Journal of Petrology Advance Access originally published online on May 20, 2005
Journal of Petrology 2005 46(10):2059-2090; doi:10.1093/petrology/egi047
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Contrasting Group I and Group II Eclogite Xenolith Petrogenesis: Petrological, Trace Element and Isotopic Evidence from Eclogite, Garnet-Websterite and Alkremite Xenoliths in the Kaalvallei Kimberlite, South Africa
1 GEOSCIENCE CENTRE, DE BEERS CONSOLIDATED MINES LTD, P.O. BOX 82232, SOUTHDALE, 2135, SOUTH AFRICA
2 BERNARD PRICE INSTITUTE OF GEOPHYSICAL RESEARCH, HUGH ALLSOPP LABORATORY, UNIVERSITY OF THE WITWATERSRAND, PRIVATE BAG 3, WITS, JOHANNESBURG, 2050, SOUTH AFRICA
3 DEPARTMENT OF GEOLOGY, UNIVERSITY OF TORONTO, ERINDALE COLLEGE, MISSISSAUGA, ONTARIO, CANADA L5L 1C6
RECEIVED MARCH 15, 2004; ACCEPTED APRIL 4, 2005
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONEclogite xenolith varieties at Kaalvallei include accessory-free bimineralic xenoliths, diamond-bearing eclogite, corundum-bearing eclogite, garnet-websterite, alkremite and spinel-bearing eclogite. The xenoliths can be accurately classified into previously defined Group I and Group II varieties on both petrographic and geochemical principles. Kaalvallei Group I eclogites (including diamond-bearing eclogite) are considered to derive from a heterogeneous protolith. Eclogite genesis might have been by residue formation associated with the dehydration and partial melting of a protolith consisting of variably mixed, subducted ocean floor basalt and sediment. Kaalvallei Group II eclogite xenoliths are likely to have formed through crystallization of small-volume melts within conduits in old, enriched subcontinental lithosphere. Kaalvallei websterite xenoliths might be petrogenetically related to Group II xenoliths. Isotopic data for Kaalvallei corundum-bearing eclogite and alkremite xenoliths do not provide constraints on petrogenesis. Spinel-bearing eclogite xenoliths are ultradepleted in virtually all trace elements, with very low light rare earth element contents, relatively high heavy rare earth element concentrations, extreme 87Sr/86Sr (
0·915) and extreme 143Nd/144Nd (0·517) isotopic compositions. These xenoliths are considered to be the residues of a partial melting event. KEY WORDS: eclogite xenoliths; trace elements; REE; isotopic composition; diamonds
|
INTRODUCTION |
|---|
Eclogites are coarse-grained rocks comprising clinopyroxene and garnet. When present as xenoliths in kimberlite they usually represent portions of the upper mantle that have been transported to the surface of the Earth. Some eclogites are now recognized as an important source rock of diamond, as evidenced by the occurrence of accessory diamond in many eclogite xenoliths (Dawson & Carswell, 1990
) as well as the presence of eclogitic mineral inclusions in diamond (Meyer, 1987). These fragments of the Earth's mantle form a significant proportion of the mantle xenolith suite of many kimberlites and have been studied at numerous xenolith localities in areas such as west Africa (e.g. Hills & Haggerty, 1989; Fung & Haggerty, 1995; Barth et al., 2001, 2002; Jacob, 2004), southern Africa (e.g. Schulze & Helmstaedt, 1988; Shervais et al., 1988; Caporuscio & Smyth, 1990; Gurney et al., 1991a; O'Reilly & Griffin, 1995; Harte & Kirkley, 1997; Pyle & Haggerty, 1998; Schulze et al., 2000; Jacob et al., 2003; Jacob, 2004; Schmickler et al., 2004), Siberia (e.g. Jerde et al., 1993a, 1993b; Jacob et al., 1994; Snyder et al., 1995, 1997; Beard et al., 1996; Jacob & Foley, 1999; Taylor et al., 2003; Jacob, 2004) and Canada (e.g. Griffin et al., 1999; Pearson et al., 1999; Heaman et al., 2002; Jacob, 2004).
A classification scheme based on the texture of eclogites was first used by MacGregor & Carter (1970) in a study of xenoliths from the Roberts Victor kimberlite in South Africa. They recognized two groups: Group I in which subhedral or rounded garnets are set in an interstitial matrix of clinopyroxene, with occasional clinopyroxene poikilitically enclosing rounded garnet; and Group II, consisting of anhedral, sometimes elongate, straight-edged garnet and pyroxene that have an interlocking fabric. A further classification, linking textural and chemical criteria, is based on the Na2O content of garnet and the K2O content of clinopyroxene. In a study of Roberts Victor eclogites that have unambiguous Group I or Group II textures, McCandless & Gurney (1989) found that most (81%) of Group I garnet contains >0·09 wt % Na2O, whereas 89% of the Group II garnet has <0·09 wt % Na2O. For K2O in clinopyroxene, 94% of Group I pyroxenes contain >0·08 wt % and 76% of Group II pyroxenes contains less than these concentrations.
