Sulfides and chalcophile elements in Roberts Victor eclogites: Unravelling a sulfide-rich metasomatic event

Chemical Geology 354 (2013) 73–92

Contents lists available at ScienceDirect

Chemical Geology journal homepage: www.elsevier.com/locate/chemgeo

Sulfides and chalcophile elements in Roberts Victor eclogites: Unravelling a sulfide-rich metasomatic event Yoann Gréau a,b,⁎, Olivier Alard a,b, William L. Griffin a, Jin-Xiang Huang a,c, Suzanne Y. O'Reilly a a b c

ARC Centre of Excellence for Core to Crust Fluid Systems and GEMOC National Key Centre, Dept of Earth and Planetary Sciences, Macquarie University, NSW 2109, Australia Géosciences Montpellier, CNRS UMR-5243, Equipe Manteau et Interfaces, Université Montpellier II, France State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, China Academy of Sciences, Beijing, China

a r t i c l e

i n f o

Article history: Received 6 February 2012 Received in revised form 14 May 2013 Accepted 13 June 2013 Available online 21 June 2013 Editor: L. Reisberg Keywords: Eclogite xenoliths Sulfides Mantle metasomatism Sub-continental lithospheric mantle Chalcophile elements

a b s t r a c t A suite of eclogite xenoliths from the Roberts Victor kimberlite (South Africa) is remarkable for its high abundances of base metal sulfides (BMS; up to 2 modal %). However, while sulfides are nearly ubiquitous in Type I eclogites (garnet N 0.07% Na2O), Type II eclogites (garnet b 0.07% Na2O) systematically lack sulfides. Two different sulfide assemblages are recognised within the Type I xenolith suite. Both populations are polyphase Cu–Ni–Fe sulfides, one characterised by the pyrrhotite (Po) + pentlandite (Pn) + chalcopyrite (Cp) assemblage and the other by the smythite/violarite (Smy/Vi) + (Ni)pyrite (Py) + Cp assemblage. The latter is the most abundant assemblage and reflects the supergene alteration of the “primary” Po + Pn + Cp assemblage. This process overprints the original composition of the sulfides by remobilising elements such as S, Fe, Ni, Se and Te. In the Type I eclogites, BMS occur as inclusions within silicates (garnet and clinopyroxene) and as interstitial grains. No chemical differences were observed between enclosed and interstitial sulfides. However, their relative abundances are correlated, indicating a similar origin. The Po + Pn + Cp assemblage is identical to eclogitic sulfides previously described in some Roberts Victor diamonds. Silicate-enclosed sulfides could appear to be early phases, but they are restricted to the outer parts of the silicate grains, suggesting a late incorporation during partial recrystallisation of silicate phases, induced by fluid-rock percolation-reaction during a metasomatic event. Positive whole-rock correlations between Se and (La/Sm)N⁎ and between Cu and ΣLREEN⁎ indicate a direct link between sulfide content and the enrichment of Mg + incompatible elements in the Type I eclogites, further supporting a metasomatic origin of the sulfide component. Similarly, a negative correlation between ΣLREEN⁎ and Cucpx implies that the metasomatic agent was at least partially composed of S-rich fluid (e.g. H2S, SO2) that reacted with the rock and leached Cu out of the silicates. The coupling between sulfides and LREE in the Type I eclogites, as well as the absence of sulfides and metasomatic features (e.g. unequilibrated grain boundaries, melt pockets, fluid inclusions, phlogopite) in the Type II eclogites, demonstrate that Type II eclogites did not undergo such a metasomatic event and therefore may represent the less-modified protoliths of Type I eclogites. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Eclogite xenoliths brought to the surface by the Roberts Victor kimberlite (South Africa) comprise the most extensively studied suite of mantle-derived eclogites in the world (e.g. MacGregor and Carter, 1970; Garlick et al., 1971; Manton and Tatsumoto, 1971; Harte and Gurney, 1975; Ozima and Saito, 1975; Hatton and Gurney, 1977; Kramers, 1979; MacGregor and Manton, 1986; Ongley et al., 1987; Sautter and Harte, 1988; Viljoen et al., 1991; Jacob et al., 2002, 2005; Huang et al., 2010; Gréau et al., 2011). For over forty years, work on their petrology and chemistry has added successive layers of detail, and these data have been used to support widely differing

⁎ Corresponding author at: ARC Centre of Excellence for Core to Crust Fluid Systems and GEMOC National Key Centre, Dept of Earth and Planetary Sciences, Macquarie University, NSW 2109, Australia. E-mail address: [email protected] (Y. Gréau). 0009-2541/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chemgeo.2013.06.015

interpretations. The main alternative hypotheses are that mantle eclogites either are rocks produced by magmatic processes in the mantle (e.g. Smyth et al., 1989; Caporuscio and Smyth, 1990; Griffin and O'Reilly, 2007) or are fragments of subducted oceanic slabs (e.g. Anderson, 1982; MacGregor and Manton, 1986; Schulze and Helmstaedt, 1988; Jacob et al., 1994, 2002). These ideas have been debated for 30 years, and the origin of these rocks is still not well understood. Recently, some studies have shown that numerous mantle eclogites have witnessed important metasomatism (Jacob et al., 2009; Smart et al., 2009; Gréau et al., 2011; Huang et al., 2012), which may have overprinted the original microstructural and geochemical characteristics of some of these rocks, blurring away essential features that would be critical to constrain the origin of mantle eclogites. In this regard, even the extensively studied Roberts Victor eclogite suite, from which have been elaborated most of the hypotheses proposed in the literature, has been shown to be intensively microstructurally, chemically and isotopically overprinted (Gréau et al., 2011; Huang et al., 2012).

74

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

However, these studies have also identified specimens apparently free of such overprinting and which would be better candidates to (re-)evaluate the nature of mantle eclogites. A complete understanding of Earth's evolution requires that we constrain the origins of all the components of the heterogeneous lithospheric mantle, and mantle-derived eclogites are one of these significant components. Although eclogite xenoliths were one of the first mantle lithologies in which the sulfide component was carefully described (Czamanske and Desborough, 1968; Desborough and Czamanske, 1973; Frick, 1973; Tsai et al., 1979) our chemical knowledge of the chemistry of the sulfide phases in eclogites remains limited. This early interest probably reflected the high abundance of sulfide in some xenoliths and the significant size of some sulfide grains (N500 μm). Since then sulfide minerals have been systematically described in peridotites from many geological contexts but eclogite xenoliths have not benefited from the same level of interest. In the last decades progress has been made in understanding the significance of sulfide phases in Earth's mantle, and in using sulfide chemistry to constrain petrological processes (e.g. Dromgoole and Pasteris, 1986; Lorand, 1989a, 1989b, 1989c; Szabo and Bodnar, 1995; Guo et al., 1999). It is now clear that the sulfides are sensitive to mantle processes such as melting and metasomatism (Alard et al., 2000; Lorand and Gregoire, 2006; Lorand et al., 2008; Lorand and Alard, 2010; Alard et al., 2011), but also to crustal contamination (Lorand and Alard, 2010, 2011). Here we document the mineralogical and geochemical features of base metal sulfides (BMS) in the Roberts Victor eclogites and present

new whole-rock data for sulfide-hosted elements such as S, Se, Te, Cu and Ni (chalcophile elements). Although this study is focused primarily on the origin and the petrogenesis of the BMS assemblage in the Roberts Victor xenolith suite, the data also bring important information on the history of these eclogites, and shed some light on the origin of the diamonds they contain. 2. Geological setting and samples The Group II kimberlite pipe in the Roberts Victor mine (South Africa: 25° 34′E, 28° 27′S) is one of many kimberlite intrusions in the Kaapvaal craton. Rb–Sr dating of mica indicates an eruption age of 128 ± 15 Ma (Smith et al., 1985). This kimberlite has been extensively studied, not only because of its diamond yield but also because between 80 and 98% of the abundant, large xenoliths that it carries are eclogites (MacGregor and Carter, 1970; Hatton, 1978). We have studied 29 eclogite xenoliths collected in 2006 at the Roberts Victor mine and 9 more samples from an older GEMOC collection. The sample suite includes 28 Type I eclogites, and 10 Type II eclogites, as defined by McCandless and Gurney (1989). Using the Na2O contents in garnet and the K2O contents in clinopyroxene, McCandless and Gurney recognised that the two Roberts Victor eclogite groups defined by MacGregor and Carter (1970) on the basis of their microstructures, were also two chemically distinct groups of eclogites. In this chemical scheme, eclogites that have Na2O contents in garnet N 0.07 wt.% and/or K2O contents in clinopyroxene N 0.08 wt.% belong to Type I,

Table 1 Sulfide abundances and assemblages of Roberts Victor eclogites. Sample

Host rock

N sulf

e-Gt

e-cpx

i

Po (mss1)

Pn

Cp

Smy/Vi

Ni–Py

Py

FeOOH

RV07-1 RV07-2 RV07-3 RV07-5a RV07-5b RV07-7 RV07-8 RV07-9a RV07-9b RV07-10 RV07-11 RV07-12 RV07-14 RV07-16 RV07-17 RV07-18 RV07-19 RV07-20 RV07-23a RV07-23b RV07-24 RV07-25 RV07-26 RV07-30 RV07-31 RV07-33 RV07-34 RV07-36 RV07-37 Rva RVb RV5 RV6 BD1191 BD3699 RV-1G RV-2G RV73-12

Eclogite-Type I Eclogite-Type I Eclogite-Type I Kyanite eclogite-Type I Kyanite eclogite-Type I Eclogite-Type I Eclogite-Type II Phlogopite eclogite-Type I Phlogopite eclogite-Type I Eclogite-Type I Eclogite-Type I Eclogite-Type II Eclogite-Type I Eclogite-Type I Eclogite-Type I Eclogite-Type I Eclogite-Type I Eclogite-Type I Kyanite eclogite-Type I Kyanite eclogite-Type I Eclogite-Type I Kyanite eclogite-Type I Kyanite eclogite-Type I Eclogite-Type II Eclogite-Type II Eclogite-Type II Eclogite-Type II Eclogite-Type II Eclogite-Type II Kyanite eclogite-Type I Kyanite eclogite-Type I Kyanite eclogite-Type I Eclogite-Type I Eclogite-Type II Kyanite eclogite-Type I Eclogite-Type I Eclogite-Type I Eclogite-Type II

4 1 2 1 1 0 0 14 23 0 19 0 28 10 14 37 0 23 3 2 9 0 9 0 0 0 0 0 0 7 37 1 36 0 5 117 9 0

1 − 1 − − − − 5 − − 1 − 9 − − 2 − 5 − − 5 − 3 − − − − − − 5 10 − − − 1 14 − −

1 1 − − − − − 2 8 − 2 − 5 1 5 17 − 18 − − − − 3 − − − − − − − 4 − − − − 32 7 −

2 − 1 1 1 − − 7 15 − 16 − 14 9 9 18 − − 3 2 4 − 3 − − − − − − 2 23 1 36 − 4 71 2 −

− − − + − − − − − − +++ −

+ − + + − − − −

+ − + + + − − ++ ++ − +++ − ++ − ++ ++ − ++ − − ++ − + − − − − − − +++ +++ ++ ++ − ++ ++ ++ −

− − − − + − − − ++ − + − ++ − +++ +++ − − − − ++ − − − − − − − − +++ +++ ++ − − + − + −

++ − ++ − + − − +++ +++ − − − +++ ++ +++ +++ − + − − +++ − ++ − − − − − − + ++ − +++ − +++ +++ +++ −

− − ++ + + − − ++ − − − ++

++ +++ + − − − − ++ − +++ − − − + + − − +++ − − ++ + + − − − − − − − − − − − − + + −

− − − − +++ ++ − − − − − − − − − − +++ +++ − − − + − − −

− +++ − − − − − − − − − − − − − − − − − − +++ ++ − ++ − − − − −

+++ + − − − + ++ − ++ − − − − − − ++ ++ +++ +++ − +++ +++ +++ −

N sulf = number of sulfide per thin section, e-Gt = garnet enclosed sulfide, e-Cpx = clinopyroxene enclosed sulfide, I = interstitial sulfide. Po = pyrrhotite, mss1 = mono-sulfide solution 1, Pn = pentlandite, Cp = chalcopyrite, Smy/V = smythite/violarite, Ni–Py = nickel rich pyrite, Py = Pyrite, FeOOH = iron (oxy)hydroxide. ‘−’ = absent or not observed, ‘+’ = rare, ‘++’ = present, ‘+++’ = very abundant.

