Eclogite xenoliths from Orapa: Ocean crust recycling, mantle metasomatism and carbon cycling at the western Zimbabwe craton margin

Available online at www.sciencedirect.com

ScienceDirect Geochimica et Cosmochimica Acta 213 (2017) 574–592 www.elsevier.com/locate/gca

Eclogite xenoliths from Orapa: Ocean crust recycling, mantle metasomatism and carbon cycling at the western Zimbabwe craton margin S. Aulbach a,⇑, D.E. Jacob b, P. Cartigny c,d, R.A. Stern e, S.S. Simonetti f, G. Wo¨rner c, K.S. Viljoen g a Inst. Geosciences, Goethe-Universita¨t, Altenho¨ferallee 1, Frankfurt am Main, Germany Australian Research Council Centre of Excellence for Core to Crust Fluid Systems and Department of Earth and Planetary Sciences, Macquarie University, North Ryde, New South Wales 2109, Australia c Abteilung Geochemie, Geowissenschaftliches Zentrum (GZG), Universita¨t Go¨ttingen, Goldschmidtstr. 1, 37077 Go¨ttingen, Germany d Laboratoire de Ge´ochimie des Isotopes Stables, Institut de Physique du Globe de Paris, Universite´ Paris Diderot, Centre National de la Reserche Scientifique, UMR 7154, Sorbonne Paris-Cite´, 75005 Paris, France e Canadian Centre for Isotopic Microanalysis, Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, AB T6G 2E3, Canada f Department of Civil & Environmental Engineering & Earth Sciences, University of Notre Dame, IN 11 46556, USA g Department of Geology, University of Johannesburg, PO Box 524, Auckland Park, 2006 Johannesburg, South Africa b

Received 9 November 2016; accepted in revised form 21 June 2017; available online 1 July 2017

Abstract Major- and trace-element compositions of garnet and clinopyroxene, as well as 87Sr/86Sr in clinopyroxene and d18O in garnet in eclogite and pyroxenite xenoliths from Orapa, at the western margin of the Zimbabwe craton (central Botswana), were investigated in order to trace their origin and evolution in the mantle lithosphere. Two groups of eclogites are distinguished with respect to 87Sr/86Sr: One with moderate ratios (0.7026–0.7046) and another with 87Sr/86Sr >0.7048 to 0.7091. In the former group, heavy d18O attests to low-temperature alteration on the ocean floor, while 87Sr/86Sr correlates with indices of low-pressure igneous processes (Eu/Eu*, Mg#, Sr/Y). This suggests relatively undisturbed long-term ingrowth of 87Sr at near-igneous Rb/Sr after metamorphism, despite the exposed craton margin setting. The high-87Sr/86Sr group has mainly mantle-like d18O and is suggested to have interacted with a small-volume melt derived from an aged phlogopite-rich metasome. The overlap of diamondiferous and graphite-bearing eclogites and pyroxenites over a pressure interval of 3.2 to 4.9 GPa is interpreted as reflecting a mantle parcel beneath Orapa that has moved out of the diamond stability field, due to a change in geotherm and/or decompression. Diamondiferous eclogites record lower median 87Sr/86Sr (0.7039) than graphite-bearing samples (0.7064) and carbon-free samples (0.7051), suggesting that interaction with the – possibly oxidising – metasomederived melt caused carbon removal in some eclogites, while catalysing the conversion of diamond to graphite in others. This highlights the role of small-volume melts in modulating the lithospheric carbon cycle. Compared to diamondiferous eclogites, eclogitic inclusions in diamonds are restricted to high FeO and low SiO2, CaO and Na2O contents, they record higher

⇑ Corresponding author.

E-mail address: [email protected] (S. Aulbach). http://dx.doi.org/10.1016/j.gca.2017.06.038 0016-7037/Ó 2017 Elsevier Ltd. All rights reserved.

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

575

equilibrium temperatures and garnets have mostly mantle-like O isotopic composition. We suggest that this signature was imparted by a sublithospheric melt with contributions from a clinopyroxene-rich source, possibly related to the ca. 2.0 Ga Bushveld event. Ó 2017 Elsevier Ltd. All rights reserved. Keywords: Kimberlite-borne xenoliths; Archaean oceanic crust; Diamond stability; Mantle metasomatism; Strontium isotopes; Oxygen isotopes; Mantle heterogeneities

1. INTRODUCTION Kimberlite-borne mantle eclogite xenoliths with oceanic crustal protoliths have been the subject of intense investigation (reviews in Jacob, 2004; Aulbach and Jacob, 2016), as their presence in the cratonic lithosphere is thought to support the operation of some form of plate tectonic process: Cratons may have widened and thickened by accretion of oceanic lithosphere, while loss of silicic partial melt, similar to tonalite-trondhjemite-granodiorites (TTGs) typically found in Archaean granite-greenstone belts, from subducting basalt may have contributed to growth of continental crust (Ireland et al., 1994; Rollinson, 1997; Shirey et al., 2001; Foley et al., 2002a; Rapp et al., 2003; Lee, 2006; Shirey and Richardson, 2011; Pearson and Wittig, 2008, 2014). In addition, mantle eclogites host a disproportionate amount of gem-quality diamonds relative to their subordinate abundance in the cratonic mantle (Stachel and Luth, 2015). Finally, eclogite xenoliths potentially represent our only samples of ancient oceanic crust (Helmstaedt and Doig, 1975; MacGregor and Manton, 1986; Jacob and Foley, 1999; Foley, 2011) and as such may hold clues to the thermal, redox and chemical state of the ambient convecting mantle, from which their protoliths are ultimately derived (Aulbach and Viljoen, 2015). Amongst mantle eclogite localities, the Orapa kimberlite cluster is noteworthy for several reasons: (1) It is one of the few kimberlites yielding predominantly eclogitic xenoliths, other examples being Jwaneng (also in Botswana) as well as Premier, Roberts Victor and Kaalvallei in the Kaapvaal craton (Viljoen et al., 1996, 2005; Shirey et al., 2002) and Koidu in the West African craton (Hills and Haggerty, 1989). (2) The kimberlites also yielded predominantly eclogitic inclusion-bearing diamonds (85%; Gurney et al., 1984) and eclogitic minerals occurring along with polycrystalline diamond (framesite; Jacob et al., 2011), all of which have been extensively studied, as detailed in Appendix A. (3) Orapa is located in a Palaeoproterozoic fold belt at the western margin of the craton, which has been strongly affected by Palaeo- to Midproterozoic accretionary processes (Jacobs et al., 2008). Geochronology of inclusions in diamond has revealed the presence of multiple generations of eclogitic diamond (Richardson et al., 1990; Shirey et al., 2008; Timmerman et al., 2017). The western margin of the Zimbabwe craton was exposed to a multiplicity of processes ranging from crustal accretions to rifting in the Proterozoic (Jacobs et al., 2008), and at least four compositionally distinct groups of eclogite xenoliths have been recognised (Viljoen et al., 1996). Thus, the question arises whether the different

diamond generations and eclogite groups recognised at Orapa relate to tectonic events affecting the western Zimbabwe craton. The exact nature of the relationship between the origin(s) and evolution of mantle eclogites and diamond genesis at Orapa has not been addressed in detail since early work (Viljoen et al., 1996). In the meantime, more data-sets have been produced in studies focussing on specific aspects of eclogitic diamond formation. This study combines new data (garnet and clinopyroxene major, minor and trace elements, garnet and clinopyroxene d18O and clinopyroxene 87Sr/86Sr) with published data with the aims to (1) decipher the origin(s) and chemical evolution of the mantle eclogite reservoir, (2) investigate the overprint of mantle eclogites from Orapa in the context of its proximity to the craton margin and (3) establish a link to multiple diamond formation events recognised at this locality. 2. CRUST AND MANTLE GEOLOGY The Orapa kimberlite cluster in central Botswana (Fig. 1) was emplaced into the Palaeoproterozoic Magondi

Fig. 1. Map of southern Africa (after Pearson et al., 1998; Shirey et al., 2002) showing the outline of the Kaapvaal and Zimbabwe cratons, separated by the Limpopo Belt, and approximate locations of kimberlites that have yielded abundant eclogite materials (O Orapa, B Bellsbank, JW Jwaneng, K Kimberley, L Lace, RV Roberts Victor, P Premier). The highlighted area south and west of the Limpopo belt encompasses lithosphere affected by the ca. 2 Ga Bushveld large igneous event. Proterozoic mobile belts (KheisOkwa-Magondi, Namaqua-Natal) and political boundaries (fine stipples) are also shown. Inset indicates known kimberlites and diamond mines forming part of the Orapa kimberlite field, including the main Orapa pipe, Damtshaa and Letlhakane, all of which have yielded xenoliths and/or inclusion-bearing diamonds that have been studied.

576

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

Fold Belt, which forms the northwestern margin of the Zimbabwe craton (Stiefenhofer et al., 1997). The Zimbabwe craton comprises tectonic units as old as 3.5 Ga and its collision with the Kaapvaal craton at ca. 2.6 Ga led to formation of the intervening Limpopo province and to stabilisation of the Kalahari craton (de Wit et al., 1992; Kusky, 1998). The subsequent evolution of the northwestern margin of the Zimbabwe craton comprises the accretion of Palaeoproterozoic rocks of the Magondi Fold Belt at ca. 2000 to 1850 Ma, followed by a passive margin stage until Mesoproterozoic accretion of the Choma-Kalomo Block and intense rifting and associated volcanism around ca. 1100 Ma (Jacobs et al., 2008). The Late Cretaceous emplacement age for the host kimberlites (ca. 90 Ma; Davis, 1977; Allsopp et al., 1989), combined with its situation at the margin of the Zimbabwe craton (Fig. 1), suggests that the underlying lithospheric mantle section has been extensively metasomatised, as is the case for the mantle sampled by similarly aged Group I kimberlites in the Kaapvaal craton (mostly 95 Ma) compared to an older generation of kimberlites (Group II, >110 Ma; Griffin et al., 2003; Becker and Le Roex, 2006; Kobussen et al., 2008). The Damtshaa kimberlites, which belong to the Orapa cluster and intruded some 20 km east of the Orapa mine, and the Letlhakane kimberlite, which was emplaced some 30 km southeast of Orapa, may have sampled a similar lithosphere column (Stiefenhofer et al., 1997; Stachel et al., 2004; Ickert and Stern, 2013) and will be considered together with data from the main Orapa kimberlite. Prior work has shown that, although Orapa is positioned in the Palaeoproterozoic Magondi Belt, the underlying lithospheric mantle may form part of the western subsurface extension of the Zimbabwe craton. This interpretation is based on the chemical composition of peridotite xenoliths from Letlhakane and on thermobarometry, which reveal that the lithosphere has a thickness of least 150 km, comprises samples with both fertile and highly depleted signatures (as gauged by olivine Mg# ranging from 0.898 to 0.936) and equilibrated to a geothermal gradient corresponding to a surface heat flow of 40 mW/m2 at the time of kimberlite entrainment (Stiefenhofer et al., 1997). This gradient is typical for the thermal state of mantle beneath the Kaapvaal craton affected by Group I kimberlite magmatism (e.g. Koffiefontein, Jagersfontein and Kimberley; Gru¨tter, 2009). Alternatively, it has been suggested that Orapa sits at the northward extension of the Kaapvaal craton (Shirey et al., 2008). Either way, Mesoarchaean Re-depletion model ages for peridotites from Letlhakane (Carlson et al., 1999) provide a minimum age for the lithospheric mantle beneath the region, including Orapa. Later study of a large number of peridotitic xenocrysts from the area yielded a low proportion of depleted peridotites and high proportion of fertile lherzolites at shallow depths (<120 km) compared to typically cratonic mantle sections (Griffin et al., 2003). These results are consistent with the detection of seismically slow and presumably refertilised mantle lithosphere underlying the area, a feature that has been linked to the high proportion of eclogite xenoliths and inclusions in diamond (DI) in the kimberlites