The apparent separation of eclogite xenoliths into Group I and Group II varieties is an interesting and perplexing aspect of eclogite petrology. It is now recognized that diamond is confined to eclogites with Group I textural characteristics whereas graphite occurs in both textural types (Gurney, 1991b). The mineral compositions of eclogitic inclusions in diamonds suggest that these diamonds are derived only from Group I, and such xenoliths have been found only in on-craton kimberlites (Robinson et al., 1984; Gurney, 1991b). In view of these features (as well as economic aspects), a detailed integrated petrological study of both xenolith varieties from individual localities is desirable. However, most studies of eclogite xenoliths typically deal with specific varieties rather than a full xenolith suite, and, as a consequence, there are few published geochemical data available for both Group I and Group II eclogite xenoliths from individual localities.
Collection of eclogite xenoliths from the Kaalvallei kimberlite was accomplished over an extended time span, and has resulted in assembly of a suite of >1000 eclogite specimens (Viljoen, 1994). Subsequent microprobe analysis of approximately 350 specimens, coupled with the very fresh state of most xenoliths and the varied petrological nature of the suite, allowed the selection of a set of xenoliths for isotopic study that are fully representative of the eclogite suite as a whole. The suite is also unique in that Group I and Group II eclogites are easily recognizable on both visual as well as chemical grounds. Kaalvallei therefore serves as an ideal locality on which to acquire more information on the contrasting petrological characteristics of Group I and Group II eclogite xenoliths. The purpose of this contribution is therefore to provide a compilation of trace element and isotopic data for a full suite of typical Kaalvallei eclogite xenoliths and to use this dataset to further explore the genetic history of these rocks.
|
LOCALITY AND SAMPLE DESCRIPTION |
|---|
The Kaalvallei kimberlite pipe is located in the Free State of South Africa on the farm Kaalvallei 12, approximately 7·5 km from Welkom along the Welkom–Virginia road (Fig. 1). The kimberlite occurs on the Kaapvaal craton and intrudes sedimentary rocks of the Ecca Group. The pipe is roughly circular with a surface area of 2 ha. A satellite intrusion occurs 300 m west of the main pipe, and this is filled with weathered hypabyssal-facies kimberlite breccia. The main pipe consists almost exclusively of two types of diatreme-facies tuffisitic kimberlite breccia cut by a large number of internal kimberlite dykes (Stiefenhofer, 1989
). Monticellite-bearing kimberlite varieties predominate and the body is classified petrographically and isotopically as a Group I kimberlite after Smith (1983) and Skinner (1989). It has an emplacement age of 85 Ma (Viljoen, 1994).
|
The xenolith suite contains members of most mantle rock types, including spinel peridotite, coarse and deformed garnet peridotite and eclogite, as well as a suite of Cr-poor megacrysts (Schulze, 1989
). The samples analysed here (Tables 1 and 2) were all collected from concentrates at the mine, and typically range in size from 2 cm to 5 cm. Individual specimens were carefully selected for detailed study from a much larger collection of samples for which mineral compositional data are available. Sample selection was primarily based on mineral compositions (to assemble a fully representative suite of specimens) as well as specimen size and lack of alteration. In total, constituent minerals from 23 Group I eclogites (including four diamond-bearing samples), 10 bimineralic Group II eclogites, four corundum-bearing eclogites, two spinel-bearing eclogites, two garnet-websterites and three alkremites were analysed for their trace element and Sr and Nd isotopic composition, and the oxygen isotopic compositions of constituent minerals were determined on a subset of these specimens.
|
|
The majority of eclogite xenoliths from Kaalvallei are bimineralic and are texturally identical to the Group I eclogite xenoliths (coarse rocks with rounded garnets) and Group II eclogites (finer-grained rocks with straight grain boundaries) recognized at Roberts Victor (MacGregor & Carter, 1970
; Hatton, 1978). Group I eclogite xenoliths at Kaalvallei are characterized by large (up to 8 mm), rounded, dark reddish brown garnet set in enclosing dark, greenish-blue to black clinopyroxene, which reach up to 1 cm or more in size (Fig. 2a). Garnet:clinopyroxene typically ranges between 40:60 and 60:40 (Table 1). Minute quantities of orthopyroxene (<<1 vol. %) are present in Group I eclogite xenoliths 196 and 374. All of the diamond-bearing eclogite xenoliths recovered from the Kaalvallei kimberlite are of the Group I variety (this study; Kiviets et al., 1995; Kiviets, 2000). When present, diamond is typically in the form of sharp-edged octahedra with graphite coats and surface textures comprising ribbing, serrate laminae and knob-like asperities (Kiviets, 2000). They are therefore very similar to diamonds from other eclogite xenoliths; for example, those from Orapa (Robinson et al., 1984). Minerals in the Group II eclogite variety are generally much smaller than in the Group I eclogite xenoliths, ranging in size up to 2–3 mm for both clinopyroxene and garnet (Fig. 2b). The garnet colour in these eclogites is usually shades of pink, pale purple or brown, whereas clinopyroxene is typically bright green to olive green in colour. Grain boundaries are straight, triple junctions common, and garnet and clinopyroxene show an interlocking textural relation. Clinopyroxene is typically the dominant phase and abundances as high as 79% are observed (Table 1).
|
Corundum-bearing eclogite xenoliths are characterized by clinopyroxene (often very cloudy and altered) as well as transparent, very fresh, blue–white garnet and pink corundum. Garnet and clinopyroxene are modally dominant (Table 1) with corundum accounting for 2–14% of the rock volume. The rocks are granular with a fine grain size (typically <1 mm). Corundum is often in the form of elongate, oriented laths and appears to have exsolved from clinopyroxene (Fig. 2c).