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

whereas Type II eclogites have Na2O contents in garnet ≤ 0.07 wt.% AND K2O contents in clinopyroxene ≤ 0.08 wt.%. Moreover, recent studies (Gréau et al., 2011; Huang et al., 2012) have shown that this chemical dichotomy is not restricted to major elements, but also applies to their trace elements (e.g. HFSE, LREE, Th, U, Cu), oxygen isotopes and radiogenic isotopes (e.g. Lu/Hf, Sm/Nd). The samples are mainly bimineralic eclogites with modal proportions of garnet:omphacite varying from 70:30 to 40:60. They are coarse-

a

Cpx

75

grained (1–5 mm) and show a wide range of microstructures (e.g. layering, patchy accumulations of garnet). Several samples show microstructural disequilibrium between garnet and clinopyroxene, defined by irregular grain boundaries (high-energy surfaces). Most of the rocks also show partial to complete destabilisation of the primary omphacite into a spongy heterogeneous secondary cpx, which is Na-depleted (Na2O b 1.41 wt.%) relative to the primary one (e.g. Na2O N 3.91). This spongy microstructure is mainly observed on the rims of cpx grains and

b

expansion cracks Sulf

Sulf Grt

spongy Cpx

Ccp 200µm

500µm

c Grt

Ccp

Sulf

d Grt

Ccp

Sulf

Vi Py 200µm

500µm

e

f

Ccp Grt

Grt

Sulf

Po+Pn Sulf Cpx

g

200µm

Ccp

500µm

h

Sulf

Grt

Po Cpx Pn Cpx 50µm

500µm

Fig. 1. Back-scattered electron images of sulfides in Roberts Victor eclogites. a Clinopyroxene-enclosed sulfide surrounded by secondary spongy clinopyroxene. b Polygonal sulfide enclosed in garnet; note the expansion cracks radiating from the sulfide grain. c Typical Roberts Victor eclogite sulfide displaying strongly irregular outline; note the millimetric size of the grain. d garnet-enclosed sulfide; note the violarite (light grey) along fracture; the chalcopyrite corona on the edge of the grain (white); euhedral pyrite crystals set in a uneven matrix of smythite/violarite + pyrite. e Interstitial sulfide at garnet–clinopyroxene grain boundary. f Fresh pyrrhotite grain (RV07-11) with chalcopyrite patches (white) bordering the grain and small oriented pentlandite flames exsolving from the pyrrhotite matrix. g Enlargement of pentlandite flames within the pyrrhotite matrix. h Sulfide grain enclosed in the outer part of a garnet grain; note the intricate relationship between garnet and clinopyroxene showing intensive recrystallisation. Sulf: sulfide; Cpx: clinopyroxene; Grt: garnet; Ccp: chalcopyrite; Vi: violarite; Py: pyrite; Po: pyrrhotite; Pn: pentlandite.

76

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

along cracks or cleavages. These features can be explained either by metasomatism (Jin and Taylor, 1988; Taylor and Neal, 1989; Wulff-Pedersen et al., 1996) or by in situ melting of the cpx resulting from decompression and heating during the ascent of the host kimberlite (Donaldson, 1978; Pearson et al., 1995). Garnet is usually fresher than the cpx and is in general chemically homogeneous at the thin-section scale. However, some specimens (e.g. ky-bearing xenoliths) show variable garnet composition within a single thin section. Accessory minerals such as phlogopite, rutile, amphibole, feldspar, spinel, barite, calcite and sulfide are also commonly present. Nevertheless, minerals like kyanite or phlogopite can sometimes constitute up to several modal % of the rock. When this is the case the terminologies kyanite eclogite or phlogopite eclogite have been used. However, this does not affect the primary classification and all our kyanite and phlogopite eclogites belong to Type I eclogites as defined previously. This is especially true for phlogopite-bearing eclogites, which are macroscopically and microscopically identical to “normal” Type I eclogites aside from the local abundance of phlogopite (modal metasomatism in fluid pathways?). The case of the kyanite-bearing samples is slightly different because when kyanite is present it is always associated with a change in garnet composition (strong increase of grossular content) and the breakdown of clinopyroxene into a sub-micrometric white turbid matrix presumably composed of augite and albite or an analcime-like zeolite (Berg, 1968; Chinner and Cornell, 1974). Although this transformation

a

produces decimetric layers/patches of kyanite-bearing eclogite in contact with typical Type I eclogite, the outlines of a Type-I microstructure can still be seen within the kyanite-rich zones. T estimates for these samples calculated by Gréau et al. (2011) show a narrow range, with Type I eclogites ranging between 1079 and 1243 °C; depths (estimated by projection to a xenolith-defined geotherm) lie between 181and 194 km (~6 GPa). The range of temperatures obtained for Type II eclogites is slightly wider, between 869 and 1305 °C, corresponding to depths between 145 and 199 km. These values are consistent with those reported by Griffin and O'Reilly (2007) and indicate that the Roberts Victor eclogites were emplaced and equilibrated close to the base of the sub-continental lithospheric mantle (SCLM). 3. Analytical techniques Major-element contents of sulfides were analysed using a CAMECA Camebax SX100 electron microprobe at GEMOC (Macquarie University, Australia) fitted with five crystal spectrometers. Analyses were done with an accelerating voltage of 15 kV and a 20 nA beam current. Counting times were 10 s for peaks and 5 s for backgrounds on either side of the peak and corrections were done by the method of Pouchou and Pichoir (1984). Elements analysed are S, Fe, Cu, Ni, Co, Mg, Si, K and O. Internal calibrations for these elements were done using different

b spongy Cpx

Phl Sulf

Rt

Cpx

500µm

200µm

c

d

Crack FeOOH

Grt

Sulf spongy Cpx

Shielded sulfide Cpx 50µm

200µm

e

f

Cpx sulfide strings

spongy Cpx

FeOOH

spongy Cpx Cpx Sulf + FeOOH 500µm

500µm

Fig. 2. Back-scattered electron images and reflected-light images of sulfides in Roberts Victor eclogites. a Sulfide grain associated with a phlogopite patch in phlogopite-bearing eclogite. b Clinopyroxene-enclosed rutile surrounded by secondary spongy clinopyroxene. c Sulfide associated with highly hydrated sheeted silicate. d Sulfide bleb shielded by garnet and oxidised open inclusions (RV07-20). d Oxidised sulfide showing replacement by FeOOH and sulfide strings radiating out from the main grain. f Same as e except that all sulfide (grain and strings) has been replaced by FeOOH. Rt: rutile; Phl: phlogopite; FeOOH: iron-(oxy)hydroxide.

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

examined, 24 have sulfide, 2 contain (only) iron-(oxy)hydroxide after sulfides, and 2 lack both sulfides and iron-(oxy)hydroxide. In the 24 sulfide-bearing samples, the number of individual sulfide grains (excluding sulfide veins, see below) has been estimated by optical grain counting (diameter ≥ 20 μm) on polished thin sections (400 by 250 mm; 2 per sample) and varies between 1 and 117, with a mean abundance of 7 grains per thin section (Table 1). The sulfide occurrences can be divided on microstructural criteria into three types.

4.1. Enclosed sulfides Sulfide inclusions in silicate minerals are mostly rounded blebs (Fig. 1-a) or subhedral polygons (Fig. 1-b) with sizes varying from 30 to 350 μm (Fig. 1-c). They are found in both garnet and omphacite, and typically occur in the outer part of each grain but rarely in the core; this is especially true of garnet. Around some enclosed grains, particularly in garnet, cracks radiate from the sulfide (Fig. 1-b, d), probably reflecting melting and accompanying volume expansion of the sulfide during the ascent of the eclogite in the kimberlite. Sulfide inclusions are more abundant in the clinopyroxene, where they are usually surrounded by the destabilised cpx (Fig. 1-a).

