Journal of Petrology Advance Access published online on October 22, 2008
Journal of Petrology, doi:10.1093/petrology/egn048
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Melt Depletion and Enrichment beneath the Western Kaapvaal Craton: Evidence from Finsch Peridotite Xenoliths
Department of Earth Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EQ, UK
Received December 6, 2007; Revised typescript accepted September 12, 2008
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONWe present major- and trace-element analyses of mineral phases present in a suite of 16 garnet-peridotite xenoliths from the western terrane of the Kaapvaal craton. The xenoliths were entrained by a Group II Finsch kimberlite at 118 Ma, shortly prior to a major metasomatic event that caused widespread enrichment of the Kaapvaal lithospheric mantle. Compositionally homogeneous grains of olivine, orthopyroxene, garnet and clinopyroxene and coarse–equant textures indicate equilibrium relationships between mineral phases in the majority of xenoliths. Pressure and temperature estimates suggest that clinopyroxene-bearing garnet peridotites last equilibrated at 1130–1270°C and 45–59 kbar whereas clinopyroxene-free xenoliths record temperatures of 1000–1070°C and pressures of 34–42 kbar. The Finsch xenoliths plot on a conductive palaeogeotherm with a surface heat flux of
46 mW m2. Combined Ca and Cr abundances of Finsch pyrope garnets suggest that both lherzolitic and harzburgitic parageneses are present. Samples bearing sub-calcic (harzburgitic) garnets are from the shallowest depths. The lherzolitic garnets are depleted in light rare earth elements (LREE) relative to the middle and heavy REE (MREE and HREE) and have ‘smooth’ chondrite-normalized patterns. In contrast, the sub-calcic garnets are characterized by sinusoidal chondrite-normalized REE patterns that peak at Nd and Lu and exhibit lows at La and Er. The sub-calcic garnets also have lower Zr, Hf, Ti and HREE, and higher LREE and Sr, than lherzolitic garnets. The variations in REE ratios correlate with temperature and pressure and also Cr/Ca ratio. The high Cr content of harzburgitic and some lherzolitic Finsch garnets may have a significant effect on the crystal framework. Substitution of the larger Cr3+ ion for the smaller Al3+ ion increases with decreasing temperature and pressure and distorts the crystal lattice; this permits a greater substitution of Ca by large cations, such as Sr and the LREE, but also limits the replacement of Al by Ti, Zr and Hf. Positive HREE slopes displayed by harzburgitic garnets on chondrite-normalized plots are believed to result from metasomatic enrichment by a melt that had already undergone significant garnet fractionation during ascent through the lithospheric mantle. The low-temperature Finsch peridotites are characterized by much lower orthopyroxene (< 17%) and higher olivine (up to 96%) modal abundances than have been reported from xenolith suites elsewhere in the Kaapvaal craton. Significantly, they resemble residues generated in partial melting experiments. The Finsch harzburgites have very low Al2O3 (0·18 wt %) and CaO (0·38 wt %) and high MgO contents (49·75 wt %) and appear to be highly refractory. They also have high bulk-rock Mg/(Mg + Fe) and high modal olivine contents, and in this respect resemble some of those recently described from NW Canada and Greenland. We suggest that some of the Finsch low-temperature peridotites represent Kaapvaal lithospheric mantle that formed as a residue of adiabatic decompression melting between 4·5 and 1·5 GPa. The inferred mantle potential temperature of 1550°C would have been similar to that of ambient Archaean mantle. Importantly, it appears that the sub-Finsch lithospheric mantle has remained unmodified by the silica enrichment that has been so prevalent elsewhere in the craton. This may reflect the remoteness from the subduction zone that is believed to have been in existence at 2·9 Ga on the eastern margin of the craton. KEY WORDS: Kaapvaal craton; peridotite; harzburgite; mantle enrichment, melt depletion
|
INTRODUCTION |
|---|
Mantle xenoliths provide a unique insight into the composition of the Earth's deep interior and invaluable information on the thermal structure of the lithosphere. They also offer an opportunity to constrain the melt depletion and enrichment processes that have been involved in the evolution of the Earth's subcontinental lithospheric mantle from Archaean to Recent times. Many mantle xenoliths have, however, undergone extensive modification since the time of their formation such that mineralogical and geochemical evidence relating to processes involved in the very early stages of lithosphere formation is overprinted. For example, kimberlites from all of the major cratons have entrained peridotite xenoliths that have experienced mantle metasomatism (e.g. Dawson & Smith, 1977
, 1988; Harte et al., 1987; Pearson et al., 1995a; Kopylova & Caro, 2004). This is widely believed to have resulted from percolation of small-fraction volatile-rich melts derived from the convecting mantle. Sometimes metasomatism is evident from the growth of ‘new’ phases such as garnet, clinopyroxene and/or phlogopite but in other cases only the mineral chemistry has changed. The latter, ‘cryptic’ metasomatism (Dawson, 1984), is especially apparent in incompatible trace element concentrations and ratios. Studies of pyrope garnets have shown that those belonging to the harzburgite paragenesis (i.e. low Ca and often high Cr), and believed to represent depleted mantle, are characterized by sinusoidal chondrite-normalized rare earth element (REE) patterns with a peak at Nd or Sm and sometimes Lu. In contrast, lherzolitic garnets, thought to typify fertile or metasomatized mantle, have smooth chondrite-normalized REE patterns with a positive slope for the light to middle REE (LREE to MREE) and a fairly flat MREE to heavy REE (HREE) slope. Furthermore, the low-Ca harzburgitic garnets usually have higher LREE concentrations than the lherzolitic garnets, which are often more Ca-rich. This paradox between the major element and REE concentrations of mantle garnets has provoked much discussion in the recent literature (e.g. Stachel et al., 1998; Griffin et al., 1999a; Wang et al., 2000; Bell et al., 2005). Although invoking a wide array of processes to explain the variations in REE, most interpretations suggest that the source of the sub-calcic garnets has undergone metasomatism. If this is correct then much of the subcratonic lithospheric mantle has been modified by metasomatic enrichment since the time of its formation. Indeed, trace-element and isotopic ratios of peridotitic diamond inclusions suggest that enrichment of cratonic lithospheric mantle may have been operational since at least 3·3 Ga (Richardson et al., 1984; Pearson et al., 1995b).
Incompatible trace-element enrichment of mantle xenoliths is also thought to occur as a result of the percolation of melts associated with the host kimberlite (Kinny & Dawson, 1992). Diffusion of both major and trace elements through grains in mantle xenoliths is significantly slower (hundreds of years; e.g. Van Orman et al., 2002) than the timescale of kimberlite emplacement (days; Kelley & Wartho, 2000) and this type of enrichment, immediately prior to or during ascent, results in compositionally heterogeneous grains. This is a common characteristic of peridotite xenoliths entrained by Group I South African kimberlites (e.g. Griffin et al., 1999c; Simon et al., 2003; Burgess & Harte, 2004). As well as overprinting the original mantle ‘signature’, chemical disequilibrium within and between grains considerably restricts the application of geothermometers and barometers.
A different style of metasomatism, involving SiO2-enriched fluids and/or melts, has been attributed by some to low-temperature (<1150°C) peridotite xenoliths from the Kaapvaal and Siberian cratons. The orthopyroxene contents and Si/Mg ratios of these xenoliths are higher than those of residues generated in partial melting experiments (e.g. Boyd, 1989). Recent studies have reinstated the proposal of Kesson & Ringwood (1989) that the silica enrichment results from the invasion of cratonic lithosphere by hydrous fluids and/or melts derived from a subducting slab (Rudnick et al., 1994; Kelemen et al., 1998; Bell et al., 2005; Simon et al., 2007). This contrasts with models which suggest that the ubiquitous high modal orthopyroxene in low-temperature peridotites from the Kaapvaal and Siberian cratons resulted from high-pressure melting processes that occurred at the time of lithosphere formation (Boyd, 1989; Herzberg, 1999).
In this study we report on a suite of peridotite xenoliths entrained by the Finsch kimberlite from beneath the western part of the Kaapvaal craton. The Finsch peridotites are of particular interest because: (1) they display a continuum of compositions ranging from harzburgites to lherzolites; (2) unlike most peridotites from other parts of the craton, all of the mineral phases are in chemical equilibrium; (3) the modal abundance of orthopyroxene in the low-temperature peridotites is much less than previously described from locations further east in the Kaapvaal craton. It is, however, similar to that reported in recent studies of xenoliths from Greenland and NW Canada (e.g. Bernstein et al., 1998; Schmidberger & Francis, 1999). We have combined our new petrographic and geochemical results with those previously described for 3·3 Ga Finsch peridotitic diamond inclusions (Richardson et al., 1984) and xenoliths (Gurney & Switzer, 1973; Shee et al., 1982; Skinner, 1989; Viljoen et al., 1992) to gain further understanding of the long-term lithospheric mantle evolution of the Kaapvaal craton. Our detailed petrological and geochemical study of all phases present in the Finsch xenolith suite complements recent work on the Kaapvaal craton that has focused on specific mineral phases, such as pyrope garnet (e.g. Griffin et al., 1999b; Burgess & Harte, 2004).
|
GEOLOGICAL SETTING AND PREVIOUS INVESTIGATIONS OF FINSCH PERIDOTITES |
|---|
The Finsch kimberlite cluster is located at 28°21'S, 23°28'E, on the western margin of the Kimberley Block (Fig. 1). It represents one of the westernmost kimberlite localities described from the Kaapvaal craton. Kimberlite was first identified in 1960 and diamond production at Finsch Mine has been in operation since 1964. Fresh peridotite xenoliths were not, however, discovered until the late 1970s. Detailed mapping has revealed the presence of three kimberlite pipes and three dykes (Ekkerd et al., 2003
). These intrude the Karoo and Griqualand West Groups. Erosion has removed the Karoo sediments and the upper, crater-facies parts of the pipes and exposed hypabyssal-facies kimberlites. The F1 kimberlite is the dominant intrusion but the rare mantle xenoliths (peridotites and eclogites) were found in a small intrusion of melilite-rich phlogopite kimberlite (F7) and a kimberlite breccia (F8), which represents the highest-grade ore-body at Finsch Mine (Ekkerd et al., 2003).
|
In terms of their petrography and bulk-rock chemistry, the Finsch kimberlites resemble other Group II southern African kimberlites (Fraser & Hawkesworth, 1992). Rb–Sr phlogopite ages suggest that Finsch was emplaced at 118·4 ± 2·2 Ma, at a similar time to other Group II southern African kimberlites; for example, those from Barkley West and Boshof (Smith et al., 1985
).
Combined petrographic descriptions and analyses of major-element mineral chemistry are available in the published literature for 15 garnet peridotites and a garnet websterite (Gurney & Switzer, 1973; Shee et al., 1982; Smith et al., 1985; Skinner, 1989; Viljoen et al., 1992). Radiogenic isotope determinations have also been made on a small number of peridotite xenoliths (Pearson et al., 1995a; Griffin et al., 2004). In addition, geochemical studies have been undertaken on the Finsch kimberlite concentrate (Gurney & Switzer, 1973) and on peridotite and eclogite diamond-inclusion suites (Gurney et al., 1979; Richardson et al., 1984; Shimizu & Richardson, 1987; Smith et al., 1991; Griffin et al., 1992; Appleyard et al., 2004).
At least four of the peridotite xenoliths previously described from Finsch are known to be diamondiferous (Shee et al., 1982; Viljoen et al., 1992). These, together with those reported from Roberts Victor (Viljoen et al., 1994) and Udachnaya (Sobolev et al., 1984), are amongst the largest known number of samples of diamondiferous peridotites from a single locality for which there is information available in international published literature. Sm–Nd model ages suggest that sub-calcic pyrope garnet inclusions in Finsch diamonds formed at 3·3 Ga (Richardson et al., 1984). Furthermore, the incompatible trace element and Sm–Nd isotopic ratios of these garnets indicate that, prior to diamond formation, the mantle had undergone both melt depletion and enrichment events (Richardson et al., 1984; Shimizu & Richardson, 1987). Re–Os ratios for both whole-rock samples and sulphides in Finsch peridotites confirm that a melt depletion event occurred prior to 3 Ga (Pearson et al., 1995a; Griffin et al., 2004). Considerably higher and varied Nd isotopic ratios for garnets from the Finsch peridotites, relative to those in the diamond inclusions (Richardson et al., 1984; Shimizu & Richardson, 1987; Pearson et al., 1995a), suggest that lithospheric enrichment may have continued until immediately prior to their entrainment.
|
PETROGRAPHY |
|---|
We have studied a total of 16 peridotite xenoliths from Finsch Mine. Six of these were collected by J. B. Dawson from the F7 intrusion, exposed in the open pit in 1979 (sample numbers with the prefix BD), and the remainder were collected by J. Malarkey (F05JM sample numbers) during a visit to the mine in 2005. The xenoliths are rounded to ellipsoidal in shape and reach >50 cm in diameter. They exhibit variable degrees of alteration. Most xenoliths have a weathered outer surface but a relatively fresh interior, and petrographically fresh material was obtained in all but two cases (F05JM1 and BD3697A). In the latter, olivine has been completely serpentinized but the other phases are generally fresh. In the majority of the xenoliths, alteration is restricted to cracks in olivine grains.
