaria Formation and six samples of komatiite and basalt from the Roodekrans complex.
|
GENERAL GEOLOGICAL SETTING |
|---|
|
TOPABSTRACTINTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The Komati Formation occurs near the base of the volcano-sedimentary sequence of the Barberton greenstone belt, in the eastern part of South Africa (Fig. 1). The age of the Formation has been established by U–Pb zircon and Ar–Ar dating at 3·48 Ga and its thickness at c. 3·0–3·5 km (Dann, 2000). The rocks consist of interlayered komatiite (predominantly in the lower portion of the Formation) and komatiitic basalt (predominantly in the upper portion). Individual lava flows may have thicknesses between a few tens of centimetres and >10 m. The common occurrence of pillowed flows attests to a submarine eruptive setting.
|
The Westonaria Formation forms the basal flow unit of the ultramafic–mafic–felsic 2·7 Ga Ventersdorp Supergroup. The Ventersdorp Supergroup has previously been compared to Phanerozoic flood basalts by Marsh et al. (1992), based partly on its extension over some 300 000 km2 (Fig. 1), subaerial eruption, and the petrology and chemistry of the lava flows.
The Roodekrans ultramafic complex is interpreted as a 3·2 Ga greenstone remnant within the Archaean granite gneisses of the Johannesburg dome (Anhaeusser, 1977). It is located some 25 km to the NW of the Johannesburg city centre (Fig. 1). The ultramafic rocks are severely altered and consist of steeply dipping alternating serpentinite and amphibolite units that have been interpreted as lava flows by Anhaeusser (1977).
|
PREVIOUS CHEMICAL WORK |
|---|
|
TOPABSTRACTINTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The Barberton komatiites can be distinguished from most komatiites elsewhere in that they are depleted in Al2O3, TiO2, V, Sc and HREE (Nesbitt et al., 1979; Smith & Erlank, 1982). They are therefore commonly referred to as ‘Al-depleted’ or ‘Barberton-type’ komatiites, and distinguished from ‘Al-undepleted’ or ‘Munro-type’ komatiites. Takahashi & Scarfe (1985) and Herzberg (1992) showed that at a high pressure above 10–12 GPa near-solidus mantle melts may contain >30% MgO. In these conditions, the stability of orthopyroxene and garnet is increased at the expense of olivine, which could produce Al-depleted magmas. A similar conclusion was reached by Blichert-Toft & Arndt (1999), based on Hf isotopic data.
Much less is known about the composition of the Westonaria and Roodekrans lavas. The Westonaria Formation has been investigated by McIver et al. (1982) and Myers et al. (1990). The rocks are komatiites with minor sand and lithic tuff horizons that have been metamorphosed at lower greenschist facies (Myers et al., 1990). This has resulted in enrichment in mobile elements such as alkali and alkaline earth elements, as well as SiO2 (Myers et al., 1990). The published data also indicate enrichment of the lavas in the incompatible trace elements Zr, Y and Nb, i.e. elements that are normally considered immobile. The Roodekrans rocks have been analysed for major elements as well as Cr, Sr and Rb by Anhaeusser (1977).
|
SAMPLE DESCRIPTION |
|---|
|
TOPABSTRACTINTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Our samples from the Komati Formation all come from one selected outcrop, located on the bank of Spinifex Creek, at S25 59·341', E030 51·507'. The outcrop extends c. 6·5 m perpendicular to the layering and covers a surface of around 20 m2. Five fully or partially exposed flows were distinguished, of which the thickest was 2·9 m in thickness, and the thinnest 20 cm. Some, but not all, of the flows are characterized by a basal massive unit that is overlain by successive zones of coarse random or oriented spinifex, fine random spinifex, and a commonly vesicular flow-top breccia. A schematic profile through the studied sequence is shown in Fig. 2. Flow 1, the lowermost flow, consists of a basal 1·3 m massive unit, overlain by 20 cm of fine random spinifex, 30–40 cm of coarse oriented plate spinifex, 1·4 m of medium- and coarse-grained random spinifex, and a 3 cm vesicular flow-top breccia. Flow 2 is much thinner and consists of a basal 10 cm of massive peridotite, overlain by 15 cm of fine random spinifex. No flow top could be identified, but the contact to the overlying flow is sharp. The upper portion of this flow could have been eroded by the overlying flow, or the unit may be a sill. Flow 3 has a total thickness of 105 cm. It is subdivided into a basal 60 cm massive unit, overlain with a sharp contact by 20 cm of coarse oriented plate spinifex, 10 cm of fine random spinifex, and a 15 cm vesicular flow-top breccia. Flow 4 is 85 cm thick. It has a 40 cm massive unit at the base, overlain by 20 cm of coarse plate spinifex, 10 cm of medium-grained random spinifex, 10 cm of fine random spinifex, and a 5 cm vesicular flow-top breccia. Flow 5 consists of 90 cm of massive peridotite, overlain by 50 cm of coarse random spinifex. The top of this flow was not exposed.
|
The Westonaria samples are from borehole core in the Evander Goldfields. They are distributed throughout the Formation and have been described by Myers et al. (1990). The Roodekrans samples have been described by Anhaeusser (1977). For both the Westonaria and Roodekrans suites, no information relating to textures or sample position within individual flows is available.
|
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
In view of the pervasive alteration of the samples from the three South African komatiite suites their mineral chemistry could not be determined. Smith & Erlank (1982) presented some information on the mineral chemistry of the Komati Formation and found that olivines have MgO contents between 48 and 50 wt %, with Fo89–91.