4.2. Interstitial sulfides These are commonly anhedral with curved or elongated shapes (Fig. 1-e), and the grain sizes vary from 20 μm to 3 mm. They occur between the silicate grains and may be in close spatial relationship with rutile, or with phlogopite in phlogopite-rich eclogites (Fig. 2-a).

a

14

N samples

12 10 8 6 4 2 0 0

20

40

60

80

100

120

140

N sulfides/sample 80

b

70

N interstitial sulfides

standards. Ni and Co were calibrated using pure Ni and Co metal, while minerals were used in the other cases: pyrite for Fe and S, chalcopyrite for Cu, olivine for Mg, CaSiO3 for Si, andradite for O and orthoclase for K. Each of the sulfide phases making up individual sulfide grains were analysed; analyses showing high Si or Mg contents were considered as mixed silicate–sulfide analyses and were disregarded. Highly oxidised sulfides displaying high O contents were also disregarded. Bulk sulfide compositions were reconstructed using the compositions and modal abundances of the phases within each sulfide grain; modal abundances have been obtained by image processing (ImageJ© 1.38 software) of X-ray element (S, Fe, Ni, Cu, Si, O) distribution maps obtained in energy-dispersive mode on the CAMECA Camebax SX100. Major- and trace-element compositions of whole-rock samples were analysed by Geosciences Australia (Canberra). Major-element concentrations (and sulfur contents) were obtained by XRF analysis of fused discs using a Philips PW 2404 XRF spectrometer. The whole-rock sulfur contents were also determined at Geosciences Montpellier (Université de Montpellier II, France) using the hightemperature iodo-titration (HTIT) technique described by Gros et al. (2005), and by infrared absorption techniques (IRC) at Geo Labs Ontario. The latter has been used to check the values obtained with the two previous techniques. XRF and IRC data correlate very well for S values N 500 ppm, while IRC data fit better with HTIT data for S values b500 ppm. As a result, we use S* as whole rock sulfur content, where S* = average [(SIRC + SXRF)S N 500 ppm or (SIRC + SHTIT)S b 500 ppm]. Concentrations obtained with the three different techniques are available as electronic supplementary data (Supplementary Table A). Whole-rock concentrations of selenium and tellurium were analysed at Geo Labs Ontario. Samples were digested in mixed acids (hydrofluoric, hydrochloric and perchloric) in open vessels before Se and Te were separated by hydride generation and analysed by ICP-MS. Lower detection limits were 3.5 ppb for Se and 1 ppb for Te (www.mndmf.gov.on.ca/mines/ogs/labs). Lorand and Alard (2010) estimated the precision of the method for mantle rocks as being better than 2% for Se and of 4–9% for Te (samples also analysed by Geo Labs Ontario). Replicates of 2 of our samples give a similar precision for Se (agreement within 2%), but precision on Te was more variable (0–33%) as the two samples duplicated were close to the 1 ppb detection limit. Moreover, König et al. (2012) showed that Te concentrations were only representative of the sample aliquot analysed and not of the entire sample. The proposed explanation was that as a semi-metal Te tends to partition between BMS and micro tellurides phases, while Se as a chalcogen non-metal will replace S in the crystalline structure of BMS. Therefore, Te concentrations are highly sensitive to the presence or not of such microphases in the aliquot, and this is especially more susceptible to generate poor reproducibility for sample having low Te concentrations. Transition-metal concentrations in garnet and clinopyroxene have been measured in situ using the LA-ICPMS facilities at GEMOC. Ablations were carried out in a pure He atmosphere with a New Wave UP266 nm Nd:YAG laser microprobe coupled to an Agilent 7500 ICPMS. Operating conditions included a 60 μm beam diameter with a pulse frequency of 5 Hz and laser intensity around 0.150 mJ. Calcium was used as internal standard and the NIST610 glass as external standard. The ablation sequence lasted 260 s with a background segment of 120 s, and a wash-out time of 5 min was observed between analyses. The raw data were processed on-line using the software package GLITTER© (Griffin et al., 2008; www.gemoc.mq.edu.au). International standard BCR-2G and the in-house garnet standard MONGOL were analysed before and after each run.

77

RV eclogite sulfide

60 50 40 30 20 10 0

4. Sulfide petrography Most of the Type I eclogite xenoliths from Roberts Victor carry sulfides, with modal abundances up to ca 2%. However, all the Type II eclogites studied are sulfide-free. Among the 28 Type I eclogites

0

5

10

15

20

25

30

35

40

45

50

N enclosed sulfides Fig. 3. Modal distribution of sulfides in Roberts Victor eclogites. a Number of sulfides per sample. b Number of interstitial sulfides per sample versus number of enclosed sulfides per sample.

78

Table 2 Major elements representative analyses of Roberts Victor eclogites sulfides. Sulfide phase

Poh

Microstructural position

Enclosed

n=

Pom Interstitial

47

Pn

Enclosed

25

Interstitial

39

Ccp

Enclosed

17

Interstitial

2

Enclosed

12

Interstitial

111

119

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.%

0.28 38.28 57.19 0.07 0.11 2.20 0.00 0.02 0.00 98.15

0.08 0.66 1.32 0.08 0.13 1.38 0.01 0.04 0.00 0.71

0.25 38.24 57.21 0.05 0.06 2.23 0.00 0.00 0.00 98.06

0.09 0.51 1.64 0.04 0.09 1.38 0.00 0.01 0.00 0.78

0.25 39.62 56.30 0.29 0.04 2.03 0.00 0.02 0.00 98.55

0.15 0.56 2.04 0.98 0.09 0.88 0.01 0.03 0.00 0.74

0.15 39.23 57.36 0.06 0.01 1.87 0.00 0.01 0.00 98.68

0.06 0.40 1.62 0.03 0.04 1.11 0.00 0.02 0.00 0.61

0.32 33.53 38.07 0.11 0.67 26.66 0.00 0.01 0.00 99.37

0.08 0.13 0.21 0.09 0.22 0.41 0.00 0.01 0.00 0.71

0.79 33.15 30.31 1.08 1.15 32.82 0.03 0.10 0.00 99.42

1.89 0.32 5.04 2.67 0.34 4.21 0.09 0.32 0.01 1.45

0.24 34.18 29.74 34.17 0.02 0.24 0.01 0.03 0.01 98.65

0.26 0.62 0.79 0.71 0.09 0.36 0.03 0.07 0.01 1.05

0.25 34.17 29.84 34.05 0.02 0.33 0.00 0.02 0.00 98.69

0.24 0.63 0.57 1.21 0.05 0.69 0.02 0.05 0.01 0.73

O S Fe Cu Co Ni

at.% at.% at.% at.% at.% at.% M/S

0.77 52.46 45.00 0.05 0.08 1.65 0.89

0.21 0.91 1.04 0.06 0.10 1.03

0.70 52.47 45.07 0.04 0.05 1.67 0.89

0.25 0.70 1.30 0.03 0.07 1.03

0.67 53.74 43.85 0.20 0.03 1.51 0.85

0.41 0.75 1.59 0.67 0.07 0.65

0.41 53.37 44.79 0.04 0.01 1.39 0.87

0.15 0.54 1.26 0.02 0.03 0.83

0.90 47.22 30.79 0.08 0.51 20.50 1.10

0.24 0.18 0.17 0.07 0.17 0.32

2.22 46.54 24.43 0.76 0.88 25.17 1.10

5.32 0.45 4.06 1.89 0.26 3.23

0.71 49.44 24.70 24.94 0.02 0.19 1.01

0.75 0.89 0.66 0.52 0.07 0.28

0.72 49.40 24.77 24.84 0.01 0.26 1.01

0.68 0.92 0.47 0.88 0.04 0.55

Sulfide phase

Py

Microstructural position

Enclosed

n=

Ni-Py Interstitial

31

Smy

Enclosed

34

Interstitial

62

Vi

Enclosed

63

Interstitial

5

Enclosed

2

Interstitial

2

4

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

Avg.

s.d.

O S Fe Cu Co Ni Mg Si K Total

wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.%

0.44 51.59 46.36 0.64 0.02 0.30 0.01 0.08 0.01 99.43

0.26 1.02 0.79 0.68 0.05 0.25 0.02 0.13 0.01 1.10

0.48 51.20 46.30 0.44 0.01 0.32 0.02 0.05 0.01 98.84

0.23 0.85 0.54 0.43 0.03 0.21 0.03 0.07 0.01 1.15

0.57 50.84 43.10 0.14 0.38 3.67 0.02 0.05 0.00 98.77

0.42 1.12 1.94 0.08 0.82 1.92 0.05 0.10 0.00 1.25

0.62 50.61 43.47 0.14 0.19 3.19 0.02 0.08 0.01 98.34

0.23 0.77 1.75 0.13 0.19 1.82 0.03 0.09 0.01 1.00

0.32 41.07 37.94 0.08 0.34 20.58 0.00 0.00 0.00 100.35

0.23 0.41 3.78 0.07 0.31 3.84 0.00 0.01 0.00 0.33

0.46 40.96 40.97 0.01 0.42 16.88 0.00 0.01 0.02 99.71

0.01 0.09 0.94 0.01 0.11 0.19 0.00 0.00 0.01 0.58

0.76 40.35 21.03 0.46 0.91 34.25 0.10 0.28 0.01 98.15

0.20 0.17 0.45 0.18 0.76 0.31 0.10 0.11 0.00 0.05

0.29 41.12 19.03 1.49 0.33 37.37 0.00 0.01 0.01 99.64

0.21 0.27 2.01 1.92 0.19 3.24 0.00 0.01 0.01 0.19

O S Fe Cu Co Ni

at.% at.% at.% at.% at.% at.% M/S

1.11 64.82 33.45 0.40 0.01 0.21 0.53

0.66 1.28 0.57 0.43 0.03 0.17

1.22 64.68 33.59 0.28 0.01 0.22 0.53

0.59 1.08 0.39 0.28 0.02 0.15

1.45 64.34 31.32 0.09 0.26 2.54 0.53

1.05 1.41 1.41 0.05 0.57 1.33

1.58 64.28 31.70 0.09 0.13 2.22 0.53

0.59 0.98 1.27 0.08 0.13 1.26

0.85 54.79 29.06 0.06 0.25 14.99 0.81

0.62 0.55 2.90 0.05 0.23 2.79

1.23 54.72 31.42 0.00 0.30 12.32 0.80

0.04 0.12 0.72 0.00 0.08 0.14

2.08 54.98 16.45 0.32 0.68 25.49 0.78

0.55 0.24 0.35 0.13 0.56 0.23

0.79 55.59 14.77 1.02 0.24 27.59 0.78

0.57 0.37 1.56 1.31 0.14 2.39

Poh = hexagonal pyrrhotite, Pom = monoclinic pyrrhotite, Pn = pentlandite, Ccp = chalcopyrite, Py = pyrite, Ni–Py = nickel-rich pyrite, Smy = smythite, Vi = violarite. n = number of electron-probe analysis, avg. = average, s.d. = standard deviation. M/S = metal/sulfur ratio with metal = Fe + Ni + Cu + Co.

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

O S Fe Cu Co Ni Mg Si K Total

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

4.3. Veins These occur as fine ribbons cross-cutting the eclogite matrix or as discrete grains between silicate minerals (images available as electronic supplement). They are composed of pyrite (FeS2) with low Ni contents (Ni b 1 wt.%). They are similar to sulfides observed in the kimberlite matrix bordering the eclogite xenoliths. The veins thus clearly have a secondary origin, and will not be discussed further here.