Unusual spinel-bearing eclogite xenoliths at Kaalvallei are characterized by dominant pale blue garnet and pale green clinopyroxene with minor interstitial, reddish-purple spinel (Fig. 2d). The spinel has irregular (amoeboid) shapes and is commonly enclosed in garnet.
In the garnet-websterite (Fig. 2e) garnet is typically <4 mm and modally dominant (up to 70% of the total rock volume). Clinopyroxene and orthopyroxene are generally smaller (2 mm) and less abundant. Garnet and clinopyroxene are bright red and bright grass green in colour, whereas orthopyroxene is canary yellow and most commonly occurs on the boundary between these two minerals. The grains show a tightly interlocking fabric. The xenoliths are petrographically identical to websterite xenoliths from Orapa (Shee, 1978) and most of the Group II garnet-websterite xenoliths from Roberts Victor (Hatton, 1978). Hatton (1978) also described coarse-grained orthopyroxene-bearing eclogite xenoliths (with orange garnet and dark blue–green clinopyroxene; classified as Group I eclogites) from Roberts Victor, but the Kaalvallei garnet-websterite xenoliths described here are totally unlike these rocks.
Rare alkremite xenoliths occur. The specimens analysed are mineralogically variable, consisting of dominant garnet and spinel with accessory, primary corundum. Garnet is a pale flesh colour, spinel is typically black (or green in thin section) and the corundum is pinkish red. Grain boundaries are curved and the spinel usually occurs as irregular aggregates, partially or wholly enclosed in garnet (Fig. 2f). The xenoliths are petrographically identical to specimens from other kimberlites.
|
ANALYTICAL METHODS |
|---|
Minerals were analysed for major and trace elements with a fully automated Cameca SX-50 electron microprobe at the De Beers GeoScience Centre. A finely focused 3 µm diameter beam was used for all analyses. Analytical conditions employed were: accelerating voltage of 20 kV; a beam current of 20 nA; 30 s counting time for Na2O and K2O and 20 s for other elements. This resulted in detection limits of 0·02 wt % for Na2O, K2O, Cr2O3 and TiO2, and generally on the order of 0·04 wt % for MnO. The data were processed using an on-line, PAP correction procedure.
Radiogenic isotope analyses were performed in the Hugh Allsopp Laboratory of the Bernard Price Institute of Geophysical Research at the University of the Witwatersrand. Samples for isotopic analysis were crushed in a steel mortar and pestle, and sieved into a size fraction ranging from 250 to 500 µm diameter. Garnet and clinopyroxene were then separated on a magnetic separator, leached in warm 6N HCl overnight and etched in 4% HF for 30 min. Final hand cleaning of all separates was carried out under a microscope and only clear, pristine grains with no surface or internal blemishes were selected for analysis. Subsequently the separates were examined at high magnification in alcohol on a dark-field stage as a final check for purity. This aided in the detection of minute mineral inclusions and these grains were removed from the separates prior to analysis. It is stressed that all grains finally dissolved were optically pure and of uniform colour. The pure mineral separates were dissolved in pure HF in Savillex beakers with a few drops of added HNO3 to retard the formation of insoluble fluorides. After drying, the dissolved samples were taken up in 6N HCl and split into spiked and unspiked aliquots. Rb, Sr and the rare earth elements (REE) were separated first using standard cation resin techniques; further separation of Sm and Nd was performed using HDEHP-coated Teflon powder contained in thin glass columns (Richard et al., 1976). Total method blanks for these procedures never exceeded 140 pg, and were generally <100 pg. Blank contributions (with the exception of Rb), are insignificant for these samples, and blank corrections have not been applied. Concentration measurements were carried out on a single-collector Micromass 30 (Rb, Sm and Nd) and a multicollector VG354 (Sr). All isotopic compositions were determined on the VG354. Nd was analysed utilizing in situ reduction (Noble et al., 1989). Uncertainty estimates are ±0·5 relative percent for all element concentrations. 87Sr/86Sr in the Eimer & Amend standard is 0·708008 ± 32 (1 SD for 49 analyses). 143Nd/144Nd in BCR-1 basalt standard is 0·512642 ± 17 (1 SD for 10 analyses). Long-term duplicate analyses suggest an estimated uncertainty on 147Sm/144Nd of 0·28% and on 143Nd/144Nd of 0·004%.