(Shirey et al., 2002). Additional information on the state of the lithospheric mantle, at least until peridotitic diamond formation, comes from inclusion studies, which confirm that the lithosphere is less depleted and has a low harzburgite-to-lherzolite ratio compared to typically cratonic regions, such as the Kimberley area (Stachel et al., 2004). Moreover, non-touching inclusions (reflecting the thermal state at the time of entrapment) fall on a 42 mW/m2 geothermal gradient, whereas touching inclusions (hence in equilibrium at the time of kimberlite eruption) fall on a geothermal gradient of 37 mW/m2, with the difference interpreted to reflect secular cooling of the mantle lithosphere (Stachel et al., 2004). Numerous studies have been carried out on eclogitic materials from Orapa and their results are summarised in Appendix A and addressed again in the discussion. 3. SAMPLES AND ANALYTICAL TECHNIQUES The xenolith samples in this study (n = 289) were collected from coarse concentrate, with sizes typically limited to between 1 and 2 cm, owing to the nature of diamond processing at the mine, and various data were published (Appendix A). We here report new major-±trace-element data for minerals in 220 of these samples (Appendices C and D), in addition to in situ Sr isotopes for clinopyroxene (n = 55; Appendix E) and O isotopes for garnet (n = 66) and cpx (n = 29) (Appendix F). The xenoliths were embedded in resin and the blocks were polished to expose typically several to 10 s of grains each of the main constituents (garnet and omphacite), depending on grain size, which ranges from coarse to medium. Mineralogically, about one half of eclogites comprise a bimineralic variety without accessory minerals (n = 122), the remaining samples containing diamond (n = 99), graphite (32) or diamond with graphite (n = 4), plus varying occurrences of rutile, ilmenite, spinel, corundum, kyanite and amphibole (Appendix B). Orthopyroxene is absent, which contrasts with its presence in the diamond inclusion suite (Gurney et al., 1984). Textural types range from eclogites with unequilibrated textures, with roundish garnet grains set in a matrix of anhedral clinopyroxene (Group I in the classification of MacGregor and Carter, 1970), to eclogites with equilibrated microstructures indicated by straight, interlocking grain boundaries and 120° junctions (Type II) or transitional types between the two, with both curvilinear and straight grain boundaries present (Fig. 2). Mineral element concentrations of most samples were collected by electron microprobe analyser (EPMA) and laser ablation inductively-coupled plasma mass spectrometry (LAM ICPMS) at the De Beers GeoScience Centre in Johannesburg as described in Viljoen (1995) and Viljoen et al. (2009). Briefly, major and minor oxides were analysed with the Cameca SX-50 microprobe, employing a 3 lm beam size, an accelerating voltage of 20 kV, a beam current of 20 nA and 30 s counting time for Na2O and K2O and 20 s for the remaining oxides, with detection limits of 0.02 wt% for TiO2, Cr2O3, Na2O and K2O and 0.04 wt% for MnO. A PAP correction procedure was employed. Trace element concentrations were determined by a

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

577

Fig. 2. Illustration of various textures represented among the Orapa eclogite and pyroxenite xenolith suite. OE83: Medium-grained (1 mm < grain size < 5 mm) high-Ca eclogite with interlocking grain boundaries; OE52: Coarse-grained gabbroic eclogite with straight to curvilinear grain boundaries and teal-green clinopyroxene; OE72: Coarse-grained, cumulate-textured and foliated pyroxenite with apple-green clinopyroxene.

Merchantek New Wave Research Nd:YAG UV laser system (213 nm wavelength) coupled to a Perkin Elmer Sciex ELAN 6000 ICPMS, with ablation carried out at a 10 Hz repetition rate, 0.5 mJ laser pulse energy and a beam size resulting in 105 lm pits. Each analysis consisted of 80 s data collection on the ablated material plus 30 s on the background. The NIST610 glass was used as a calibration standard and Ca (determined by EPMA) as internal standard. Data quality was checked using Mongolian garnet MU53388 (Norman et al., 1996) and with the exception of Y and Hf, results are accurate within 10% (Appendix C). Strontium isotopes were measured in situ in clinopyroxene from a subset of the diamondiferous eclogites, following the method outlined in Schmidberger et al. (2003, 2007), employing a Nd:YAG UP213 nm laser system (New Wave Research) coupled to the NuPlasma multicollector-ICPMS at the Radiogenic Isotope Facility in the University of Alberta. Ablation was carried out in a He atmosphere at a 20 Hz repetition rate with a 160 lm spot size and 15 J/cm2 energy density with a 60 s ablation time; data were collected in static mode using five Faraday cups. Corrections for the isobaric interference of 87Rb on 87 Sr were <1% for most of the samples (n = 36); correction for additional seven samples ranged from 1.2 to 5.2%. Analyses for four samples with high isobaric interferences (9.5%) were discarded; the remaining samples show no correlation between 85Rb beam intensity and either total Sr beam intensity or 87Sr/86Sr. The Sr isotopic composition of an in-house standard of modern-day coral (Bizzarro et al., 2003) was analysed repeatedly throughout the course of this work in order to check accuracy. Repeated measurements (n = 40) yielded an average 84Sr/86Sr of 0.0563 ± 0.0004 (2r), 84Sr/88Sr ratios of 0.00672 ± 0.00005 and 87 Sr/86Sr value of 0.70910 ± 0.00005, respectively, identical to the average 87Sr/86Sr of 0.709098 ± 0.000019 determined by thermal ionisation mass spectrometry (2r; Bizzarro et al., 2003). Median within-run uncertainty for 87Sr/86Sr in cpx in this study is 0.00015 (2r), while two sigma standard deviations for analyses of multiple grains per sample are 0.00022, which suggests that the samples are largely homogeneous with respect to 87Sr/86Sr (Appendix E). Heterogeneity is revealed at the sample scale for a few samples where standard deviations for 87Sr/86Sr are significantly higher, ranging up to 0.0048 (2r) for sample JJG893.

Oxygen isotopes (Appendix F1) were acquired at University of Go¨ttingen for garnet and cpx in selected samples, using laser-ablation fluorination following techniques outlined in Wiechert et al. (2002). Briefly, O2 was liberated by UV-laser ablation in a F2 atmosphere with spot sizes of 300 µm, followed by data acquisition using continuous flow mass spectrometry. Oxygen isotopic ratios are expressed in permil deviation from the VSMOW standard (d18O). No correction was applied to any data and d18O values of San Carlos olivine remained within the a 1r precision of 0.15‰ of 5.25‰, a precision similar to that given by Wiechert et al. (2002). Garnet and clinopyroxene in additional samples were analysed in situ for trace elements following the methods detailed in Jacob (2006) and Dongre et al. (2015), employing an esi/NWR193 nm wavelength excimer laser coupled to an Agilent 7500ce quadrupole ICPMS at Johannes Gutenberg University, Mainz (Germany). Ablation was carried out in a He atmosphere, with the laser operating at a 10 Hz repetition rate, a 6.6 J/cm2 energy density and a 100 lm spot size. Quantification was achieved using NIST SRM 612 as a calibration standard and Ca as the internal standard, while BCR2-G was measured as an unknown in order to monitor performance. With the exception of Li, results agree with recommended values within 16% (Appendix C). Garnet in these samples was prepared and analysed at the University of Alberta for major and minor oxides as well as oxygen isotope composition, closely following the methodology described in Ickert and Stern (2013) (Appendix E2). Garnet crystals were mounted along with reference materials UAG (S0069; d18OVSMOW = +5.72) and S0088B (d18OVSMOW = +4.13) within 5 mm of the centre of a 25 mm epoxy plug. Oxygen isotope analyses were carried out using a Cameca IMS 1280 multicollector ion microprobe. Ionisation was achieved with a 133Cs+ primary beam at an impact energy of 20 keV, a beam current of 2.5 nA and a 12 lm probe operating in rastering mode (20  20 µm for 30 s during pre-analysis primary beam implantation, 5  5 µm for 75–90 s during peak counting). Negative secondary ions were collected in Faraday cups in static collection mode. Instrumental mass fractionation was monitored through repeat analyses of the two standards and first corrected to garnet standard S0068, followed by a correction taking into account matrix effects as a function of garnet

578

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

Ca# (molar Ca/(Ca + Mg + Fe)). The uncertainty for d18OVSMOW in garnet at the 95% confidence level, reflecting counting statistics, geometric effects and corrections, is estimated to be ±0.28‰.

4. ASSIGNMENT TO GROUPS, RESULTS AND THERMOBAROMETRY 4.1. Assignment to groups Eclogite xenoliths from Orapa have been shown to be largely consistent with the classification of McCandless and Gurney (1989) who suggested that Group I eclogites tend to be diamondiferous, containing garnet with Na2O  0.09 wt% and clinopyroxene with K2O  0.08 wt %, in contrast to samples with lower concentrations belonging to diamond-free Group II eclogites (Viljoen et al., 1996) (Appendix B). In this work, samples are primarily distinguished according to whether they are carbon-bearing (diamondiferous, graphite- or diamond-graphite-bearing) or carbon-free, with additional reference to their geochemical characteristics. The chemically-based assignment to groups follows that outlined in Aulbach and Jacob (2016). Figures illustrating their elemental relationships are shown in the Appendices. Briefly, given that clinopyroxene in true eclogites should have Na/(Na + Ca)  0.2, samples with clinopyroxene having lower ratios are considered pyroxenites. Samples are further described as basaltic or picritic if they have Eu* (defined here as NMORB-normalised Eu/(0.5 * Sm + 0.5 * Gd); NMORB of Sun and McDonough, 1989)  1.05, otherwise they are described as gabbroic, here implying a cumulate origin. Gabbroic samples are found in particular amongst high-Ca eclogites, but occur in all compositional groups. Thus, samples in the present study for which clinopyroxene composition was not determined may be pyroxenites, especially if garnet is MgO-rich, and

samples without REE information may have a cumulate origin (Eu* > 1.05), especially if garnet is CaO-rich. 4.2. Mineral major and trace elements Mineral major element compositions are illustrated in Fig. 3. Garnet compositions vary widely, with Mg# (100 Mg/(Mg + Fetotal) molar) ranging from 0.43 to 0.84, Ca# (100Ca/(Mg + Fetotal + Mn + Ca) molar) from 0.09 to 0.54 and Cr# (100Cr/(Cr + Al) molar) from <0.001 to 0.16. Pyroxenites have the highest median Mg# (0.77) and relatively high Cr2O3 contents (median 0.29 wt%), while gabbroic eclogites have the highest Ca# (0.24). In terms of their REE patterns (Appendix C2), garnet in gabbroic eclogites has chondrite-normalised abundances (denoted by subscript N) of LREEN < 1, HREEN > 1 and marked positive Eu anomalies. The majority of samples has flat MREEN to HREEN, which vary by more than an order of magnitude. With the exception of two samples, garnet has positive slopes in the LREE. In contrast, garnet in some basaltic/picritic eclogites shows positive slopes in the REEN throughout or has peaked REE patterns (cf. Smart et al., 2014), with negative slopes in the HREE (also observed for garnet in some gabbroic eclogites; Appendix C2). Small negative Eu anomalies (down to 0.90) are occasionally present. Median jadeite contents in clinopyroxene range from 0.05 to 0.57, the highest values being observed in gabbroic eclogites (0.34)) and the lowest values in pyroxenites (0.11) (Appendix D). As is true for garnet, clinopyroxene in the latter also have higher median Cr2O3 contents (0.25 wt%) and Mg# (0.92). REEN abundances for clinopyroxene in gabbroic eclogites vary widely (more than one order of magnitude, Appendix D2), with generally parallel patterns, and humps between Ce and Pr (that is, La/CeN ± La/PrN < 1), deceasing MREEN and HREEN to Yb or Lu, and positive Eu anomalies. Several samples cross-cut these patterns (13 of 85), having La/CeN or La/PrN > 1.

Fig. 3. Major-element variations in garnet and clinopyroxene. A. Assignment to pyroxenites and to high-Ca, high-Mg and low-Mg eclogites based on Aulbach and Jacob (2016). B. K2O in clinopyroxene as a function of Na2O in garnet (wt%) and C. Jadeite content (Jd) as a function of Mg# (molar Mg/(Mg + Fe)) in clinopyroxene. In A. eclogite and pyroxenite xenoliths (Robinson et al., 1984; O’Reilly and Griffin, 1995; Viljoen et al., 1996; this study) are compared to minerals included in diamond (ecl eclogitic, pyx pyroxenitic-websteritic: Gurney et al., 1984, 1986; Deines and Harris, 2004; Ickert et al., 2013). Stippled lines in B. separate Group I eclogites (Na2O in garnet 0.09 wt%, K2O in clinopyroxene 0.08 wt%; McCandless and Gurney, 1989) from Group II eclogites.