79

Reflected-light microscopy and backscattered-electron images show that most of these sulfides are polyphase grains. There are no obvious differences in terms of sulfide mineralogy between enclosed and interstitial sulfides. The phases observed are pyrrhotite (Po), pentlandite (Pn), chalcopyrite (Cp), pyrite and Ni-rich pyrite ((Ni)Py), “smythite/violarite” (Smy/Vi; alteration products see below). The occurrence and relative proportions of each phase vary from grain to grain and from sample to sample. 4.5. Alteration

4.4. Sulfide distribution Most of the samples contain less than 20 sulfide grains (Table 1; Fig. 3-a), which can occur in both interstitial and enclosed positions, with a predominance of interstitial over enclosed. At the scale of the xenolith suite, enclosed sulfides do not show a clear preference for garnet or clinopyroxene. Sulfides are solely interstitial in four samples and solely enclosed in two. However, there is a positive correlation between the number of enclosed sulfides and the number of interstitial sulfides (Fig. 3-b), which is consistent with a similar origin for both types. With a few exceptions that may be due to sectioning effects, this suggests a quite homogeneous distribution of sulfides at the larger scale.

a

Sulfides are altered to varying degrees. RV07-11 hosts the freshest sulfides, which show fine exsolved pentlandite flames in a pyrrhotite matrix (Fig. 1-f, -g), and chalcopyrite coronas or patches (up to ≈30%). Such relationships between Pn and Po indicate subsolidus exsolution from a previous mono-sulfide solution (mss) phase. No pyrite or Ni-rich -pyrite is present. Samples RV07-20, RV-a and RV-b also show a large proportion of such sulfides. In RV07-11, the relationships of the sulfides to the silicates are also slightly different from those in other samples. The interstitial sulfides have more curvilinear shapes and sometimes even enclose some silicates. In this sample, grains of garnet and clinopyroxene can also show highly S

S (wt.%) 60% Py

50% Fe MSS 1100°C

Ni+Co

Vi

Smy

Po

MSS 1000°C

40%

MSS 900°C

Ccp Pn

Fe (wt.%)

10%

20%

30%

40%

50%

60%

Ni+Co (wt.%)

Primary assemblage: Pyrrhotite S (wt.%) Pendlandite 60% Chalcopyrite Supergene assemblage: “Smythite/Violarite” Roberts Victor diamond 50% oxydised Smy/Vi inclusions (Deines, 1994) Pyrite Ni-rich pyrite

b Py

40%

Po Peridotitic DI

Eclogitic DI Ccp

Pn

Fe (wt.%) 10%

20%

30%

40%

50%

Ni+Co (wt.%)

60%

S

c

S (at.%)

60%

Ccp

50%

Cu

Fe

ISS 600˚C

40%

Cu (at.%) 15%

25%

35%

45%

55%

Fe (at.%)

Fig. 4. a Single electron-probe analyses of individual phases, plotted on Fe–Ni–S and Cu–Fe–S ternary diagrams. a Fe–Ni–S diagram of Fe–Ni–Cu sulfides. b Sulfide inclusions in diamonds from Roberts Victor (Deines and Harris, 1995) in the Fe–Ni–S diagram. c Cu–Fe–S diagram of chalcopyrite. Py: pyrite; Ccp: chalcopyrite; Po: pyrrhotite; Pn: pentlandite; Vi: violarite; Smy: smythite; MSS: monosulfide solid solution; DI: diamond inclusion; ISS: isobaric solid solution. Pn compositional range is from (Lorand, 1990). Open star represents smythite of hypothetical composition Fe6.75Ni2.25S11; solid star represents violarite FeNi2S11. High-temperature mss fields (Kullerud et al., 1969). Fields of eclogitic and peridotitic sulfide inclusions in diamond from Deines and Harris (1995), Bulanova et al. (1996), Thomasot et al. (2009). Single analysis data available as electronic supplement (Table B).

80

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

curvilinear shapes; clinopyroxene can be found enclosed in garnet and conversely. However, the most typical sulfide grains of this sample suite show a “cleaved” microstructure, in which finely interfingered Ni-rich/ Ni-poor pyrite and lamellar smythite/violarite (see below) coexist with chalcopyrite (Figs. 1-d, 2-a). In several cases this sulfide assemblage is finely intergrown with a silicate component, which usually makes up b 10% of the sulfide grain; in samples such as RV6, it may represent up to 80% of the sulfide volume. Microprobe analyses show that this silicate is a highly hydrated (ca 12 wt.% H2O) clinochlore-like mineral with an MgO content around 20 wt.%. Ten samples contain grains of iron-(oxy)hydroxide, which commonly are similar to the sulfides in terms of microstructures (Fig. 2-f); some of these grains have minute relics of sulfide in their cores. Sulfide grains

10 9 8 7 6 5 4 3 2 1 0

HPo

also may show iron-hydroxide rims, developed to varying degrees. Another noteworthy feature is the presence of secondary strings of sulfide radiating from some of the Fe-(oxy)hydroxide grains (Fig. 2-f, -e). For instance, RV07-20 is particularly rich in iron hydroxide grains. All the interstitial grains, as well as all sulfide inclusions with surrounding cracks, have been converted to iron hydroxides. However, tiny unfractured sulfide globules (≤70 μm) shielded in garnet and clinopyroxene are preserved from alteration (Fig. 2-d). The presence of several wt.% Ni in the cores of some of the iron hydroxide grains confirms that these are replacement products of the original sulfides. In addition, the cores of many Fe-hydroxide grains contain obvious pseudomorphs after euhedral crystals of pyrite. It seems clear that the Fe-hydroxides represent oxidised sulfide grains and that this alteration was a late event, occurring after the low-temperature re-equilibration of the sulfide assemblage.

MPo

Smy

Vi

Assemblages

RV07-11 Pyrrhotite

“Fresh” sulfides

Smythite/ Violarite

Po+Pn+Cp

(No Py)

14 12

RV07-20 “Fresh” sulfides Po+Pn+Cp

10 8

(+minor Py)

6

Violaritisation

N analyses

4 2 0

25 20

RVa-RVb “Striated”+ “Fresh”sulfides

15

Po+Pn+Cp

(Ni)Py +”Smy/Vi” +Cp

10 5 0 9 8 7 6 5 4 3 2 1 0

Other samples “Striated” only (Ni)Py +”Smy/Vi” +Cp Oxydised “Smy/Vi”

50.5

51.0

51.5

52.0

52.5

53.0

53.5

54.0

54.5

55.0

55.5

56.0

56.5

57.0

S at.% Fig. 5. Distribution of S (at.%) in pyrrhotite and effect of violaritisation on the sulfide assemblage. HPo: hexagonal pyrrhotite; MPo: monoclinic pyrrhotite; Pn: pentlandite; Ccp: chalcopyrite; Py: pyrite; Smy: smy; Vi: violarite. Solid star represents smythite of hypothetical composition Fe6.75Ni2.25S11.

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

81

Table 3 Reconstructed bulk sulfide analysis. Sample

Host rock

RV07-5a RV07-5b RV07-9b

Kyanite eclogite-Type I Kyanite eclogite-Type I Phlogopite eclogite-Type I

RV07-11

Eclogite-Type I

RV07-14

Eclogite-Type I

RV07-17

Eclogite-Type I

RV07-18

Eclogite-Type I

RV07-24 BD 3699

Eclogite-Type I Kyanite eclogite-Type I

RV-2G

Eclogite-Type I

RV-6

Eclogite-Type I

RV-a

Eclogite-Type I

RV-b

Eclogite-Type I

Sulfide ID

s5 s1a s1 s3 s4 s5 s6 s7 s8 s9 s10 s11 s14a s14b s16 s17 s18 s21 s22 s26a s1 s4a s4b s4c s7 s8 s9 s10a s10b s12a s4c s5a s6 s8a s9a s12 s12c s13 s2 s3 s4 s2 s3a s4 s5a s5b s6 s7 s8 s9 s10b s11 s5 s1 s3 s1 s2 s3 s1 s2 s2b s3 s2 s3 s4 s6 s4 s5c s6 s9 s10

Sulfide type

i e cpx e grt i i i i i i i e cpx e cpx i e cpx e cpx e cpx i e cpx e cpx i e cpx i e grt e grt e cpx i i i e grt e grt e grt i e grt i e cpx i i e grt e cpx e cpx e cpx i i e cpx i i i e cpx e cpx i i e cpx i i e grt e cpx e cpx e cpx i i i i e grt i e grt e grt e grt i e grt e grt e grt

Assemblage (%)

2 Po 75 Pn 23Cp 46 NiPy 37 Smy/Vi 17 Cp 76 NiPy 8 Smy/Vi 13 Cp 83 NiPy 9 Smy/Vi 8 Cp 83 NiPy 11 Smy/Vi 6 Cp 71 NiPy 18 Smy/Vi 12 Cp 71 NiPy 25 Smy/Vi 4 Cp 81 NiPy 9 Smy/Vi 10 Cp 78 NiPy 12 Smy/Vi 10 Cp 86 NiPy 10 Smy/Vi 4 Cp 80 NiPy 13 Smy/Vi 8 Cp 73 NiPy 20 Smy/Vi 7 Cp 73 NiPy 16 Smy/Vi 11 Cp 60 NiPy 35 Smy/Vi 16 Cp 70 NiPy 19 Smy/Vi 11 Cp 76 NiPy 15 Smy/Vi 10 Cp 70 NiPy 14 Smy/Vi 16 Cp 72 NiPy 13 Smy/Vi 14 Cp 66 NiPy 25 Smy/Vi 9 Cp 91 NiPy 3 Smy/Vi 6 Cp 56 Po 35 Pn 9 Cp 64 Po 19 Pn 18 Cp 69 Po 25 Pn 18 Cp 79 Po 16 Pn 5 Cp 74 Po 16 Pn 10 Cp 55 Po 29 Pn 16 Cp 39 Po 29 Pn 33 Cp 72 Po 17 Pn 11 Cp 65 Po 24 Pn 11 Cp 64 Po 26 Pn 11 Cp 31 NiPy 50 Smy/Vi 19 Cp 59 NiPy 29 Smy/Vi 12 Cp 5 Py 40 NiPy 39 Smy/Vi 15 Cp 6 Py 54 NiPy 34 Smy/Vi 6 Cp 6 Py 42 NiPy 41 Smy/Vi 12 Cp 57 NiPy 36 Smy/Vi 7 Cp 3 Py 39 NiPy 40 Smy/Vi 18 Cp 3 Py 43 NiPy 44 Smy/Vi 11 Cp 5 Py 45 NiPy 42 Smy/Vi 7 Cp 53 NiPy 37 Smy/Vi 11 Cp 2 Py 45 NiPy 45 Smy/Vi 8 Cp 59 NiPy 36 Smy/Vi 5 Cp 4 Py 4 NiPy 46 Smy/Vi 11 Cp 4 Py 56 NiPy 27 Smy/Vi 13 Cp 4 Py 41 NiPy 40 Smy/Vi 14 Cp 5 Py 56 NiPy 28 Smy/Vi 11 Cp 6 Py 37 NiPy 45 Smy/Vi 12 Cp 8 Py 60 NiPy 24 Smy/Vi 8 Cp 14 Py 46 NiPy 36 Smy/Vi 5 Cp 3 Py 58 NiPy 32 Smy/Vi 7 Cp 4 Py 46 NiPy 39 Smy/Vi 12 Cp 4 Py 37 NiPy 47 Smy/Vi 12 Cp 21 Py 23 NiPy 31 Smy/Vi 24 Cp 34 Py 29 NiPy 24 Smy/Vi 5 Cp 70 Smy/Vi 30 Cp 35 Py 49 NiPy 16 Cp 34 Py 45 Smy/Vi 21 Cp 24 Py 45 Smy/Vi 31 Cp 66 Py 13 Pn 20 Cp 67 Pn 33 Cp 75 Py 15 Pn 10 Cp 80 Py 16 Pn 4 Cp 4 NiPy 80 Po 13 Smy/Vi 3 Cp 4 NiPy 73 Po 20 Smy/Vi 3 Cp 1 NiPy 68 Po 28 Smy/Vi 3 Cp 1 Py 72 Po 23 Smy/Vi 5 Cp 47 Po 37 Smy/Vi 16 Cp 9 NiPy 49 Po 25 Smy/Vi 17 Cp 5 NiPy 75 Po 17 Smy/Vi 3 Cp 3 Py 71 Po 20 Smy/Vi 6 Cp 4 Py 57 Po 34 Smy/Vi 6 Cp