The Finsch mantle xenolith suite is dominated by clinopyroxene-bearing garnet peridotites. Estimates of modal mineralogy were made using image analysis software (ImageJ©) on both thin sections and hand specimens. The results are shown in Table 1. Both olivine and orthopyroxene exhibit large variations in modal abundances, varying from 60 to 96 modal % and from 1 to 33%, respectively. Small modal amounts of garnet (1–8%) and clinopyroxene (0–5%) are also present. Phlogopite is rare and tends to be restricted to the margins of kelyphitic rims that surround garnet, or as a replacement for clinopyroxene. The variations in modal mineralogy that we have observed are similar to those reported in previous studies of Finsch xenoliths (Table 1; Shee et al., 1982; Skinner, 1989).
|
The most abundant textural type of mantle xenoliths at Finsch are coarse–tabular peridotites (based on the definitions of Harte, 1977
). This texture is displayed by all of the clinopyroxene-bearing xenoliths and is defined by tabular grains of olivine and orthopyroxene that have straight grain boundaries. Olivine forms 67–89 modal % of these rocks and is up to 14 mm in diameter. Orthopyroxene contents vary between 7 and 33 modal %. The orthopyroxene grains are up to 6 mm in diameter and exhibit undulose extinction. Garnet forms 3·5–8·3 modal % of the coarse–tabular peridotites. The grains are red in hand specimen and pink in thin section, rounded and are commonly 3–4 mm in diameter, although a few reach up to 1 cm (Supplementary Images are available for downloading at http://petrology.oxfordjournals.org). The garnets show no evidence of recrystallization or shearing but usually have thin kelyphitic rims. In the coarse–tabular peridotites, clinopyroxene occurs as small (<3 mm), anhedral (often lobate), emerald green, strain-free grains and constitutes 1–4·7 modal % of the rock. The clinopyroxenes often have narrow (200 µm) ‘spongy’ corroded rims. In F05JM6, the clinopyroxenes have well-developed exsolution lamellae that are truncated by the ‘spongy’ rims. SEM back-scattered images and qualitative spectra analyses suggest that the rims and cores have the same composition; this indicates that the corroded rims are probably the result of partial melting of the clinopyroxene, perhaps during kimberlite emplacement. Only one sample (F05JM2) displays evidence of recrystallization of olivine to strain-free neoblasts and has extensively strained orthopyroxene (i.e. has a porphyroclastic texture). Neither the garnet nor clinopyroxene show evidence of recrystallization in this sample. Clinopyroxene-free xenoliths (F05JM5, 8 and 9; BD3694 and 3695) have coarse–equant textures. These xenoliths are very rich in olivine (80–96 modal %, Table 1) and have variable amounts of orthopyroxene (2–18 modal %) and low amounts of garnet (1–4 modal %). The olivine grains have straight or smoothly curving grain boundaries and are up to 12 mm in length. They often display 120° triple junctions with orthopyroxene and appear to be in textural equilibrium. They exhibit slight undulose extinction. Orthopyroxene has diameters up to 5 mm. Garnet is relatively evenly distributed through the coarse–equant xenoliths. Grains are generally anhedral and range in size from 4 to 10 mm. They have thick kelyphitic rims of spinel and phlogopite and are readily distinguished from garnets in the coarse–tabular peridotites by their lilac colour and partially altered appearance (see Supplementary Images).
|
MINERAL CHEMISTRY |
|---|
The major- and trace-element systematics of mantle phases are highly dependent upon bulk-rock assemblages as well as temperature and pressure. In harzburgites, garnet is the major host of the REE, Sc, Y, Ti, Zr, Sr, Nb and Hf whereas in lherzolites these trace elements are partitioned between both garnet and clinopyroxene. Below we summarize variations in the major- and trace-element concentrations of single phases and then consider the implications of these to our understanding of mantle processes. Representative analyses of mineral phases are shown in Table 2 and in the Supplementary Dataset (http://petrology.oxfordjournals.org/).
|
|
ANALYTICAL TECHNIQUES |
|---|
Olivine, orthopyroxene, clinopyroxene and garnet were analysed for major and some trace elements using a Cameca SX100 electronprobe microanalyser equipped with five wavelength-dispersive spectrometers at the University of Cambridge. Both rim and core measurements were taken for the garnets as well as a traverse across one garnet in each section. During each analysis the beam voltage was 15 kV and the beam diameter was 1 µm. A current of 10 nA was employed for major elements and 100 nA for trace elements. The peak counting time was 20 s and background time half this. Standards were natural silicates, pure oxides and metals. Details of detection limits are given in the Supplementary Dataset.
Trace-element concentrations (Hf, Nb, Pb, Sc, Sr, Ta, Th, U, REE) in clinopyroxene and garnet were determined using a New Wave UP213 Nd:YAG laser ablation system interfaced to a Perkin–Elmer Elan DRC II inductively coupled plasma mass spectrometer (ICP-MS) system at the University of Cambridge. The diameter of the laser beam was 120 µm, the laser repetition rate was 10 Hz and the laser power was 12 J/cm. The ICP-MS data acquisition settings were two sweeps per reading, 40 readings, one replicate. The dwell times for each mass were dependent on the isotope and concentration of the element in the samples but was typically 10–50 ms. For all data, NIST 610 was used for calibration of element sensitivity. Calibration accuracy was verified by analysing either NIST 612 or 614 and MPIDING standards as unknown samples (see Supplementary Dataset); recoveries were typically 90–110% of the values published by Pearce et al. (1997) and Jochum (2006). The CaO content of each garnet and clinopyroxene was used for internal standard normalization of the trace-element signals. ICP-MS drift during the analytical session was less than 10%. The data were processed, and concentrations calculated, using Glitter Software (GEMOC, Australia) which allows precise selection of blanks, signals, and visualization of data quality.
Garnet
Garnets in the Finsch mantle peridotites exhibit a limited range in MgO (17·3–21·9 wt %) and FeO (5·6–9·3 wt %) but show a very wide variation in CaO (1·7–8 wt %) and Cr2O3 (1·7–13·8 wt %) contents (Table 2; Supplementary Dataset; Gurney & Switzer, 1973; Gurney et al., 1979; Shee et al., 1982; Skinner, 1989; Viljoen et al., 1992; Grütter et al., 2006). Molar Mg/(Mg + Fe + Ca) ratios vary from 0.76 to 0.64 and indicate that the garnets are pyrope-rich. Single garnet grains have a restricted compositional range and do not show the pronounced systematic zonation that has been described from many other Kaapvaal peridotites (e.g. Smith & Boyd, 1992; Griffin et al., 1999a; Burgess & Harte, 2004).
CaO vs Cr2O3 plots have been widely employed to distinguish between pyrope garnets of different parageneses and we have followed this approach (Fig. 2). On the basis of CaO and Cr2O3 contents and the recently updated classification scheme of Grütter et al. (2006), we have divided the Finsch garnets into ‘lherzolitic’ and ‘harzburgitic’ types and subsequently we refer to the host xenoliths according to these parageneses. Most of the Finsch garnets analysed in our study plot in the field of lherzolitic garnets in Fig. 2 and are G9 garnets according to the nomenclature of Dawson & Stephens (1976). This ‘lherzolitic’ trend indicates that the garnets were in equilibrium with clinopyroxene, although this phase was not observed in samples containing garnets with >8 wt % Cr2O3. We have, therefore, further subdivided the lherzolitic garnets into ‘Low-Cr’ (<8 wt % Cr2O3) and ‘High-Cr’ types (>8 wt % Cr2O3). This subdivision also correlates with the textural variations in the Finsch xenoliths; the Low-Cr lherzolitic garnets generally occur in xenoliths with coarse–tabular textures whereas the High-Cr lherzolitic garnets occur in samples that have coarse–equant grains (Table 1). The highest Cr2O3 (13·8 wt %) and CaO (8 wt %) contents occur in garnets in dunite F05JM5, which contains virtually no ortho- or clinopyroxene (Table 1). Zonation has been observed only in two of the Low-Cr lherzolitic garnets, where cores have lower CaO and Cr2O3 contents (by 0·5 wt %) than the rims. The positive slope of the CaO vs Cr2O3 trend displayed by the Finsch lherzolitic garnets is defined by the equation Cr2O3 = 2·7CaO – 8·6 and is thought to be both temperature and pressure dependent (Griffin et al., 1999a).
|
The least abundant type of garnets in the Finsch peridotites are those that plot in the field of G10 harzburgitic garnets (e.g. F05JM9, Fig. 2). These are sub-calcic garnets that contain 1·7–5·7 wt % CaO and 13·5 wt % Cr2O3. Their undersaturation in Ca suggests that they were not in equilibrium with clinopyroxene (Brey et al., 1990
). In Fig. 2, the harzburgitic garnets, and also those from dunite F05JM5, exhibit an almost continuous positive linear trend; the slope of this is slightly steeper than the isobaric trend predicted by Grütter et al. (2006). The Finsch garnets display a wide variation in Al2O3 (12–22·5 wt %) and exhibit a negative correlation between this oxide and Cr2O3. Cr-number [Cr/(Cr + Al)] varies from 0·43 in the dunite, to 0·40–0·35 in the harzburgitic garnets, 0·34–0·26 in the High-Cr lherzolitic garnets, and to 0·19–0·09 in the Low-Cr lherzolitic garnets.
Peridotitic inclusions in Finsch diamonds also contain pyrope garnets. These have higher MgO contents (20–27·5 wt %; Richardson et al., 1984) than garnets in the Finsch peridotites. The pyrope garnet inclusions have some of the lowest recorded CaO contents (<1·42 wt %) and have a wide range of Cr2O3 contents (5·4–15 wt %) and Cr-number (44–15). Molar Mg/(Mg + Fe + Ca) ratios vary from 0·93 to 0·73. On a CaO vs Cr2O3 plot (Fig. 2), the garnets from peridotitic suite inclusions plot in the field of harzburgitic garnets but lie away from the linear trend displayed by sub-calcic garnets in the xenolith suite.
In addition to exhibiting large differences in their major-element chemistry, the Finsch pyrope garnets have varied trace-element concentrations and ratios. The lherzolitic garnets are characterized by high Ti (500–2000 ppm), Y (2·5–16·5 ppm), Lu (0·08–0·27 ppm) and Zr contents (4–40 ppm) but low La (0·01–0·14 ppm), Sr (<0·5 ppm) and Sc (57–117 ppm) relative to the harzburgitic garnets in which Ti = 500 ppm, Y = 1·7 ppm, Lu = 0·07, Zr = 5–10 ppm, La = 0·14 ppm, Sr 1 ppm and Sc = 135 ppm (Fig. 3). The highest contents of Sr (6–9 ppm) and lowest contents of Ti (<200 ppm) and Zr (5–8 ppm) in Finsch garnets occur in the diamond inclusions (Shimizu & Richardson, 1987). On mantle-normalized multi-element plots, all of the garnets have relative depletions at Sr and most are also depleted in Ti (Fig. 3). Additionally, the sub-calcic harzburgitic garnets have troughs at Zr and Hf.
|
As has previously commonly been observed in peridotite xenoliths and diamond inclusions from the Kaapvaal and other regions of cratonic lithosphere, sinuous- or ‘humped’-shaped chondrite-normalized REE patterns that peak at Nd (or Sm) and have depletions at La and Er (and Tm) are typically found in the harzburgitic (sub-calcic) garnets whereas much flatter patterns with no pronounced peaks or troughs are restricted to the lherzolitic garnets (Nixon et al., 1987
; Shimizu & Richardson, 1987; Hoal et al., 1994; Shimizu et al., 1997; Stachel et al., 1998). The strong LREE/HREE fractionation of the lherzolitic garnets resembles pyrope garnets that crystallize in high-temperature melting experiments in which clinopyroxene is also present (Fig. 4) (e.g. Hauri et al., 1994; Salters et al., 2002; Tuff & Gibson, 2007). The Finsch garnets exhibit a continuum from sinuous harzburgitic to smooth lherzolitic REE patterns with the High-Cr lherzolitic garnets displaying intermediate REE patterns (Fig. 4). The lherzolitic garnets exhibit increasing LREE and decreasing HREE with CaO content but the harzburgitic garnets plot away from these trends (Fig. 5). LREE/MREE ratios are relatively constant (e.g. [La/Nd]n 0·5–0·15) for garnets in all lithologies. We have used chondrite-normalized Nd/Er and Er/Lu ratios as a measure of the degree of sinuosity of the REE patterns (Fig. 6). [Nd/Er]n ratios are 6–8 in the Low-Cr harzburgitic garnets, two in the dunite and High-Cr lherzolitic garnets and <1 in the Low-Cr lherzolitic garnets; we note that the highest [Nd/Er]n ratio (12) is for a garnet from a Finsch diamond inclusion analysed by Shimizu & Richardson (1987). [Er/Lu]n ratios show the opposite variation to [Nd/Er]n, being highest in the Low-Cr lherzolitic garnets (0·5–1) and lowest in the Low-Cr harzburgitic garnets (0·4). There are no Lu concentrations available for garnet inclusions in Finsch diamonds.
|
|
|
|
Olivine
In the Finsch mantle xenolith suite, olivine Mg-number ([Mg/(Mg + Fe)] x 100) varies from 89·6 to 93·7 (Table 2; Supplementary Dataset; Shee et al., 1982
; Richardson et al., 1984; Skinner, 1989; Viljoen et al., 1992). The most Fe-rich olivines are present in xenoliths bearing the Low-Cr lherzolitic garnets (Mg-number = 89·6–92·1) and the dunite (91·8); Mg-rich olivines occur in samples with High-Cr lherzolitic (91·36–93·67) and harzburgitic garnets (92·6–92·94) but by far the most magnesian olivines at Finsch are those present as diamond inclusions (92·1–95·3; Table 2; Shee et al., 1982; Richardson et al. 1984; Skinner, 1989; Viljoen et al., 1992). CaO (0·04 wt %) and NiO (0·36–0·45 wt %) exhibit little variation in Finsch olivines.
Orthopyroxene
Single orthopyroxene grains in the Finsch peridotites show no evidence of compositional zonation. They are enstatite rich and have Mg-numbers that are slightly higher than those of coexisting olivine. Orthopyroxene Mg-number varies from 91· 4 in the xenoliths with Low-Cr lherzolitic garnets to 93·7 in those bearing High-Cr lherzolitic garnets. Orthopyroxene in the harzburgite (F05JM9) also has a high Mg-number (92·8) but the most magnesium-rich orthopyroxenes observed at Finsch are those present in peridotitic diamond inclusions (Mg-number = 94· 4–95·5; Richardson et al., 1984).