The routine technique used for oxygen extraction from silicates and oxides in the Department of Geological Sciences at the University of Cape Town has been described in detail by Vennemann & Smith (1990). Hand-picked pure garnet and clinopyroxene mineral separates were crushed in an agate mortar and pestle of known 
18O composition until finer than 200 µm. Samples were leached in 6N HCl and 4% HF prior to grinding to remove impurities. Analyses of leached and unleached clinopyroxene were within error (1 SD = 0·1
) of each other, indicating that the leaching procedure has no effect on 18O compositions. Approximately 10 mg of sample was loaded into seamless nickel reaction vessels, which were overpressurized above atmospheric by dry nitrogen. Once loaded, the reaction vessels were evacuated at 200°C to 10–5 mbar for a minimum of 2 h. The reaction vessels were then charged with about seven times the stoichiometrically required amount of ClF3, isolated and then heated to 550°C (clinopyroxene) or 650°C (garnet) for c. 13 h. The extracted oxygen was converted to CO2 by passing it over a hot platinized carbon rod. Manometric yields were monitored relative to theoretical values and were close to 100% for all clinopyroxene separates. Oxygen yields for garnets were initially never close to 100% and in some cases were as low as 77%, with difficulty experienced in duplicating runs. This was subsequently solved by regrinding to at least 75 µm and reacting at 650°C for up to 16 h. CO2 was analysed for its oxygen isotopic composition using a VG Micromass 602E, double-collector ratio mass spectrometer in the Department of Archaeology at the University of Cape Town. All oxygen isotope values are reported in -notation in per mil units relative to V-SMOW. The average 18O of 13 analyses of quartz standard NBS-28 is 9·64 ± 0·09 (1 SD; 0·93%), which is identical to the certified value for this standard.
Concentrations of selected trace elements in garnet and clinopyroxene were determined by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) analysis at the De Beers GeoScience Centre in Johannesburg, utilizing a frequency-quadrupled Cetac LSX200 Nd-YAG UV laser beam of 266 nm wavelength at a 4 Hz repetition rate, and operating at maximum power (Norman et al., 1996; Norman, 1998; Jackson, 2001). The ablated material was carried to a Perkin Elmer Sciex ELAN 6000 inductively coupled plasma mass spectrometer in a high-purity Ar–He mixture gas stream. A typical analysis consists of a data collection period of 60 s during ablation, and an additional 40 s for background counting. All analytical data were examined in real-time to preclude incorporation of data for mineral inclusions or alteration products. Typical pit diameters were of the order of 100 µm. The NIST 610 glass standard was used to calibrate relative element sensitivities for the analyses, using the data of Pearce et al. (1997). Each analysis was normalized using Si or Ca values determined by electron microprobe. Accuracy, based on LA-ICP-MS measurements of garnet MU53388 (Norman et al., 1996; a megacryst from a Pliocene–Pleistocene alkali basalt at Savaryn-Tscharam in NW Mongolia) is always better than 15% and in most cases better than 10% for all elements analysed (Table 3). Typical detection limits are in the range 10–20 ppb for REE, Sr, Zr, Hf, Y, Ga, and 2 ppm for Ni. The choice of Si as an internal standard for Ni and Ga resulted in significantly improved accuracy relative to MU53388 for these two elements when compared with Ca as internal standard (used for all other elements analysed). This is probably due to laser-induced elemental fractionation, which may be significant for elements such as Li, Si, V, Fe, Ni, Zn and Ga (Jackson, 2001; Günther & Hattendorf, 2001). Norman et al. (1996) showed that relatively constant and reproducible Ca-normalized signals can be obtained for refractory elements such as Sr, Nb, REE, Th and U under a variety of LA-ICP-MS operating conditions. However, more volatile elements such as Pb, Ni, Zn, Ga, Cu, Sn, Sb, Te, Ag, Bi, Rb and Si are especially prone to laser-induced element fractionation. Successful determination of these elements could be achieved by careful matching with internal standards with similar behaviour during ablation (Fryer et al., 1995).
|
|
MAJOR AND MINOR ELEMENT MINERAL CHEMISTRY |
|---|
Averaged microprobe analyses of constituent minerals in the eclogite xenoliths studied are presented in Table 2. Chemical heterogeneity was not detected in garnet and clinopyroxene in Group I eclogite xenoliths. Analyses of individual grains from the rest of the xenoliths tend to be more variable, with ranges up to 0·5 wt % in oxides such as FeO, MgO and CaO in both garnet and clinopyroxene as well as Al2O3 and Na2O in clinopyroxene. Variations in CaO and MgO content <3 wt % are present in garnet from xenoliths 276 (websterite), 306 (spinel eclogite), 325 (corundum eclogite) and 415 (bimineralic Group II eclogite).