Reconstructed bulk sulfide (wt.%) O

S

Fe

Cu

Co

Ni

M/S (at.%)

0.41 0.44 0.84 0.67 0.94 1.02 1.18 0.89 0.94 0.94 0.96 1.07 0.93 1.31 1.05 0.98 0.93 0.94 1.14 0.87 0.35 0.29 0.31 0.27 0.34 0.34 0.31 0.34 0.28 0.25 0.36 0.50 0.43 0.47 0.55 0.48 0.40 0.41 0.61 2.04 1.25 0.59 0.62 0.69 0.56 0.61 0.75 0.64 0.66 0.61 0.61 0.59 1.57 1.22 0.47 0.27 0.51 0.60 0.68 0.75 0.72 0.40 0.28 0.24 0.32 0.22 0.19 0.25 0.22 0.28 0.38

39.43 39.24 44.75 45.33 47.10 45.27 45.77 46.62 46.20 47.49 46.57 45.87 45.98 44.46 45.29 45.99 44.90 45.32 44.94 49.00 37.96 36.80 37.24 37.35 37.22 36.35 35.47 36.47 36.12 36.12 41.81 45.47 43.66 45.93 44.06 45.84 43.28 43.79 40.71 42.00 42.09 43.75 41.13 43.71 41.20 43.71 41.38 44.58 43.76 43.41 42.03 40.61 42.84 42.96 37.54 49.80 43.40 41.23 45.83 45.07 46.46 47.74 38.28 37.96 37.78 38.71 39.34 39.58 39.76 39.37 40.51

38.31 36.59 36.94 37.94 38.48 36.13 35.34 38.32 37.55 38.72 37.75 36.25 37.25 33.10 35.89 37.06 36.44 36.79 34.87 40.13 47.46 48.33 52.74 51.95 51.37 48.17 44.32 50.86 49.55 49.05 37.51 39.40 38.79 41.14 38.76 39.75 38.53 38.33 34.12 39.34 37.47 39.22 37.67 39.91 38.07 38.75 37.02 40.40 40.43 40.30 38.29 37.32 35.74 37.16 32.47 43.19 31.51 29.56 41.02 41.88 42.70 43.21 54.55 52.76 50.83 50.28 48.12 46.74 52.70 51.82 47.61

7.80 6.78 4.91 3.28 2.66 5.16 2.95 4.00 4.28 2.09 3.49 3.72 5.12 4.00 4.91 4.29 6.57 5.82 4.68 2.32 3.26 6.15 2.08 1.64 3.27 5.32 11.33 3.93 3.65 3.66 6.65 4.52 5.57 2.32 4.27 2.55 6.35 3.90 2.86 4.34 3.13 2.01 4.03 4.86 5.34 3.93 3.74 2.99 1.98 2.73 4.36 4.47 8.99 1.74 10.24 5.91 7.67 10.79 7.11 0.29 3.62 1.56 0.94 1.27 1.19 1.90 5.40 5.77 1.11 2.00 2.27

0.32 0.09 0.23 0.21 0.16 0.36 0.47 0.25 0.29 0.27 0.30 0.39 0.42 0.58 0.38 0.32 0.30 0.30 0.45 0.07 0.50 0.29 0.19 0.28 0.32 0.33 0.29 0.25 0.33 0.28 0.36 0.29 0.34 0.27 0.43 0.38 0.26 0.28 3.31 0.46 1.94 0.64 0.57 0.39 0.46 0.37 0.55 0.39 0.46 0.40 0.50 0.59 0.09 0.28 0.76 0.05 0.76 0.76 0.15 0.37 0.17 0.34 0.12 0.15 0.19 0.15 0.27 0.14 0.09 0.08 0.03

12.81 15.75 6.39 7.85 7.80 8.87 11.09 6.90 7.68 7.47 7.87 9.55 7.81 13.22 9.33 8.26 7.75 7.73 10.69 5.65 8.35 5.71 4.61 6.35 4.84 6.88 5.91 5.72 7.65 8.61 12.60 8.83 10.58 9.02 11.33 10.46 10.42 12.31 12.87 11.90 13.91 12.59 14.61 9.01 12.69 10.79 15.18 8.69 11.38 10.71 12.71 15.01 6.89 7.52 17.93 0.90 15.31 15.35 4.13 10.03 4.52 5.45 4.26 6.10 8.27 7.30 6.42 6.93 5.36 5.54 8.71

0.84 0.84 0.61 0.61 0.59 0.63 0.61 0.60 0.61 0.58 0.60 0.61 0.62 0.64 0.63 0.61 0.64 0.63 0.63 0.56 0.89 0.93 0.91 0.92 0.91 0.94 0.97 0.94 0.96 0.97 0.76 0.66 0.71 0.65 0.70 0.66 0.72 0.70 0.73 0.75 0.75 0.70 0.78 0.70 0.77 0.69 0.77 0.67 0.70 0.70 0.75 0.79 0.67 0.62 0.91 0.57 0.71 0.76 0.64 0.66 0.62 0.60 0.89 0.90 0.91 0.88 0.86 0.85 0.85 0.86 0.82

N sulf = number of sulfide per thin section, e-grt = garnet enclosed sulfide, e-cpx = clinopyroxene enclosed sulfide, I = interstitial sulfide, M/S = metal/sulfur. Po = pyrrhotite, mss1 = mono-sulfide solution 1, Pn = pentlandite, Cp = chalcopyrite, Smy/Vi = smythite/violarite, Ni–Py = nickel rich pyrite, Py = pyrite, FeOOH = iron (oxy)hydroxide.

82

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

5. Sulfide mineral chemistry

from pyrrhotite-rich sulfide assemblages of worldwide provenance. It is noteworthy that pentlandite only occurs in samples in which pyrrhotite is present.

The major-element compositions of the sulfide phases have been analysed by electron microprobe and are summarised in Table 2 and Fig. 4. The microprobe analyses do not reveal any obvious differences in the composition of enclosed and interstitial sulfides.

5.3. Chalcopyrite Chalcopyrite (CuFeS2) is quite homogeneous within the sample suite and is close to the stoichiometric composition (Fig. 4c), with average Cu/Fe of 1.01 and metal/sulfur ratio (M/S) of 1.01. The average Ni content is b0.3 wt.%. There are no obvious compositional differences between interstitial and enclosed chalcopyrite.

5.1. Pyrrhotite Two types of pyrrhotite are observed in this eclogite suite. The atomic proportion of sulfur distinguishes between hexagonal pyrrhotite (Fe9S10; S ≈ 52.5 at.%) and monoclinic pyrrhotite (Fe7S8, S ≈ 53.5 at.%; Fig. 5). The Ni content of Po averages 2.1 wt.% but values range from 0.8 to 5 wt.%. Ni substitution in pyrrhotite is temperature-dependent and it has been shown that monoclinic pyrrhotite can take up to 5 wt.% Ni at 250 °C (Craig, 1973; Dromgoole and Pasteris, 1986; Szabo and Bodnar, 1995). The wide range of Ni contents observed in pyrrhotite at ambient temperature therefore indicates either a wide range of final equilibration temperatures, contamination by sub-micron sized pentlandite flames or local metastability of the “original” high-temperature sulfide (i.e. mss). However, if this was due to pentlandite contamination, correlations between S and Ni and M/S and Ni should be obtained (Lorand et al., 2005) but this was not the case. RV07-11, which shows the freshest sulfides and in which pyrite is absent from the sulfide assemblage, has Po with sulfur contents ranging from 51 to 55 at.% with a maximum between 51.5 and 52.5, compatible with a hexagonal structure. Sulfide inclusions in silicates from sample RV07-20 have Po with S contents ranging between 52.2 and 53.5 at.%, intermediate between the hexagonal and monoclinic structure; minor pyrite is observed in this sample. RV-a and RV-b, which show a larger proportion of “cleaved” sulfides with significant contents of pyrite/Ni-rich pyrite, contain Po with a maximum S content between 52.5 and 53.5 at.% (i.e., mostly monoclinic).

5.4. Pyrite Two types of pyrite occur in Roberts Victor eclogites, pyrite s.s. (FeS2) and Ni-rich pyrite ((Fe,Ni)S2). Both types can occur within a single sulfide grain. We have arbitrary classified as Ni-rich any pyrite with Ni N1 wt.%. Ni-rich pyrite grains commonly have Ni contents well above the expected values for pyrite, ranging up to 9 wt.% with an average of 3.4 wt.%. The M/S ratio of both pyrite and Ni-pyrite is 0.53, indicating a sulfur-deficient composition for both types, as well as Fe–Ni substitution. There is no significant chemical difference between intergranular and enclosed grains. Pyrite is characteristic of the “cleaved” sulfides. 5.5. Smythite and Violarite The wide variations in the compositions of the Ni-rich phases present some difficulties in the determination of the mineral species present. Depending on their metal/sulfur ratio (M/S), three groups can be distinguished. The phase with M/S = 0.81 is very similar to the rare smythite (Fe,Ni)13S16-(Fe,Ni)9S11 (S ≈ 55 at.%; e.g. Taylor and Williams, 1972; Nickel et al., 1974; Watmuff, 1974; Krupp, 1994; Furukawa and Barnes, 1996). However, until now the few natural occurrences described in the literature did not contain more than 7–8 wt.% Ni (Taylor and Williams, 1972; Nickel et al., 1974) whereas those described here have Ni contents ranging from 16 to 26 wt.%. The second group shows M/S = 0.78 and S ≈ 56 at.%, which is close to violarite (FeNi2S4). Ni contents range from 32 to 41 wt.%. Using X-ray diffraction, Desborough and Czamanske (1973) found that although this is not a known phase in the Fe–Ni–S system, its structure and composition are very close to violarite. Smythite and violarite are always found in (Ni-)

5.2. Pentlandite Pentlandite ss ((Fe,Ni)9S8), which is found only in samples RV07-11, RV-a and RV-b, ranges between 25 and 37 wt.% Ni. The Ni/ (Ni + Fe) of pentlandite varies from 0.40 to 0.56, with an average about 0.50 ± 0.05; it is similar to pentlandite from other mantle xenoliths (Szabo and Bodnar, 1995; Guo et al., 1999) and to pentlandite

S (wt.%)

“Fresh” Sulfides

“Cleaved” Sulfides

RV-a S

RV-b RV07-11 Py

50%

RV07-14

RV-2G

RV07-17

RV-6

RV07-18

RV07-5a

RV07-23b

RV07-5b

RV07-24

RV07-9b

RV07-26

Ni+Co

Fe

Supergene alteration (violaritisation)

BD3699

Smy

Po

40% Peridotitic DI

Fe (wt.%)

Sulfides enclosed in eclogitic diamonds 10%

20%

30%

40%

Ni+Co (wt.%)

Fig. 6. Reconstructed bulk sulfide compositions plotted in the Fe–Ni–S ternary diagram. Each symbol represents sulfide grains from a single sample. Fields of eclogitic and peridotitic sulfide inclusions in diamond are from Deines and Harris (1995); Bulanova et al. (1996); and Thomasot et al. (2009). Arrow represents the effect of the supergene alteration on the bulk sulfide composition. Po: pyrrhotite; Py: pyrite.