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

Clinopyroxene in two gabbroic eclogites has conspicuously higher HREE contents. Clinopyroxene in basaltic/picritic eclogites generally follows these patterns, but several samples are LREE-richer and HREE-poorer, cross-cutting the dominant REE pattern. Three samples have HREEN > 1 and relatively flat, little fractionated REE patterns. 4.3. Oxygen isotopes in garnet and Sr isotopes in clinopyroxene New d18O of eclogitic and pyroxenitic garnet range from 4.83 to 8.36‰, while those in cpx range from 5.05 to 8.18‰ (Appendix F). Together with published data, garnet in diamondiferous eclogites spans the entire range of values, with 17 of 35 samples above the canonical mantle range of 5.1– 5.9‰, whereas carbon-free eclogites and pyroxenites plot dominantly in the mantle field (Fig. 4). 87 Sr/86Sr in clinopyroxene varies from 0.7026 to 0.7091 (Appendix D). This range of values is similar to that determined on clinopyroxene separates from Orapa eclogites by thermal ionisation mass spectrometry (0.7023 –0.7083; Viljoen et al., 1996). Clinopyroxene in diamondiferous samples records lower median ratios (0.7039; n = 54) than that in graphite-bearing samples (0.7064; n = 3) or in carbonfree samples (0.7051; n = 11) (Table 1, Fig. 5). 4.4. Geothermobarometry Temperatures were calculated using the garnetclinopyroxene Mg-Fe exchange thermometer of KroghRavna (2000; TK00) at a pre-set pressure of 4.5 GPa, which roughly corresponds to the pressure where the diamondgraphite boundary intersects the peridotite-derived conductive geotherm of 40 mW/m2 (Letlhakane; Stiefenhofer et al., 1997). Temperatures show modes at 1050 °C and 1350 °C (Fig. 6). Iterative solution of TK00 with a

579

geothermal gradient of 40 mW/m2 yields results consistent with occurrence of diamond for many of these samples, but, if some samples did equilibrate to a locally higher geotherm, their pressures will be overestimated. Higher median pressures are recorded for diamondiferous eclogites (5.8 GPa) than for graphite eclogites and carbon-free eclogites (4.0 and 4.3 GPa, respectively) (Table 1). Eclogites occur almost throughout the entire pressure interval, with carbon-free eclogites and pyroxenites dominating at low pressures and temperatures (median 4.3 GPa and 1040 °C vs. 5.8 GPa and 1180 °C for diamondferous eclogites). Since a depth to the lithosphere-asthenosphere boundary of 220 km has been determined for Orapa (Miensopust et al., 2011), which corresponds to a pressure of 7.1 GPa, values in excess of this could be due to disequilibrium, but may also reflect lack of equilibration to the conductive geotherm, for example due to late heating. Such deviation from the geotherm is also evident in some peridotite xenoliths investigated by Stiefenhofer et al. (1997). 5. RECONSTRUCTED BULK ROCKS AND TRACEELEMENT DISTRIBUTION Whole rocks, reconstructed from garnet and clinopyroxene compositions assuming equal modal abundance (details in Appendix A), show a large range of MgO from 7.1 to 21.3 wt%. This corresponds to basaltic to picritic compositions in eclogites, with the highest values observed for pyroxenites (Appendix G). Concentrations of other oxides are also highly variable – more variable than in measured bulk rocks (McDonald and Viljoen, 2006) – with nearly the entire range covered by gabbroic eclogites. Some lowMg eclogites extend to higher FeO and lower CaO values at a given MgO content, and pyroxenites form a marked anti-correlation of FeO with MgO (Appendix G2). This variability far exceeds the expected uncertainty due to bulk

Fig. 4. A. Histogram of oxygen isotopic composition (d18O, in ‰ relative to VSMOW) determined for garnet in eclogite and pyroxenite xenoliths. Black bars show results for inclusions in diamond, with numbers per bin given. B. d18O as a function of Ca# (Ca/(Ca + Mn + Mg + Fe) molar) in garnet. Horizontal grey bar encompasses mantle range (5.1–5.9‰; Mattey et al., 1994), with heavier and lighter O ascribed to low- and high-temperature seawater alteration, respectively (Gregory and Taylor, 1981; Jacob et al., 1994; Jacob and Foley, 1999). Vertical bars show range for gabbros, serpentinites and metasediments (Philippot et al., 2007) and for zircon (zr) in mica-amphibole-rutile-ilmenitediopside (MARID) xenoliths (Giuliani et al., 2015). Data sources: this study, Deines et al. (1991), Viljoen et al. (1996) and Ickert et al. (2013).

580

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

Table 1 Median compositions (wt%) for eclogites and pyroxenites grouped according to carbon mineralogy. Garnet

SiO2

TiO2

Al2O3

Cr2O3

FeO

MnO

MgO

CaO

Na2O

Mg#

Cr#

Ca#

d18O

Diamondiferous 1r Count Dia+graph-bearing 1r Count Graphite-bearing 1r Count Carbon-free 1r Count Eclogitic DI 1r Count Websteritic DI 1r Count Framesite 1r Count

40.7 0.6 102 40.3 0.9 4 40.4 0.9 32 40.8 0.8 151 39.8 0.5 66 40.4 0.6 9 40.8 0.7 32

0.4 0.2 102 0.3 0.2 4 0.2 0.1 32 0.3 0.2 151 0.5 0.1 66 0.7 0.3 9 0.3 0.1 32

22.6 0.5 102 22.8 0.6 4 22.9 0.7 32 22.5 0.7 151 22.4 0.4 66 20.2 1.3 9 23.0 0.4 32

0.08 0.06 102 0.11 0.09 4 0.05 0.05 32 0.08 0.69 151 0.07 0.09 66 3.0 1.4 9 0.15 0.29 32

12.3 3.8 102 11.2 3.7 4 12.9 3.9 32 11.9 3.6 151 15.7 2.0 66 13.5 3.4 9 13.5 3.9 32

0.3 0.1 102 0.3 0.1 4 0.3 0.1 32 0.3 0.1 151 0.3 0.1 66 0.5 0.1 9 0.4 0.0 32

13.8 2.4 102 11.7 4.5 4 10.6 3.4 32 14.5 3.5 151 10.9 2.2 66 18.2 2.0 9 16.5 2.7 32

9.1 3.5 102 10.7 4.4 4 10.9 4.3 32 7.7 3.6 151 9.6 2.7 66 4.0 1.0 9 4.6 0.7 32

0.13 0.05 102 0.16 0.04 4 0.10 0.04 32 0.09 0.06 151 0.18 0.05 62 0.04 0.05 2 0.11 0.03 31

0.7 0.1 102 0.7 0.1 4 0.6 0.1 32 0.7 0.1 151 0.6 0.1 66 0.7 0.1 9 0.7 0.1 32

0.0 0.0 102 0.0 0.0 4 0.0 0.0 32 0.0 0.0 151 0.0 0.0 66 0.1 0.0 9 0.0 0.0 32

0.2 0.1 102 0.3 0.1 4 0.3 0.1 32 0.2 0.1 151 0.3 0.1 66 0.1 0.0 9 0.1 0.0 32

6.15 1.15 20 6.20 0.00 2 6.20 0.00 2 5.52 0.48 15 5.66 1.21 15 na

Clinopyroxene

SiO2

TiO2

Al2O3

Cr2O3

FeO

MnO

MgO

CaO

Na2O

K2O

Mg#

Cr#

Jd

87

Diamondiferous 1r Count Dia+graph-bearing 1r Count Graphite-bearing 1r Count Carbon-free 1r Count Eclogitic DI 1r Count Websteritic DI 1r Count Framesite 1r Count

55.0 0.6 83 54.8 0.4 4 55.2 0.5 24 54.5 0.7 151 54.8 6.8 41 54.7 0.9 2 54.5 0.4 8

0.4 0.2 83 0.4 0.1 4 0.4 0.1 24 0.4 0.1 151 0.5 0.2 40 0.2 0.0 2 0.5 0.2 8

9.7 2.2 83 9.8 2.7 4 10.2 3.9 24 8.1 3.8 151 9.2 3.9 41 1.4 0.5 2 3.7 20.3 8

0.07 0.04 82 0.08 0.15 4 0.07 0.06 24 0.07 0.43 151 0.07 0.06 40 0.56 0.52 2 0.75 0.77 8

3.9 1.4 83 3.5 0.9 4 3.0 1.5 24 3.6 1.2 151 5.0 1.5 40 5.1 1.7 2 3.2 1.9 8

0.0 0.0 82 0.0 0.0 4 0.0 0.0 22 0.0 0.0 150 0.1 0.0 40 0.2 0.1 2 0.1 0.1 8

10.6 1.5 83 10.9 2.3 4 9.5 2.6 24 11.9 2.6 151 10.4 3.3 40 18.5 1.0 2 14.4 1.5 8

14.8 2.0 83 15.1 1.8 4 14.6 2.6 24 16.0 2.7 151 13.1 3.3 41 18.3 1.1 2 18.9 1.4 8

5.3 1.0 83 5.1 1.2 4 5.6 1.7 24 4.5 1.7 151 4.8 1.7 40 0.9 0.4 2 2.8 0.3 8

0.06 0.04 83 0.13 0.05 3 0.03 0.05 21 0.02 0.06 151 0.18 0.25 40 0.17 0.13 2 0.06 0.06 8

0.8 0.1 83 0.8 0.1 4 0.9 0.1 24 0.9 0.0 151 0.8 0.0 40 0.9 0.0 2 0.9 0.1 8

0.0 0.0 83 0.0 0.0 4 0.0 0.0 24 0.0 0.1 151 0.0 0.0 40 0.2 0.1 2 0.1 0.1 8

0.3 0.1 83 0.3 0.1 4 0.4 0.1 24 0.3 0.1 151 0.3 0.1 41 0.0 0.0 2 0.1 0.0 8

0.7039 0.0014 54 0.7081 na 1 0.7064 0.0009 3 0.7051 0.0019 11

na

Sr/86Sr

Whole rock

SiO2

TiO2

Al2O3

Cr2O3

FeO

MnO

MgO

CaO

Na2O

[email protected] GPaa

TK00@40 mW*m 2

P40mW*m 2@TK00a

Diamondiferous 1r Count Dia+graphbearing 1r Count Graphitebearing 1r Count Carbon-free 1r Count

47.8 0.5 79 47.4

0.4 0.2 79 0.3

16.2 1.2 79 16.3

0.08 0.04 77 0.09

8.5 2.6 79 7.4

0.2 0.1 76 0.1

12.1 1.7 79 11.3

12.4 2.3 79 12.9

2.6 0.5 79 2.6

1193 120 66 1133

1292 228 66 1180

5.8 1.3 66 5.1

0.3 4 47.8

0.1 4 0.3

0.7 4 16.4

0.12 4 0.07

2.2 4 8.4

0.1 3 0.2

3.5 4 10.3

1.7 4 12.1

0.6 4 2.9

170 4 1002

287 4 968

1.6 4 4.0

0.6 24 47.4 0.6 144

0.1 24 0.3 0.2 144

1.8 24 15.7 2.0 144

0.05 22 0.08 0.59 137

2.8 24 7.9 2.4 144

0.0 15 0.2 0.1 124

3.1 24 13.5 3.4 144

1.5 24 11.6 1.5 144

0.7 20 2.2 0.9 125

72 18 1049 164 144

114 18 1037 282 144

0.6 18 4.3 1.6 144

Line missing

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592 Eclogitic DI 1r Count

47.0 0.7 18

0.0 0.2 68

16.0 2.1 18

0.08 0.51 18

11.0 1.8 18

0.2 0.1 18

12.1 3.4 18

10.6 1.3 18

2.3 0.6 12

1175 103 16

1253 210 16

581 5.6 1.2 16

All Fe as ferrous iron; Mg# is molar Mg/(Mg + Fe); Cr# is molar Cr/(Cr + Al); Ca# is molar Ca/(Mg + Fe + Mn + Ca); Jadeite component (jd) is molar Na-Cr-2 * Ti. Dia diamond, graph graphite, DI inclusion in diamond. a Temperatures (Krogh Ravna, 2000: TK00) calculated at a pre-set pressure of 4.5 GPa, and solved iteratively with the parameterisation of a conductive geotherm corresponding to a surface heat flow of 40 mW/m2 (Hasterok and Chapman, 2011).

Fig. 5. Strontium isotope composition (87Sr/86Sr) in clinopyroxene as a function of A. Eu/Eu* (chondrite-normalised Eu/(Sm * Gd)0.5; Rudnick and Fountain, 1994) in cpx, B. Sr/Y in cpx and C. Ca# in garnet. Inset in B. shows Sr/Y-Rb/Sr relationships for complete fossil oceanic crustal sections from Oman (Benoit et al., 1996; Godard et al., 2003) and Gabal Gerf (Zimmer et al., 1995). Samples with 87 Sr/86Sr < 0.7048 (below dashed line) show co-variations consistent with increasing olivine + plagioclase accumulation in the protoliths, which would be accompanied by decreasing Rb/Sr, and may reflect ingrowth of radiogenic Sr due to in situ decay of 87Rb. D. 87Sr/86Sr in clinopyroxene as a function of d18O. Arrows with tick marks (numbers denote % melt added) are for bulk addition of sediment (Viljoen et al., 1996) and a phlogopite-clinopyroxenite metasome-derived melt (distribution coefficients from Pilet et al., 2011). The latter was calculated for bulk mixing of (1) a median gabbro composition with a 87Sr/86Sr of 0.7034, Sr concentration of 63 ppm and d18O of 5.3‰ (white diamond), and (2) a 1% batch melt derived from a phlogopite clinopyroxenite (similar to sample FW12 reported in Gre´goire et al., 2002) with 87Sr/86Sr of 0.726 (see text for details), Sr concentration of 1050 ppm and mantle like d18O. Tick marks are for melt addition in 0.01% increments. Modelling shows that at such small melt additions, Rb concentrations would increase from 0.01 ppm (assumed) to 0.04 ppm for addition of 0.02% melt, which is negligible. Vertical grey bar shows canonical mantle range of Mattey et al. (1994). Data sources as in Fig. 3.