For the Komati suite, samples weighing c. 1 kg were ground to <75 mm in a C-steel milling vessel. The mill has been checked for PGE contamination by milling quartz and determining PGE in the quartz powder. No PGE were detected in the quartz. X-ray fluorescence (XRF) analyses were carried out using an ARL 9400XP+ spectrometer at the University of Pretoria (UP). Sulphur was analysed with an Eltra CS-500 Carbon Sulfur Determinator at UP, at a temperature of 1350°C. The samples from the Westonaria and Roodekrans suites were analysed for major and lithophile trace elements by XRF at the University of the Witwatersrand. Analytical details have been given by Anhaeusser (1977) and Myers et al. (1990).
The platinum-group elements, Re and Au were determined by instrumental neutron activation analysis (INAA) at the University of Quebec at Chicoutimi, after pre-concentration in a Ni-sulphide bead from 50 g of rock powder. Sample irradiation was carried out at the École Polytechnique in Montreal in a SLOWPOKE II reactor.
The precision and accuracy of the analyses may be estimated by considering the relative standard deviations for five determinations of five different NiS beads of the CANMET standard WGB-1 (Table 1). For all of the elements except Au the relative standard deviations are 9–17%. The relative standard deviation for Ir in the komatiites is probably better than this as Ir is present at an order of magnitude higher in the komatiites than in the standards and will approach the relative standard deviation in standard UTM-1 at c. 4%. The relative standard deviation for Au is large at 44%.
|
For Au, Pt and Pd the accuracy of the analyses may be assessed by comparing the results obtained at UQAC for standards UTM-1 and WGB-1 with the certified values. The results are in good agreement with both the high- and low-level standard. For Rh, Ru and Ir certified values are available only for UTM-1. Our results agree with CANMET results. For the low-level standard WGB-1 only informational values are available for Rh, Ru, Ir and Re. Our results agree with these when the standard deviation on the CANMET informational value is considered.
No noble metals were detected in the blank, except Ir and Au. These were present at 0·02 and 0·1 ppb, respectively. As both values are far lower than the levels present in the samples, no significant contamination is believed to have occurred in preparing the samples, and no blank correction was made to the samples.
As a further test to make sure that the INAA technique is reliable at the low levels of PGE found in our samples, we also determined PGE in three of the Komati samples by inductively coupled plasma mass spectrometry (ICP-MS) at the Geological Survey of Finland (GSF; Tables 1 and 2). Preconcentration was carried out by nickel sulphide fire assay and tellurium co-precipitation. The instrument used was a Perkin–Elmer SCIEX Elan 5000, equipped with a cross-flow nebulizer and a standard Scott-type double-pass spray chamber. Aliquot sizes applied were 15 g. Results were controlled by frequent analysis of reagent blank samples and reference samples. The result of one analysis of CANMET low-PGE standard WGB-1 is given in Table 1. A repeat analysis of sample SC11 produced little variation (Table 1). Details of the fusion procedure and ICP-MS determination have been given by Juvonen et al. (2002).
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Petrography
The Komati samples are pervasively altered to an assemblage of serpentine, magnetite, chlorite and tremolite, rendering the identification of primary textures and mineral assemblages difficult. The main constituent in the massive peridotites that form the base of the flows is anhedral to subhedral pseudomorphs of olivine. The grains are between 0·5 and 1 mm in size and form a loosely connected framework, set in a fine-grained matrix of chlorite and amphibole. Within individual samples, there are areas where the olivines are denser packed, giving rise to near adcumulate textures. The base of the massive units may be chilled, as at the base of flow 2. Here, the basal 0·5 cm of the flow consists of elongated olivines, 0·1–0·2 mm in length, that are loosely set in a matrix of devitrified glass.
The coarse plate spinifex contains elongated serpentine pseudomorphs of olivine, up to 10 cm in length, that are oriented broadly perpendicular to the layering. The serpentines are set in a matrix of devitrified volcanic glass containing fine-grained acicular serpentine and amphibole. In places, the matrix serpentines and amphiboles are oriented perpendicular to the coarse serpentine blades.
The fine random spinifex contains plates of serpentine between 1·5 and 5 mm in length and 0·3–0·5 mm in width set in a matrix of finer-grained needles of serpentine and amphibole, and devitrified glass.
Some flows have zones of between 1 and 5 cm at their top that consist of angular fragments of devitrified volcanic glass set in a matrix of volcanic glass. The glass contains small (<0·1 mm) acicular crystals of amphibole, chlorite and serpentine, as well as anhedral magnetite. In samples SC9 and SC15, vesicles up to 4 mm in size and filled by chlorite are found in the fine-grained flow top.
The Westonaria komatiites are severely altered to chlorite, amphibole and carbonate (Myers et al., 1990), and the Roodekrans rocks are altered to serpentine–tremolite–chlorite–talc–carbonate schists. Antigorite is the dominant phase in the serpentinites, whereas the amphibolite interlayers contain mainly tremolite and Mg-rich chlorites (Anhaeusser, 1977).