Garnet
Garnet from the Group I eclogite xenoliths (including the diamond-bearing xenoliths) are Fe-rich and contain relatively low amounts of CaO whereas those from Group II eclogite, alkremite- and spinel-bearing eclogite xenoliths typically show a trend away from MgO towards CaO enrichment at constant FeO (Fig. 3). Websterite garnet is very magnesian (e.g. 21 wt % MgO), and garnet from corundum eclogite xenoliths is very calcic (e.g. 22 wt % CaO). The garnet in the spinel eclogite xenoliths has very low FeO (e.g. 3 wt %).
|
With respect to minor element chemistry, the garnet from xenoliths classified visually as Group I eclogites (including the diamond-bearing specimens) have low Cr2O3 contents (typically <0·1 wt %), whereas those from xenoliths classified visually as Group II eclogite, corundum eclogite, spinel eclogite and websterite have higher Cr2O3 contents, ranging up to 1·6 wt % Cr2O3 in garnets from sample 171 (a Group II eclogite). Elevated levels of Na2O (Fig. 4) and TiO2 are found in garnet from Group I eclogite xenoliths, whereas the rest of the xenolith garnets have concentrations typically <0·06 wt %.
|
Clinopyroxene
Clinopyroxene compositions mirror those of garnet. Clinopyroxenes from xenoliths classified visually as Group I eclogites (including the diamond-bearing eclogite xenoliths) are the most iron-rich but are depleted in CaO and MgO relative to those from the rest of the eclogite variants. Clinopyroxene from corundum eclogite xenoliths has extremely high concentrations of Na2O (e.g. 7 wt %) and Al2O3 (e.g. 16 wt %) whereas clinopyroxene from Group I eclogites tends to have elevated levels of Na2O relative to the rest of the xenoliths.
In terms of minor elements, the highest Cr2O3 contents are present in clinopyroxene from xenoliths classified visually as Group II eclogites as well as in clinopyroxene from websterite xenoliths. Potassium and titanium levels of clinopyroxene from most of the xenoliths are low, with the exception of the Group I eclogites (including the diamond eclogite xenoliths) where K2O contents ranging from 0·04 wt % to >0·1 wt % (Fig. 4), as well as TiO2 contents of the order of 0·4 wt %, have been found.
Primary accessory minerals
Spinel in the alkremite xenoliths is aluminous (61–68 wt % Al2O3; Table 2) with up to 14 wt % FeO and 22 wt % MgO. The primary spinel in the spinel eclogite xenoliths is similar in composition, but the grains are comparatively iron-poor (FeO up to 3·6 wt %) and magnesium-rich (MgO up to 25 wt %). Cr2O3 concentrations in the spinel generally vary between 0·3 and 8 wt %, whereas Ni contents range up to 1 wt %.
Large, primary corundum grains from the corundum-bearing eclogite and alkremite assemblages are relatively pure (minimum 99·13 wt % Al2O3; Table 2), with only minor amounts of FeO (up to 0·56 wt %) and Cr2O3 (up to 0·49 wt %). Corundum in the alkremite variants are generally more iron-rich than those from corundum eclogite xenoliths.
Orthopyroxene from Group I eclogite xenoliths 196 and 374 is iron-rich, with FeO contents of 9 and 13·6 wt %, respectively (Table 2). Orthopyroxene from websterites has between 0·36 and 0·82 wt % Al2O3. FeO concentrations range from 5·06 to 7·08 wt %, and MgO varies from 34·30 to 35·79 wt %. CaO contents are low (0·12–1·11 wt %).
|
TRACE ELEMENT MINERAL CHEMISTRY |
|---|
Garnet
Garnet from Group I eclogite xenoliths (including the diamond-bearing eclogite xenoliths) generally has higher levels of Zr and Hf than garnet in most other xenolith types (Table 3). Ga, Ni and Y concentrations show wide scatter. The garnet is generally depleted in light rare earth elements (LREE) (Table 3; Fig. 5), with the most extreme LREE depletion in garnet from spinel-bearing eclogite xenoliths. The REE concentrations and chondrite-normalized abundance patterns in garnet from Group I and Group II eclogite xenoliths are generally comparable. In contrast, garnets from the corundum-bearing eclogite xenoliths are enriched in the LREE.
|
Clinopyroxene
Ni contents of clinopyroxene from eclogite and websterite xenoliths are highly variable between specimens (Table 3). Three bimineralic Group II eclogites (samples 384, 413 and 415) have unusually high Ni. Sr concentrations in all clinopyroxene vary between
20 and 180 ppm. Concentrations are lowest in the spinel-bearing eclogite xenoliths. Most clinopyroxene is characterized by chondrite-normalized LREE-enriched abundance patterns, with generally lower concentrations of the LREE in clinopyroxene from Group I eclogite xenoliths (including diamond-bearing specimens), as well as spinel-bearing eclogite xenoliths, relative to the rest (Fig. 5). |
XENOLITH MAJOR ELEMENT WHOLE-ROCK CHEMISTRY |
|---|
The eclogite xenoliths collected in this study were all too small for whole-rock chemical analysis. Consequently, the whole-rock chemistry of each sample was calculated by combining the chemistry of constituent minerals in ratios according to the mineral abundance estimates. Abundance estimates were obtained by weighing each mineral fraction after crushing and separation on a Frantz magnetic separator (Table 1).
The eclogite xenoliths have a wide range in modal abundances with a bias towards garnet-rich compositions (particularly for the Group I eclogites). This is probably due to the preferential abrasion of clinopyroxene, both during transport in the kimberlite and during crushing at the mine processing plant. In addition, eclogites are often layered rocks, with alternating garnet- or clinopyroxene-rich bands (Dawson & Carswell, 1990). Such layering will generally not be detectable in the xenoliths collected for this study because of the small specimen size. The calculated whole-rock compositions are, therefore, at best only crude estimates, and detailed analysis and modelling of compositional trends are not warranted.