Table 4 Whole-rock and silicate chemistry. S/Se

δ18OGrt

Ni*

MgO

Na2O

ΣLREEN*

(La/Sm)N*

Zr*

Cucpx

Cugrt

Srcpx

Pb

(a)

(a)

(a)

(a)

(a)

(a)

(a)

(a)

(a)

(a)

(a)

(b)

ppm

ppm

wt.%

wt.%

ppm

ppm

ppm

ppm

ppm

‰

6600 9000 4500 5750 8294 7000 4598 3233 4000 6286 4200 3700 2500 3631 3833 – 2228 – – – – –

5 7 4 8 4 9 4 6 5 4 6 2 11 6 7 11 3 9 5 3 3

208 224 81 157 59 106 117 124 181 213 145 145 81 197 222 138 192 193 251 170 75 75

13.42 12.96 13.23 13.87 7.61 14.37 13.08 14.42 11.61 14.95 14.53 13.18 14.82 13.77 12.78 14.30 13.79 16.68e 14.39e 11.13e 12.52e 11.07e

2.67 3.08 1.62 2.07 2.51 2.08 2.36 1.39 2.98 2.29 1.55 1.89 1.14 2.78 2.51 1.42 1.91 1.59e 2.72e 3.07e 1.45e 1.42e

17.78 23.30 28.27 12.51 28.89 12.65 21.33 9.91 7.67 18.87 10.02 7.77 24.88 14.19 8.49 12.36 17.33 43.71 29.37 17.71 19.75 19.98

0.484 0.594 0.476 0.408 0.645 0.429 0.493 0.293 0.138 0.761 0.277 0.173 0.609 0.428 0.125 0.294 0.413 0.226 0.727 0.503 0.423 0.237

16.80 19.99 53.21 10.85 12.38 11.58 31.00 16.88 14.24 22.16 17.36 14.85 16.47 17.19 17.89 18.00 19.00 62.05 15.93 14.87 28.01 29.24

9 9 n.a. 10 13 8 15 10 14 6 11 17 3 21 12 18 18 5 11 8 10 10

n.a. 0.99 n.a. n.a. n.a. 0.85 0.74 1.55 n.a. 0.88 n.a. n.a. 0.30 1.09 1.15 0.98 0.98 0.34 0.77 0.82 0.70 1.28

256 234 329 226 332 238 251 228 129 210 211 144 252 226 121 263 274 401 343 208 271 271

0.408 0.961 0.627 0.423 3.151 0.474 0.563 0.490 0.261 0.643 0.429 0.285 0.822 0.470 0.187 0.530 0.540 3.120 0.810 0.493 0.464 0.464

6.69 6.84 5.68 4.22 5.25 6.3 n.a. 6.13 4.13 7.68 5.93 6.02 6.09 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.

50.8 30.8

7300 4966

3 3

132 225

14.55 14.03

1.84 2.10

31.74 34.93

0.827 0.840

26.95 28.33

5 4

0.89 1.91

319 307

0.894 0.843

7.7 n.a.

300,762 264,227 71,164 134,158 205,317 58,710

9.9 10.3 5.1 21.6 21.6 8.1

7000 4000 3667 8000 5000 2800

1 1 6 4 1 3

155 161 200 135 89 68

8.16 9.56 11.28 13.61 12.51 9.19

3.17 1.52 3.43 1.08 1.56 2.65

12.76 9.68 9.68 9.22 14.54 13.64

0.138 0.210 0.210 0.115 0.106 0.081

28.41 35.19 14.95 21.85 76.57 24.14

2 6 11 11 7 9

n.a. n.a. 0.89 0.82 n.a. 0.50

40 138 204 171 143

0.144 2.054 0.273 0.352 0.289 0.125

n.a. n.a. n.a. 5.26 5.41 2.24

300,000 248,168 29,300 – – – – – – –

18.5 10.2 1.3 – – – – – – –

13,379 27,778 1513 – – – – – – –

34 24 26 20 19 29 60 24 19 22

409 96 324 364 190 216 130 405 210 343

13.08 8.26 11.96 10.24 9.93 9.13 7.59 10.95e 14.53e 10.09e

2.36 3.01 1.84 3.26 2.37 3.26 3.92 2.35e 0.70e 4.87e

2.35 1.59 10.60 1.18 1.60 0.89 0.95 7.72 1.87 7.00

0.069 0.006 – – – – 0.011 – – –

3.91 1.68 11.12 2.35 1.21 0.53 0.91 8.00 5.00 11.50

53 41 48 33 40 39 83 51 43 23

2.31 n.a. 1.83 1.88 2.38 2.54 5.07 2.23 1.93 1.12

26 11 60 15 21 19 11 92 16 204

0.031 0.032 0.050 0.115 0.200 0.074 0.028 b.d.l. b.d.l. 0.259

3.52 1.89 3.04 2.32 2.38 2.15 2.4 n.a. n.a. n.a.

S

Cu

Ni

Se

Te

S/Te

Se/Te

Type I

ppm

ppm

ppm

ppb

ppb

RV07-1 RV07-2 RV07-3 RV07-7 RV07-10 RV07-11 RV07-13 RV07-14 RV07-16 RV07-17 RV07-18 RV07-19 RV07-20 RV07-22 RV07-24 RV07-29a RV07-29b RV07-40 RV07-41 HRV77bimin XRV6dusty XRV6clean

226 204 225 420 157 88 440 260 536 783 110 224 321 530 300 50 90 n.a. n.a. n.a. n.a. n.a.

33 81 9 23 141 14 64 97 36 88 63 37 125 84 46 27 82 n.a. n.a. n.a. n.a. n.a.

324 462 112 259 151 162 271 394 298 360 369 197 325 359 338 211 300 n.a. n.a. n.a. n.a. n.a.

126 209 23 68 73 29 76 457 126 348 220 48 280 305 84 n.a. 121 n.a. n.a. n.a. n.a. n.a.

5 9 2 4 17 2 14 30 9 14 15 10 50 23 12 n.a. 37 n.a. n.a. n.a. n.a. n.a.

1793 977 9924 6190 2139 3049 5782 569 4243 2249 500 4631 1146 1739 3576 – 743 – – – – –

45,231 22,679 112,637 104,920 9235 43,755 31,429 8671 59,594 55,895 7338 22,415 6414 23,043 25,000 – 2432 – – – – –

25.2 23.2 11.4 17.0 4.3 14.4 5.4 15.2 14.0 24.9 14.7 4.8 5.6 13.3 7.0 – 3.3 – – – – –

Type I phlogopite RV07-9a 800 RV07-9b 2000

73 144

576 542

508 892

10 29

1574 2242

80,000 68,966

Type I kyanite RV07-5a RV07-5b RV07-23a RV07-23b RV07-25 RV07-26

301 264 213 134 205 294

7 4 11 8 5 14

155 92 241 126 138 141

10 10 15 22 22 41

1 1 3 1 1 5

30,380 25,653 14,046 6211 9505 7230

Type II RV73-12 RV07-8 RV07-12 RV07-30 RV07-31 RV07-33 RV07-34 RV07-36 RV07-37 HRV77macrocx

300 223 498 n.a. n.a. n.a. n.a. n.a. n.a. n.a.

13 25 26 n.a. n.a. n.a. n.a. n.a. n.a. n.a.

348 119 347 n.a. n.a. n.a. n.a. n.a. n.a. n.a.

19 9 23 n.a. n.a. n.a. n.a. n.a. n.a. n.a.

1 0.9 17 n.a. n.a. n.a. n.a. n.a. n.a. n.a.

16,216 24,277 21,846 – – – – – – –

Cu/Te

cpx

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

Cu*

Sample

– not calculable; n.a. not analysed; b.d.l. below detection limit. (a) Data from Gréau et al. (2011); (b) Data from Huang et al. (2012). * Reconstructed whole-rock from LA-ICPMS and silicate modal abundances. e Reconstructed whole-rock from EMP analyses and silicate modal abundances. 83

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

2500

a

2000

S* (ppm)

pyrite-bearing samples. Violarite typically occurs along cracks within the sulfide grains or on the edge of a sulfide grain. The third group is more difficult to characterise. Indeed, analyses represented in a Fe–Ni–S diagram, the analyses plot in a field similar to the Ni-rich Mss or mss2 ((Fe,Ni)1-xS) (Fig. 4a). However, the strongly variable M/S (0.70–0.96) is inconsistent with mss2. Moreover, the presence of significant levels of oxygen (0.5 to 5 wt.%), which is correlated with a decrease in the sulfur content, indicates that these phases are more likely to represent an oxidised state of the previously described smythite and violarite and are possibly an early stage of transformation into FeOOH. These three groups are optically indistinguishable and X-ray diffraction would be necessary to solve this conundrum. However, because the small grain size prevents hand-picking of crystals, no light can really be shed on this problem at this time. As a result, and also because smythite and violarite share a continuum of compositions (Fig. 4a), these two phases will be treated as a unique group and referred as “smythite/violarite”.