The Al2O3 contents of the Finsch orthopyroxenes are low (0· 47–0·68 wt %) relative to those analysed from most other Kaapvaal xenolith suites (Fig. 7). They are, however, generally higher than those analysed in orthopyroxenes from Finsch diamond inclusions, which have extremely low Al2O3 contents (0·11–0·66 wt %; Fig. 7).
|
The CaO contents of Finsch orthopyroxenes vary with rock type in the same way as CaO contents in garnets; they are lowest in the sample with harzburgitic garnets (
0· 4 wt %) and highest in xenoliths containing lherzolitic garnets (0·5–0·8 wt %; Fig. 7). Orthopyroxenes from harzburgitic diamond inclusions have the lowest CaO contents (<0·3 wt %) observed at Finsch. This is consistent with experimental data suggesting that CaO contents of orthopyroxene increase with temperature (Brey & Kohler, 1990) and/or increasing degree of metasomatism (see below).
TiO2 contents of all orthopyroxenes are low (0·7–0·8 wt%). Cr2O3 contents of orthopyroxene are low in xenoliths bearing lherzolitic garnets (0·3 wt %) and highest in the rare grains that occur in the dunite (F05JM5) and harzburgite (F05JM9; 0·5 wt %). A negative relationship exists between Cr2O3 and Al2O3 for orthopyroxene in the lherzolitic xenoliths and reflects the substitution of Al3+ for Cr3+ in the M1 cation site. A different substitution mechanism appears to occur in orthopyroxene in the dunite and harzburgite, where Cr2O3 contents are high for a given Al2O3 content. This relationship is similar to that observed in the garnets (see above) and reflects the absence of Cr-diopside in these samples such that all of the Cr partitions into garnet and orthopyroxene.
The high CaO contents of orthopyroxene in Finsch coarse–tabular garnet lherzolites, relative to those found in coarse–equant garnet lherzolites (Fig. 7), suggest that they equilibrated at higher temperatures. Furthermore, the low Al2O3 of orthopyroxene in the diamond inclusions and harzburgites suggests that the geothermal gradient beneath Finsch peridotites is unusually low for the Kaapvaal lithosphere and/or that they equilibrated in mantle with a low Al2O3 content (see below).
Clinopyroxene
At Finsch, clinopyroxene is present only in xenoliths bearing Low-Cr lherzolitic garnets and has not been observed in peridotitic diamond inclusions. As has been described for other phases, there is no evidence of compositional zonation within grains. In the Finsch peridotite xenolith suite, however, Mg-number ranges from 91·7 to 93·5 and is higher and lower, respectively, than in coexisting olivine and orthopyroxene. Partitioning of Mg and Fe between clinopyroxene and olivine is similar to that displayed by orthopyroxene and olivine, and suggests that, for the major elements at least, these phases are in equilibrium. Ca/(Ca + Mg) in the clinopyroxene exhibits very little variation, ranging from 0· 435 to 0· 445. Cr2O3 contents vary from 1 to 2 wt %, indicating that they are Cr-diopsides (Stephens & Dawson, 1977). The Cr-diopside is rich in Na2O (1·3–2· 4 wt %) and has low to moderate contents of Al2O3 (1· 4–3 wt %) and TiO2 (0·04–0·17 wt %). Cr-diopsides in xenoliths bearing very low-Cr lherzolitic garnets (BD3692, 3693, BD3697A) also have low Ca but are enriched in Cr, Al, Na and Ti relative to Cr-diopsides from samples with higher Ca lherzolitic garnets (F05JM1, 2, 3, 4, 6, 7, 10 and BD3697B; Fig. 8). Concentrations of CaO and Al2O3 in clinopyroxene are strongly temperature dependent but also vary with metasomatism, as do FeO, TiO2 and Na2O. The latter also increases with pressure (see below).
|
Compatible trace elements, such as Sc, are present in similar concentrations in the Cr-diopsides but the incompatible trace elements exhibit more variation. Clinopyroxenes with the lowest Ca contents (i.e. from BD3692 and 3693) have the highest abundances of incompatible trace elements in addition to Cr, Al, Na and Ti. For example, concentrations of La range from 8·5 ppm in the ‘Low-Ca’ Cr-diopsides to 1·2 ppm in the ‘High-Ca’ Cr-diopsides. On normalized multi-element plots, the Finsch clinopyroxenes are characterized by variable (and sometimes large) negative Zr anomalies and slight but variable negative Ti and Nb anomalies (Fig. 9). These elements show variable positive anomalies on the corresponding normalized plots for coexisting lherzolitic garnets; this feature suggests that the garnet–clinopyroxene partition coefficients for Nb, Ti and Zr are greater than those for adjacent elements (K, La, Sm, Hf, Eu, Gd).
|
The Cr-diopsides display a large and variable fractionation of LREE and HREE with [La/Yb]n ranging from 100 to 800. Chondrite-normalized REE patterns generally peak at Ce; BD3693, 3692, 3697A and F05JM1 exhibit less fractionation of La and Ce ([La/Ce]n = 0·8–1·0) than the other Cr diopsides (0·6–0·8). We have also identified subtle differences in [Er/Lu]n, which, as for the garnets, are highest in the BD3693 and 3692. This suggests that, where these phases coexist, they are in chemical equilibrium. The Low-Ca Cr-diopsides (i.e. BD3692 and 3693) display peaks at Th and U relative to adjacent elements (i.e. Rb, Ba and K) on mantle-normalized multi-element plots (Fig. 9). In the High-Ca Cr-diopsides, mantle-normalized concentrations of Th and U are less than K, and [Th/U]n >1 whereas in the Low-Ca Cr-diopsides [Th/U]n <1. Our observations of Cr-diopside trace-element chemistry are consistent with the results of theoretical studies of pyroxenes (e.g. Wood & Blundy, 1997
, 2001) which have shown that the REE, Ba, K, Na, Rb, Sr, Th and U reside in the large M2 cation site and that their partitioning is dependent upon the amount of replacement of Ca2+ by Na+ (i.e. REE3+ + Na+ = 2Ca2+). Fractionation of Th and U may reflect the temperature dependence of the lattice to incorporate highly charged (4+) ions. |
THERMOBAROMETRY OF FINSCH XENOLITHS |
|---|
Pressure and temperature (PT) estimates calculated from published analyses of minerals present in Finsch xenoliths suggest that some of the samples last equilibrated at high pressures (
6 GPa) but relatively low temperatures (1250°C). These pressure estimates are some of the highest recorded for the Kaapvaal craton. Furthermore, these ‘cold’ and very ‘deep’ mantle xenoliths have been used as strong evidence that the conductive geotherm beneath the Kaapvaal craton has not been perturbed by heating events immediately prior to xenolith entrainment (Bell et al., 2003).
We have employed various combinations of the most commonly used two-pyroxene solvus geothermometers (TBKN, Brey & Kohler, 1990; TFB, Finnerty & Boyd, 1987) together with Al-in-opx barometers (PBBG, Brey et al., 2008; PBKN, Brey & Kohler, 1990; PMcG, MacGregor formulation published by Finnerty & Boyd, 1984) to estimate the pre-eruption equilibration conditions of clinopyroxene-bearing Finsch peridotites. Pressure and temperatures were calculated iteratively and the results are shown in Table 3 and Fig. 10. Pressures and temperatures for clinopyroxene-free peridotites were similarly estimated using the olivine–garnet Fe–Mg exchange thermometer of O’Neill & Wood (1979) (TONW) with PBBG, PBKN and PMcG Al-in-opx barometers. Different combinations of these geobarometers (excluding PBBG) and thermometers have been used in recent studies (e.g. Bell et al., 2003; Burgess & Harte, 2004; Simon et al., 2007), and allow comparison of our results from Finsch with those from other Kaapvaal xenolith suites.
|
The recently published Al-in-opx barometer of Brey et al. (2008
) includes results from new high-pressure (6–10 GPa) multi-anvil experiments. Pressure estimates for the Finsch lherzolites derived by combining this barometer (PBBG) with TBKN are up to 13 and 16 kbar lower than those calculated using PBKN vs TBKN and PMcG vs TBKN, respectively (Table 3). This equates to differences in equilibration depths of up to 50 km and has a huge effect on estimates of geothermal gradients and lithospheric thickness (see below). Despite the fact that PBBG relies on the same low- to medium-pressure experimental results as PBKN (2·8–6 GPa; Brey et al., 1990) we note that, for Finsch lherzolites that equilibrated at shallow depths (equivalent to 50 kbar), the pressure estimates from PBBG are significantly lower than those from PBKN. The PBBG vs TBKN/TONW combination places a diamond-bearing lherzolite (XM48; Shee et al., 1982) and two diamond-bearing harzburgites (865 and 866; Viljoen et al., 1992) shallower than the graphite–diamond stability field boundary (Fig. 11); only diamond-bearing xenolith XM46 (Shee et al., 1982) plots in the diamond stability field. The 1
error associated with PBBG (3 kbar; Brey et al., 2008) is such that the P estimates for XM48 and 865 are within error of the diamond stability field. This is not the case for harzburgite 866, which plots at a significantly lower pressure than the graphite–diamond phase boundary of Kennedy & Kennedy (1976). This deviation may, however, reflect the sensitivity of TONW to Fe3+ (Canil et al., 1994; Canil & O’Neill, 1996). The effects on PT estimates associated with errors in the calculation of Fe3+ are shown in Fig. 11. Here we have used Fe3+ contents calculated from Mössbauer spectroscopy to calculate equilibration pressures and temperatures for the one Finsch sample (865) for which there are published data (Canil & O’Neill, 1996). This diagram shows that increasing the Fe3+ content of both garnet and orthopyroxene to those accurately determined for this harzburgite will increase estimates of both temperature and pressure derived from PBBG vs TONW. Nevertheless, TBKN is insensitive to calculation of Fe3+ and problems associated with the estimation of this from microprobe analyses cannot explain why a diamond-bearing lherzolite from Lesotho (BD2125; Dawson & Smith, 1975) plots in the graphite stability field (Fig. 11).
|
Using the previously widely used PBKN vs TBKN combination, four literature samples (XM46, SK869, 691, 725) gave very-high pressure estimates (63–68 kbar). These xenoliths contain orthopyroxenes with a low jadeite component (because of low Na and/or high Ti; Shee et al., 1982
; Skinner, 1989), which causes an overestimate of pressure by 6 kbar and the xenoliths plot well below the geotherm (see below). Using the PBBG vs TBKN combination, we note that lherzolites SK725, SK871 and SK869 (together with JJG470; Gurney et al., 1979) have much shallower pressure estimates but plot well below the geotherm defined by other Finsch samples (Fig. 10a); this may indicate analytical problems or the presence of disequilibrium phases.
Temperature and pressure estimates for Finsch peridotites with Low-Cr lherzolitic garnets are in the range of 1130–1265°C (TBKN) and 45–59 kbar (PBBG), respectively (Table 3), and the xenoliths show a systematic change in temperature with depth (Fig. 10b). According to PT estimates determined using PBBG vs TONW, the cpx-free peridotites (which include xenoliths with High-Cr lherzolitc garnets) equilibrated at the lowest temperatures (1000–1070°C) and pressures (34–42 kbar) and all plot above the diamond–graphite phase boundary (Fig. 10b). As beneath other parts of the Kaapvaal craton (e.g. Jagersfontein and northern Lesotho), PT estimates suggest that the harzburgites are derived from the shallowest depths (Burgess & Harte, 1999; Simon et al., 2003; S. A. Gibson et al., unpublished data). We note that several of the cpx-free peridotites plot above the geotherm (Fig. 10b). This may be due to the sensitivity of the olivine–garnet Fe–Mg exchange thermometer of O’Neill & Wood (1979) to the calculation of Fe3+ as described above.
We have used the results of our PT calculations for the Finsch xenoliths to estimate the thickness of the mechanical and thermal boundary layers beneath the west of the Kaapvaal craton. This part of the craton was unaffected by the 2 Ga Bushveld magmatic event. Seismic studies have produced varied thickness estimates for the lithosphere beneath Finsch, ranging from 250 to 300 km (James & Fouch, 2002) and 180 km (Li & Burke, 2006; Priestley & McKenzie, 2006). If we assume a crustal thickness of 38 km (Nair et al., 2006) and a mantle potential temperature of 1315°C, the Finsch samples fall on a geotherm with a surface heat flux of 44·5 mW m2 (PBKN vs TONW/TBKN; Fig. 10a) or 45·7 mW m2 (PBBG vs TONW/TBKN; Fig. 10b). These palaeogeotherms intersect the diamond–graphite phase boundary at depths of 135 and 150 km, respectively. This geothermal gradient is slightly hotter than has sometimes been calculated in previous studies (35–40 mW m2, e.g. Finnerty & Boyd, 1987; Griffin et al., 2003) and reflects the fact that we have used the constant lithospheric mantle heat-flux values of McKenzie et al. (2005) to calculate the conductive geotherm rather than the commonly used variable conductivity values of Pollack & Chapman (1977). Assuming that the temperature of the convecting mantle was 1315°C immediately prior to eruption of the Finsch kimberlite, our calculations show that the base of the Cretaceous mechanical boundary layer (MBL) would have been at a depth of 181 km or 165 km for conductive mantle geotherms of 44·5 mW m2 (PBKN vs TONW/TBKN) or 45·7 mW m2 (PBBG vs TONW/TBKN), respectively. If, however, we assume that the Tp of the sub-Kaapvaal convecting mantle was 1500°C at the time of Finsch kimberlite genesis (i.e. the region was underlain by a mantle plume), then the base of the Cretaceous Kaapvaal MBL may have been at a depth of 208 km (PBKN vs TONW/TBKN) or 190 km (PBBG vs TONW/TBKN). These estimates of Cretaceous lithospheric thickness, regardless of the different combinations of geobarometers and thermometers used in the xenolith PT estimates, are similar to those of 180 ± 20 km suggested for the present-day Kaapvaal keel by Li & Burke (2006) and Priestley & McKenzie (2006).
|
MINERAL–MELT PARTITION COEFFICIENTS |
|---|
Mineral–melt partition coefficients (D values) for trace elements are strongly linked to the major-element composition of an individual phase, which is controlled by the prevailing PT conditions and the bulk-rock composition. D values are usually determined from the results of experimental studies but most of these are undertaken at much higher temperatures than those existing in the lithospheric mantle. Partition coefficients derived from natural lithospheric mantle assemblages of equilibrium phases are therefore believed to be more appropriate for mantle xenolith studies (Harte et al., 1996
). Burgess & Harte (2004) calculated DGrt–melt values over a range of temperatures (900–1400°C) by combining ion-probe analyses of coexisting garnets and clinopyroxenes with the experimentally determined low-alumina DCpx–melt values of Grutzeck et al. (1974) and Hart & Dunn (1993). Their calculations used DCpx–melt values determined at 1380°C (Hart & Dunn, 1993) and 1265°C (Grutzeck et al., 1974) and ignored any dependence of D values on temperature. A recent study by Tuff & Gibson (2007) has, however, shown that DCpx–melt values decrease by 20% with a 100°C increase in temperature.