Group I eclogite bulk compositions are tholeiitic and are broadly comparable with mid-ocean ridge basalts, although the eclogite MgO contents are too high. A much improved compositional correspondence is obtained when Group I eclogite xenoliths are compared with the more magnesian primary magma compositional estimates for ocean floor basalts (Basaltic Volcanism Study Project, 1981).
Group II eclogite and the spinel-bearing eclogite xenoliths are characterized by consistently high MgO and CaO contents and therefore have compositional affinities with clinopyroxenite xenoliths in alkali basalt diatremes (e.g. the Bou Ibalrhatene Maar in Morocco; Moukadiri & Kornprobst, 1984).
Kaalvallei corundum-bearing eclogite xenoliths have chemical similarities to metagabbroic and anorthositic rocks, particularly in their high Al2O3 and CaO contents. The average corundum eclogite major element composition is almost perfectly matched by metagabbroic rocks, for example, from the Sakhakot-Qila ophiolite in Pakistan (Ahmed & Hall, 1984).
Garnet-websterite xenoliths from the Kaalvallei kimberlite typically contain up to 22 wt % MgO, corresponding compositionally to picrite rather than komatiite, which generally has higher MgO (e.g. 30 wt %) and lower Al2O3.
Kaalvallei alkremite xenoliths are characterized by very low SiO2 and extremely high Al2O3. They are compositionally not matched by any known igneous rock type, although they do have broad chemical similarities to laterites, particularly in terms of high Al2O3 content. Alkremite specimens from Kaalvallei are compositionally identical to alkremite xenoliths from Jagersfontein (Mazzone & Haggerty, 1989a, 1989b).
In terms of minor elements, the highest MnO and TiO2 contents are observed in Group I eclogite xenoliths. Group II eclogite, spinel eclogite and websterite xenoliths have the highest Cr2O3 contents. TiO2 and K2O contents of all the xenoliths are generally much lower than in typical ocean floor basalts but are in some cases (e.g. the Group I eclogites) comparable with concentration estimates for primitive, near-primary basalts (e.g. oceanic primary tholeiites and picrite basalts; Basaltic Volcanism Study Project, 1981, tables 1.4.2.1
[EC]
and 1.4.2.2
[EC]
).
|
XENOLITH EQUILIBRATION CONDITIONS |
|---|
Equilibration temperatures were calculated with the garnet–clinopyroxene geothermometer formulations of Ellis & Green (1979)
, Powell (1985) and Krogh (1988) for the Kaalvallei eclogites selected for detailed study. All Fe is assumed to be Fe2+, which results in minimum temperatures (Jacob, 2004). Calculated temperatures obtained for an assumed pressure of 50 kbar from the three formulations differ by up to 130°C from each other. Application of the Krogh (1988) formulation to the corundum eclogite and the garnet-websterite xenoliths results in significantly lower equilibration temperatures for these xenoliths relative to temperatures produced by the other two thermometric formulations.
Kaalvallei eclogite xenoliths exhibit a large range in equilibration temperatures (Ellis & Green formulation at 50 kbar), extending from 800°C to 1400°C at 50 kbar (Fig. 6).
|
Group I eclogite xenoliths (including the diamond-bearing xenoliths) range from 1100°C to 1350°C, with an average at 1260°C. Group II eclogite xenoliths exhibit equilibration temperatures ranging from 900°C to 1300°C, averaging at
1100°C. Therefore, on average, Group I eclogites are characterized by equilibration temperatures that are 160°C higher than that of Group II eclogites.
Spinel-bearing eclogite xenoliths have equilibration temperatures in the range 800–900°C. Garnet-websterite xenoliths are characterized by a restricted range of temperatures, on average 1050°C. Corundum-bearing eclogite xenoliths give the highest temperatures of equilibration, averaging at 1350°C. The average equilibration temperature of the Group II eclogite and websterite xenoliths are comparable, perhaps indicating similar levels of mantle residence for these two xenolith types (assuming a comparable geothermal gradient).
P–T estimates obtained for the two Group I eclogite xenoliths that contain orthopyroxene are in the range 1037–1329°C and 34·5–65·4 kbar, utilizing the two-pyroxene thermometer and the orthopyroxene–garnet barometer combination of Brey & Kohler (1990) and Brey et al. (1990). Garnet-websterite xenoliths yield temperatures of 898°C and 939°C and pressures of 43·5 and 41·5 kbar, respectively.
Pressure calculations for eclogite xenoliths are generally not possible because of the absence of suitable coexisting mineral phases (e.g. orthopyroxene coexisting with garnet). The Al2O3 contents of clinopyroxene in equilibrium with garnet in the eclogite-facies mineral assemblage are markedly pressure–temperature dependent (Gasparik, 1984). However, difficulties over the evaluation of the quantity and activity of Ca-Tschermaks molecule in clinopyroxene seriously impair the extrapolation and application of this particular barometer, based on CaO–MgO–Al2O3–SiO2 system data, to more chemically complex natural garnet–clinopyroxene equilibria (Carswell & Harley, 1990).