Type I RV07-10 & RV07-20 RV07-11 Type I phlogopite Type I kyanite Type II

1500 1000 500

Important S loss FeOOH-rich

0

b

140

Cu (ppm)

84

100

60

5.6. Reconstructed bulk sulfide compositions

6. Whole-rock chalcophile budget We have measured the whole-rock contents of some critical chalcophile elements: S, Se, Te, Cu and Ni (Table 4). The whole-rock sulfur contents of 25 analysed xenoliths range from 88 to 2131 ppm. Although the average value is 420 ppm, the median is 264 ppm with 52% of the samples lying between 250 and 400 ppm. There is a weak correlation (Fig. 7-a) between sulfide modal abundance and whole-rock S content. The large scatter probably reflects (1) the variations in the sulfide compositions (Po/Py/Vi ratios); (2) the extremely heterogeneous distribution of sulfide (as shown by the large difference

20 0

c

900

Se (ppb)

700 500 300 100 0

d

50 40

Te (ppb)

The phase assemblage described above represents the lowtemperature re-equilibration of high-temperature Fe–Ni–Cu sulfides. To find the compositions of the original sulfides, we need to reconstruct the bulk composition of each sulfide grain using the compositions of the individual low-temperatures phases weighted by their modal abundances as imaged on element-distribution maps. The reconstructed bulk sulfide compositions obtained (Table 3) in this way can be divided in two groups (Fig. 6). The first is made up mostly of the “fresh” sulfides from samples RV07-11, RV-a and RV-b. All bulk sulphide compositions from these sample fall into the field of Ni-poor Mss, and correspond to samples with a high modal proportion of Po and low Py contents (Py b 9%). This group is relatively homogeneous (especially for RV07-11); the scatter toward high S contents is due to the variable occurrence of monoclinic Po and the presence of Py in the assemblage (e.g. RV-a, RV-b). Ni contents range from 4.2 to 8.7 wt.%, with a mean of 6.5 wt.%. Copper contents mostly vary from 1 to 6 wt.%, but Cu reaches 11 wt.% in one sulfide grain that shows a high modal proportion of chalcopyrite, which is most likely a bias due to random sanctioning of the sulfide grain. The other group (“cleaved” sulfides) defines a trend from the Ni-rich Mss field toward a pyrite-like end-member and shows a wide spread of compositions. This second group includes samples with higher modal abundances of nickel-rich and/or S-rich sulfides (Smy, Vi, Py and Ni-rich Py). The sulfides in this group range between Fe40Ni5S49 (“pyrite” end-member) and Fe35Ni13S40 (“Smy/Vi” end-member); the average Cu content is 4.4 wt.% and the average Ni content is 9.3 wt.%. Besides the large variations within this group, there is also wide variation within single samples. These differences in the bulk composition can probably be attributed to the image processing technique used, the sectioning effect and differential alteration of the various sulfide phases. These heterogeneities prevent a more precise petrographic estimate of sulfide abundances and limit the accuracy of the estimated bulk sulfide compositions.

30 20 10 0 0

10

20

30

40

50

N Sulfides* Fig. 7. Covariation of N*sulfide and some selected whole-rock contents. a S* versus N*sulfide. b Cu versus N*sulfide. c Se versus N*sulfide. d Te versus N*sulfide. Note the low S contents of FeOOH-rich samples (RV07-10; RV07-20). Dashed lines outline possible correlation trends. S* is the whole-rock S content recalculated from the three analytical techniques used (XRF, HTIT and IRC). N sulfides* = number of sulfides grains + number of FeOOH grains per sample.

between thin sections a/b of sample RV07-9 or RV-a and RV-b; Table 1); (3) the extremely heterogeneous size distribution of the sulfides (e.g. all (N = 23) of the preserved sulfides in RV07-20 are small (Ø b 100 μm) while RV07-1 has a much smaller number (N = 4) of much larger sulfides (250 N Ø N 1500 μm)); (4) the large variations in the proportion of Fe-oxy hydroxide (implying the loss of S) from sample to sample and from grain to grain. There is a broad correlation between S and Cu (Fig. 8-a), with concentrations from 4 to 144 ppm Cu. The Ni content of this eclogite suite varies from 92 to 576 ppm with most of the values between 150 and 350 ppm. However, Cu and Ni can also be hosted in silicates. The reconstructed whole-rock compositions of Type II and Type I

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92 Type I RV07-10 & RV07-20 RV07-11 Type I phlogopite Type I kyanite Type II Pyroxenites - Lherz

Cu (ppm)

a

200 150

700

b

600

Ni (ppm)

250

85

500

Type I RV07-10 & RV07-20 RV07-11 Type I phlogopite Type I kyanite Type II

gabbros

400 300

100

200

metagabbros

50 100

metabasalts

S (ppm) 0 0 40

1000

1500

2000

2500

20

N sulfides/sample

25

15 10 5 0

0

500

S (ppm)

1500

2500

30 25 20 15 10 5

6

7

8

9

10

11

12

13

14

15

0

16

0

1

2

3

4

5

Na2O wt.% whole-rock 600

30 25

f

500

20 15

1

Type I RV07-10 & RV07-20 RV07-11 Type I phlogopite Type I kyanite Type II

Ni (ppm) WR*

35

1=1

e

1=

40

400 300 Type I RV07-10 & RV07-20 RV07-11 Type I phlogopite Type I kyanite Type II

200

10 100

5 0

2000

Type I RV07-10 & RV07-20 RV07-11 Type I phlogopite Type I kyanite Type II

MgO wt.% whole-rock

Cu (ppm) WR*

1000

d

35 Type I RV07-10 & RV07-20 RV07-11 Type I phlogopite Type I kyanite Type II

30

0 40

c

35

N sulfides/sample

500

0

20

40

60

80

100

120

140

160

Cu (ppm) whole-rock

0

0

100

200

300

400

500

600

700

800

Ni (ppm) whole-rock

Fig. 8. Selected whole-rock chemistry plots. a Copper content versus sulfur content; solid circles, Roberts Victor eclogites; open diamonds, pyroxenites from Lherz (Bodinier et al., 1987; Lorand, 1989b). b Nickel versus sulfur; solid circles, Roberts Victor eclogites; grey field, Zermatt–Saas gabbros; dashed field, Zermatt–Saas metagabbros; open field, Zermatt– Saas metabasalt. Zermatt–Saas lithologies from Dale et al. (2009). c Number of sulfides per sample versus whole-rock MgO content. d Number of sulfides per sample versus whole-rock Na2O content. e Reconstructed whole-rock Cu contents (from grt and cpx LA-ICPMS analyses) versus XRF-analysed whole-rock Cu contents. f Reconstructed whole-rock Ni contents (from grt and cpx LA-ICP-MS analyses) versus XRF-analysed whole-rock Ni contents.

eclogites have quite distinct Cu contents. While Cu varies from 0.8 to 7.75 ppm in Type I, it reaches a minimum of 24 ppm in Type II (Fig. 8-e); this high level is explained by a copper-rich cpx that hosts 33–83 ppm Cu, vs. 2–17 ppm for Type I cpx. Garnets in Type II eclogites contain 1.7–5 ppm Cu; those in Type I eclogites usually have less than 1.5 ppm (Table 4). Comparison between the whole-rock analyses and the whole-rock compositions reconstructed from the silicate analyses allows an assessment of which elements are carried by the main silicate assemblage, and which reside in accessory phases and/or secondary minerals associated with the host kimberlite. In Fig. 8(e–f), data from Type II eclogites lie close to the 1:1 line, which means that Cu and Ni are entirely hosted by the silicates; this is consistent with the observed absence of sulfide in Type II eclogites. Type I eclogites show a distinct trend away from the 1:1 line, which suggests that most of the Cu, and to a lesser extent Ni, is carried by sulfide phases.

The fact that Cu vs. NSulf (Fig. 7-b) defines a better correlation than Cu vs. S (Fig. 8-a) suggests that Cu is more resistant to weathering processes than S, which is easily leached out as SO4 (Lorand, 1989a; Handler et al., 1999). This seems to be the case for some of the Type I xenoliths, which have abnormally low sulfur relative to their Cu contents, suggesting sulfur loss during alteration (e.g. RV07-10, RV07-20). Selenium and tellurium usually behave like sulfur in magmatic processes, but are less easily mobilised than sulfur during hydrothermal alteration (Lorand and Alard, 2001; Hattori et al., 2002; Lorand et al., 2003). The abundances of these elements show some differences depending on the type of eclogite, but as expected Se and Te are positively correlated with both sulfur and copper (Fig. 9). Selenium contents vary from 10 to 890 ppb in Type I eclogites but only from 9 to 23 ppb in Type II. Similarly, Te ranges from 1 ppb to N 50 ppb (RV07-20, maximum detection limit) in Type I and from b1 ppb to 17 ppb in Type II. Kyanite-bearing Type I eclogites also have low Se

86

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

100

a

b

5 I=2

C Se

50

S/

Te (ppb) whole-rock

S* (ppm)

2000

60

71

50 00

00

2500

1500

1000

1000

500

40

S loss

30 20 10

0 0

200

400

600

0

800

0

500

1000

1000

800

Te (ppb) whole-rock

Se (ppb) whole-rock

d

2500

Type I RV07-10 & RV07-20 RV07-11 Type I phlogopite Type I kyanite Type II

50

600

Se loss

400

200

40 30 20 10

0

20

40

60

80

100

120

140

0

160

0

20

40

Cu (ppm) whole-rock

e

3.104

S/Se kimberlitic Py=90000

f

3.104

Kimberlite contamination 2.104

1.104 5.103

60

80

100

120

140

160

Cu (ppm) whole-rock

S/Se whole-rock

S/Se whole-rock

2000

60

c

0

1500

S* (ppm)

Se (ppb) whole-rock

S/Se CI=2571

0

Se/Te CI=9

2.104

1.104 5.103

S/Se CI=2571

0 0

500

1000

1500

2000

2500

S* (ppm)

0

10

20

30

40

50

60

Se/Te whole-rock

Fig. 9. Chalcophile whole-rock chemistry. a Selenium versus sulfur. b Tellurium versus sulfur. c Selenium versus copper. d Tellurium versus copper. e S/Se versus sulfur. f S/Se versus Se/Te. Chondritic values from McDonough and Sun (1995).

and Te contents, on the same order as Type II eclogites, which may reflect their limited sulfide contents. However, in plots of Se or Te vs. Cu (Fig. 9-c, d), kyanite-bearing Type I eclogites follow the trends defined by Type I and phlogopite-bearing Type I eclogites, whereas Type II eclogites tend to scatter from those correlations. These are in agreement with the apparent absence of sulfides in Type II eclogites and their limited number in kyanite-bearing eclogites. Phlogopitebearing eclogites are among those most enriched in Se, Te and S. Se has been shown to be quite immobile during S-leaching processes (Lorand and Alard, 2001; Hattori et al., 2002; Lorand et al., 2003), but the Se–Nsulf relationship shows considerable scatter when compared to the Cu–Nsulf correlation (Fig. 7-b, c). However, RV07-11 has a much lower Se content relative to sulfide abundance than the other Type I samples rich in cleaved sulfides; the same is true for RV07-20. The S/Se of RV07-11 is much higher than in RV07-20 and this is inversely related to the ratio of iron-(oxy)

hydroxides to sulfides. However, since some of the sulfides in RV07-20 have been almost completely transformed into iron-(oxy) hydroxides, its S/Se should be treated with caution. The best correlation with Cu is obtained for Te (Fig. 9-d), which seems unaffected by alteration (i.e. proportion of iron (oxy)hydroxides), the ratio of the various sulfide types (i.e. fresh/cleaved) or the variability of sulfide phase abundances across the RV eclogite suite. Selenium is weakly correlated with the LREE enrichments of the reconstructed garnet–clinopyroxene assemblage (Fig. 10-a); Type II samples are less enriched ((La/Sm)N⁎ = 0.07 (* = reconstructed WR from main silicate assemblage), Se b25 ppm) and phlogopite eclogite is the most enriched with (La/Sm)N⁎ = 0.83 and Se = 892 ppm. Selenium also correlates with the ZrWR*, (Fig. 10-b), although kyanite-bearing eclogite do not follow this trend, which is possibly linked to the altered aspect of these samples; Selenium also correlates with other incompatible elements such as Srcpx and Pbcpx