Major element geochemistry
Concentrations of some of the major element oxides in the Komati suite are plotted vs MgO in Fig. 3 and listed in Table 3. It is evident that for Al2O3, TiO2 and, to a somewhat lesser degree, SiO2, the Komati rocks plot on olivine control lines, indicating that these elements behaved in an immobile manner during alteration. The trend of CaO is oblique to the olivine control line and intersects the MgO axis at c. 40 wt %. This is because CaO is selectively lost from the massive komatiites (see also Barnes, 1983; Arndt, 1994), as there are no secondary alteration phases present that accommodate CaO. In contrast, MgO seems to have been relatively immobile. The massive rocks have between 30 and 39·8 wt % MgO. The next most MgO-rich rocks are the fine-grained spinifex textured samples, with 28·4–28·8 wt % MgO. The medium-grained spinifex has 24·2–27·1 wt % MgO, and the coarse-grained spinifex 21·4–26·3 wt % MgO. A similar pattern of the coarse spinifex forming the least magnesian portion of individual flows has been found by Viljoen et al. (1983) and may be explained by larger proportions of more evolved melt. The flow tops have variable MgO contents, between 23·9 and 29·7 wt %. Smith & Erlank (1982) assigned this variation to different degrees of partial melting of the mantle source. Alternatively, it may be due to fractionation of olivine. Our data are in accord with the findings of Nesbitt et al. (1979) and Smith & Erlank (1982), in that the komatiites of the Barberton greenstone belt are relatively depleted in Al2O3 and TiO2 compared with those from many other localities, e.g. at Munro Township or Kambalda.
|
|
The Westonaria and Roodekrans komatiites are not plotted in Fig. 3 because they are more pervasively altered and because very little is known about their geochemistry, field relations and petrology. The Westonaria samples have MgO contents between 21 and 24·6 wt %, slightly higher Al2O3 contents than the Barberton komatiites, but significantly higher TiO2, at c. 1 wt % (Myers et al., 1990). By analogy with other Ti-rich komatiites (Clarke, 1970; Saverikko, 1983; Barnes & Often, 1990) this may suggest smaller degrees of partial melting than for the Barberton magmas. The Roodekrans komatiites have between 23·6 and 30·0 wt % MgO, relatively high Al2O3 (c. 7 wt %) and low TiO2 contents (c. 0·2 wt %) (Anhaeusser, 1977). These compositional traits are reminiscent of komatiites from Newfoundland (Upadhyay, 1982) and could indicate larger degrees of partial melting than for the Barberton komatiites.
Trace element geochemistry
Trace element concentrations of the analysed samples are listed in Tables 3 and 4. Selected lithophile and moderately chalcophile trace elements of the Komati samples are plotted vs MgO in Fig. 4. Nickel and Co are compatible with respect to olivine and thus both elements show broadly positive correlations with MgO, with regression intercepts at approximately 3500 ppm Ni and 130 ppm Co. Zirconium shows a negative correlation with MgO and samples broadly plot along the olivine control line, indicating that we can use the Zr contents of the rocks as an index of crystallization. Concentrations of Cr also appear to be controlled by the relative proportions of olivine and melt, indicating that chromite was not a liquidus phase during the crystallization of the lavas. Regression analysis indicates that the olivine contains some 1500 ppm Cr, which is a broadly similar value to that found by Puchtel & Humayun (2000).
|
|
The highly chalcophile elements (Cu, Au and PGE) are plotted vs MgO in Fig. 5. Platinum, Pd and Os show the largest amount of scatter, probably because of their relatively high analytical errors in the INAA technique (Table 1). All PGE display a tendency to increase in the less magnesian samples, suggesting incompatible behaviour during crystallization (Fig. 5). This is confirmed by the broadly constant Pd/Ir ratios with falling MgO (Fig. 6). For Ir and Rh, bulk partition coefficients between the crystallizing assemblage and the magma may be calculated by regression analysis indicating a bulk D value of 0·75 for Ir and 0·5 for Rh. For the other elements, the analytical scatter is too large to compute meaningful regressions. However, it appears that for Cu, Au, Pd, Pt and Ru most of the data fall between tie-lines from the flow tops/fine spinifex to Cu–Au–Pd-depleted olivine, suggesting bulk D values near zero.
|
|
The enrichment of Pt and Pd in the less magnesian samples is in agreement with previously published data on komatiites from Barberton (Keays, 1995) and elsewhere (Puchtel & Humayun, 2000). However, in most of the previous studies, the IPGE show a positive correlation with MgO indicating compatible behaviour during fractionation (Crocket and McRae, 1986; Brügmann et al., 1987; Keays, 1995). The only other komatiite study where the IPGE show a negative correlation with MgO is that of Puchtel & Humayun (2000) on lavas from Kostomuksha.
PGE data for the Westonaria and Roodekrans komatiites are compared with those of the Komati lavas in Table 5 and Fig. 7. It is evident that the mantle-normalized patterns of the three South African komatiite suites have similar shapes, reflecting relatively low Pd/Ir, and, for two of the three suites, high Cu/Pd ratios. But perhaps the most notable characteristic of the South African komatiites is that their PGE contents, and Pt and Pd in particular, appear to be markedly lower than those of most other komatiites (Table 5; Fig. 8). Similarly low Pd and Ir values were reported by Keays (1983, 1995) in his study of Barberton komatiites. Tredoux & McDonald (1996) found significantly higher levels of PGE in a bulk sample collected in the Geluk Formation, but Pearson & Woodland (2000) have recently reanalysed the sample and found markedly lower PGE contents more in line with the present data.
|
|
|
The basaltic samples from the Roodekrans Complex have more fractionated PGE patterns, with the IPGE and Rh being depleted and Pd and Au being enriched, relative to the associated komatiites. This confirms that the IPGE, Rh and Pt become compatible during advanced magmatic fractionation (see also Puchtel & Humayun, 2000), whereas Pd and Au tend to remain incompatible in the absence of S saturation (Philipp et al., 2001).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Reasons for the relatively low PGE concentrations in the South African komatiites
Several reasons for the PGE depletion of the South African komatiites relative to most other komatiites elsewhere may be considered, as follows.