We analysed adjacent grains of Cr-diopside and garnet in four Finsch peridotites that last equilibrated between 1150 and 1230°C for 26 trace elements by LA-ICP-MS. To calculate DGrt–melt values for all of the 14 REE, Y and Sc we estimated, using the lattice-strain model of Blundy & Wood (1994, 2003) and Wood & Blundy (1997), and the data of Hart & Dunn (1993), the theoretical DCpx–melt values at 1200°C. The lattice-strain model of Blundy & Wood (1994) is able to predict REE partitioning behaviour, which is commonly expressed by plotting the partition coefficient as a parabolic function of the ionic radius of a given REE (Fig. 12). The parabola varies in height and width with varying pressure, temperature and composition. Hart & Dunn (1993) published analyses for only eight REE in their original experimental dataset and so we extrapolated the results of our lattice strain model to estimate D values for the remaining REE (Fig. 12 and Table 4). We used the published DCpx–melt values of Hart & Dunn (1993) for the remaining trace elements and then followed the method of Harte et al. (1996) and Burgess & Harte (2004) to estimate DGrt–melt values (Table 4). For the REE, our results are similar to those determined for natural (Burgess & Harte, 2004) and synthetic pyrope garnets (Van Westrenen et al., 2000) (Fig. 13). We note that our DGrt–melt values for Lu, which range from 6·5 to 10·2, are 50% lower than those estimated by Burgess & Harte (2004) but are consistent with those predicted by lattice-strain models (Fig. 12).
|
|
|
When plotted against ionic radii, the DGrt–melt values that we have obtained for the REE, Sc and Y from the Finsch xenoliths display a parabolic pattern (Fig. 12) and conform closely to the partition coefficients predicted by the lattice-strain model of Blundy & Wood (1994
, 2003). The parabolic pattern of DGrt–melt values displayed in Fig. 12 confirms that all of the REE were in equilibrium in the adjacent grains of garnet and clinopyroxene. In xenolith suites that exhibit REE disequilibria between these phases (as has been shown for northern Lesotho; e.g. Simon et al. 2003) the calculated DGrt–melt values display a sigmoidal trend when plotted against ionic radii (S. A. Gibson et al., unpublished data). We have used the equations of Blundy & Wood (1994) to calculate the strain-free partition coefficient (Di) for an X-cation site with an effective radius of 0·0912 nm and a Young's modulus of 550 GPa. The (Di) value (11·7) that we have obtained for the parabolic dependence of the REE and Sc in the octahedral site in Finsch garnets is higher than that predicted for pyrope garnets in experimental studies (4·8) but reflects the lower equilibration temperatures of the former (1150–1250°C) relative to the latter (1425–1550°C; Van Westrenen et al., 1999; Tuff & Gibson, 2007). |
CAUSE OF SINUSOIDAL REE PATTERNS IN SUB-CALCIC GARNETS |
|---|
The high LREE concentrations in sub-calcic harzburgitic garnets, relative to those of lherzolitic garnets, are inconsistent with their formation as single-stage melting residues of a fertile lherzolite source. Various hypotheses have been proposed to account for this discrepancy and also the difference in shape of the REE patterns exhibited by lherzolitic and harzburgitic garnets.
(1) Slow rates of diffusion of HREE. Hoal et al. (1994) and Shimizu et al. (1997) suggested that only the lherzolitic garnets had reached full equilibrium with an infiltrating metasomatic melt. They interpreted variations in REE patterns as a result of chemical modification or fractionation of garnet, and suggested that sinusoidal patterns reflected decreasing rates of REE diffusion with decreasing ionic radius. Such a ‘disequilibrium metasomatism’ hypothesis is not, however, supported by more recent work, which has shown that REE with different ionic radii have similar diffusion coefficients (Van Orman et al., 2002).
(2) Limited equilibrium with a metasomatic carbonatite melt. Griffin et al. (1999b) proposed that sinuous REE patterns are due to the reduced number of cation sites available for LREE substitution in low-Ca (harzburgitic) garnets relative to those in high-Ca (lherzolitic) garnets. In the Finsch peridotites only the lherzolitic garnets show a good correlation between Ca and LREE concentration (Fig. 5) and incorporation of LREE into the sub-calcic garnets appears to involve a different substitution mechanism.
(3) Garnet fractionation from ascending metasomatic silicate melts. Burgess & Harte (2004) suggested that the compositions of ascending metasomatic melts control the REE patterns of mantle garnets. They proposed that high-pressure garnet fractionation caused depletion of HREE in the metasomatic melt, which subsequently ascended through the lithospheric mantle and crystallized sub-calcic garnets with ‘humped’ REE patterns.
(4) Spatial and temporal variations in melt/rock ratios. Wang et al. (2000) proposed that the behaviour (open or closed system) of infiltrating metasomatic melts would depend on the distribution and proportion of veins relative to the host rock. Chemical equilibrium of garnet and an infiltrating LREE-rich carbonatite melt will be achieved only by the open-system behaviour that occurs when there is a high melt/rock ratio. Wang et al. (2000) proposed that sub-calcic garnets form in systems where the melt/rock ratio is low and that do not achieve this equilibrium.
(5) Multi-stage tectonomagmatic processes. It has been suggested that the sinuous REE patterns displayed by harzburgitic garnets reflect their multi-stage formation. Stachel et al. (1998) proposed a tectonomagmatic model involving the following sequence of events: (a) polybaric melting, initially in the garnet stability field and extending into the spinel stability field, resulting in harzburgite formation and removal of the LREE and MREE from the residue; (b) lithospheric thickening to levels at which garnet is the stable aluminous phase, causing HREE fractionation; (c) metasomatism of harzburgite by percolating methane-rich fluids with variable Si and high LREE/HREE ratios, which were also involved in diamond formation. Similar models have been invoked by Bell et al. (2005), Westerlund et al. (2006) and Simon et al. (2007), who proposed that the subduction-related fluids were Si-rich and associated with the transformation of olivine to orthopyroxene.
Relationship between major elements (Ca and Cr) and REE partitioning in mantle garnets
Few data are available on the substitution mechanisms of trivalent cations, such as the REE, in natural garnets. In synthetic systems (e.g. CMAS and FCMAS), the REE are thought to substitute for Ca in the garnet X-cation site. Experimental studies of pyrope garnets have shown that a large change in grossular content (Py82Gr18 to Py15Gr85) has a greater effect on D values for the LREE than for the HREE. This is because the solid solution between pyrope and grossular garnet is non-ideal and causes non-linear elasticity, which has a more significant influence on the partitioning of larger ions; that is, the LREE (Ballaran et al., 1999). However, over the range of grossular contents present in pyrope garnets from mantle peridotites (Py70–80Al17–20Gr8–14) fractionation of D
values is small. In the Finsch lherzolitic garnets there is a reasonable correlation between Ca and LREE and HREE contents (Fig. 5) confirming that REE substitution is dependent on the amount of Ca in the X-cation site of these garnets. Such a relationship does not exist, however, when sub-calcic (harzburgitic) garnets are included in the dataset. This suggests that it is not simply the Ca content and hence pyrope:grossular ratio that controls the REE patterns of peridotitic garnets.
When garnets present in both lherzolitic and harzburgitic parageneses at Finsch are considered, the LREE show a much better correlation with Cr/Ca than Ca content (Figs 5 and 6). The relationship between Cr3+ and REE fractionation does not appear to be a charge-balancing effect of the trivalent REE in the garnet X-cation site, by substitution of Al3+ in the Z site, because there is no Si4+ deficiency and Al3++ Cr3+ 
2 cations per formula unit. Significant substitution of Al by the larger Cr cation will, however, increase the effective radius of the garnet octahedral Y site and cause a large amount of distortion of the garnet framework. This will result in enlargement of the X-cation site (Van Westrenen et al., 2000). Such behaviour appears to be exemplified by the Finsch garnets, where there is greater preferential substitution of large cations (i.e. Sr and the LREE) in the high-Cr garnets. It may also explain the apparent restriction of high Sr contents (>30 ppm) to G10 garnets (Pearson et al., 2003).
Our findings agree with the results of the experimental study of Wang et al. (1998), who examined the influence of Cr on garnet–melt REE partitioning by undertaking high-temperature (1900–2100°C) and high-pressure (7·5 GPa) experiments on garnets doped with variable amounts of Cr2O3 (0–14 wt %). The results of their study suggest that D
values increase with, and are more affected by, Cr content than D
values. Furthermore, the large amount of substitution of Al3+ by Cr3+ in the sub-calcic garnets present in the Finsch xenoliths and diamond inclusions would limit the replacement of Al in the garnet Y-cation site by Ti, Zr and Hf. This is consistent with the observations that we have outlined above and would account for the relative depletions of these trace elements on a normalized multi-element plot (Fig. 3).
P–T–X dependence
Experiments producing lherzolite assemblages have been conducted at temperatures between 1425 and 1750°C (e.g. Salters et al., 2002; Tuff & Gibson, 2007) and have generated pyrope garnets with variable CaO (3–6 wt %), low Cr2O3 (< 1·7 wt %) and high Al2O3 contents (22 wt %). These studies have shown that partitioning of the REE into garnet is dependent on temperature as well as Ca content. This is because contraction of the garnet lattice during cooling causes an increase in elasticity, which allows enhanced substitution of large cations into the garnet X site. Hence the amount of Ca substitution by the LREE in pyrope garnets crystallizing in equilibrium with clinopyroxene decreases with increasing temperature. We have shown above, however, that Ca shows a good correlation with temperature only if the harzburgitic garnet data are excluded, so, although this might explain variations in LREE in the lherzolitic garnets, a different mechanism is necessary to explain the high partitioning of LREE into sub-calcic high-Cr garnets, such as those present in F05JM9.
The Finsch garnets display a good correlation between Cr, the LREE and also Cr/Ca ratio with temperature (Fig. 14). The dependence of LREE concentrations on Cr/Ca ratio is not readily apparent in lherzolitic garnets because Cr contents are generally low. As a consequence, lherzolitic garnets show reasonable correlations between Ca and LREE, but in harzburgitic garnets, where Cr contents are high and Ca contents low, this relationship breaks down. Substitution of Cr3+ for Al3+ in the garnet Y-cation site is known to decrease with pressure and slightly with temperature (Nickel, 1989; Ryan et al., 1996; Grütter et al., 2006). As temperature and pressure increase, the radius of the garnet X-cation site will decrease and inhibit substitution of large for small ions (such as Cr3+ for Al3+); Sr and the LREE will also fit less readily into the garnet structure than the HREE. This may explain, to some extent, the relatively low LREE concentrations in the high-pressure and -temperature Finsch lherzolitic garnets (Figs 6c,d and 14d).
|
Another important factor that may have influenced trace-element concentrations and ratios in the Finsch garnets may be bulk-rock composition. The subcalcic garnets occur in harzburgites, which are present at shallower depths than the lherzolites. It might be argued that the lack of clinopyroxene results in the LREE and Cr partitioning more readily into the harzburgitic garnets. Griffin et al. (1999c
) showed that there is a negative relationship between garnet Cr2O3 content and bulk-rock Al2O3. Bulk-rock Al2O3 will be influenced by the degree of partial melting involved in the initial formation of the lithospheric mantle, but it will also be affected by the degree of subsequent metasomatism (see below). The amount of metasomatism is dependent upon pressure and temperature as well as the volume and composition of the infiltrating melt or fluid. Small volumes of metasomatic melts or fluids will readily lose heat to their surroundings such that their ascent may be limited and enrichment will be concentrated at the base of the lithosphere. At Finsch, where we see a continuous variation in both REE patterns and also equilibration pressures and temperatures, we believe that both crystal chemistry and bulk-rock composition are important in determining the partitioning of LREE into garnets. Elsewhere, the situation may be much more complex, and it may be fortuitous that at Finsch, where there have not been multi-phase metasomatic events, we have been able to see this relationship. Further studies of unzoned garnets from xenoliths with well-constrained PT estimates are required to establish the significance of one or both of these processes in LREE partitioning.