582

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

Fig. 6. A. Histogram of temperatures (Krogh Ravna, 2000: TK00) calculated at a pre-set pressure of 4.5 GPa for clinopyroxene-garnet pairs in eclogite and pyroxenite xenoliths and included in diamond (DI) from Orapa. B. Histogram of pressures obtained by iterative solution of TK00 with the peridotite-derived geotherm corresponding to 40 mW/m2 surface heatflow (Stiefenhofer et al., 1997). Thick dashed line in A. shows the approximate graphite-diamond boundary (Day, 2012) assuming this conductive geotherm. Thin dashed lines in B. show the pressure below which the peridotitic mantle lithosphere becomes noticeably more fertile and above which it is dominated by depleted-metasomatised rocks (corresponding to a depth of 120 km), the pressure interpreted as the chemical lithosphere-asthenosphere boundary (depth of 190 km), both from Griffin et al. (2003), and the pressure corresponding to the base of the electrically resistive lithosphere (depth of 220 km, Miensopust et al., 2011).

rock reconstruction that would derive from varying garnet and clinopyroxene modal abundances, but is similar to that observed in natural gabbros, albeit with higher MgO (median 12.3 wt% vs. 9.6 wt% for modern gabbros from PetDB: www.earthchem.org/petdb). This may reflect hotter formation conditions typical for Archaean protoliths (Aulbach and Jacob, 2016). Carbon-free eclogites show a similar range of compositions to that in diamondiferous and graphite-bearing samples, but dominate the high-MgO end. The REE patterns of reconstructed gabbroic eclogites show the positive Eu anomalies by which they are classified and most have relatively flat and parallel MREEN to HREEN varying by an order of magnitude, with depletions in LaN to PrN (Appendix G3). Similar flat to mildly enriched HREE patterns, with prominent positive Eu anomalies (Eu/Eu* up to 6), have been determined for measured bulk rocks from Orapa, which have been preserved despite the effects of kimberlite infiltration, which led to strong enrichments in the more incompatible REE and decreasing Eu/Eu* (McDonald and Viljoen, 2006). The patterns for the basaltic/picritic eclogites vary widely, ranging from those similar to REE-rich gabbroic eclogites (minus the Eu-anomaly) to those having positive or negative slopes in the REE. Apparent clinopyroxene-garnet trace-element distribution coefficients (clinopyroxene/garnetDElement; n = 75) are generally consistent with experimentally determined partitioning (Green et al., 2000; Klemme et al., 2002; Barth et al., 2002a; Adam and Green, 2006), with V, Ga, Sr and the LREE partitioned into clinopyroxene, whereas Sc, Y, Zr and the HREE prefer garnet. Large variations can ensue because partitioning of TiO2, Sr, Sc, Y and the LREE varies strongly as a function of Ca# in garnet (Appendix H), as previously observed (O’Reilly and Griffin, 1995; Harte and Kirkley, 1997). The dependence

of DSr,Sc,Y,Ce on Ca# has been parameterised for a large eclogite suite (Aulbach et al., 2016) and can be used to calculate the expected D as a function of garnet Ca# for eclogites from Orapa, taking into account the dependence of D on pressure and temperature. This exercise shows that the majority of samples are in trace-element equilibrium (Appendix H). Judging by iteratively calculated pressures and temperatures placing them within the lithospheric mantle column (Appendix B), samples that appear unequilibrated with respect to trace elements nevertheless seem to be in major-element equilibrium. 6. DISCUSSION 6.1. Elemental and O isotope signatures inherited from lowpressure protoliths Most mantle eclogite suites carry signatures of an origin as recycled oceanic crust (Jacob, 2004; Aulbach and Jacob, 2016), and there is increasing evidence that this recycling involved some form of plate tectonics as early as the Mesoarchaean (Shirey and Richardson, 2011; Smart et al., 2016; Tang et al., 2016). Both elemental and isotopic relationships suggest that the Orapa eclogite suite also preserved evidence for a crustal origin that was not destroyed during multiple accretionary events at the craton margin. The anti-correlation between Mg# and Y in garnet (Fig. 7A), and the positive correlation between Sr* and Eu/Eu* (Fig. 7B) and between Ti and RHREE in calculated whole rocks (Fig. 7C) strongly argue against a highpressure cumulate origin, where both Mg and Y would otherwise be enriched in crystallising garnet. Instead, these systematics are consistent with fractional crystallisation on the ocean floor, as early crystallising olivine and plagioclase exclude Ti, Y and the REE, leading to their enrichment,

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

583

Fig. 7. Major (oxides in wt%) and trace elements (ppm) in A. garnet and B.–C. reconstructed whole rocks illustrating relationships suggestive of low-pressure accumulation and fractional crystallisation processes involving olivine and plagioclase. A. High-pressure cumulates would be expected to have high Y contents at high Mg#. Inset in B. shows REE patterns for complete fossil oceanic crustal sections from Oman (Benoit et al., 1996; Godard et al., 2003), illustrating the link between the size of the Eu anomaly and total REE contents. Also shown in B. and C. are the compositions of olivine, plagioclase and cpx (white crosses) in equilibrium with a melt (distribution coefficients from McCoy-West et al., 2015), the composition of which was calculated by 20% batch melting of a Depleted Mantle source (Salters and Stracke, 2004). Natural gabbros are not pure cumulates, but originate by reaction of olivine-plagioclase cumulates with residual melts, forming cpx (Sanfilippo et al., 2015). Eu/Eu* as defined in Fig. 5; Sr* is NMORB-normalised Sr/(0.5 * Pr + 0.5 * Nd); NMORB of Sun and McDonough (1989); RHREE is the sum of rare earths from Tb to Lu (ppm) except Tm. Data sources as in Fig. 3.

whereas Eu and Sr are sequestered into plagioclase and consequently depleted in the residual melt (Aulbach and Viljoen, 2015). Modelling shows that accumulation of at least 25% of plagioclase is required in order for positive Eu anomalies to appear (Schmickler et al., 2004). This is again consistent with measured whole rocks having high normative plagioclase contents of 50% and olivine contents of 21 –31% which feature prominent positive Eu anomalies (McDonald and Viljoen, 2006). The conspicuous anti-correlation of FeO and MgO in pyroxenites (Appendix G2) suggests olivine-clinopyroxene control during formation of the protoliths, which could occur in the deep crust or uppermost mantle at an oceanic ridge (e.g. Barth et al., 2002b; Smart et al., 2012, 2017), but may also point to a high-pressure cumulate or hybrid origin (e.g. Smart et al., 2009). These two possibilities cannot be distinguished with the data at hand. The only known process that could generate non-mantle d18O in eclogite xenoliths is water-rock interactions where seawater alteration results in elevated d18O at low temperatures and in lower d18O at hydrothermal temperature (Gregory and Taylor, 1981; Jacob et al., 1994; Jacob and Foley, 1999). This is supported by the strong similarity of the global sampling distribution of d18O in mantle eclogites and ocean crust sections (Ickert et al., 2013). If mantle-like d18O ranges from 5.1 to 5.9‰ (Mattey et al., 1994), garnet and clinopyroxene in many Orapa eclogites have nonmantle d18O consistent with a low-temperature (crustal) origin (Fig. 4A) (Deines et al., 1991; Viljoen et al., 1996). Interestingly, low Ca# in garnet appears to be associated with supra-mantle d18O (>5.9‰ to 9.2‰) typical of lowtemperature seawater alteration (Fig. 4B). If d18O values are related to major-element variation in these minerals, we conclude that the eclogites preserve chemical compositions mostly inherited from crustal conditions.

6.2. Sr isotopic insights into protolith formation, metamorphic processing and metasomatism The lack of covariation of Rb/Sr with 87Sr/86Sr in clinopyroxene (Appendix E2) suggests that radiogenic isotope ratios are not due to in situ decay of Rb in clinopyroxene. Thus, radiogenic compositions either require addition of unsupported radiogenic Sr during metasomatism by isotopically enriched melts, or they reflect time-integrated 87Sr ingrowth in a bulk-rock with variable Rb/Sr. Fig. 5 shows that two groups may be distinguished with respect to 87 Sr/86Sr: (1) A group with ratios <0.7047 that shows broadly correlated behaviour with clinopyroxene majorand trace-elements, and (2) a group that shows scatter and is off-set to higher ratios. 6.2.1. Protolith formation: Preservation of near-igneous Rb/ Sr Inheritance during low-pressure processes gains support from the observation that the low-87Sr/86Sr group of samples shows a broad anti-correlation between 87Sr/86Sr and Eu/Eu* and Sr/Y in clinopyroxene (Fig. 5A and B) as well as Ca# in garnet (Fig. 5C), which reflect the effect of plagioclase accumulation and fractionation. This, therefore, excludes a significant role for contamination by the host kimberlite, which has 87Sr/86Sr of 0.7030 –0.7036 (as estimated from pristine perovskite; Sarkar et al., 2014). Plagioclase-rich cumulates, such as the protoliths to gabbroic eclogites, are expected to have low Rb/Sr (e.g. Oman cumulates: median 0.002 vs. lavas: 0.018; Benoit et al., 1996; Godard et al., 2003) and accordingly, within the low-87Sr/86Sr group, the lowest Sr isotope ratios are recorded in clinopyroxene from gabbroic eclogites. In this scenario, variable 87Sr/86Sr result from variable Rb/Sr inherited during igneous processes on the ocean floor,

584

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

followed by ingrowth of radiogenic Sr (Fig. 5B, inset). The positive trend between 87Sr/86Sr and d18O (Fig. 5D) would be due to the increasing effect of low-temperature seawater alteration in the shallow crust, consisting of the crystallisation products of residual melts (high Rb/Sr), after accumulation of plagioclase-rich rocks with low Rb/Sr in the deep crust. Regression of a broad trend formed by 87Sr/86Sr and ln (Gd/Eu) (as a proxy for the Eu anomaly) for the group with 87 Sr/86Sr < 0.7048 yields an intercept with a value of 0.7021 2 (r =0.65, n = 37), interpreted as the initial 87Sr/86Sr during protolith formation. This is higher than the 87Sr/86Sr of the chondritic mantle 3 Ga ago (0.7009, for an initial 87Sr/86Sr of 0.69898, age of the Earth of 4.566 Ga and Rb/Sr of 0.0307; Alle`gre et al., 1975; Workman and Hart, 2005), which corresponds to the oldest diamond formation event beneath Orapa, but could reflect interaction with Archaean seawater with 87Sr/86Sr of 0.7025 to 0.7031 (Veizer, 1989; Eglington et al., 2003) or a younger formation age. Given the uncertain nature of the mantle source and the possibility of a seawater imprint, it is not possible to obtain firm age constraints from 87Sr/86Sr, but modelling supports a Palaeoproterozoic or older protolith age. 6.2.2. Metamorphic processing: a role for continental crust formation? Given d18O evidence for seawater alteration, it is likely that the protoliths were hydrated on the ocean floor, then underwent dehydration at shallow depths upon recycling into the mantle, possibly accompanied or followed by partial melt loss (Schmidt and Poli, 2014). If this process had been significant, higher solubility of Rb compared to Sr in both aqueous fluids and hydrous melts in equilibrium with MORB (Kessel et al., 2005) should lead to low Rb/ Sr, and would therefore obliterate coherent behaviour of Sr isotope compositions in the low-87Sr/86Sr group with indicators of low-pressure igneous processes, such as Eu/ Eu* or Mg# (previous section). Instead, this indicates near-undisturbed time-integrated ingrowth of 87Sr for billions of years and suggests that Rb/Sr ratios (which could not be reliably determined in situ because Rb concentrations in both garnet and clinopyroxene are too close to the detection limit) were not significantly disturbed during metamorphism. Depletions in LREE in NMORB-normalised patterns of eclogites, also observed for Orapa, are often ascribed to loss of a broadly tonalitic partial melt and associated formation of cratonic continental crust, as suggested by trace element modelling (Ireland et al., 1994; Barth et al., 2001; Tappe et al., 2011; Smart et al., 2014; Smit et al., 2014; Dongre et al., 2015). However, LREE-depletion in gabbroic eclogites cannot be unambiguously ascribed to melt loss because this is also a feature of cumulates (Appendix G3). In contrast, the LREE-depletion of some of the basaltic/picritic eclogites with flat HREE patterns is not consistent with igneous differentiation and suggests that they have lost a partial melt, in which case a link to continental crust formation can be explored. Tonalite-trondhemite-granodiorite of 2.65 –2.7 Ga age occur at the southwestern margin of the Zimbabwe craton and are suggested to have formed by