(1) Sulphide segregation during ascent and/or emplacement. There are numerous examples of mafic–ultramafic rocks depleted in PGE relative to Cu and Ni (e.g. Duluth, Theriault et al., 2000; Pechenga, Barnes et al., 2001). In most cases this seems to be the result of sulphide segregation in the crust. The only example of a komatiite with similarly low PGE contents to the present samples is from Mt Keith in Australia, where sulphide segregation in the crust is the preferred model (Dowling & Hill, 1992). Removal of a small percentage of sulphide and accumulation of olivine could account for the distribution of the Cu and Pd in the Barberton rocks (Fig. 9), but is an unlikely proposition in view of the uniform PGE depletion observed in all South African komatiite suites (Table 5, Fig. 7). More importantly, this model cannot explain the relatively low Pd/Ir ratios of the South African komatiites.
|
(2) Sulphide retention in the source. Based on the concentration of Zr in the chilled flow tops, and assuming batch melting of primitive mantle with 10·5 ppm Zr (McDonough & Sun, 1995), we can model between 26 and 33% partial melting of the mantle for the Komati lavas. The S solubility of basaltic–picritic magmas depends strongly on the depth of melting (Mavrogenes & O'Neill, 1999). Assuming that the primitive mantle contained c. 200 ppm S (Lorand, 1990) and that low-pressure basalts may dissolve up to 1000 ppm S, some 20–25% partial melting is required for such magmas to consume all the sulphides in the mantle source (e.g. Barnes, 1987). The S solubilities of komatiites are less well constrained. In their experiments on picrites, Mavrogenes & O'Neill found that at 14 GPa and 2000°C the magma may dissolve some 386 ppm, whereas at 10 GPa and 1810°C, it dissolves 685 ppm S (J. Mavrogenes, personal communication, 2003). If these results are representative of komatiites, then late Archaean Al-undepleted komatiites require some 30% melting to consume all sulphides in the source. Early Archaean Al-depleted komatiites would require some 50% partial melting to consume all sulphides in the source. Thus, at the high pressure at which the Barberton komatiites are believed to have separated from the mantle source it is likely that some sulphides remained in the source. Bulk partition coefficients of the PGE into the source are given in Table 6. It is evident that some phase has retained all the PGE and Au in the mantle. The only phase known to host Pd in the mantle is Cu-rich sulphide (e.g. Bulanova et al., 1996; Lorand & Alard, 2001). Therefore, we argue that some Cu-rich sulphides have been retained in the source. If Dsul/sil = 10 000 then 0·0127% sulphides were retained (or c. 44 ppm S). The other PGE have higher D values into the source than Pd suggesting another phase (or phases) in addition to the Cu-rich sulphide is/are retaining them [possibly monosulphide solid solution (mss) and metal alloys, Mitchell & Keays, 1981; Bulanova et al., 1996; Lorand et al., 1999; Alard et al., 2000]. This phase shows the greatest preference for Os, Ir and Ru, and a slightly lower preference for Rh and Pt, i.e. an order of compatibility with regard to the mantle source that has also been found in previous studies (Barnes & Picard, 1993; Philipp et al., 2001). If it is assumed that the partition coefficient for all PGE with regard to sulphide is similar then the partition coefficient with regard to the other phase(s) is c. 2–4 for Os, Ir and Ru, and unity for Rh and Pt.
|
(3) Distinct mantle sources. Mantle samples (including xenoliths, alpine-type, ophiolitic and abyssal peridotites) have higher PGE contents than expected assuming core formation with quantitative removal of HSE (Chou, 1978). This phenomenon is often explained by addition of a late chondritic veneer (Jagoutz et al., 1979). It could be argued that at 3·5 Ga different types of chondrites had insufficiently mixed with the whole mantle and hence the PGE were inhomogeneously distributed (Spettel et al., 1991; Schmidt et al., 2000; Morgan et al., 2001). For example, the lower mantle could have been initially depleted in PPGE (Rh, Pt, Pd) relative to the upper mantle (Morgan et al., 2001). However, based on the similarity in PGE concentration patterns between the Komati suite and the much younger Westonaria suite, age appears to be of limited importance in controlling the PGE patterns of the South African lavas. Pattou et al. (1996) proposed the existence of more local heterogeneities in the mantle. The data of Morgan et al. (2001) may support this model. They show that parts of the mantle underlying the USA and France are relatively Pd rich. Accordingly, one may use the PGE data from the South African komatiites to suggest that the mantle underlying the Kaapvaal Craton shows a local Pt and Pd depletion. However, this would seem unlikely in view of the fact that the world's largest PGE deposits (i.e. in the Bushveld Complex) have formed there subsequently. Finally, there could have been incomplete separation of the core (Jones & Drake, 1986), with the lower mantle having experienced a more complete segregation of PPGE-rich sulphides than the upper mantle. This possibility could be tested by examining Al-depleted komatiites elsewhere for PGE.