Garnet crystallization and HREE fractionation
Sinusoidal chondrite-normalized REE patterns with positive HREE slopes characterize pyrope garnets that occur in clinopyroxene-free Finsch peridotites and all occur in xenoliths that last equilibrated at temperatures < 1100°C (Fig. 4). As we shall show below, the harzburgites appear to represent the residue of large-degree polybaric melt extraction predominantly in the spinel stability field (1·5 GPa). Because of the absence of residual garnet, at such low pressures and large degrees of partial melting, we believe that it is unlikely that the positive HREEn slope is linked to the mantle melting event that caused the initial formation of the harzburgite (see Stachel et al., 1998). Instead, we examine the role of metasomatic agents that may have infiltrated the harzburgites after their formation and caused the HREE fractionation.
Figure 12 shows that the partitioning of Yb, Tm and Lu relative to Er is very different for garnet and clinopyroxene. In clinopyroxenes these elements, and also the LREE and MREE, have lower D values than Er, whereas garnets exhibit a negative correlation between ionic radius and D value for all of the REE. This is because the effective radius (ro) of the cation site into which the REE partition in clinopyroxene is 0·1 nm (i.e. Er) whereas in pyrope garnets ro is 0·09 nm (i.e. Lu). Additionally, Harte et al. (1996) have shown from studies of mantle xenoliths that DCpx/Grt ratios for most of the REE are strongly temperature dependent; DCpx/Grt values for large ions such as Nd exhibit a negative correlation with temperature whereas elements with a smaller ionic radius (such as Lu) have a positive correlation. DCpx/Grt for Er, however, exhibits very little variation so that [Nd/Er]n and [Er/Lu]n ratios for garnets in equilibrium with clinopyroxene show opposite trends when plotted against temperature (Fig. 15). At first sight this may explain why pyrope garnets from Finsch xenoliths exhibit differences in their HREE ratios. It might be assumed that when clinopyroxene ceased to crystallize (i.e. in the lower temperature Finsch peridotite) there would be more Tm, Yb and Lu relative to Er in the melt available for partitioning into garnet; this would produce sinusoidal chondrite-normalized REE patterns with [Nd/Er]n 
1 and [Er/Lu]n 1. However, because the HREE are incompatible in clinopyroxene (i.e. DCpx–melt values are < 1, Fig. 12), the lack of clinopyroxene crystallization from the melt has little effect on HREE ratios. What might be more important is the amount of garnet that fractionates from the melt because D values vary between one and 10. As the amount of garnet crystallization from the metasomatic melt increases so does the HREE fractionation both in the melt and in the garnets that are in equilibrium with it. Hence garnets that equilibrate at low temperatures with percolating and crystallizing metasomatic melts may have relatively fractionated HREE ratios. This is consistent with evidence from LREE that the sub-calcic garnets in the Finsch harzburgite xenoliths have undergone some metasomatism.
|
|
EVOLUTION OF THE LITHOSPHERIC MANTLE BENEATH THE WESTERN MARGIN OF THE KAAPVAAL CRATON |
|---|
We have reconstructed the bulk-rock compositions of the Finsch xenoliths by combining estimates of modal mineralogy with major- and trace-element mineral chemistry (Table 5). This type of bulk-rock reconstruction is sensitive to any heterogeneity in modal mineralogy but was chosen in preference to bulk-rock analysis because it overcomes: (1) secondary metasomatic effects related to infiltration of kimberlite melt (e.g. Hawkesworth et al., 1983
; Simon et al., 2007); (2) hydrothermal alteration and serpentinization, which causes loss of K, Rb, etc. and introduction of FeO, CaO, TiO2 (Boyd et al., 1997) and SiO2. As might be expected from their variations in modal mineralogy and mineral chemistry, the Finsch xenoliths exhibit a wide range of bulk-rock compositions. The harzburgites have the lowest concentrations of CaO, Al2O3, FeO, TiO2 and LREE, and are the most ‘depleted’ (e.g. Fig. 16). Concentrations of these magmaphile oxides and elements are greatest in the lherzolites, especially BD3692 and BD3693, but are less than those present in fertile mantle (Table 5).
|
|
Melt depletion: formation of Finsch harzburgites
Finsch garnet harzburgite F05JM9 is characterized by high bulk-rock Mg-number (0·923), moderate FeO (7·22 wt %), and low SiO2 (42·6 wt %), Al2O3 (0·18 wt %) and CaO (0·11 wt %; Table 5) relative to fertile mantle peridotite. Herzberg (2004
) has recently parameterized the results of experimental studies on fertile peridotite to calculate the compositions of residues that form during polybaric melting of upwelling mantle (Fig. 17). A comparison of the estimated abundances of Al2O3, SiO2, FeO and MgO in the Finsch harzburgite with the melt trajectories calculated by Herzberg (2004) indicates that F05JM9 may represent the residue of extensive (40%), polybaric melting of fertile peridotite (Fig. 17). Figure 17 suggests that in upwelling convecting mantle such large-degree melting would start at 4·5 GPa and continue until 1·5 GPa. The high-pressure melts would be in equilibrium with garnet whereas at the top of the melting column spinel would be the stable aluminous phase, although it is likely that most of this would have been exhausted at such large degrees of partial melting. If we assume that the upwelling mantle had a composition similar to a fertile peridotite, such as KR4003, then intersection of the solidus at 4·5 GPa corresponds to a potential temperature (Tp) of 1550°C (Fig. 18). During the Archaean, such a Tp would have been similar to that of ambient convecting mantle (Richter, 1988) and does not support hypotheses that invoke harzburgites as residues of high-pressure and -temperature melting in upwelling mantle plumes (Pearson et al., 1995a; Herzberg, 1999; and see below).
|
|
The extremely low contents of CaO in garnet (< 1· 42 wt %) and Al2O3 in orthopyroxene (<0·66 wt %), together with the high Mg-number of olivine and orthopyroxene (up to 95·5) in Finsch diamond inclusions (Gurney et al., 1979
; Tsai et al., 1979; Richardson et al., 1984), indicate that the lithospheric mantle beneath the western Kaapvaal may have undergone even larger amounts of melting than we have estimated for harzburgite F05JM9. The diamond inclusions would have been shielded from any post 3·3 Ga melt enrichment events and it may be that harzburgite F05JM9 has undergone slight modification by cryptic metasomatism, post diamond formation as discussed above. Re–Os ages, determined on whole-rock samples and sulphide inclusions from both lherzolitic and harzburgitic Finsch xenoliths, confirm that this melt depletion event occurred prior to 3 Ga (Pearson et al., 1995a; Griffin et al., 2004).
We used the following equation to calculate the concentration of elements in the aggregate fractional melt that generated a residue with the composition of the Finsch harzburgite:
|
|
), F is the mass fraction of the aggregate melt, Cl is the concentration in the aggregate melt and Cs is the concentration in the solid residue (i.e. F05JM9, Table 5). Because the amount of melting we have estimated from harzburgite F05JM9 is a minimum, estimates of the composition of the extracted melt will contain less MgO but be richer in CaO and Al2O3 than the actual melts. Our calculations suggest that the aggregate melt, which left F05JM9 as a final residue, contained 20 wt % MgO, 9·5 wt % FeO, 9·66 wt % CaO, 11· 84 wt % Al2O3 and 49 wt % SiO2. This is similar to a Hawaiian picrite, although subtle differences in some oxides (e.g. Al2O3) reflect the shallower top of the melt column and slightly higher Tp involved in the formation of the Finsch residue. We envisage that the residue formed in an Archaean spreading-ridge environment.
The most plausible explanation for the origin of the high Cr-number, sub-calcic garnets present both in Finsch diamond inclusions and peridotite xenoliths (e.g. F05JM9) is that they formed during lithospheric thickening, rather than that they are residual phases of an early melting event (i.e. they are metamorphic). In rare peridotite xenoliths, pyrope garnet can be seen exsolving from orthopyroxene (Cox et al., 1987; Dawson, 2004) but these garnets are characterized by low Cr2O3 contents (2·5 wt %) and thus unlike the ones that occur in the Finsch harzburgites. Previous studies have shown that high-Cr pyrope garnets must have formed from a melt residue with a high Cr-number (>30). Because the K d Cr/Al Grt-melt value is 1 (Canil & Wei, 1992), residues with such high Cr-number are not able to form as a result of melting in the garnet stability field (Stachel et al., 1998). Melt depletion in the spinel stability field can, however, produce a residue with a Cr-number as high as 42 (Tainton & McKenzie, 1994). The wide variation of garnet Cr contents might indicate variable modal amounts of Cr spinel in the original melt residue, although we believe that, in general, the modal amount of Cr-spinel would be low, such that only a small amount of garnet would form. Such pressure-induced transformation of the aluminous phase would theoretically also consume two pyroxenes and produce mantle rich in olivine and low in orthopyroxene (Johnson et al., 1990):
|
|
|
|
|
|
|
|
|
|
|
|
Significance of Low-T Finsch peridotites
The Kaapvaal lithospheric mantle is believed to predominantly consist of Mg- and enstatite-rich (20–45%), coarse garnet peridotites that last equilibrated at low temperatures (below 1150°C); these have been termed Low-T peridotites (Boyd & Mertzman, 1987; Boyd, 1989). Samples of Low-T peridotites from the Kaapvaal and Siberian cratons have been noted for their high Mg-number at a given modal olivine content relative to samples from oceanic lithosphere (abyssal peridotites and ophiolite tectonites; Fig. 19). The high Mg-number of Low-T cratonic peridotites has been widely attributed to their origin as residues of very high amounts of partial melting, as a consequence of elevated Archaean mantle temperatures (Boyd, 1989). The Low-T Finsch peridotites are significant because they have much higher modal olivine (up to 96%) than has previously been estimated for Low-T peridotites (<85%) from other parts of the Kaapvaal craton (Hawkesworth et al., 1983; Boyd & Mertzman, 1987; Boyd et al., 1997; Grégoire et al., 2003; Simon et al., 2007). Nevertheless, Low-T peridotites with very high modal olivine contents (up to 98·5 modal %) have recently been discovered from the Greenland and Slave cratons (Bernstein et al., 1998; Schmidberger & Francis, 1999). These are similar to the Finsch Low-T peridotites although, with the exception of the garnet harzburgite (F05JM09), the Finsch samples have slightly less Mg-rich olivine at a given modal per cent olivine (Fig. 19).
|
The high modal olivine content of the Finsch Low-T peridotites is compensated by low modal orthopyroxene (< 17%). Other Kaapvaal (and also Siberian) craton Low-T peridotites have an average of
30% orthopyroxene (Boyd, 1989). This is higher than has been observed in residues of partial melting experiments of anhydrous fertile peridotites (<25%; e.g. Walter, 1998), and in numerical models of polybaric mantle melting of anhydrous peridotites (Herzberg, 2004).
The large variation in modal olivine (40–85%) and high contents of orthopyroxene in Low-T Kaapvaal peridotites have been attributed to: (1) large-degree melting of peridotite at high pressures (Boyd, 1989; Herzberg, 1999); (2) metamorphic differentiation (Boyd, 1989); (3) coarse layering in the lithospheric mantle as a result of cumulate sorting (Herzberg, 1993); (4) metasomatic replacement of olivine by orthopyroxene (Boyd et al., 1993; Kelemen et al., 1998). In contrast, the unusually low modal orthopyroxene contents of the Finsch Low-T peridotites are consistent with those calculated for residues of melt extraction.
We believe that the Low-T Finsch peridotites are important end-members in models of the multi-stage evolution of the Kaapvaal lithospheric mantle. Figure 17 shows that, if the bulk-rock composition of Finsch harzburgite (F05JM9) is representative of the initial Kaapvaal lithospheric mantle residue, the increase in modal orthopyroxene evident in other Low-T peridotites may be associated with an increase of up to 4 wt % SiO2. A simple increase in modal orthopyroxene with a composition similar to that present in the Kaapvaal Low-T peridotites (as a result of SiO2 addition) does not, however, satisfactorily account for either the elevated contents of Al2O3 (0·7–2· 46 wt %) and CaO (0·31–1·36 wt %) of these xenoliths relative to the Finsch harzburgite (0·18 wt % Al2O3 and 0·11 wt % CaO; Fig. 16) or the depletion in FeO and MgO. The Al2O3 contents of the orthopyroxene in the Kaapvaal Low-T peridotites are similar to those at Finsch (<0·8 wt %), presumably because they equilibrated along similar geothermal gradients in the garnet stability field. The increase in Al2O3 of the Kaapvaal Low-T peridotites may be due to either addition of orthopyroxene that originally contained 4 wt % Al2O3 [i.e. similar to the composition of orthopyroxene in equilibrium with spinel rather than garnet peridotite (Fig. 17b)] or introduction of additional phases, such as spinel, garnet and/or phlogopite (Bell et al., 2003).
|
MELT–ROCK REACTION: METASOMATISM AND LHERZOLITE FORMATION |
|---|
The role of metasomatic melts vs fluids involved in the LREE enrichment of sub-calcic garnets remains controversial (e.g. Stachel & Harris, 1997
; Burgess & Harte, 2004; Stachel et al., 2004). Stachel and co-workers have argued against the role of melts on the grounds that they would freeze upon equilibration with the surrounding peridotite and have proposed, instead, that the high LREE, and low Ti, Zr, Y and HREE that characterize sub-calcic garnets are related to a C–H–O-rich fluid. For H2O-rich fluids, interconnectivity along grain boundaries is increased considerably with increasing temperature and pressure (Watson et al., 1990) and Stachel & Harris (1997) have argued that metasomatism of sheared peridotites, within the diamond stability field, could take place at subsolidus conditions by H2O-rich fluids. Nevertheless, experiments by Ono et al. (2002) have indicated that dihedral angles for aqueous fluids are only <60° at pressures greater than 8–9 GPa. Percolation of fluids would therefore occur only at pressures greater than those observed in the lithospheric mantle. The experiments of Ono et al. (2002) involved a pyrope-rich matrix and may not be wholly appropriate for peridotites. Nevertheless, the low equilibration pressures (3·5 GPa) that we have estimated for the Finsch harzburgites and their coarse–equant textures do not appear to be conducive to the percolation of fluids. We therefore currently favour a mechanism involving metasomatic melts to explain the trace-element characteristics of sub-calcic Finsch garnets. We believe that the migration of these melts through the lithospheric mantle would be enhanced if they are volatile-rich and are of such a large volume that they would not fully react with the surrounding peridotite. Our model implies that this metasomatism is not directly linked to Finsch diamond formation.