partial melting of continental and oceanic crust (Zhai et al., 2006). Currently available age constraints for eclogitic materials from Orapa include Palaeoproterozoic and Mesoarchaean ages (Shirey et al., 2002, 2008; Timmerman et al., 2017), and therefore are permissive of a link between continental crust formation and eclogite melting at Orapa, but more data are required to firmly established such a connection. In summary, major and trace elements in gabbroic eclogites are inherited from low-pressure differentiation, with little evidence for strong subduction-related modification. In contrast, it is possible that basaltic/picritic eclogites underwent partial melting. 6.2.3. Mantle metasomatism: Interaction with aged, metasome-derived melt Orapa diamonds show great isotope variability, from typical mantle values (including d13C, d15N and d18O of diamond and its inclusions) to non-mantle like values (including d13C, d15N, D33S and d18O; Deines et al., 1991, 1993; Viljoen et al., 1996; Cartigny et al., 1999; Farquhar et al., 2002; Deines and Harris, 2004; Ickert et al., 2013; Chinn et al., 2016). There is still debate as to whether positive d15N and low d13C-values reflect subduction or hightemperature fractionation, but clear evidence for a subduction signature comes from the study of their inclusions. This raises the question whether the carbon and nitrogen were inherited or introduced by metasomatism. For example, diamonds with inclusions analysed for S-isotopes have not been yet analysed for C-isotopes and the relationship between d13C-d18O, inferred to imply that carbon was inherited in high-d18O eclogitic diamonds, was subsequently questioned (Cartigny et al., 2014). Whether carbon in eclogite is primary or secondary is key to understand the origin and fate of volatiles and can be discussed in light of Srisotope compositions. Samples with 87Sr/86Sr > 0.7048 do not lie on arrays consistent with control by igneous processes on the ocean floor (Fig. 5) and are additionally characterised by mostly mantle-like O. These relationships require addition of a contaminant that has high concentrations of radiogenic Sr and can quickly shift the Sr isotopic composition without much affecting oxygen isotopes. This could be subducted carbonate-rich sediment, the bulk addition of a few percent of which suffices to explain the radiogenic Sr-O isotope relationships of most Group II eclogites (Viljoen et al., 1996), but would predict non-mantle d13C and d15N values in diamonds that are not observed. The strong association of unsupported radiogenic 87 Sr/86Sr with mantle-like d18O (this data-set; Fig. 5D) and mantle-like C and N isotopic ratios in the diamonds (Cartigny et al., 1999; Ickert et al., 2013) may alternatively reflect addition of a mantle-derived contaminant. Phlogopite-rich metasomes, such as mica-amphibole-rutil e-ilmenite-diopside (MARID), have high Rb and Sr contents (median 356 and 413 ppm, respectively, with median Rb/Sr of 0.66; Waters, 1987; Gre´goire et al., 2002). Once formed, they will not only quickly evolve to radiogenic Sr isotope compositions, but they are also potent expressions and sources of mantle metasomatism. Such metasomes are directly observed in xenolith suites, e.g. the Kimberley

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

area in the Kaapvaal craton (Waters, 1987; Gre´goire et al., 2002). Moreover, they are inferred in the sources of smallvolume volatile-rich melts intruded in cratons around the globe (e.g. Foley et al., 2002b; Foley, 2008; Tappe et al., 2007, 2012; Rosenthal et al., 2009), and provide the mantle ‘‘conditioning” necessary for successful kimberlite eruption (Giuliani et al., 2014). It is worth noting that MARID and glimmerite xenoliths also occur within kimberlites of the Orapa cluster (Letlhakane; Stiefenhofer et al., 1997). Relatively young (170–150 Ma ago), enriched MARID rocks from Kimberley have 87Sr/86Sr at the time of eruption (90 Ma) up to 0.709 (Giuliani et al., 2015). Older ages (1.25–1.0 Ga) have been determined for phlogopite in mantle xenoliths from southern Africa (Hopp et al., 2008). If emplaced 1 Ga ago with a Rb/Sr of 0.66, average MARID would evolve to present-day 87Sr/86Sr of 0.73 (or 0.726 at the time of kimberlite eruption). This process would therefore account for radiogenic Sr and mantle-like O in the inclusions combined with mantle-like d13C in the diamonds at Orapa (if the fluids are also the source of carbon). Modelling shows that for a bulk-mixing scenario minute amounts of a metasome-derived melt (e.g. a proportion of 0.0002 of 0.1% batch melt derived from phlogopiteclinopyroxenite) would suffice to significantly raise 87 Sr/86Sr in the eclogites (Fig. 5D; parameters given in caption). A call for phlogopite-rich metasomes producing 87 Sr-enriched melts, for which we argue above, and for interaction with sediments called for in earlier work (Viljoen et al., 1996) need not represent two completely unrelated scenarios, as phlogopite is known to form by reaction of siliceous hydrous melts or fluids, such as those typically released from crustal rocks during dehydration reactions in subduction zones, with the mantle (Wyllie and Sekine, 1982; Sato et al., 1997; Konzett and Ulmer, 1999). For example, in the Kaapvaal craton, Mesoproterozoic phlogopite ages were linked to the Namaqua-Natal orogeny (Hopp et al., 2008), while in the Slave craton rare

585

phlogopitite has yielded ages consistent with formation during 2.7 Ga craton amalgamation (Aulbach et al., 2007). 6.3. Diamond formation and destruction in the mantle eclogite reservoir beneath Orapa In this section, the chemical compositions and radiogenic and stable isotope signatures of diamondiferous, graphite-bearing and carbon-free eclogites and of inclusions in diamond are scrutinised with the aim to constrain carbon sources (originally present vs. metasomatically introduced) and to understand processes leading to diamond destruction or complete removal of carbon. 6.3.1. Carbon-free vs. carbon-bearing eclogites and pyroxenites Carbon-free eclogites span nearly the whole range of diamondiferous and graphite-bearing eclogites. Given that there are no significant geochemical differences between carbon-bearing and carbon-free eclogites, it is tempting to ascribe the absence of accessory carbon minerals in many of these samples to sectioning effects combined with small sample size. In contrast, carbon-free pyroxenites plot at low pressures (<4 GPa) and have low K2O content in clinopyroxene and Na2O content in garnet (Fig. 8). They show no overlap with diamondiferous samples and likely represent a largely barren component in the mantle lithosphere beneath Orapa, implying that there was no carbon originally present, or that it was removed during subsequent processes. 6.3.2. Diamondiferous vs. graphite-bearing eclogites and pyroxenites High Na2O contents in garnet and high K2O in clinopyroxene from typically diamondiferous Group I eclogite have been related to high pressure (>4.5 GPa), which favours not only diamond stabilisation, but also the incorporation of alkalis in eclogitic minerals

Fig. 8. A. Na2O in garnet and B. K2O in clinopyroxene as a function of pressure obtained by iterative solution of TK00 with the peridotitederived geotherm (see Fig. 6 for details), showing inclusions in diamond for comparison. Vertical dashed line shows the approximate graphitediamond (G-D) boundary (as in Fig. 6) while horizontal lines separate Group I and Group II diamonds (as in Fig. 3). Offsets from the main array in B. may reflect metasomatism during diamond formation in the DI and/or re-equilibration of the matrix eclogite after decompression. Data sources as in Fig. 3.

586

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

(McCandless and Gurney, 1989). At Orapa, the highest Na2O contents in garnet are clearly observed in diamondiferous eclogites (up to 0.25 wt%), whereas graphite-bearing eclogites are restricted to lower Na2O contents (up to 0.16 wt%). The broad correlation with pressure (Fig. 8A) suggests a dominant pressure control. Diamond in many of these samples may therefore have formed during introduction of carbon- and alkali-bearing fluids (e.g. Weiss et al., 2014), perhaps as early as during Archaean oceanic crust subduction. Clinopyroxene in some samples shows offsets to high K2O content at a given pressure, indicating an additional compositional effect (K addition; Fig. 8B). It is interesting to note the overlap of graphite- and diamond-bearing samples with regard to temperature (Fig. 6) and pressure (if equilibrated to the same conductive geotherm), and the occurrence of samples containing both carbon polymorphs. This suggests that the mantle in this pressure interval was once in the diamond stability field, but did not remain there. Indeed, a given parcel of lithospheric mantle can move with respect to the graphitediamond boundary, for example due to transient heating by advected melts (O’Reilly and Griffin, 2010). Decompression can ensue due to lithosphere stretching during plume impingement (McKenzie, 1989), such as that responsible for heating, strong refertilisation of the deep root and kimberlite emplacement in the Kaapvaal craton (Kobussen et al., 2008). While only little carbonate is expected in melts in the highly reducing, deeper reaches of the cratonic lithosphere (>180 km), some shallow lithospheric mantle regions (150–180 km) are oxidising enough to permit the percolation of carbonated melts (FMQ-1 to -1.5; Stagno et al., 2013). Such melts may have interacted with reduced eclogite lithologies (e.g. Smart et al., 2017) and oxidised and destroyed part of the diamond/graphite inventory beneath Orapa. Under lower melt-rock ratios, they may precipitate diamond at high pressure and, after heating and/or decompression, catalyse the conversion of pre-existing metastable

diamond to graphite. It is worth noting that diamondiferous eclogites record lower median 87Sr/86Sr (0.7039) than graphite-bearing samples (0.7064) and carbon-free samples (0.7051) (Table 1; Viljoen et al., 1996). This observation provides a link to the proposed introduction of unsupported radiogenic Sr during mantle metasomatism precursory to kimberlite emplacement and sourced in an aged phlogopite-bearing metasome, highlighting the role of oxidising, small-volume melts in the lithospheric carbon cycle. 6.3.3. Diamondiferous eclogites vs. inclusions in diamond It has long been recognised that eclogite xenoliths are overall compositionally different from DI at Orapa (Gurney et al., 1984; Viljoen et al., 1996), as there are diamondiferous orthopyroxene-bearing rocks that were either not sampled by the kimberlite or not targeted for xenolith studies. The specific characteristics (C-isotopes, Ncontents and aggregation state) of inclusion-free diamonds (Chinn et al., 2016) is another indication that diamonds form under a much wider range of conditions (fluid speciation, time of formation, depth) than would be deduced from eclogite studies solely. This is also evident in the extended data-base (this study): (1) While diamondiferous eclogites span the entire range of FeO and CaO contents in reconstructed bulk rocks, inclusions in diamonds are compositionally restricted to high FeO and low CaO, SiO2 and Na2O contents (Fig. 9). (2) Garnet DI do not overlap xenolithic garnet in high Ca#-high Mg# space, which is populated by garnet in diamondiferous eclogite (Fig. 3A). (3) Clinopyroxene DI are mostly restricted to low Mg# and high K2O, whereas clinopyroxene in eclogite xenoliths shows a much wider range (Figs. 3C and 8B). It is further noteworthy that garnet DI have a mode at lower d18O than eclogite minerals, though scattering to similarly high values (Fig. 4A), and that they appear to have equilibrated at higher temperatures (Fig. 6). Multiple evidence presented above indicates that elemental and isotope systematics of Orapa eclogites,

Fig. 9. Major element oxides in eclogites and pyroxenites as a function of MgO (wt%) designed to illustrate the difference between diamondiferous eclogites and DI, the latter restricted to high FeO and low SiO2, CaO and Na2O. Shown for comparison are natural Fe picrites (Milidragovic and Francis, 2016) and experimental pyroxenite-derived melts (Kogiso et al., 2003; Keshav et al., 2004; Tuff et al., 2005; Sobolev et al., 2007; Lambart et al., 2009, 2012; Gerbode et al., 2010; Rosenthal et al., 2014) along with suggested trends for interaction with, or formation from, such magmas. Error crosses reflect uncertainty estimates derived by varying the modal abundance of garnet and clinopyroxene by 10% (as given in Appendix G). Data sources as in Fig. 3.