Fractionation of the PGE during cooling and crystallization
The incompatible behaviour of all PGE observed during crystallization of the Komati lavas is relatively uncommon in komatiite suites. In most cases, the IPGE behave in a compatible manner, with only the Kostomuksha komatiites showing a similar incompatible behaviour of the IPGE to the Komati lavas. We considered whether the lavas could have been sulphide-saturated, with finely dispersed sulphide droplets concentrating all PGE but being sufficiently small to remain suspended in the convecting magma rather than accumulating in the massive parts of the flow. Fluid dynamic calculations by Lesher & Groves (1986) and experiments by de Bremond d'Ars et al. (2001) have indicated that sulphide entrainment in turbulently flowing magma is realistic, but for the Komati lavas the model is impractical as the concentration of Ir, i.e. the element with the highest analytical precision, can be explained as mixtures between olivine and the chilled flow tops/fine random spinifex, suggesting Ir control by the proportion of silicate melt.
The incompatible behaviour of the IPGE and Rh with regard to olivine found in the present study is in accord with the data of Mitchell & Keays (1981) and Lorand et al. (1993) on mantle olivines and Maier & Barnes (1999) on olivines in cumulate rocks within the Bushveld Complex. In contrast, Brügmann et al. (1987) and Brenan et al. (2002) proposed that Ru and Rh are compatible with respect to olivine at elevated oxygen fugacities (>QFM + 1, where QFM is quartz–fayalite–magnetite).
Some workers have suggested that the IPGE may partition into chromite (Capobianco et al., 1994; Peach & Mathez, 1996; Righter, 2001). The effect of chromite crystallization on IPGE concentration cannot be tested in the present study, as chromite is not a liquidus phase (Fig. 4). Crocket & McRae (1986) found little correlation between Cr and Ir in komatiite flows at Munro Township, and Handler & Bennett (1999) found no Ru enrichment in mantle spinel separates. These studies therefore provide no support for IPGE partitioning into chromite.
Direct crystallization of PGM from silicate magma has been proposed by Merkle (1992), Lorand et al. (1993) and Tredoux et al. (1995), but Peach & Mathez (1996) argued that the high solubilities of the PGE in basaltic magma would preclude direct nucleation and crystallization of PGM from silicate melt. Experimental work by Brenan & Andrews (2001) suggests that PGM crystallization from basaltic magma may be feasible, but it is possible that the extrusion temperature of komatiites exceeds the maximum thermal stability of the PGM. For example, laurite has been shown to be stable up to c. 1275°C in mafic magmas (e.g. Brenan & Andrews, 2001). The present study indicates that the minerals crystallizing from the komatiite host some Ir and Rh, but whether these PGE are controlled by olivine or PGM remains unclear.
PGE host phases in the mantle
Significant advances have recently been made concerning our knowledge of the phases hosting the PGE in the Earth's upper mantle. One of the key observations is that in both the primordial and the lithospheric mantle the PGE are largely hosted by different types of sulphides (Bulanova et al., 1996; Alard et al., 2000; Lorand & Alard, 2001; Luguet et al., 2001). The IPGE and Rh are mainly accommodated by mss that is often included in silicate minerals, whereas Pd is mainly hosted by interstitial Cu-rich sulphides. The nature of the phase(s) hosting Pt remains less clear. A discrete Pt-rich phase has been postulated by Alard et al. (2000) and Luguet et al. (2001), but whether this is primary or has formed during alteration remains unclear. Evidence for the existence of additional refractory metal phases that may accommodate Ir, Ru and Rh has been summarized by Lorand et al. (1999).
Sulphide and PGE abundances and inter-element ratios may vary on a local scale. Mss-rich mantle samples may represent residual rocks after partial melting (Bulanova et al., 1996), whereas Cu–Pd-rich samples may have been metasomatized by secondary melts (Lorand & Alard, 2001). Our results are in accord with sulphide control of the PGE during partial melting. We suggest that the low PGE contents in the South African magmas are the result of some of the Cu–Pd-rich interstitial sulphides having been retained in the mantle. The silicate-included mss would also have been largely refractory, resulting in the relatively low Pd/Ir ratios. We would further argue that at the relatively larger degrees of melting applicable to Al-undepleted komatiites, the interstitial Cu–Pd-rich sulphides were largely consumed, resulting in generally higher PPGE contents. Pd/Ir ratios are higher than in the Al-depleted komatiites because some of the mss remained shielded from interaction with the silicate melt, possibly producing IPGE-enriched harzburgites similar to those described by Lorand & Alard (2001). The model does not explain why Pt/Pd ratios of komatiites and basalts worldwide are sub-chondritic and broadly similar (e.g. Crocket, 2002), although the basaltic samples from Roodekrans do not confirm this trend. Clearly, some Pt is retained in the mantle, irrespective of the degree of melting. Notably, alkaline basalts tend to have higher Pt/Pd than tholeiitic basalts and komatiites (McDonald et al. 1995; Crocket, 2002), and we suggest that this is because the Pt alloys may be more fusible during the wet melting conditions proposed for many alkaline magmas (e.g. Francis & Ludden, 1995). We feel that this relationship may be interpreted to support the primary nature of the discrete Pt phases postulated by Alard et al. (2000) and Luguet et al. (2001).