The wide variations in bulk-rock composition that we have observed in the Finsch peridotite xenolith suite appear to be the result of different degrees of metasomatism of a harzburgite host. Our results indicate that, prior to 120 Ma, the lithosphere beneath the western part of the Kaapvaal craton had been pervasively metasomatized between 180 and 130 km depth (Fig. 10). Lherzolites from the base of the lithospheric mantle (e.g. BD3692, F05JM6) appear to have undergone significant enrichment in Fe, Ca, Al, Si, Mn, Na, Ti, Cr, REE, Sr, Y and Zr relative to the ‘shallow’ harzburgites. The style of modal metasomatism beneath Finsch also varies with depth: both Cr-diopside and garnet (G9) are present at depths >130 km but Cr diopside is absent between 100 and 130 km. The continuous Ca vs Cr trends that we have observed in the Finsch garnets have been reported from other mantle xenolith suites and interpreted as evidence either of progressive major-element depletion such that high-Cr sub-calcic garnets (i.e. G10 garnets) found in harzburgites are residua of melt extraction from primary lherzolite compositions (Griffin et al., 1992; Grütter et al., 1999), or of refertilization of harzburgites to form lherzolites (Schulze, 1991; Griffin et al., 1999a; Burgess & Harte, 2004).
Metasomatic melt compositions
Previous studies have highlighted the fact that Cr-diopsides and pyrope garnets in the same mantle xenolith appear to have crystallized from melts of different compositions (e.g. Jones, 1987; Simon et al., 2003). In some cases, this may be an artefact of errors in garnet–melt partition coefficients (Tainton & McKenzie, 1994). We have used the DGrt–melt and DCpx–melt values defined in Table 4 to calculate equilibrium melt compositions responsible for lithospheric mantle enrichment and lherzolite formation beneath Finsch. Mantle-normalized compositions of these melts are plotted in Fig. 20. This shows that the estimated compositions of the equilibrium melts are extremely similar for both the Cr-diopsides and pyrope garnets present in lherzolite F05JM7. The metasomatic melts are strongly enriched in LREE (up to 500 x primitive mantle) and large ion lithophile elements (LILE) and have depletions at Th, Nb and Ti on mantle-normalized multi-element plots. The LREE/HREE ratios of the melts appear to increase with decreasing equilibration pressure such that the compositions of melts estimated to be in equilibrium with garnets in the diamond inclusions, High-Cr lherzolitic and sub-calcic harzburgitic garnets have much steeper normalized LREE/HREE patterns than those of the Low-Cr lherzolitic garnets. When normalized to the composition of melts in equilibrium with the least enriched xenolith (F05JM9) it can be seen that the greatest enrichments in the high-pressure lherzolites are in the LREE, HREE, Zr, Sr and perhaps Ti (Figs 21 and 22). The most plausible interpretation of this difference in equilibrium melt compositions is that the melts have undergone fractional crystallization of both garnet and clinopyroxene during their ascent. Garnets, and where present Cr-diopsides, record evidence of this melt–rock reaction in single xenoliths. A similar percolation fractionation model was proposed by Burgess & Harte (2004) to account for variations in garnet chemistry in Jagersfontein peridotite xenoliths.
|
|
|
Age of lithospheric enrichment events
Richardson et al. (1984
) have proposed that the Sr-isotopic ratios of sub-calcic garnet inclusions from Finsch diamonds required a two-stage melting model, involving melt depletion and enrichment (immediately prior to 3·3 Ga diamond formation). Concentrate sub-calcic garnets (i.e. harzburgite paragenesis) from Finsch have higher 143Nd/144Nd (0·5112–0·512) and 87Sr/86Sr ratios (0·707–0·732) than garnet inclusions in diamond (0·5109, 0·7035; Richardson et al. 1984). 143Nd/144Nd and 87Sr/86Sr ratios for a Low-Cr lherzolitic garnet are even higher (0·51295 and 0·7638; Pearson et al., 1995a). Sm–Nd model ages are slightly younger for the concentrate sub-calcic garnets (tCHUR = 1·3–3·1 Ga) than the diamond inclusions but even younger for the lherzolitic garnet (tCHUR = 181 Ma). Ages of the latter are very close to the time of entrainment and emplacement of the Finsch kimberlite. The Sm–Nd model ages and wide range of 87Sr/86Sr ratios indicate that lithospheric enrichment, of variable styles, has occurred both before and after 3·3 Ga diamond formation beneath the western margin of the Kaapvaal craton. Sm–Nd model ages for diopsides suggest that the Kaapvaal lithospheric mantle was subjected to a major metasomatic event at 1 Ga (Pearson, 1999) but if we assume that the Sm–Nd model age for the Low-Cr lherzolitic garnet analysed by Pearson et al. (1995a) is representative of this paragenesis, then significant refertilization of the base of the western Kaapvaal lithosphere may have occurred more recently.
We note that the predicted compositions of metasomatic melts in equilibrium with the clinopyroxene-free samples have much higher concentrations of LREE and lower concentrations of Ti and HREE than the host 120 Ma Finsch kimberlite (Fig. 20). Those in equilibrium with the lherzolites are more comparable. Nevertheless, variations in garnet and clinopyroxene chemistry and modal abundance of the peridotites must have occurred prior to entrainment and ascent to the surface, otherwise more evidence of textural and chemical disequilibrium would be apparent. In this respect, the metasomatic enrichment that we have identified at Finsch is distinctly different from that which caused the formation of strongly zoned garnets found in xenoliths entrained by the 90 Ma, Group 1, Jagersfontein kimberlite (Burgess & Harte, 2004). Differences between lithospheric mantle sampled by Group I and II southern African kimberlites from the SW of the Kaapvaal craton have previously been proposed by Griffin et al. (1999b, 2003). They suggested, from their study based predominantly on garnet concentrates, that Group II kimberlites sample less metasomatized lithospheric mantle than Group I kimberlites. Our findings for Finsch are broadly consistent with this hypothesis.
|
CONCLUSIONS |
|---|
(1) Mantle peridotite xenoliths from Finsch Mine provide a rare insight into the evolution of the lithosphere beneath the west of the Kaapvaal craton. They are important because they were entrained by Group II kimberlites and have not been subjected to the metasomatic event that is evident in mantle xenoliths entrained by Group I kimberlites (e.g. Jagersfontein, Wesselton, NW Lesotho). This recent metasomatism appears to have caused major- and trace-element disequilibrium in lithospheric mantle phases, such as garnet and clinopyroxene (Griffin et al., 1999b
; Simon et al., 2003; Burgess & Harte, 2004; S. A. Gibson et al., unpublished data).
(2) Our study shows that the amount of Cr-diopside present in the sub-Finsch lithospheric mantle decreases consistently with decreasing depth. We interpret this as evidence of reduced amounts of mantle metasomatism as low-volume small-fraction volatile-rich melts ‘froze’ during their ascent through the lithospheric mantle. As has been observed from other parts of the Kaapvaal craton, the deepest samples from Finsch are all garnet lherzolites whereas garnet harzburgites were entrained from shallower depths. Pressure estimates using the Al-in-Opx barometer of Brey et al. (2008) place the lherzolite–harzburgite transition at 35–40 kbar.
(3) Variations in incompatible trace elements in garnet and clinopyroxene from lherzolite xenoliths correlate with those of major elements, such as Ca. However, this correlation is not apparent in the harzburgitic garnets, where Ca contents are low and concentrations of LREE are high. The REE patterns of the garnets exhibit a systematic variation from ‘smooth’ to ‘sinusoidal’ with increasing Cr/Ca ratio. The variation in LREE may be due to changes in crystal chemistry related to variations in pressure (and temperature) and/or accompanying differences in bulk-rock composition. The latter may reflect decreasing amounts of mantle metasomatism with depth.
(4) The unusually low modal abundances of orthopyroxene in the Finsch peridotites, compared with those from elsewhere in the Kaapvaal craton, suggest that this part of the lithosphere has not undergone the pervasive SiO2 enrichment experienced further east. The reaction of silica-rich melt with olivine to form orthopyroxene is believed to have occurred at 2·9 Ga and has been linked with subduction zones, associated with the amalgamation of the Kimberley and Witwatersrand blocks, and stabilization of the Kaapvaal craton (e.g. Schmitz et al., 2004; Simon et al., 2007). This silica enrichment has also been linked by some to diamond formation (Bell et al., 2005). In contrast, the low modal orthopyroxene contents in the diamondiferous Finsch peridotites (Shee et al., 1982; Viljoen et al., 1992) suggest that these processes are independent of one another. The sub-Finsch lithospheric mantle would have been remote from the subduction zone on the eastern margin of the Kimberley block and this may explain the lower modal amounts of orthopyroxene. The lithospheric mantle sampled by the Kimberley kimberlites (Fig. 23) would have been much closer to the subduction zone and hence more susceptible to modification by slab-derived melts and/or fluids. We believe that Finsch harzburgite F05JM9 represents one of the most highly refractory xenoliths sampled from the Kaapvaal craton.
|
|
SUPPLEMENTARY DATA |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
We are extremely grateful to Mathew Field, William MacDonald, Jock Robey and Ken Tainton of DeBeers Consolidated Mines who granted permission to sample the xenoliths at Finsch Mine and provided logistical support to J.M. We thank Barry Dawson for his encouragement and for generously donating his collection of Finsch mantle xenoliths. Chris Hayward is thanked for his assistance with the electron microprobe, and Dan McKenzie for allowing us to use his PTMIN and FITPLOT software. Research on Finsch xenoliths was initiated by J.M. as an MSc project at the University of Cambridge and was funded by the Department of Earth Sciences. We are indebted to G. Pearson, T. Stachel and K. S. Viljoen for their thorough and constructive reviews of an earlier version of the manuscript. Their thought-provoking comments have greatly improved the paper. A further set of comments was provided by an anonymous reviewer. This is Department of Earth Sciences Contribution ES9309.
|
FOOTNOTES |
|---|
Present address: Department of Earth Sciences, Durham University, Science Labs, Durham DH1 3LE, UK. 
*Corresponding author: Telephone: +44 1223 333400. Fax: +44 1223 333450. E-mail: sally{at}esc.cam.ac.uk
|
REFERENCES |
|---|
Appleyard CM, Viljoen KS, Dobbe R. A study of eclogitic diamonds and their inclusions from the Finsch kimberlite pipe, South Africa. Lithos (2004) 77(1–4):317–332.[CrossRef][Web of Science]
Ballaran TB, Carpenter MA, Geiger CA, Koziol AM. Local structural heterogeneity in garnet solid solutions. Physics and Chemistry of Minerals (1999) 26(7):554–569.[CrossRef][Web of Science]
Becker M, Le Roex AP. Geochemistry of South African on- and off-craton, Group I and Group II kimberlites: petrogenesis and source region evolution. Journal of Petrology (2006) 47(4):673–703.
Bell DR, Schmitz MD, Janney PE. Mesozoic thermal evolution of the southern African mantle lithosphere. Lithos (2003) 71:273–287.[CrossRef][Web of Science]
Bell D, Grégoire M, Grove T, Chatterjee N, Carlson R, Buseck P. Silica and volatile-element metasomatism of Archean mantle: a xenolith-scale example from the Kaapvaal Craton. Contributions to Mineralogy and Petrology (2005) 150(3):251–367.[CrossRef][Web of Science]
Bernstein S, Kelemen PB, Brooks CK. Depleted spinel harzburgite xenoliths in Tertiary dykes from East Greenland: Restites from high degree melting. Earth and Planetary Science Letters (1998) 154(1–4):221–235.[CrossRef][Web of Science]
Bernstein S, Hanghøj K, Kelemen PB, Brooks CK. Ultra-depleted, shallow cratonic mantle beneath West Greenland: dunitic xenoliths from Ubekendt Ejland. Contributions to Mineralogy and Petrology (2006) 152(3):335–347.[CrossRef][Web of Science]
Blundy J, Wood B. Prediction of crystal–melt partition coefficients from elastic moduli. Nature (1994) 372(6505):452–454.[CrossRef][Web of Science]
Blundy J, Wood B. Partitioning of trace elements between crystals and melts. Earth and Planetary Science Letters (2003) 210(3–4):383–397.[CrossRef][Web of Science]
Boyd FR. Compositional distinction between oceanic and cratonic lithosphere. Earth and Planetary Science Letters (1989) 96(1–2):15–26.[CrossRef][Web of Science]
Boyd FR, Mertzman SA. Composition and structure of the Kaapvaal lithosphere, southern Africa. Magmatic Processes Physiochemical Principles (1987) 1:13–24.
Boyd FR, Pearson DG, Nixon PH, Mertzman SA. Low-calcium garnet harzburgites from Southern Africa—their relations to craton structure and diamond crystallization. Contributions to Mineralogy and Petrology (1993) 113(3):352–366.[CrossRef][Web of Science]
Boyd FR, Pokhilenko NP, Pearson DG, Mertzman SA, Sobolev NV, Finger LW. Composition of the Siberian cratonic mantle: evidence from Udachnaya peridotite xenoliths. Contributions to Mineralogy and Petrology (1997) 128(2):228–246.[CrossRef][Web of Science]
Brey GP, Kohler T. Geothermobarometry in four-phase lherzolites II. New thermobarometers, and practical assessment of existing thermobarometers. Journal of Petrology (1990) 31(6):1353–1378.