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

including diamondiferous eclogites, were largely inherited during the low-pressure stage. This also applies to the high CaO and low FeO contents of gabbroic eclogites, which reflect their cumulate character. Thus, the difference in composition between inclusions in diamond and the same minerals in diamondiferous eclogites cannot be ascribed to modification of the eclogite bulk compositions and mineralogy after diamond formation. Rather, the distinctiveness of DI and part of the eclogite xenoliths with respect to CaO and FeO is suggested to relate to the site, mode and timing of diamond formation. Diamond formation itself is associated with metasomatic processes involving a range of agents from solutepoor volatile-rich fluids to silicate melts (Stachel and Luth, 2015). Thus, silicate-included eclogitic diamond from Orapa may have formed, or grown, during a diamond formation episode involving percolation of an FeO-rich and SiO2-, Na2O- and CaO-poor melt, similar to those generated by pyroxenite-melting (Lambart et al., 2009) (see comparison to natural Fe-picrites and experimental pyroxenitederived melts in Fig. 9). Inclusion dating of individual zones of eclogitic diamond from Orapa has revealed a growth history spanning two billion years (Timmerman et al., 2017). This metasomatism must have similarly affected a portion of the eclogites, as indicated by the partial compositional overlap with DI (Fig. 9). Garnet inclusions in diamonds have a mode at mantlelike O isotopic composition (Fig. 4A) and isolated inclusions in diamond have a mode at higher temperatures than garnet-cpx pairs in the eclogite xenoliths that continued to equilibrate to lower temperatures in response to subsequent

587

cooling (Fig. 6A). These may indicate diamond formation during interaction with a predominantly mantle-derived melt. Combined with the FeO-rich and SiO2-poor compositions of DI, these observations provide circumstantial evidence for a link to lithosphere modification during the Bushveld event, the surface expression of which reaches near Orapa (Fig. 1), and which involved a pyroxenitecontaminated source (Wilson, 2012) and coincides with a diamond formation episode at Orapa (Shirey et al., 2008; Timmerman et al., 2017). Summary and implications The major- and trace-element compositions of garnet and clinopyroxene in diamondiferous, graphite-bearing and carbon-free eclogite xenoliths from the Orapa kimberlite, at the western Zimbabwe craton margin, were investigated along with d18O in garnet and clinopyroxene, and 87 Sr/86Sr in clinopyroxene. In combination with data from the literature, encompassing eclogitic inclusions in diamond, the new results allow constraints on the evolution of the mantle eclogite source and its diamond inventory, as illustrated in Figure 10:  The eclogite xenolith suite is predominantly gabbroic (Eu* > 1.05) and retains strong evidence for lowpressure igneous fractionation and seawater alteration, such as non-mantle d18O, anti-correlated Y and Mg# in garnet, and anti-correlated Eu/Eu* and RHREE in garnet and whole rocks. These systematics were preserved despite multiple tectonomagmatic processes at

Fig. 10. Cartoon illustrating the suggested evolution of Orapa eclogites and associated diamonds, from protolith formation on an Archaean ocean floor to exhumation in the host kimberlite, based on this work and that reviewed and cited in section ‘‘Geology and samples”. A. Protoliths to Orapa eclogites formed by igneous differentiation in the oceanic crust ca. 3 Ga ago, followed by subduction during collision of continental nuclei and formation of the earliest generation of eclogitic diamonds at ca. 2.9 Ga. If Orapa sits at the northern subsurface extension of the Kaapvaal craton, this may relate to suturing of the Kimberley and Witwatersrand cratonic blocks. B. The giant Bushveld event, the northwestern surface expression of which nearly reaches Orapa (Fig. 1), was accompanied by extensive mantle metasomatism and crystallisation (or recrystallisation?) of a new diamond generation. Accretion of Palaeoproterozoic rocks of the Magondi Fold Belt (2.0– 1.85 Ga) initiated at approximately the same time. C. Accretion at ca. 1.1 Ga was again accompanied by diamond formation or recrystallisation and may have led to introduction of phlogopite-bearing metasomes, which over time evolved to strongly radiogenic Sr isotope compositions. D. Remobilisation of oxidising melts from earlier-emplaced metasomes, heating and decompression during lithosphere stretching leading up to kimberlite emplacement led to partial destruction of the diamond inventory.

588

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

the craton margin and in the context of kimberlite magmatism. The correlation of 87Sr/86Sr in a group of eclogites with moderate 87Sr/86Sr (0.7026–0.7046) with indices of low-pressure accumulation and fractionation (Eu*, Mg#, Sr/Y) suggests a link to low-pressure igneous processes and relatively undisturbed long-term ingrowth of 87 Sr from 87Rb. This includes samples with heavy d18O attesting to low-temperature alteration on the ocean floor.  Several eclogites have clinopyroxene with unsupported, highly radiogenic Sr (87Sr/86Sr > 0.7048 to 0.7091, combined with mantle-like d18O, suggesting interaction with a melt derived from an aged phlogopite-rich mantle metasome, such as those involved in lithosphere conditioning ahead of kimberlite magmatism. Of note, graphite-bearing eclogites and carbon-free samples record higher median 87Sr/86Sr (0.7064 and 0.7051, respectively) than diamondiferous eclogites (0.7039), suggesting a link to this metasomatic event.  Diamondiferous, graphite-bearing and carbon-free eclogites and pyroxenites coexist over a calculated pressure interval of 3.2–4.9 GPa (along a cratonic geotherm). This overlap may reflect late heating and/ or decompression during lithosphere stretching and magmatism precursory to kimberlite emplacement. Thus, the 87Sr-rich, phlogopite metasome-derived melt may have facilitated conversion of metastable (after decompression/heating) diamond to graphite in some samples, while causing carbon removal in others. This may occur in the shallower lithosphere (150–180 km) where oxidising, carbonated melts can exist, highlighting the role of small-volume melts as modulators of the lithospheric carbon cycle.  Inclusion-bearing eclogitic diamonds yield reconstructed bulk rocks restricted to high FeO and low SiO2, CaO and Na2O contents compared to the range observed in diamondiferous eclogites, possibly reflecting interaction with, or crystallisation from, a melt with contributions from clinopyroxene-rich heterogeneities. Circumstantial evidence for a link to nearby ca. 2.0 Ga Bushveld magmatism includes the reported identification of pyroxenite-contributions to the Bushveld magma source and recognition of a 2.0 Ga eclogitic diamond formation episode at Orapa, and predominantly mantle-like O and higher average temperatures recorded by DI compared to eclogites in this study. ACKNOWLEDGMENTS Detailed and constructive reviews by Roberta Rudnick and two anonymous referees significantly improved this work. The Deutsche Forschungsgemeinschaft is gratefully acknowledged for their support to SA (DFG-grant AU356/8), to PC and GW (DFG Wo362/31-1, Leibniz Award 1997). The research of KSV is supported with funds from the South African Department of Science and Technology under their Research Chairs Initiative, as administered by the National Research Foundation. KSV also acknowledges financial support from the Centre of Excellence for Integrated Mineral and Energy Resource Analysis (CIMERA) at the University of Johannesburg. The management of the De Beers

Group of Companies are thanked for assistance in the collecting of the samples, for access to the electron microprobe as well as the LA-ICP-MS, and for permission to publish. This is contribution 981 from the ARC Centre of Excellence for Core to Crust Fluid Systems (http://www.ccfs.mq.edu.au) and 1160 in the GEMOC Key Centre (http://www.gemoc.mq.edu.au), Macquarie University.

APPENDIX A. SUPPLEMENTARY MATERIAL Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.gca.2017.06.03. REFERENCES Adam J. and Green T. (2006) Trace element partitioning between mica- and amphibole-bearing garnet lherzolite and hydrous basanitic melt: 1. Experimental results and the investigation of controls on partitioning behaviour. Contrib. Miner. Petrol. 152, 1–17. Alle`gre C. J., Birck J. L., Fourcade S. and Semet M. P. (1975) 87 Rb/87Sr age of Juvinas basaltic achondrite and early igneous activity in solar system. Science 187, 436–438. Allsopp, H.L., Bristow, J.W., Smith, C.B., Brown, R., Gleadow, A. J.W., Kramers, J.D. and Garvie, O.G., 1989. A summary of radiometric dating methods applicable to kimberlites and related rocks. In: Kimberlites and Related Rocks: Their Composition, Occurrence, Origin and Emplacement. Geological Society of Australia Special Publication No. 14, Blackwell Scientific. (eds.) J. Ross, A.L. Jacques, J. Ferguson, D.H. Green, S.Y. O’Reilly, R.V. Danchin, A.J.A. Janse, pp. 343–357. Aulbach S., Pearson N. J., O’Reilly S. Y. and Doyle B. J. (2007) Origins of xenolithic eclogites and pyroxenites from the central slave craton, Canada. J. Petrol. 48, 1843–1873. Aulbach S., Gerdes A. and Viljoen K. S. (2016) Formation of diamondiferous kyanite-eclogite in a subduction melange. Geochim. Cosmochim. Acta 179, 156–176. Aulbach S. and Jacob D. E. (2016) Major- and trace-elements in cratonic mantle eclogites and pyroxenites reveal heterogeneous sources and metamorphic processing of low-pressure protoliths. Lithos 262, 586–605. Aulbach S. and Viljoen K. S. (2015) Eclogite xenoliths from the Lace kimberlite, Kaapvaal craton: from convecting mantle source to palaeo-ocean floor and back. Earth Planet. Sci. Lett. 431, 274–286. Barth M. G., Rudnick R. L., Horn I., McDonough W. F., Spicuzza M. J., Valley J. W. and Haggerty S. E. (2001) Geochemistry of xenolithic eclogites from West Africa, Part I: A link between low MgO eclogites and Archean crust formation. Geochim. Cosmochim. Acta 65, 1499–1527. Barth M. G., Foley S. F. and Horn I. (2002a) Partial melting in Archean subduction zones: constraints from experimentally determined trace element partition coefficients between eclogitic minerals and tonalitic melts under upper mantle conditions. Precambr. Res. 113, 323–340. Barth M. G., Rudnick R. L., Horn I., McDonough W. F., Spicuzza M. J., Valley J. W. and Haggerty S. E. (2002b) Geochemistry of xenolithic eclogites from West Africa, part 2: Origins of the high MgO eclogites. Geochim. Cosmochim. Acta 66, 4325–4345. Becker M. and Le Roex A. P. (2006) Geochemistry of South African on- and off-craton, Group I and Group II kimberlites: petrogenesis and source region evolution. J. Petrol. 47, 673– 703.

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592 Benoit M., Polve M. and Ceuleneer G. (1996) Trace element and isotopic characterization of mafic cumulates in a fossil mantle diapir (Oman ophiolite). Chem. Geol. 134, 199–214. Bizzarro M., Baker J. A. and Ulfbeck D. (2003) A new digestion and chemical separation technique for rapid and highly reproducible determination of Lu/Hf and Hf isotope ratios in geological materials by MC-ICP-MS. Geostandandards Newsletter. 27, 133–145. Carlson R. W., Pearson D. G., Boyd F. R., Shirey S. B., Irvine G., Menzies A. H. and Gurney J. J. (1999) Regional age variation of the southern African mantle: significance for models of lithospheric mantle formation. In Proceedings of the 7th International Kimberlite Conference (eds. J. J. Gurney, J. L. Gurney, M. D. Pascoe and S. H. Richardson). Red Roof Design, Cape Town, South Africa, pp. 99–108. Cartigny P., Harris J. W. and Javoy M. (1999) Eclogitic, peridotitic, metamorphic diamonds and the problems of carbon recycling: the case of Orapa (Botswana). In 7th International Kimbrlite Conference. red roof design, Cape Town (eds. J. J. Gurney, J. L. Gurney, M. D. Pascoe and S. H. Richardson), pp. 117–124. Cartigny P., Palot M., Thomassot E. and Harris J. W. (2014) Diamond formation: a stable isotope perspective. Annu. Rev. Earth Planet. Sci. 42(42), 699–732. Chinn I., Perritt S. and Stiefenhofer J. (2016) Carbon Isotope, Nitrogen Abundane and Nitrogen Aggregation State Data for Diamonds from Orapa mine: a clear subduction signature? In 2nd European Mineralogical Conference, Rimini, Italy. Davis G. L. (1977) The ages and uranium content of zircons from kimberlites and associated rocks, 2nd International Kimberlite Conference. In 2nd International Kimberlite Conference. Carnegie Institute of Washington Yearbook, Santa Fe, USA, pp. 631–635. Day H. W. (2012) A revised diamond-graphite transition curve. Am. Miner. 97, 52–62. de Wit M. J., Roering C., Hart R. J., Armstrong R. A., de Ronde C. E. J., Green R. W. E., Tredoux M., Peberdy E. and Hart R. A. (1992) Formation of an Archaean continent. Nature 357, 553–562. Deines P. and Harris J. W. (2004) New insights into the occurrence of C-13-depleted carbon in the mantle from two closely associated kimberlites: Letlhakane and Orapa, Botswana. Lithos 77, 125–142. Deines P., Harris J. W., Robinson D. N., Gurney J. J. and Shee S. R. (1991) Carbon and oxygen isotope variations in diamond and graphite eclogites from Orapa, Botswana, and the nitrogencontent of their diamonds. Geochim. Cosmochim. Acta 55, 515– 524. Deines P., Harris J. W. and Gurney J. J. (1993) Depth-related carbon-isotope and nitrogen concentration variability in the mantle below the Orapa kimberlite, Botswana, Africa. Geochim. Cosmochim. Acta 57, 2781–2796. Dongre A. N., Jacob D. E. and Stern R. A. (2015) Subductionrelated origin of eclogite xenoliths from the Wajrakarur kimberlite field, Eastern Dharwar craton, Southern India: Constraints from petrology and geochemistry. Geochm. et Cosmochim. Acta 166, 165–188. Eglington B. M., Talma A. S., Marais S., Matthews P. E. and Dixon J. G. P. (2003) Isotopic composition of Pongola Supergroup limestones from the Buffalo River gorge, South Africa: constraints on their regional depositional setting. S. Afr. J. Geol. 106, 1–10. Farquhar J., Wing B. A., McKeegan K. D., Harris J. W., Cartigny P. and Thiemens M. H. (2002) Mass-independent sulfur of inclusions in diamond and sulfur recycling on early earth. Science 298, 2369–2372.