The possible effects of alteration on PGE concentration patterns
All our komatiite samples are pervasively altered and thus one could suggest that the relatively large amount of scatter observed for some of the PGE, particularly Os and Pd as well as Cu and Au (Fig. 5), could be due to remobilization of the elements during alteration. Mobilization of the PGE would compromise efforts to model the PGE distributions by primary magmatic processes. The available data on PGE mobility during alteration of rocks have been reviewed by Wood (2002). The data show that the PGE are relatively immobile under most conditions, with Pd showing the greatest mobility. Bandyayera (1997) reported that all PGE are immobile during low-grade metamorphism of ultramafic rocks, but Au, Pd and, to a lesser degree, Pt may be remobilized during lateritization. Together with the similarity in PGE concentrations and inter-element ratios found in all South African komatiites analysed by us, this information would suggest that the PGE patterns of the komatiites are of a primary nature.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
South African komatiites have lower PGE concentrations and Pd/Ir ratios than Al-undepleted komatiites from elsewhere. In view of the distinct melting regime of the Barberton komatiites (i.e. relatively small degrees of batch melting with garnet remaining in the source), this suggests that (1) sulphides remained in the source and retained PGE, and (2) that the PGE are fractionated during partial melting of the mantle, with the IPGE, Rh and Pt being hosted by more refractory mantle phases than Pd.
In the Komati Formation, the PGE and Cu tend to increase in the less magnesian samples. This is a function of the incompatible behaviour of the PGE during magmatic fractionation, and PGE control by the proportion of melt relative to cumulus olivine. Enhanced partitioning of the IPGE into olivine is not indicated. Partitioning of IPGE and Rh into chromite cannot be constrained by the present study, as chromite is not a liquidus phase.
|
ACKNOWLEDGEMENTS |
|---|
This study was funded by the Centre for Research on Magmatic Ore Deposits at the University of Pretoria, which receives funding from Falconbridge Ventures of Africa, Anglovaal Ltd., Impala Platinum, and Anglo American Base Metals. S.-J. B. receives funding from NSERC. C. Anhaeusser helpfully provided the samples from Roodekrans and Westonaria. We thank N. T. Arndt for his helpful comments, and J. P. Lorand and two anonymous referees for their careful reviews.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONGENERAL GEOLOGICAL SETTINGPREVIOUS CHEMICAL WORKSAMPLE DESCRIPTIONANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Alard, O., Griffin, W. L., Lorand, J.-P., Jackson, S. E. & O'Reilly, S. Y. (2000). Non-chondritic distribution of the highly siderophile elements in mantle sulfides. Nature 407, 891–894.[CrossRef][Medline]
Anhaeusser, C. R. (1977). Geological and geochemical investigations of the Roodekrans Ultramafic Complex and surrounding Archaean volcanic rocks, Krugersdorp district. Transactions of the Geological Society of South Africa 80, 17–28.
Arndt, N. T. (1994). Archean komatiites. In: Condie, K. C. (ed.) Archaean Crustal Evolution. Amsterdam: Elsevier, pp. 11–44.
Bandyayera, D. (1997). Formation des laterites nickelifères et mode de distribution des éléments du groupe du platine dans les profils lateritiques du complexe de Musongati, Burundi. Ph.D. thesis, Université du Québec à Chicoutimi, 440 pp.
Barnes, S.-J. (1983). A comparative study of olivine and clinopyroxene spinifex flows from Alexo, Abitibi greenstone belt, Canada. Contributions to Mineralogy and Petrology 83, 293–308.[CrossRef][ISI]
Barnes, S.-J. (1987). Unusual base to noble element ratios in the Rana Layered Intrusion, Northern Norway. Norsk Geologisk Tydskrift 67, 215–231.
Barnes, S.-J. & Maier, W. D. (1999). The fractionation of Ni, Cu and the noble metals in silicate and sulfide liquids. In: Keays, R. R., Lesher, C. M., Lightfoot, P. C. & Farrow, C. E. G. (eds) Dynamic Processes in Magmatic Ore Deposits and their Application to Mineral Exploration. Geological Association of Canada, Short Course Notes 13, 69–106.
Barnes, S.-J. & Often, M. (1990). Ti-rich komatiites from northern Norway. Contributions to Mineralogy and Petrology 105, 42–54.[CrossRef][ISI]
Barnes, S.-J. & Picard, C. P. (1993). The behaviour of platinum-group elements during partial melting, crystal fractionation, and sulphide segregation: an example from the Cape Smith Fold Belt, northern Quebec. Geochimica et Cosmochimica Acta 57, 79–87.[CrossRef][ISI]
Barnes, S.-J., Melezhik, V. A. & Sokolov, S. V. (2001). The composition and mode of formation of the Petchenga nickel deposits, Kola peninsula, northwestern Russia. Canadian Mineralogist 39, 447–471.
Blichert-Toft, J. & Arndt, N. T. (1999). Hf isotope compositions of komatiites. Earth and Planetary Science Letters 171, 439–451.[CrossRef][ISI]
Brenan, J. M. & Andrews, D. (2001). High-temperature stability of laurite and Ru–Os–Ir alloy and their role in PGE fractionation in mafic magmas. Canadian Mineralogist 39, 341–360.
Brenan, J. M., Dalpe, C. & McDonough, W. F. (2002). PGEs are fractionated by olivine–melt partitioning. Abstracts, Goldschmidt Conference, Davos, A103.