Higher pressures and temperatures lead to increasing solubility of TiO2 in garnet (Green & Sobolev, 1965; Ringwood, 1975). There are also some data (Sobolev & Lavrent'yev, 1971; Bishop et al., 1978) indicating that the coupled substitution of Na+ in site X in garnet (divalent cations) with Ti4+ in site Y (trivalent cations) or P5+ in site Z (primarily Si4+) is favoured by high pressure (Carswell & Harley, 1990). Sodium-rich, aluminium-poor garnet can also form at very high pressure as a result of solution of clinopyroxene in the garnet structure (Irifune et al., 1989). The higher Na and Ti content of garnet in the Group I eclogite xenoliths relative to the rest of the Kaalvallei eclogites could therefore indicate that these rocks equilibrated at higher pressure (Jacob, 2004; or it might simply be due to differences in eclogite bulk composition; Grütter & Quadling, 1999), and this is also in line with the higher equilibration temperatures calculated for garnet–clinopyroxene for the Group I eclogites. The graphite–diamond stability curves of Bundy (1980) and Kennedy & Kennedy (1976) intersect a continental geothermal gradient of 40 mW/m2 (Pollack & Chapman, 1977) at approximately 42 and 45 kbar pressure, respectively. This could be a suitable minimum pressure estimate for the Kaalvallei Group I eclogite xenoliths as all of the diamond-bearing eclogites recovered from Kaalvallei (this study; Kiviets, 2000) are of the Group I variety. If the simple assumption is made that the 50 kbar garnet–clinopyroxene temperature estimates can be directly related in P–T space to a continental geothermal gradient then Group I eclogite xenoliths derive from mantle at a pressure of 50–80 kbar. The higher temperatures, however, could very well represent frozen mineral equilibria from a previous high-temperature (igneous?) event, or, alternatively are a reflection of the exposure of these xenoliths to the underlying, hotter asthenosphere. In the same way, Group II eclogite xenoliths are inferred to have equilibrated over a pressure interval of 20 kbar, ranging from 40 kbar upwards for an assumed pressure correction in the geothermometer of 50 kbar. An assumed pressure correction of 30 kbar gives an approximate minimum pressure of equilibration for the Group II eclogite xenoliths of 37 kbar. The real lower limit of equilibration pressure for the Group II eclogites is in all probability much lower. It is possible that a number of these xenoliths record fossil equilibria of high-temperature events (possibly related to eclogite petrogenesis) and that ambient thermal conditions of the mantle at the time of xenolith sampling are not reflected at all. This could be a consequence of faster cooling as a result of lower ambient mantle temperatures relative to the Group I eclogite xenoliths.
The very high equilibration temperatures of the corundum-bearing eclogite xenoliths are compatible with equilibration pressures of 60–80 kbar. However, none of these xenoliths contains diamond and the garnet is characterized by Na2O and TiO2 concentrations comparable with that in garnet from the Group II eclogite xenoliths (i.e. low relative to garnets from Group I eclogites). These very high temperatures are therefore in all probability due to the inability of all formulations of the garnet–clinopyroxene Fe–Mg thermometer to accommodate very calcic garnet, and mantle residence in the same area as Group II eclogite and websterite xenoliths is preferred.
Estimates of equilibration conditions for the alkremite xenoliths are not possible but the low concentrations of TiO2 and Na2O in analysed garnets could indicate that they are also not spatially associated with the Group I eclogites, but rather derived from shallower regions of the mantle.
The relatively low equilibration temperatures of the spinel-bearing eclogite xenoliths (800°C) are compatible with an equilibration pressure of 32 kbar on a cratonic geothermal gradient of 40 mW/m2. The minerals in these rocks are chemically heterogeneous, however, and this is taken as evidence that constituent minerals did not fully equilibrate to ambient mantle conditions prior to sampling by the kimberlite. The estimated P–T of equilibration is therefore viewed as a maximum and the rocks are probably derived from the lower crust or from shallow upper mantle.
In summary, Group I eclogite xenoliths probably derive from well within the diamond stability field, possibly at depths ranging from at least 150 km to the base of the lithosphere or deeper. Group II eclogite, garnet-websterite and corundum eclogite xenoliths probably all derive from the same region of the mantle. A shallower depth of mantle residence relative to the Group I eclogite xenoliths is indicated, and this region probably extends from the lower boundary of the diamond stability field to the shallow upper mantle (spinel peridotite stability field) or even the base of the crust. Alkremite xenoliths could also possibly derive from this region, or alternatively are crustal rocks. Spinel-bearing eclogite xenoliths appear to have been sampled from the shallow upper mantle or from the base of the crust.
|
STABLE ISOTOPIC COMPOSITIONS |
|---|
Group I eclogite xenoliths (including the diamond-bearing xenoliths) and garnet-websterite xenoliths are characterized by garnet and clinopyroxene
18O compositions of 4·3–5·9, 4·7–7·32 and 5·6–6·1 (Table 1; Fig. 7; Kiviets, 2000). This therefore approximates to the primary mantle value of +5·5 (Kyser, 1986; Mattey et al., 1994a, 1994b; see Table 1). Group II eclogite, corundum eclogite, spinel eclogite and alkremite xenoliths typically have, on average, similar oxygen isotopic compositions to Group I eclogite xenoliths, but with 18O values ranging down to 3·7 for sample 394 (Table 1; Fig. 7).