589

Foley S., Tiepolo M. and Vannucci R. (2002a) Growth of early continental crust controlled by melting of amphibolite in subduction zones. Nature 417, 837–840. Foley S. F. (2008) Rejuvenation and erosion of the cratonic lithosphere. Nat. Geosci. 1, 503–510. Foley S. F. (2011) A reappraisal of redox melting in the earth’s mantle as a function of tectonic setting and time. J. Petrol. 52, 1363–1391. Foley S. F., Andronikov A. V. and Melzer S. (2002b) Petrology of ultramafic lamprophyres from the Beaver Lake area of Eastern Antarctica and their relation to the breakup of Gondwanaland. Mineral. Petrol. 74, 361–384. Gerbode C. and Dasgupta R. (2010) Carbonate-fluxed Melting of MORB-like Pyroxenite at 2.9 GPa and Genesis of HIMU Ocean Island Basalts. J. Petrol. 51, 2067–2088. Giuliani A., Phillips D., Maas R., Woodhead J. D., Kendrick M. A., Greig A., Armstrong R. A., Chew D., Kamenetsky V. S. and Fiorentini M. L. (2014) LIMA U-Pb ages link lithospheric mantle metasomatism to Karoo magmatism beneath the Kimberley region, South Africa. Earth Planet. Sci. Lett. 401, 132–147. Giuliani A., Phillips D., Woodhead J. D., Kamenetsky V. S., Fiorentini M. L., Maas R., Soltys A. and Armstrong R. A. (2015) Did diamond-bearing orangeites originate from MARID-veined peridotites in the lithospheric mantle? Nature Commun. 6, 1–10. Godard M., Dautria J. M. and Perrin M. (2003) Geochemical variability of the Oman ophiolite lavas: Relationship with spatial distribution and paleomagnetic directions. Geochem. Geophys. Geosyst. 4, R609. Green T. H., Blundy J. D., Adam J. and Yaxley G. M. (2000) SIMS determination of trace element partition coefficients between garnet, clinopyroxene and hydrous basaltic liquids at 2–7.5 GPa and 1080–1200 °C. Lithos 53, 165–187. Gre´goire M., Bell D. R. and Le Roex A. P. (2002) Trace element geochemistry of phlogopite-rich mafic mantle xenoliths: their classification and their relationship to phlogopite-bearing peridotites and kimberlites revisited. Contrib. Miner. Petrol. 142, 603–625. Gregory R. T. and Taylor H. P. (1981) An oxygen isotope profile in a section of cretaceous oceanic crust, Samail ophiolite, Oman: evidence for d18O buffering of the oceans by deep (>5 km) seawater-hydrothermal circulation at mid-ocean ridges. J. Geophys. Res. 86, 2737–2755. Griffin W. L., O’Reilly S. Y., Natapov L. M. and Ryan C. G. (2003) The evolution of lithospheric mantle beneath the Kalahari Craton and its margins. Lithos 71, 215–241. Gru¨tter H. S. (2009) Pyroxene xenocryst geotherms: techniques and application. Lithos 112, 1167–1178. Gurney J. J., Harris J. W. and Rickard R. S. (1984) Silicate and oxide inclusions in diamonds from the Orapa Mine, Botswana. In Kimberlites II: The mantle and crust-mantle relationships (ed. J. Kornprobst). Elsevier, Amsterdam, pp. 1–9. Gurney J. J. et al. (1986) UCT Internal Report No. 2, unpaginated. Appendix to: Gurney, J.J., Harris, J.W. and Rickard, R.S. (1984) Silicate and oxide inclusions in diamonds from the Orapa Mine, Botswana. In Kimberlites II: The Mantle and Crust-mantle Relationships (ed. J. Kornprobst). Elsevier, Amsterdam, pp. 1–9. Harte B. and Kirkley M. B. (1997) Partitioning of trace elements between clinopyroxene and garnet: data from mantle eclogites. Chem. Geol. 136, 1–24. Hasterok D. and Chapman D. S. (2011) Heat production and geotherms for the continental lithosphere. Earth Planet. Sci. Lett. 307, 59–70.

590

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

Helmstaedt H. and Doig R. (1975) Eclogite nodules from kimberlite pipes of the Colorado Plateau – samples of subducted Franciscan-type oceanic lithosphere. Phys. Chem. Earth 9, 95– 110. Hills D. V. and Haggerty S. E. (1989) Petrochemistry of eclogites from the Koidu kimberlite complex, Sierra-Leone. Contrib. Miner. Petrol. 103, 397–422. Hopp J., Trieloff M., Brey G. P., Woodland A. B., Simon N. S. C., Wijbrans J. R., Siebel W. and Reitter E. (2008) 40Ar39Ar-ages of phlogopite in mantle xenoliths from South African kimberlites: evidence for metasomatic mantle impregnation during the Kibaran orogenic cycle. Lithos 106, 351–364. Ickert R. B., Stachel T., Stern R. A. and Harris J. W. (2013) Diamond from recycled crustal carbon documented by coupled d18O-d13C measurements of diamonds and their inclusions. Earth Planet. Sci. Lett. 364, 85–97. Ickert R. B. and Stern R. A. (2013) Matrix corrections and error analysis in high-precision SIMS 18O/16O measurements of CaMg-Fe Garnet. Geostand. Geoanal. Res. 37, 429–448. Ireland T. R., Rudnick R. L. and Spetsius Z. (1994) Trace elements in diamond inclusions from eclogites reveal link to Archean granites. Earth Planet. Sci. Lett. 128, 199–213. Jacob D., Jagoutz E., Lowry D., Mattey D. and Kudrjavtseva G. (1994) Diamondiferous eclogites from Siberia - remnants of Archean oceanic crust. Geochim. Cosmochim. Acta 58, 5191– 5207. Jacob D. E. (2004) Nature and origin of eclogite xenoliths from kimberlites. Lithos 77, 295–316. Jacob D. E. (2006) High sensitivity analysis of trace element poor geological reference glasses by laser-ablation inductively coupled plasma mass spectrometry (LA- ICP-MS). Geostand. Geoanal. Res. 30, 221–235. Jacob D. E. and Foley S. F. (1999) Evidence for Archean ocean crust with low high field strength element signature from diamondiferous eclogite xenoliths. Lithos 48, 317–336. Jacob D. E., Wirth R., Enzmann F., Kronz A. and Schreiber A. (2011) Nano-inclusion suite and high resolution micro-computed-tomography of polycrystalline diamond (framesite) from Orapa, Botswana. Earth Planet. Sci. Lett. 308, 307–316. Jacobs J., Pisarevsky S., Thomas R. J. and Beckere T. (2008) The Kalahari Craton during the assembly and dispersal of Rodinia. Precambr. Res. 160, 142–158. Keshav S., Gudfinnsson G. H., Sen G. and Fei Y. W. (2004) Highpressure melting experiments on garnet clinopyroxenite and the alkalic to tholefitic transition in ocean-island basalts. Earth Planet. Sci. Lett. 223, 365–379. Kessel R., Schmidt M. W., Ulmer P. and Pettke T. (2005) Trace element signature of subduction-zone fluids, melts and supercritical liquids at 120–180 km depth. Nature 437, 724– 727. Klemme S., Blundy J. D. and Wood B. J. (2002) Experimental constraints on major and trace element partitioning during partial melting of eclogite. Geochim. Cosmochim. Acta 66, 3109– 3123. Kobussen A. F., Griffin W. L., O’Reilly S. Y. and Shee S. R. (2008) Ghosts of lithospheres past: imaging an evolving lithospheric mantle in southern Africa. Geology 36, 515–518. Kogiso T., Hirschmann M. M. and Frost D. J. (2003) Highpressure partial melting of garnet pyroxenite: possible mafic lithologies in the source of ocean island basalts. Earth Planet. Sci. Lett. 216, 603–617. Konzett J. and Ulmer P. (1999) The stability of hydrous potassic phases in lherzolitic mantle – an experimental study to 9 center dot 5 GPa in simplified and natural bulk compositions. J. Petrol. 40, 629–652.

Krogh Ravna E. J. (2000) The garnet-clinopyroxene Fe2+-Mg geothermometer: an updated calibration. J. Metamorph. Geol. 18, 211–219. Kusky T. M. (1998) Tectonic setting and terrane accretion of the Archean Zimbabwe craton. Geology 26, 163–166. Lambart S., Laporte D. and Schiano P. (2009) An experimental study of pyroxenite partial melts at 1 and 1.5 GPa: implications for the major-element composition of Mid-Ocean Ridge Basalts. Earth Planet. Sci. Lett. 288, 335–347. Lambart S., Laporte D., Provost A. and Schiano P. (2012) Fate of pyroxenite-derived melts in the peridotitic mantle: thermodynamic and experimental constraints. J. Petrol. 53, 451–476. Lee C. T. (2006) Geochemical/petrologic constraints on the origin of cratonic mantle, archean geodynamics and environments. Geophys. Monogr. Ser., 98–114. MacGregor I. D. and Carter J. L. (1970) The chemistry of clinopyroxenes and garnets of eclogite and peridotite xenoliths from the Roberts Victor mine, South Africa. Phys. Earth Planet. Inter. 3, 391–397. MacGregor I. D. and Manton W. I. (1986) Roberts-victor eclogites: ancient oceanic crust. J. Geophys. Res.-Solid Earth Planets 91, 14063–14079. Mattey D., Lowry D. and Macpherson C. (1994) Oxygen-isotope composition of mantle peridotite. Earth Planet. Sci. Lett. 128, 231–241. McCandless, T.E., Gurney, J.J., 1989. Sodium in garnet and potassium in clinopyroxene: Criteria for classifying mantle eclogites. In: Kimberlites and Related Rocks, Vol 2: Their Mantle/Crust Setting, Diamonds and Diamond Exploration. (eds.) J. Ross, A.L. Jacques, J. Ferguson, D.H. Green, S.Y. O’Reilly, R.V. Danchin, A.J.A. Janse. Geological Society of Australia Special Publication No. 14. Blackwell Scientific, pp. 827–832. McCoy-West A. J., Bennett V. C., O’Neill H. S., Hermann J. and Puchtel I. S. (2015) The interplay between melting, refertilization and carbonatite metasomatism in off-cratonic Lithospheric Mantle under Zealandia: an Integrated Major, Trace and Platinum Group Element Study. J. Petrol. 56, 563–604. McDonald I. and Viljoen K. S. (2006) Platinum-group element geochemistry of mantle eclogites: a reconnaissance study of xenoliths from the Orapa kimberlite, Botswana. Appl. Earth Sci. (Transactions of the Institution of Mining and Metallurgy B) 115, 81–93. McKenzie D. (1989) Some remarks on the movement of small melt fractions in the mantle. Earth Planet. Sci. Lett. 95, 53–72. Miensopust M. P., Jones A. G., Muller M. R., Garcia X. and Evans R. L. (2011) Lithospheric structures and Precambrian terrane boundaries in northeastern Botswana revealed through magnetotelluric profiling as part of the Southern African Magnetotelluric Experiment. J. Geophys. Res.-Solid Earth 116. Milidragovic D. and Francis D. (2016) Ca. 2.7 Ga ferropicritic magmatism: a record of Fe-rich heterogeneities during Neoarchean global mantle melting. Geochim. Cosmochim. Acta 185, 44–63. Norman M. D., Pearson N. J., Sharma A. and Griffin W. L. (1996) Quantitative analysis of trace elements in geological materials by laser ablation ICPMS: instrumental operating conditions and calibration values of NIST glasses. Geostand. Newslett.-J. Geostand. Geoanal. 20, 247–261. O’Reilly S. Y. and Griffin W. L. (1995) Trace-element partitioning between garnet and clinopyroxene in mantle-derived pyroxenites and eclogites – P-T-X controls. Chem. Geol. 121, 105–130. O’Reilly S. Y. and Griffin W. L. (2010) The continental lithosphere-asthenosphere boundary Can we sample it? Lithos 120, 1–13.