Brügmann, G. E., Arndt, N. T., Hofmann, A. W. & Tobschall, H. J. (1987). Noble metal abundances in komatiite suites from Alexo, Ontario, and Gorgona Island, Colombia. Geochimica et Cosmochimica Acta 51, 2159–2169.[CrossRef][ISI]
Bulanova, G. P., Griffin, W. L., Ryan, C. G., Shestakova, O. Ye. & Barnes, S.-J. (1996). Trace elements in sulfide inclusions from Yakutian diamonds. Contributions to Mineralogy and Petrology 124, 111–125.[CrossRef][ISI]
Capobianco, C. J., Hervig, R. L. & Drake, M. J. (1994). Experiments on crystal/liquid partitioning of Ru, Rh and Pd for magnetite and hematite solid solutions crystallized from silicate melt. Chemical Geology 113, 23–43.[CrossRef][ISI]
Chou, C. L. (1978). Fractionation of siderophile elements in the earth's upper mantle. Proceedings of the Lunar and Planetary Science Conference. Geochimica et Cosmochimica Acta Supplement 219–230.
Clarke, D. B. (1970). Tertiary basalts of Baffin Bay: possible primary magma from the mantle. Contributions to Mineralogy and Petrology 25, 203–224.[CrossRef]
Crocket, J. H. (2002). Platinum-group element geochemistry of mafic and ultramafic rocks. In: Cabri, L. J. (ed.) The Geology, Geochemistry, Mineralogy and Mineral Beneficiation of Platinum-Group Elements. Canadian Institute of Mining, Metallurgy and Petroleum, Special Volume 54, 177–210.
Crocket, J. H. & McRae, W. E. (1986). Platinum-group element distribution in komatiitic and tholeiitic volcanic rocks from Munro Township, Ontario. Economic Geology 81, 1242–1251.[Abstract]
Crocket, J. H. & Skippen, G. B. (1966). Radioactivation determination of palladium in basic and ultramafic rocks. Geochimica et Cosmochimica Acta 30, 129–141.[CrossRef][ISI]
Dann, J. C. (2000). The 3·5 Ga Komati Formation, Barberton Greenstone Belt, South Africa, Part I: New maps and magmatic architecture. South African Journal of Geology 103, 47–68.
de Bremond d'Ars, J., Arndt, N. T. & Hallot, E. (2001). Analog experimental insights into the formation of magmatic sulfide deposits. Earth and Planetary Science Letters 186, 371–381.[CrossRef][ISI]
Dowling, S. E. & Hill, R. R. T. (1992). The distribution of PGE in fractionated Archean komatiites, Western and Central ultramafic units, Mt. Keith region, Western Australia. Australian Journal of Earth Science 39, 349–363.
Fleet, M. E., Crocket, J. H. & Stone, W. E. (1996). Partitioning of platinum-group elements (Os, Ir, Ru, Pt, Pd) and gold between sulfide liquid and basalt melt. Geochimica et Cosmochimica Acta 60, 2397–2412.[CrossRef][ISI]
Francis, D. & Ludden, J. (1995). The signature of amphibole in mafic alkaline lavas, a study in the northern Canadian Cordillera. Journal of Petrology 36, 1171–1191.
Handler, M. R. & Bennett, V. C. (1999). Behaviour of platinum-group elements in the subcontinental mantle of eastern Australia during variable metasomatism and melt depletion. Geochimica et Cosmochimica Acta 63, 3597–3618.[CrossRef][ISI]
Herzberg, C. (1992). Depth and degree of melting of komatiite. Journal of Geophysical Research 97, 4521–4540.
Herzberg, C. (1995). Generation of plume magmas through time: an experimental approach. Chemical Geology 126, 1–16.[CrossRef][ISI]
Jagoutz, E., Palme, H., Baddenhausen, H., Blum, K., Cendales, M., Breibus, G., Spettel, B., Lorenz, V. & Wänke, H. (1979). The abundances of major, minor and trace elements in the earth's mantle as derived from primitive ultramafic nodules. Proceedings of the 10th Lunar and Planetary Science Conference. Geochimica et Cosmochimica Acta Supplement 2031–2050.
Jones, J. H. & Drake, M. J. (1986). Geochemical constraints on core formation in the Earth. Nature 322, 221–228.[CrossRef]
Juvonen, R., Lakomaa, T. & Soikkeli, L. (2002). Determination of gold and platinum group elements in geological samples by ICP-MS after nickel sulphide fire assay: difficulties encountered with different types of geological samples. Talanta 58, 595–603.[CrossRef]
Keays, R. R. (1983). Archean gold deposits and their source rocks: the upper mantle connection. In: Foster, R. P. (ed.) Gold '82—the Geology, Geochemistry and Genesis of Gold Deposits. Rotterdam: Balkema, pp. 17–51.
Keays, R. R. (1995). The role of komatiitic and picritic magmatism and S-saturation in the formation of ore deposits. Lithos 34, 1–18.[CrossRef][ISI]
Keays, R. R. & Scott, R. B. (1976). Precious metals in ocean-ridge basalts: implications for basalts as source rocks for gold mineralization. Economic Geology 71, 705–720.[Abstract]
Lesher, C. M. & Groves, D. I. (1986). Controls on the formation of komatiite-associated nickel–copper sulfide deposits. In: Friedrich, G. H., Genkin, A., Naldrett, A. J., Ridge, J. D., Sillitoe, R. H. & Vokes, F. M. (eds) Proceedings, 27th International Geological Congress. Berlin: Springer, pp. 43–62.