Brey GP, Kohler T, Nickel KG. Geothermobarometry in four-phase lherzolites I. Experimental results from 10 to 60 kb. Journal of Petrology (1990) 31:1313–1352.
Brey GP, Bulatov VK, Girnis AV. Geobarometry for peridotites: experiments in simple and natural systems from 6 to 10 GPa. Journal of Petrology (2008) 49(1):3–24.
Brunelli D, Seyler M, Cipriani A, Ottolini L, Bonatti E. Discontinuous melt extraction and weak refertilization of mantle peridotites at the Vema Lithospheric Section (Mid-Atlantic Ridge). Journal of Petrology (2006) 47(4):745–771.
Burgess SR, Harte B. Tracing lithosphere evolution through the analysis of heterogeneous G9/G10 garnets in peridotite xenoliths, 1: Major element chemistry. In: Proceedings of the 7th International Kimberlite Conference—Gurney JJ, Gurney JL, Pascoe MD, Richardson SH, eds. (1999) Cape Town: Red Roof Design. 66–80.
Burgess SR, Harte B. Tracing lithosphere evolution through the analysis of heterogeneous G9–G10 garnets in peridotite xenoliths, II: REE chemistry. Journal of Petrology (2004) 45(3):609–634.
Canil D, O’Neill H. SC. Distribution of ferric iron in some upper-mantle assemblages. Journal of Petrology (1996) 37(3):609–635.
Canil D, Wei K. Constraints on the origin of mantle-derived low Ca garnets. Contributions to Mineralogy and Petrology (1992) 109:421–430.[CrossRef][Web of Science]
Canil D, O’Neill H. SC, Pearson DG, Rudnick RL, McDonough WF, Carswell DA. Ferric iron in peridotites and mantle oxidation states. Earth and Planetary Science Letters (1994) 123(1–3):205–220.[CrossRef][Web of Science]
Cox KG, Smith MR, Beswetherick S. Textural studies of garnet lherzolites: evidence of exsolution origin from high-temperature harzburgites. In: Mantle Xenoliths—Nixon PH, ed. (1987) New York: John Wiley. 537–550.
Dawson JB. Contrasting types of upper mantle metasomatism? The mantle and crust–mantle relationship. In: Proceedings of the 3rd International Kimberlite Conference—Kornprobst J, ed. (1984) Amsterdam: Elsevier. 289–294.
Dawson JB. A fertile harzburgite–garnet lherzolite transition: possible inferences for the roles of strain and metasomatism in upper mantle peridotites. Lithos (2004) 77(1–4):553–569.[CrossRef][Web of Science]
Dawson JB, Smith JV. Occurrence of diamond in a mica–garnet lherzolite xenolith from kimberlite. Nature (1975) 254(5501):580–581.[CrossRef][Web of Science]
Dawson JB, Smith JV. The MARID (mica–amphibole–rutile–ilmenite–diopside) suite of xenoliths in kimberlite. Geochimica et Cosmochimica Acta (1977) 41(2):309–310.[CrossRef][Web of Science]
Dawson JB, Smith JV. Metasomatized and veined upper-mantle xenoliths from Pello Hill, Tanzania—Evidence for anomalously ight mantle beneath the Tanzanian sector of the East African Rift Valley. Contributions to Mineralogy and Petrology (1988) 100(4):510–527.[CrossRef][Web of Science]
Dawson JB, Stephens WE. Statistical classification of garnets from kimberlite and associated xenoliths. Journal of Geology (1976) 53:589–607.
de Wit MJ, de Ronde C. EJ, Tredoux M, Roering C, Hart RJ, Armstrong RA, Green R. WE, Peberdy E, Hart RA. Formation of an Archaean continent. Nature (1992) 357(6379):553–562.[CrossRef][Web of Science]
Ekkerd J, Stiefenhofer J, Field M, Lawless P. The geology of the Finsch Mine, Northern Cape Province, South Africa. In: In: 8th International Kimberlite Conference 8th International Kimberlite Conference Long Abstract (2003) Victoria: BC.
Finnerty AA, Boyd FR. Evaluation of thermobarometers for garnet peridotites. Geochimica et Cosmochimica Acta (1984) 48:15–27.[CrossRef][Web of Science]
Finnerty AA, Boyd FR. Thermobarometry for garnet peridotites: basis for the determination of thermal and compositional structure of the upper mantle. In: Mantle Xenoliths—Nixon PH, ed. (1987) New York: John Wiley. 381–402.
Fouch MJ, James DE, VanDecar JC, van der Lee S, Kaapvaal Seismic Group. Mantle seismic structure beneath the Kaapvaal and Zimbabwe Cratons. South African Journal of Geology (2004) 107(1–2):33–44.
Grégoire M, Bell DR, Le Roex AP. Garnet lherzolites from the Kaapvaal craton (South Africa): Trace element evidence for a metasomatic history. Journal of Petrology (2003) 44(4):629–657.
Griffin WL, Gurney JJ, Ryan CG. Variations in trapping temperatures and trace elements in peridotite-suite inclusions from African diamonds: evidence for two inclusion suites, and implications for lithosphere stratigraphy. Contributions to Mineralogy and Petrology (1992) 110:1–15.[CrossRef][Web of Science]
Griffin WL, Fisher NI, Friedman J, Ryan CG, O’Reilly SY. Cr-pyrope garnets in the lithospheric mantle. I. Compositional systematics and relations to tectonic setting. Journal of Petrology (1999a) 40:679–704.[CrossRef][Web of Science]
Griffin WL, Shee SR, Ryan CG, Win TT, Wyatt BA. Harzburgite to lherzolite and back again: metasomatic processes in ultramafic xenoliths from the Wesselton kimberlite, Kimberley, South Africa. Contributions to Mineralogy and Petrology (1999b) 134(2–3):232–250.[CrossRef][Web of Science]
Griffin WL, O’Reilly SY, Ryan CG. The composition and origin of subcontinental lithospheric mantle. In: Mantle Petrology: Field Observations and High Pressure Experimentation: a Tribute to Francis R. (Joe) Boyd. Geochemical Society, Special Publications—Fei Y, Bertka CM, Mysen BO, eds. (1999c) 6:13–45.
Griffin WL, O’Reilly SY, Natapov LM, Ryan CG. The evolution of lithospheric mantle beneath the Kalahari Craton and its margins. Lithos (2003) 71:215–241.[CrossRef][Web of Science]
Griffin WL, Graham S, O’Reilly SY, Pearson NJ. Lithosphere evolution beneath the Kaapvaal Craton: Re–Os systematics of sulfides in mantle-derived peridotites. Chemical Geology (2004) 208(1–4):89–118.[CrossRef][Web of Science]
Grütter HS, Apter DB, Kong J. Crust–mantle coupling: evidence from mantle-derived xenocrystic garnets. In: Proceedings of the 7th International Kimberlite Conference—Gurney JJ, Gurney JL, Pascoe MD, Richardson SH, eds. (1999) Cape Town: Red Roof Design. 307–313.
Grütter H, Latti D, Menzies A. Cr-saturation arrays in concentrate garnet compositions from kimberlite and their use in mantle barometry. Journal of Petrology (2006) 47(4):801–820.
Grutzeck MW, Kridelbough SJ, Weill DF. The distribution of Sr and REE between diopside and the silicate liquid. Geophysical Research Letters (1974) 1:273–275.
Gurney JJ. A correlation between garnets and diamonds. In: Kimberlite Occurrence and Origin: a Basis for Conceptual Models in Exploration—Glover JE, Harris PG, eds. (1984) Perth, Australia: University of Western. 143–166.
Gurney JJ, Switzer GS. Discovery of garnets closely related to diamonds in Finsch Pipe, South Africa. Contributions to Mineralogy and Petrology (1973) 39(2):103–116.[CrossRef][Web of Science]
Gurney JJ, Harris JW, Rickard RS. Silicate and oxide inclusions in diamonds from the Finsch kimberlite pipe. In: Proceedings of the 2nd Kimberlite Conference. Geophysical Monograph, American Geophysical Union—Boyd FR, Meyer H. OA, eds. (1979) 1:1–15.
Harte B. Rock nomenclature with particular relation to deformation and recrystallisation textures in olivine-bearing xenoliths. Journal of Geology (1977) 85:297–288.
Hart SR, Dunn T. Experimental cpx/melt partitioning of 24 trace elements. Contributions to Mineralogy and Petrology (1993) 113:1–8.[CrossRef][Web of Science]
Harte B, Winterburn PA, Gurney JJ. Metasomatic phenomena in garnet peridotite facies mantle xenoliths from the Matsoku kimberlite pipe, Lesotho. In: Mantle Metasomatism—Menzies MA, Hawkesworth CJ, eds. (1987) London: Academic Press. 145–220.
Harte B, Fitzsimons ICW, Kinny PD. Clinopyroxene–garnet trace element partition coefficients for mantle peridotite and melt assemblages. In:. (1996) 235. V. M. Goldschmidt Conference Proceedings. Journal of Conference Abstracts 1.
Hauri EH, Wagner TP, Grove TL. Experimental and natural partitioning of Th, U, Pb and other trace elements between garnet, clinopyroxene and basaltic melts. Chemical Geology (1994) 117:149–166.[CrossRef][Web of Science]
Hawkesworth CJ, Erlank AJ, Marsh JS, Menzies MA, van Calsteren P. Evolution of the continental lithosphere: evidence from volcanics and xenoliths in Southern Africa. In: Continental Basalts and Mantle Xenoliths—Hawkesworth CJ, Norry MJ, eds. (1983) Nantwich: Shiva. 111–138.
Hervig RL, Smith JV, Steele IM, Dawson JB. Fertile and barren Al–Cr-spinel harzburgites from the upper mantle: Ion and electron probe analyses of trace elements in olivine and orthopyroxene: Relation to lherzolites. Earth and Planetary Science Letters (1980) 50(1):41–58.[CrossRef][Web of Science]
Herzberg CT. Lithosphere peridotites of the Kaapvaal craton. Earth and Planetary Science Letters (1993) 120:13–29.[CrossRef][Web of Science]
Herzberg C. Phase equilibrium constraints on the formation of cratonic mantle. Mantle Petrology: Field Observations and High Pressure Experimentation: a Tribute to Francis R. (Joe) Boyd. Geochemical Society, Special Publications—Fei Y, Bertka CM, Mysen BO, eds. (1999) 6:241–257.
Herzberg C. Geodynamic information in peridotite petrology. Journal of Petrology (2004) 45(12):2507–2530.
Herzberg C, O’Hara MJ. Plume-associated ultramafic magmas of Phanerozoic age. Journal of Petrology (2002) 43(10):1857–1883.
Hoal K. EO, Hoal BG, Erlank AJ, Shimizu N. Metasomatism of the mantle lithosphere recorded by rare earth elements in garnets. Earth and Planetary Science Letters (1994) 126(4):303–313.[CrossRef][Web of Science]
Ionov DA, Ashchepkov IV, Stosch HG, Witt-Eickschen G, Seck HA. Garnet peridotite xenoliths from the Vitim volcanic field, Baikal Region: the nature of the garnet–spinel peridotite transition zone in the continental mantle. Journal of Petrology (1993) 34(6):1141–1175.
James DE, Fouch MJ. Formation and evolution of Archaean cratons: insights from southern Africa. In. The Early Earth: Physical, Chemical and Biological Development—Fowler CMR, Ebinger CJ, Hawkesworth CJ, eds. (2002) 199:1–26. Geological Society, London, Special Publications.
Jochum KP. MPI-DING reference glasses for in situ microanalysis: New reference values for element concentrations and isotope ratios. Geochemistry, Geophysics, Geosystems (2006) 7. paper number Q02008.
Johnson K. TM, Dick H. JB, Shimizu N. Melting in the oceanic upper mantle: an ion microprobe study of diopsides in abyssal peridotites. Journal of Geophysical Research (1990) 95:2661–2678.
Jones RA. Strontium and neodymium isotopic and rare earth evidence for the genesis of megacrysts in kimberlites of southern Africa. In: Mantle Xenoliths—Nixon PH, ed. (1987) New York: John Wiley. 711–724.
Kelemen PB, Hart SR, Bernstein S. Silica enrichment in the continental upper mantle via melt/rock reaction. Earth and Planetary Science Letters (1998) 164(1–2):387–406.[CrossRef][Web of Science]
Kelley SP, Wartho JA. Rapid kimberlite ascent and the significance of Ar–Ar ages in xenolith phlogopites. Science (2000) 289(5479):609–611.
Kennedy C, Kennedy G. The equilibrium boundary between graphite and diamond. Journal of Geophysical Research (1976) 81:2467–2470.
Kesson SE, Ringwood AE. Slab–mantle interactions. 2. The formation of diamonds. Chemical Geology (1989) 78:97–118.[CrossRef][Web of Science]
Kinny PD, Dawson JB. A mantle metasomatic injection event linked to late Cretaceous kimberlite magmatism. Nature (1992) 360(6406):726–728.[CrossRef][Web of Science]
Kopylova MG, Caro G. Mantle xenoliths from the southeastern Slave craton: Evidence for chemical zonation in a thick, cold lithosphere. Journal of Petrology (2004) 45(5):1045–1067.
Li A, Burke K. Upper mantle structure of southern Africa from Rayleigh wave tomography. Journal of Geophysical Research (2006) 111. (B10303), doi:10.1029/2006JB004321.
McDonough W, Sun S.-s. The composition of the Earth. Chemical Geology (1995) 120:223–253.[CrossRef][Web of Science]
McKenzie D, Bickle MJ. The volume and composition of melt generated by extension of the lithosphere. Journal of Petrology (1989) 29:625–679.[Web of Science]
McKenzie D, Jackson J, Priestley K. Thermal structure of oceanic and continental lithosphere. Earth and Planetary Science Letters (2005) 233:337–349.[CrossRef][Web of Science]
Nair SK, Gao SS, Liu KH, Silver PG. Southern African crustal evolution and composition: constraints from receiver function studies. Journal of Geophysical Research (2006) 111. B02304.