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592 Pearson D. G., Shirey S. B., Harris J. W. and Carlson R. W. (1998) Sulphide inclusions in diamonds from the Koffiefontein kimberlite, S Africa: constraints on diamond ages and mantle ReOs systematics. Earth Planet. Sci. Lett. 160, 311–326. Pearson D. G. and Wittig N. (2008) Formation of Archaean continental lithosphere and its diamonds: the root of the problem. J. Geol. Soc. 165, 895–914. Pearson D. G. and Wittig N. (2014) The Formation and Evolution of Cratonic Mantle Lithosphere – Evidence from Mantle Xenoliths. In Treatise on Geochemistry (eds. H. Holland and K. Turekian), second ed. Elsevier, Amsterdam, pp. 255–292. Philippot P., Busigny V., Scambelluri M. and Cartigny P. (2007) Oxygen and nitrogen isotopes as tracers of fluid activities in serpentinites and metasediments during subduction. Mineral. Petrol. 91, 11–24. Pilet S., Baker M. B., Muntener O. and Stolper E. M. (2011) Monte Carlo Simulations of Metasomatic Enrichment in the Lithosphere and Implications for the Source of Alkaline Basalts. J. Petrol. 52, 1415–1442. Rapp R. P., Shimizu N. and Norman M. D. (2003) Growth of early continental crust by partial melting of eclogite. Nature 425, 605–609. Richardson S. H., Erlank A. J., Harris J. W. and Hart S. R. (1990) Eclogitic diamonds of Proterozoic age from Cretaceous kimberlites. Nature 346, 54–56. Robinson D. N., Gurney J. J. and Shee S. R. (1984) Diamond eclogite and graphite eclogite xenoliths from Orapa, Botswana. In Kimberlites, II. The Mantle and Crust-Mantle Relationships (ed. J. Kornprobst). Elsevier, Amsterdam, pp. 11–24. Rollinson H. (1997) Eclogite xenoliths in west African kimberlites as residues from Archaean granitoid crust formation. Nature 389, 173–176. Rosenthal A., Foley S. F., Pearson D. G., Nowell G. M. and Tappe S. (2009) Petrogenesis of strongly alkaline primitive volcanic rocks at the propagating tip of the western branch of the East African Rift. Earth Planet. Sci. Lett. 284, 236–248. Rosenthal A., Yaxley G. M., Green D. H., Hermann J., Kovacs I. and Spandler C. (2014) Continuous eclogite melting and variable refertilisation in upwelling heterogeneous mantle. Sci. Rep. 4, 1–6. Rudnick R. L. and Fountain D. M. (1995) Nature and composition of the continental crust – a lower crustal perspective. Rev. Geophys. 33, 267–309. Salters V. J. M. and Stracke A. (2004) Composition of the depleted mantle. Geochem. Geophys. Geosyst. 5. Sanfilippo A., Tribuzio R., Tiepolo M. and Berno D. (2015) Reactive flow as dominant evolution process in the lowermost oceanic crust: evidence from olivine of the Pineto ophiolite (Corsica). Contrib. Miner. Petrol. 170. Sarkar C., Storey C. D. and Hawkesworth C. J. (2014) Using perovskite to determine the pre-shallow level contamination magma characteristics of kimberlite. Chem. Geol. 363, 76–90. Sato K., Katsura T. and Ito E. (1997) Phase relations of natural phlogopite with and without enstatite up to 8 GPa: implication for mantle metasomatism. Earth Planet. Sci. Lett. 146, 511–526. Schmickler B., Jacob D. E. and Foley S. F. (2004) Eclogite xenoliths from the Kuruman kimberlites, South Africa: geochemical fingerprinting of deep subduction and cumulate processes. Lithos 75, 173–207. Schmidberger S. S., Simonetti A. and Francis D. (2003) Small-scale Sr isotope investigation of clinopyroxenes from peridotite xenoliths by laser ablation MC-ICP-MS – implications for mantle metasomatism. Chem. Geol. 199, 317–329. Schmidberger S. S., Simonetti A., Heaman L. M., Creaser R. A. and Whiteford S. (2007) Lu-Hf, in-situ Sr and Pb isotope and trace element systematics for mantle eclogites from the Diavik

591

diamond mine: evidence for Paleoproterozoic subduction beneath the Slave craton, Canada. Earth Planet. Sci. Lett. 254, 55–68. Schmidt M. W. and Poli S. (2014) Devolatilization during subduction. In Treatise on Geochemistry (eds. H. D. Holland, K. K. Turekian and R. L. Rudnick). Elsevier Ltd, pp. 669– 701. Shirey S. B., Carlson R. W., Richardson S. H., Menzies A., Gurney J. J., Pearson D. G., Harris J. W. and Wiechert U. (2001) Archean emplacement of eclogitic components into the lithospheric mantle during formation of the Kaapvaal Craton. Geophys. Res. Lett. 28, 2509–2512. Shirey S. B., Harris J. W., Richardson S. H., Fouch M. J., James D. E., Cartigny P., Deines P. and Viljoen F. (2002) Diamond genesis, seismic structure, and evolution of the KaapvaalZimbabwe craton. Science 297, 1683–1686. Shirey S. B. and Richardson S. H. (2011) Start of the Wilson Cycle at 3 Ga Shown by Diamonds from Subcontinental Mantle. Science 333, 434–436. Shirey S. B., Richardson S. H. and Harris J. W. (2008) Mesoarchean to Mesoproterozoic Re-Os ages for sulfide inclusions in Orapa diamonds and implications for KaapvaalZimbabwe craton development. In 9th International Kimberlite Conference, Frankfurt, Germany, pp. Extended Abstract No. 9IKC-A-00365, 02008. Smart K. A., Heaman L. M., Chacko T., Simonetti A., Kopylova M., Mah D. and Daniels D. (2009) The origin of high-MgO diamond eclogites from the Jericho Kimberlite, Canada. Earth Planet. Sci. Lett. 284, 527–537. Smart K. A., Chacko T., Stachel T., Tappe S., Stern R. A., Ickert R. B. and EIMF (2012) Eclogite formation beneath the northern Slave craton constrained by diamond inclusions: oceanic lithosphere origin without a crustal signature. Earth Planet. Sci. Lett. 319, 165–177. Smart K. A., Chacko T., Simonetti A., Sharp Z. D. and Heaman L. M. (2014) A Record of Paleoproterozoic Subduction Preserved in the Northern Slave Cratonic Mantle: Sr-Pb-O Isotope and Trace-element Investigations of Eclogite Xenoliths from the Jericho and Muskox Kimberlites. J. Petrol. 55, 549–583. Smart K. A., Tappe S., Stern R. A., Webb S. J. and Ashwall L. D. (2016) Early Archaean tectonics and mantle redox recorded in Witwatersrand diamonds. Nat. Geosci. 9, U255–U296. Smart K. A., Tappe A., Simonetti S. S., Simonetti A. B., Woodland A. B. and Harris C. (2017) Tectonic significance and redox state of Paleoproterozoic eclogite and pyroxenite components in the Slave cratonic mantle lithosphere, Voyageur kimberlite, Arctic Canada. Chem. Geol. 455, 97–118. Smit K. V., Stachel T., Creaser R. A., Ickert R. B., DuFrane S. A., Stern R. A. and Seller M. (2014) Origin of eclogite and pyroxenite xenoliths from the Victor kimberlite, Canada, and implications for Superior craton formation. Geochim. Cosmochim. Acta 125, 308–337. Sobolev A. V., Hofmann A. W., Kuzmin D. V., Yaxley G. M., Arndt N. T., Chung S.-L., Danyushevsky L. V., Elliott T., Frey F. A., Garcia M. O., Gurenko A. A., Kamenetsky V. S., Kerr A. C., Krivolutskaya N. A., Matvienkov V. V., Nikogosian I. K., Rocholl A., Sigurdsson I. A., Sushchevskaya N. M. and Teklay M. (2007) The amount of recycled crust in sources of mantle-derived melts. Science 316, 412–417. Stachel T. and Luth R. W. (2015) Diamond formation – where, when and how? Lithos 220, 200–220. Stachel T., Viljoen K. S., McDade P. and Harris J. W. (2004) Diamondiferous lithospheric roots along the western margin of the Kalahari Craton – the peridotitic inclusion suite in diamonds from Orapa and Jwaneng. Contrib. Miner. Petrol. 147, 32–47.

592

S. Aulbach et al. / Geochimica et Cosmochimica Acta 213 (2017) 574–592

Stagno V., Ojwang D. O., McCammon C. A. and Frost D. J. (2013) The oxidation state of the mantle and the extraction of carbon from Earth’s interior. Nature 493, 84–88. Stiefenhofer J., Viljoen K. S. and Marsh J. S. (1997) Petrology and geochemistry of peridotite xenoliths from the Letlhakane kimberlites, Botswana. Contrib. Miner. Petrol. 127, 147–158. Sun, S.-S., McDonough, W.F., 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle compositions and processes. In: Geological Society Special Publication, (eds.) N.M. Saunders, A.D. Norris, pp. 313–345. Tang M., Chen K. and Rudnick R. L. (2016) Archean upper crust transition from mafic to felsic marks the onset of plate tectonics. Science 351, 372–375. Tappe S., Foley S. F., Stracke A., Romer R. L., Kjarsgaard B. A., Heaman L. M. and Joyce N. (2007) Craton reactivation on the Labrador Sea margins: 40Ar/39Ar age and Sr-Nd-Hf-Pb isotope constraints from alkaline and carbonatite intrusives. Earth Planet. Sci. Lett. 256, 433–454. Tappe S., Smart K. A., Pearson D. G., Steenfelt A. and Simonetti A. (2011) Craton formation in Late Archean subduction zones revealed by first Greenland eclogites. Geology 39, 1103–1106. Tappe S., Steenfelt A. and Nielsen T. (2012) Asthenospheric source of Neoproterozoic and Mesozoic kimberlites from the North Atlantic craton, West Greenland: new high-precision U-Pb and Sr-Nd isotope data on perovskite. Chem. Geol. 320, 113–127. Timmerman S., Koornneef J. M., Chinn I. L. and Davies G. R. (2017) Dated eclogitic diamond growth zones reveal variable recycling of crustal carbon through time. Earth Planet. Sci. Lett. 463, 178–188. Tuff J., Takahashi E. and Gibson S. A. (2005) Experimental constraints on the role of garnet pyroxenite in the genesis of high-Fe mantle plume derived melts. J. Petrol. 46, 2023–2058. Veizer J. (1989) Strontium isotopes in seawater through time. Annu. Rev. Earth Planet. Sci. 17, 141–167. Viljoen K. S. (1995) Graphite-bearing and diamond-bearing eclogite xenoliths from the Bellsbank kimberlites, northern Cape, South Africa. Contrib. Miner. Petrol. 121. Viljoen K. S., Smith C. B. and Sharp Z. D. (1996) Stable and radiogenic isotope study of eclogite xenoliths from the Orapa kimberlite, Botswana. Chem. Geol. 131, 235–255.

Viljoen K. S., Schulze D. J. and Quadling A. G. (2005) Contrasting group I and group II eclogite xenolith petrogenesis: petrological, trace element and isotopic evidence from eclogite, garnetwebsterite and alkremite xenoliths in the Kaalvallei kimberlite, South Africa. J. Petrol. 46, 2059–2090. Viljoen F., Dobbe R. and Smit B. (2009) Geochemical processes in peridotite xenoliths from the Premier diamond mine, South Africa: evidence for the depletion and refertilisation of subcratonic lithosphere. Lithos 112, 1133–1142. Waters F. G. (1987) A suggested origin of MARID xenoliths in kimberlites by high-pressure crystallisation of an ultrapotassic rock such as lamproite. Contrib. Miner. Petrol. 95, 523– 533. Weiss Y., Kiflawi I., Davies N. and Navon O. (2014) High-density fluids and the growth of monocrystalline diamonds. Geochim. Cosmochim. Acta 141, 145–159. Wiechert U., Fiebig J., Przybilla R., Xiao Y. and Hoefs J. (2002) Excimer laser isotope-ratio-monitoring mass spectrometry for in situ oxygen isotope analysis. Chem. Geol. 182, 179–194. Wilson A. H. (2012) A Chill Sequence to the Bushveld Complex: insight into the first stage of emplacement and implications for the parental magmas. J. Petrol. 53, 1123–1168. Workman R. K. and Hart S. R. (2005) Major and trace element composition of the depleted MORB mantle (DMM). Earth Planet. Sci. Lett. 231, 53–72. Wyllie P. J. and Sekine T. (1982) The formation of mantle phlogopite in subduction zone hybridisation. Contrib. Miner. Petrol. 79, 375–380. Zhai M., Kampunzu A. B., Modisi M. P. and Bagai Z. (2006) Sr and Nd isotope systematics of Francistown plutonic rocks, Botswana: implications for Neoarchaean crustal evolution of the Zimbabwe craton. Int. J. Earth Sci. 95, 355–369. Zimmer M., Kroner A., Jochum K. P., Reischmann T. and Todt W. (1995) The Gabal-Gerf complex – a Precambrian N-MORB ophiolite in the Nubian Shield, NE Africa. Chem. Geol. 123, 29– 51. Associate editor: Shichun Huang