Lorand, J.-P. (1990). Are spinel lherzolite xenoliths representative of the abundance of sulfur in the mantle? Geochimica et Cosmochimica Acta 54, 1487–1492.[CrossRef][ISI]
Lorand, J.-P. & Alard, O. (2001). Platinum-group element abundances in the upper mantle: new constraints from in situ and whole-rock analyses of Massif Central xenoliths (France). Geochimica et Cosmochimica Acta 65, 2789–2806.[CrossRef][ISI]
Lorand, J.-P., Keays, R. R. & Bodinier, J. L. (1993). Copper and noble metal enrichments across the lithosphere–asthenosphere boundary of mantle diapirs: evidence from the Lanzo lherzolite massif. Journal of Petrology 34, 1111–1140.
Lorand, J.-P., Pattou, L. & Gros, M. (1999). Fractionation of platinum-group elements and gold in the upper mantle: a detailed study in Pyrenean orogenic lherzolites. Journal of Petrology 40, 957–981.[CrossRef][ISI]
Luguet, A., Alard, O., Lorand, J.-P., Pearson, N. J., Ruan, C. & O'Reilly, S. Y. (2001). Laser-ablation microprobe (LAM)–ICPMS unravels the highly siderophile element geochemistry of the oceanic mantle. Earth and Planetary Science Letters 189, 285–294.[CrossRef][ISI]
Maier, W. D. & Barnes, S.-J. (1999). Platinum-group elements in silicate rocks of the Lower, Critical and Main Zones at Union Section, western Bushveld Complex. Journal of Petrology 40, 1647–1671.[CrossRef][ISI]
Mainville, M. (1994). Les komatiites et tholeiites à la base du Groupe de Baby, Temiscamingue. M.Sc. thesis, Université du Québec à Chicoutimi, 146 pp.
Marsh, J. S., Bowen, M. P., Rogers, N. W. & Bowen, T. B. (1992). Petrogenesis of Late Archean flood-type basic lavas from the Klipriviersberg Group, Ventersdorp Supergroup, South Africa. Journal of Petrology 33, 817–847.
Mavrogenes, J. A. & O'Neill, H. St. C. (1999). The relative effects of pressure, temperature and oxygen fugacity on the solubility of sulphide in mafic magma. Geochimica et Cosmochimica Acta 63, 1173–1180.[CrossRef][ISI]
McDonald, I., de Wit, M., Smith, C. B., Bizzi, L. A. & Viljoen, K. S. (1995). The analysis of the platinum-group elements in Brazilian and southern African kimberlites. Geochimica et Cosmochimica Acta 59, 2883–2903.[CrossRef][ISI]
McDonough, W. F. & Sun, S.-s. (1995). The composition of the Earth. Chemical Geology 120, 223–253.[CrossRef][ISI]
McIver, J. R., Cawthorn, R. G. & Wyatt, B. A. (1982). The Ventersdorp Supergroup—the youngest komatiite sequence in South Africa. In: Arndt, N. T. & Nesbitt, E. G. (eds) Komatiites. London: George Allen and Unwin, pp. 81–90.
Merkle, R. K. W. (1992). Platinum-group minerals in the middle group of chromitite layers at Marikana, western Bushveld Complex: indications for collection mechanisms and postmagmatic modification. Canadian Journal of Earth Sciences 29, 209–221.
Mitchell, R. H. & Keays, R. R. (1981). Abundance and distribution of gold, palladium and iridium in some spinel and garnet lherzolites. Implications for the nature and origin of precious-metal rich intergranular components in the upper mantle. Geochimica et Cosmochimica Acta 45, 2425–2442.[CrossRef][ISI]
Morgan, J. W., Walker, R. J., Brandon, A. D. & Horan, M. F. (2001). Siderophile elements in Earth's upper mantle and lunar breccias: data synthesis suggests manifestations of the same late influx. Meteoritics and Planetary Science 36, 1257–1275.
Myers, R. E., McCarthy, T. S., Bunyard, M., Cawthorn, R. G., Falatsa, T. M., Hewitt, T., Linton, P., Myers, J. M., Palmer, K. J. & Spencer, R. (1990). Geochemical stratigraphy of the Klipriviersberg Group volcanic rocks. South African Journal of Geology 93, 224–238.[Abstract]
Nesbitt, R. W., Sun, S.-S. & Purvis, A. C. (1979). Komatiites: geochemistry and genesis. Canadian Mineralogist 17, 165–186.
Pattou, L., Lorand, J.-P. & Gros, M. (1996). Non-chondritic platinum-group element ratios in the Earth's mantle. Nature 379, 712–715.[CrossRef]
Peach, C. L. & Mathez, E. A. (1996). Constraints on the formation of platinum-group element deposits in igneous rocks. Economic Geology 91, 439–450.[Abstract]
Pearson, D. G. & Woodland, S. J. (2000). Solvent extraction/anion exchange separation and determination of PGEs (Os, Ir, Pt, Pd, Ru) and Re–Os isotopes in geological samples by isotope dilution ICP-MS. Chemical Geology 165, 87–107.[CrossRef][ISI]
Philipp, H., Eckhardt, J.-D. & Puchelt, H. (2001). Platinum-group elements (PGE) in basalts of the seaward-dipping reflector sequence, SE Greenland coast. Journal of Petrology 42, 407–432.