|
Garnet from Kaalvallei Group I eclogite xenoliths is generally slightly depleted in 18O relative to coexisting clinopyroxene (negative garnet–clinopyroxene fractionation), whereas most of the Group II eclogites are characterized by 18O-enriched garnet relative to coexisting clinopyroxene. Garnet–clinopyroxene oxygen isotopic distribution for the rest of the xenoliths shows wide scatter, and a consistent pattern is not observed. Oxygen isotopic data for constituent minerals in eclogite xenoliths from the Orapa and Roberts Victor kimberlites show that decreasing temperature favours the incorporation of small amounts of 18O in garnet relative to clinopyroxene (Deines et al., 1991
), and the Kaalvallei bimineralic eclogite data therefore compare similarly.
The 13C isotopic compositions of diamonds from Kaalvallei eclogites are in the range –4·06 to –11·45 (Kiviets, 2000).
|
RADIOGENIC ISOTOPIC COMPOSITIONS |
|---|
The generally low Sr content of garnet complicates isotopic analysis, and high-quality Sr data for garnet were generally not obtained. Most of the Sr isotopic data for coexisting garnet and clinopyroxene show isotopic disequilibrium between the two minerals, even after rigorous hand cleaning and leaching of garnet separates. Most measured Rb concentrations for garnet and clinopyroxene are barely above the level of the analytical blank (<100 pg) and the observed garnet–clinopyroxene 87Sr/86Sr isotopic heterogeneity is therefore not due to the growth of radiogenic strontium through radioactive decay. The very low Sr concentration of garnets from mantle xenoliths makes them very susceptible to contamination through the precipitation of a ubiquitous, high-Sr contaminant (possibly related to zeolites) in cracks within grains (Richardson et al., 1985
), and Sr isotopic heterogeneity of minerals in mantle xenoliths is therefore commonly observed (Jagoutz et al., 1980). Garnet 143Nd/144Nd isotopic compositions are generally higher than that of coexisting clinopyroxene, a consequence of the higher Sm/Nd of garnet.
Clinopyroxene and garnet mineral separates from Kaalvallei eclogite xenoliths selected for detailed study show considerable variation in Sr and Nd isotope composition (Tables 4–6). Present-day Sr isotope compositions of clinopyroxene from Group I eclogite xenoliths (including the diamond-bearing eclogites) range from 0·702 to 0·722 (Table 4). The 87Sr/86Sr compositions correlate broadly with xenolith Fe content. Iron-poor xenoliths have 87Sr/86Sr <0·704 whereas Fe-rich xenoliths generally are characterized by higher 87Sr/86Sr (Fig. 8). Clinopyroxene 87Sr/86Sr compositions for Group II eclogite xenoliths range from 0·7037 to 0·7105 (Table 5) whereas compositions for corundum eclogite and garnet-websterite xenoliths cluster around values of 0·705 and 0·703, respectively (Table 6). The 87Sr/86Sr isotopic compositions of clinopyroxene from spinel-bearing eclogite xenoliths are extremely radiogenic, with 87Sr/86Sr 
0·915 (Table 6).
|
|
|
|
Clinopyroxene in most of the xenoliths is characterized by present-day 143Nd/144Nd values that are, in most cases, only slightly higher than that of Bulk Earth (Tables 4–6; Fig. 9). The xenoliths show little variation in 143Nd/144Nd isotope composition, with the exception of the spinel-bearing eclogite xenoliths, which have constituent garnet and clinopyroxene with extremely high 143Nd/144Nd (e.g. 0·5174 for clinopyroxene in xenolith 306; Table 6).
|
The Sr and Nd isotopic data for Group I eclogite xenoliths from Kaalvallei (including the diamond-bearing xenoliths) indicate that at least two populations of xenoliths are present. One variety (population A) is characterized by low Sm/Nd (e.g <0·3; Fig. 10), no correlation of Sr concentration with 87Sr/86Sr (Fig. 10), relatively Fe-poor mineral compositions (Fig. 8) and (with one exception) have 87Sr/86Sr
0·704. The other xenolith type (population B) is characterized by lower Sm and Nd concentrations, higher Sm/Nd (Fig. 10), a strong positive correlation of Sr with 87Sr/86Sr isotopic composition (Fig. 11), relatively Fe-rich mineral compositions and 87Sr/86Sr >0·704 (Fig. 8).
|
|
Clinopyroxene from the Kaalvallei Group II xenoliths defines a negative correlation of Sm/Nd with 87Sr/86Sr isotopic composition (Fig. 10) and positive correlation of Sr with Nd (Fig. 11). Two groups can be defined in terms of Sr content and 87Sr/86Sr isotopic composition (Fig. 12). The two groups differ in Ni content (Fig. 12) and can also be separated on a plot of
18O vs 87Sr/86Sr (Fig. 13).
|