Nickel KG. Garnet–pyroxene equilibria in the system SMACCr (SiO2–MgO–Al2O3–CaO–Cr2O3); the Cr geobarometer. In: Kimberlites and Related Rocks. Geological Society of Australia, Special Publications —Ross J, Jaques AL, Ferguson J, et al, eds. (1989) 2:901–912.
Nixon PH, van Calsteren P. WC, Boyd FR, Hawkesworth CJ. Harzburgites with garnets of diamond facies from southern African kimberlites. In: Mantle Xenoliths—Nixon PH, ed. (1987) New York: John Wiley. 523–534.
O’Neill H. SC, Wood BJ. An experimental study of Fe–Mg partitioning between garnet and olivine and its calibration as a geothermometer. Contributions to Mineralogy and Petrology (1979) 70(1):59–70.[CrossRef][Web of Science]
Ono S, Mibe K, Yoshino T. Aqueous fluid connectivity in pyrope aggregates: water transport into the deep mantle by a subducted oceanic crust without any hydrous minerals. Earth and Planetary Science Letters (2002) 203(3–4):895–903.[CrossRef][Web of Science]
O’Reilly S, Griffin WL, Djomani Y, Morgan P. Are lithospheres forever? GSA Today (2001) 11:4–10.
Pearce N. JG, Perkins WT, Westgate JA, Gorton MP, Jackson SE, Neal CR, Chenery SP. A compilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials. Geostandards Newsletter (1997) 21:115–144.[CrossRef][Web of Science]
Pearson DG. The age of continental roots. Lithos (1999) 48(1–4):171–194.[CrossRef][Web of Science]
Pearson DG, Boyd FR, Haggerty SE, Pasteris JD, Field SW, Nixon PH, Pokhilenko NP. The characterisation and origin of graphite in cratonic lithospheric mantle: a petrological carbon isotope and Raman spectroscopic study. Contributions to Mineralogy and Petrology (1994) 115:449–466.[CrossRef][Web of Science]
Pearson DG, Carlson RW, Shirey SB, Boyd FR, Nixon PH. The stabilisation of Archaean lithospheric mantle: a Re–Os isotope study of peridotite xenoliths from the Kaapvaal craton. Earth and Planetary Science Letters (1995a) 134:341–357.[CrossRef][Web of Science]
Pearson DG, Shirey SB, Carlson RW, Boyd FR, Pokhilenko NP, Shimizu N. Re–Os, Sm–Nd, and Rb–Sr isotope evidence for thick Archaean lithospheric mantle beneath the Siberian craton modified by multistage metasomatism. Geochimica et Cosmochimica Acta (1995b) 59:959–977.[Web of Science]
Pearson DG, Irvine GJ, Carlson RW, Kopylova MG, Ionov DA. The development of lithospheric keels beneath the earliest continents: time constraints using PGE and Re-Os isotope systematics. Geological Society, London, Special Publications (2002) 199:65–90.
Pearson DG, Canil D, Shirey SB. Mantle samples included in volcanic rocks: xenoliths and diamonds. In: Treatise on Geochemistry—Holland HD, Turekian KK, eds. (2003) Amsterdam: Elsevier. 171–275.
Pollack HN, Chapman DS. On the regional variation of heat flow, geotherms, and lithospheric thickness. Tectonophysics (1977) 38(3–4):279–296.[CrossRef][Web of Science]
Priestley K, McKenzie D. The thermal structure of the lithosphere from shear wave velocities. Earth and Planetary Science Letters (2006) 244:285–301.[CrossRef][Web of Science]
Richardson SH, Gurney JJ, Erlank AJ, Harris JW. Origin of diamonds in old enriched mantle. Nature (1984) 310(5974):198–202.[CrossRef][Web of Science]
Richardson SH, Shirey SB, Harris JW, Carlson RW. Archean subduction recorded by Re–Os isotopes in eclogitic sulfide inclusions in Kimberley diamonds. Earth and Planetary Science Letters (2001) 191(3–4):257–266.[CrossRef][Web of Science]
Richter FM. A major change in the thermal state of the Earth at the Archean–Proterozoic boundary: consequences for the nature and preservation of continental lithosphere. Journal of Petrology, Special Lithosphere Volume (1988) 39–52.
Rudnick RL, McDonough WF, Orpin A. Northern Tanzania peridotite xenoliths: a comparison with Kaapvaal xenoliths and inferences of metasomatic reactions. Meyer H. OA, Leonardos OH, eds. (1994) Companhia de Pesquisa de Recursos Minerais. 336–353. Proceedings of the 5th International Kimberlite Conference, 2.
Ryan CG, Griffin WL, Pearson NJ. Garnet geotherms: Pressure–temperature data from Cr-pyrope garnet xenocrysts in volcanic rocks. Journal of Geophysical Research (1996) 101(B3):5611–5626.[CrossRef]
Salters VJM, Longhi JE, Bizimis M. Near mantle solidus trace element partitioning at pressures up to 3· 4 GPa. Geochemistry, Geophysics, Geosystems (2002) 3. doi:10.1029/2001GC000148.
Schmidberger SS, Francis D. Nature of the mantle roots beneath the North American craton: mantle xenolith evidence from Somerset Island kimberlites. Lithos (1999) 48(1–4):195.[CrossRef][Web of Science]
Schmitz MD, Bowring SA, de Wit MJ, Gartz V. Subduction and terrane collision stabilize the western Kaapvaal craton tectosphere 2·9 billion years ago. Earth and Planetary Science Letters (2004) 222(2):363–376.[CrossRef][Web of Science]
Schulze D. Low Ca garnet harzburgite xenoliths from southern Africa: abundance composition and bearing on the structure and evolution of the subcratonic lithosphere. In: In: 5th International Kimberlite Conference, Extended Abstracts. CPRM Special Publication (1991) 2:350–352.
Shee SR, Gurney JJ, Robinson DN. Two diamond-bearing peridotite xenoliths from the Finsch kimberlite, South Africa. Contributions to Mineralogy and Petrology (1982) 81(2):79–87.[CrossRef][Web of Science]
Shimizu N, Richardson SH. Trace-element abundance patterns of garnet inclusions in peridotite-suite diamonds. Geochimica et Cosmochimica Acta (1987) 51(3):755–758.[CrossRef][Web of Science]
Shimizu N, Sobolev NV, Yefimova ES. Chemical heterogeneities of inclusion garnets and juvenile character of peridotitic diamonds from Siberia. In: Dobrestov, N. L. (ed.) Proceedings of the 6th International Kimberlite Conference. Russian Geology & Geophysics (1997) 2:356–372.
Simon N. SC, Irvine GJ, Davies GR, Pearson DG, Carlson RW. The origin of garnet and clinopyroxene in ‘depleted’ Kaapvaal peridotites. Lithos (2003) 71(2–4):289–322.[CrossRef][Web of Science]
Simon N. SC, Carlson RW, Pearson DG, Davies GR. The origin and evolution of the Kaapvaal cratonic lithospheric mantle. Journal of Petrology (2007) 48(3):589–625.
Skinner CP. The petrology of xenoliths from the Finsch kimberlite. South African Journal of Geology (1989) 92:197–206.[Abstract]
Smith CB, Gurney JJ, Skinner E. MW, Clement CR, Ebrahim N. Geochemical character of southern African kimberlites: a new approach based on isotopic constraints. Transactions of the Geological Society of South Africa (1985) 88:267–280.
Smith CB, Gurney JJ, Harris JW, Otter ML, Kirkley MB, Jagoutz E. Neodymium and strontium isotope systematics of eclogite and websterite paragenesis inclusions from single diamonds, Finsch and Kimberley Pool, RSA. Geochimica et Cosmochimica Acta (1991) 55(9):2579–2590.[CrossRef][Web of Science]
Smith D, Boyd FR. Compositional zonation in garnets in peridotite xenoliths. Contributions to Mineralogy and Petrology (1992) 112(1):134–147.[CrossRef][Web of Science]
Sobolev NV, Pokhilenko NP, Efimova ES. Xenoliths of diamond-bearing peridotites in kimberlites and the problem of diamond origin. Geologia i Geofizica (1984) 12:63–80.
Stachel T, Harris JW. Diamond precipitation and mantle metasomatism—evidence from the trace element chemistry of silicate inclusions in diamonds from Akwatia, Ghana. Contributions to Mineralogy and Petrology (1997) 129(2):143–154.[CrossRef][Web of Science]
Stachel T, Viljoen KS, Brey G, Harris JW. Metasomatic processes in lherzolitic and harzburgitic domains of diamondiferous lithospheric mantle: REE in garnets from xenoliths and inclusions in diamonds. Earth and Planetary Science Letters (1998) 159(1–2):1–12.[CrossRef][Web of Science]
Stachel T, Viljoen KS, McDade P, Harris JW. Diamondiferous lithospheric roots along the western margin of the Kalahari Craton—the peridotitic inclusion suite in diamonds from Orapa and Jwaneng. Contributions to Mineralogy and Petrology (2004) 147(1):32–47.[CrossRef][Web of Science]
Stephens WE, Dawson JB. Statistical comparison between pyroxenes from kimberlites and their associated xenoliths. Journal of Geology (1977) 85:433–449.[Web of Science]
Tainton KM, McKenzie D. The generation of kimberlites, lamproites, and their source rocks. Journal of Petrology (1994) 35(3):787–817.
Thompson RN, Gibson SA. Extremely magnesian olivines in Phanerozoic picrites signify transient high temperatures during mantle plume impact. Nature (2000) 407:502–506.[CrossRef][Web of Science][Medline]
Tsai HM, Meyer H. OA, Moreau J, Milledge HJ. Mineral inclusions in diamond: Premier, Jagersfontein and Finsch kimberlites, South Africa, and Williamson Mine, Tanzania. In: Proceedings of the 2nd Kimberlite Conference. Geophysical Monograph, American Geophysical Union —Meyer H. OA, Boyd FR, eds. (1979) 1:16–26.
Tuff J, Gibson SA. Trace-element partitioning between garnet, clinopyroxene and Fe-rich picritic melts at 3 to 7 GPa. Contributions to Mineralogy and Petrology (2007) 153(4):369–387.[CrossRef][Web of Science]
Van Orman JA, Grove TL, Shimizu N, Layne GD. Rare earth element diffusion in a natural pyrope single crystal at 2·8 GPa. Contributions to Mineralogy and Petrology (2002) 142:416–424.[Web of Science]
Van Westrenen W, Blundy JD, Wood BJ. Crystal-chemical controls on trace element partitioning between garnet and anhydrous silicate melts. American Mineralogist (1999) 84:838–847.[Abstract]
Van Westrenen W, Blundy JD, Wood BJ. Effect of Fe2+ on garnet–melt trace element partitioning: Experiments in FCMAS and quantification of crystal-chemical controls in natural systems. Lithos (2000) 53:191–203.
Viljoen KS, Swash PM, Otter ML, Schulze DJ, Lawless PJ. Diamondiferous garnet harzburgites from the Finsch Kimberlite, Northern Cape, South Africa. Contributions to Mineralogy and Petrology (1992) 110(1):133–138.[CrossRef][Web of Science]
Viljoen KS, Robinson DN, Swash PM, Griffin WL, Otter ML, Ryan CG, Win TT. Diamond- and graphite-bearing peridotite xenoliths from the Roberts Victor kimberlite, South Africa. In: Kimberlites, Related Rocks and Mantle Xenoliths, Vol. 1. Proceedings of the Fifth International Kimberlite Conference. Companhia de Pesquisa de Recursos Minerais, Special Publication—Meyer H. OA, Leonardos OH, eds. (1994) 1, 285–303.
Walter MJ. Melting of garnet peridotite and the origin of komatiite and depleted lithosphere. Journal of Petrology (1998) 39(1):29–60.[CrossRef][Web of Science]
Wang W, Sueno S, Takahashi E. Influence of Cr on REE partitioning between garnet and silicate melt: application to metasomatism of mineral inclusions in diamonds. Reviews of High Pressure Science and Technology (1998) 7:92–94.
Wang W, Sueno S, Takahashi E, Yurimoto H, Gasparik T. Enrichment processes at the base of the Archean lithospheric mantle: observations from trace element characteristics of pyropic garnet inclusions in diamonds. Contributions to Mineralogy and Petrology (2000) 139(6):720–733.[CrossRef][Web of Science]
Watson EB, Brenan JM, Baker DR. Distribution of fluids in the continental mantle. In: Continental Mantle—Menzies MA, ed. (1990) Oxford: Clarendon Press. 111–125.
Westerlund KJ, Shirey SB, Richardson SH, Carlson RW, Gurney JJ, Harris JW. A subduction wedge origin for Paleoarchean peridotitic diamonds and harzburgites from the Panda kimberlite, Slave craton: evidence from Re–Os isotope systematics. Contributions to Mineralogy and Petrology (2006) 152(3):275–294.[CrossRef][Web of Science]
Wood BJ, Blundy J. A predictive model for rare earth element partitioning between clinopyroxene and anhydrous silicate melt. Contributions to Mineralogy and Petrology (1997) 129:166–181.[CrossRef][Web of Science]
Wood BJ, Blundy JD. The effect of cation charge on crystal–melt partitioning of trace elements. Earth and Planetary Science Letters (2001) 188(1–2):59–71.[CrossRef][Web of Science]
Xue X, Baadsgaard H, Irving AJ, Scarfe CM. Geochemical and isotopic characteristics of lithospheric mantle beneath West Kettle River, British Columbia: evidence from ultramafic xenoliths. Journal of Geophysical Research (1990) 95:15879–15891.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||