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

(Fig. 10-c, -d); Finally selenium also shows an interesting apparent correlation with the oxygen isotopic compositions of garnet (Fig. 10-e). It is interesting to note that in all these correlations (whether they are weak or strong), Type II eclogites and Phlogopite-rich Type I eclogites are positioned at both ends of these trends. Similarly to Se, Cu is broadly correlated to LREE (Fig. 10-f). Type II eclogites have lower ΣLREEN⁎ (Avg. = 1.25) than Type I eclogites (Avg. = 18.5) and phlogopite-bearing Type I (Avg. = 33.5). Fig. 8(-c, -d) shows that the number of sulfide grains (including iron (oxy)hydroxide) in each sample is linked to the MgO and Na2O content (and to a lesser extent with SiO2, FeO, and CaO) of the whole-rock. The number of sulfides and the MgO contents of the rocks show a positive logarithmic correlation from the less magnesium-rich, sulfide-free Type II eclogites toward the MgO-rich Type I that host several tens of sulfides per thin section. The whole-rock contents of SiO2, CaO and Na2O in the Type I eclogites vary inversely with the number of sulfide grains. 7. Discussion 7.1. Supergene alteration 7.1.1. Replacement of the primary assemblage As described above, we observe a progressive increase in the proportion of the Py-rich “cleaved” sulfides from RV07-11 (free of cleaved sulfides), followed by RV-a and RV-b (showing both types) to the samples displaying only “cleaved” sulfides. This progressive change is characterised by a change of structural formula, going from hexagonal (S ≈ 52 at.%, RV07-11, Po + Pn + Cp, fresh sulfides) to monoclinic pyrrhotite (S ≈ 53.5 at.%, RV-a-RV-b, Po + Pn + Cp and Po + smythite/violarite + Py/Ni-Py + Cp, fresh and “cleaved” sulfides) and finally to smythite/violarite (S N54 at.%, most of the samples, smythite/violarite + Py/Ni-Py + Cp, “cleaved” sulfides only; Fig. 5). These observations can be interpreted as the result of a sulfidation of the sulfide assemblages, which could simply be described by the following reactions: Hexagonal Po þ Pn þ S→Monoclinic Po þ Pnð Smy=Vi  PyÞ Monoclinic Po þ Pnð Smy=Vi  PyÞ þ S→Smy=Vi þ Py: However, this is also similar to the supergene alteration of pyrrhotite– pentlandite ore observed in Western Australian massive nickel sulphide deposits (Nickel et al., 1974; Watmuff, 1974). In these, the replacement sequence of the original Po + Pn (+Cp) assemblage exhibits all the different assemblages and phases observed in the Roberts Victor eclogites. This weathering sequence is function of the depth and is the result of a gradient in oxidation potential (liberation of electrons) between the surface, the water table and the rocks lying directly below the water table (Nickel et al., 1974). Below the water table, when alteration starts (“transition zone”), Pn + Po are first transformed into Vi and monoclinic Po, liberating Ni2+ (and Fe2+). Part of the liberated Ni2+ goes into the margins of pyrrhotite grains, where Ni-smythite is formed (Nickel et al., 1974; Watmuff, 1974). In the Roberts Victor eclogites, nickeliferous smythite has a much higher Ni content (≈17 wt.%) than the precursor pyrrhotite (≈ 2.1 wt.%). However, Nickel (1972) described nickeliferous smythite that accommodates up to 9 times more Ni than the pyrrhotite from which it formed. Considering that in this weathering process, smythite is an intermediate stage in the transformation of pyrrhotite into violarite, the observation of smythite with a high Ni content seems expectable and most likely reflects Fe–Ni substitution. Higher in the replacement profile, once all the pentlandite has been replaced by violarite, the replacement of pyrrhotite by smythite and violarite stops, as no more Ni2+ is liberated (Nickel et al., 1974).

87

At this stage (“violarite–pyrite zone”), any unreplaced pyrrhotite will be quickly transformed into pyrite, which can show Ni contents between 1 and 12 wt.% (Nickel et al., 1974). It most likely is during this process that the sulfides acquire their “cleaved” microstructures. Indeed, the replacement of pyrrhotite in both the Roberts Victor eclogites and the Western Australian deposits produces a violarite with a lamellar microstructure, which is inherited from the twin lamellae (001) of pyrrhotite (Nickel et al., 1974). Therefore, as this phase transformation is also marked by a volume contraction, the weathering process almost certainly is responsible for the “cleaved” microstructure commonly observed in this sample suite. Above the water table (“oxide zone”), all the sulphides are pseudomorphously replaced by iron (oxy)hydroxides, potentially with minor pyrite surviving. This shallow reaction leaches away Ni and S, which are potentially remobilised deeper beneath the water table and consumed by the reaction transforming the original assemblage into smythite, violarite and pyrite. The overall reaction (excluding FeOOH) implies either a sulfur addition or an iron loss in order to go from a Po + Pn(+Cp) to Smy/ Vi + Py(+Cp). Nickel et al. (1974) suggested that the reaction occurs at constant S content because it would require a considerable amount of S, as well as a volume expansion (Watmuff, 1974) whereas observations suggest a volume contraction. Therefore a significant amount of Fe was lost during the process and most likely was carried away by water. Eclogitic sulfide inclusions shielded in diamonds generally consist of Po + Pn + Cp (e.g. Deines and Harris, 1995; Bulanova et al., 1996), which as noted above corresponds to the fresh sulfide assemblage observed in RV07-11. We can therefore confirm that the alteration process may have been a late process relative to the formation of diamond. Similarly, this alteration process must have occurred after the entrainment of the eclogite xenoliths in the kimberlite because Po + Pn exsolution occurs at near-surface temperatures (Lorand, 1987) and the secondary products (violarite, S-deficient pyrite) are not stable above 450 °C. Therefore, there seems to be little doubt about the supergene origin of the secondary sulfide assemblage. In this case the gradient of alteration observed in the xenolith suite most likely reflects the former position/depth of the xenoliths within the pipe before they were excavated during the exploitation of the kimberlite. RV07-11 probably came from a deep zone, away from extensive oxidation; RV-a and RV-b are characteristic of the “transition zone”; most of the samples fall into the “violarite–pyrite zone” and finally samples like RV07-10, in which only iron-(oxy)hydroxides persist, represent the “oxide zone” (above the water table).

7.1.2. Iron (oxy)hydroxides and the relative mobility of chalcophile elements The variations of the contents in chalcophile elements can also provide some clues about the supergene alteration of BMS. Indeed, numerous studies of xenoliths have described the loss of sulfur during the ascent of the xenolith in the host magma (Lorand, 1990; O'Neill et al., 1995; Handler et al., 1999; Lorand et al., 2003) or during superficial processes post-dating eruption (Lorand, 1990). This loss of sulfur leads to the formation of iron-(oxy)hydroxides. In the case of magmatic entrainment, the S loss may occur through volatilisation as the xenoliths are heated, decompressed and oxidised. Roberts Victor eclogites often show iron (oxy)hydroxides and/or secondary strings of sulfide radiating from sulfide grains. These microstructures suggest that the iron (oxy)hydroxides represent former sulfide grains that underwent an intensive S loss. This S loss has to be a late event, because many grains of iron (oxy)hydroxide display euhedral pseudomorphs after pyrite crystals, and pyrite should be one of the latest sub-solidus phases to form, appearing after alteration of the Po + Pn + Cp assemblage which is stable below 200 °C (Lorand, 1987). Therefore it is most likely that the sulfur loss observed in Roberts Victor eclogites was not driven mainly by entrainment of the xenoliths in the magma, but reflects the

88

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

1000

1000

a

e

100

Se (ppb)

Se (ppb)

100

10

10

1 0.0

1 0.2

0.4

0.6

0.8

0

2

4

6

8

δ18OGrt

(La/Sm)N* 1000

40

f Metasomatism

b

30

Se (ppb)

ΣLREEN*

100 20

10 10 sulfide addition

1

0 0

20

40

60

0

80

40

80

Zr* (ppm)

120

160

Cu (ppm)

1000

50

c

g Metasomatism

40

Se (ppb)

ΣLREEN*

100

10

30

20

10 Cu leaching

1

0 0

100

200

1

300

10

Srcpx (ppm) 50

d

40

Se (ppb)

N*sulfide

100

30

h

50

sulfide addition

1000

100

Cucpx (ppm)

25

0

0

40

80

20

10 10 Type II

Cu leaching

1

0 0

1

2

3

0

5

Pbcpx (ppm) cryptic

Type II

10

15

20

Cucpx (ppm) Metasomatism intensity (Gréau et al., 2011)

HRV77macrocx (Type II/I)

Type I

modal

Type I phlogopite

25

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

near-surface alteration sequence described above, in which the iron (oxy)hydroxides represent the most altered members. The S loss is well constrained in terms of chemistry. RV07-20 and RV07-10 are the samples with the highest abundances of iron (oxy) hydroxides, and are also the samples that systematically plot furthest from the well-defined chalcophile-element correlations (e.g. S vs. Cu, S vs. Se, Se vs. Cu; Figs. 8-a and 9-a, -c). From these samples, it appears that S, Cu, Se and Te did not respond equally to the alteration process. 7.2. Sulfides: a fingerprint of metasomatic processes Early studies recognised the unusually high abundance of sulfides in eclogite xenoliths compared to other mantle rocks (Czamanske and Desborough, 1968; Desborough and Czamanske, 1973; Frick, 1973; Meyer and Boctor, 1975). Desborough and Czamanske (1973) and Frick (1973) concluded that the sulfides probably originated through the immiscibility of sulfide liquids in silicate melt. However, it remained unclear whether this sulfide liquid was cogenetic with the silicate assemblages or if it was introduced into the rock later. The obvious correlation between the number of enclosed and interstitial sulfides (Fig. 3-b), and the absence of clear mineralogical or chemical differences between these two sulfide populations, suggest that they share a similar origin. But is that origin primary or secondary? First of all, the correlations between the whole-rock chemistry, mineral chemistry and sulfide abundance are strong evidence against the late addition of sulfides linked to the host kimberlite magma. The spherical shape of most sulfides enclosed in garnet or clinopyroxene suggests that the silicates crystallised (or recrystallised) around the sulfides while the latter were still liquid. Therefore, enclosed sulfides might represent an immiscible sulfide melt trapped during crystallisation or recrystallisation of the silicates, while interstitial grains represent the non-trapped fraction of the same sulfide melt. However, it is very unlikely that these immiscible sulfide melts are genetically directly related to the eclogites in which they are found, simply because the garnet + clinopyroxene assemblage observed in the eclogites is not primary. Indeed, if the eclogites have formed by the subduction of oceanic crust, they have re-equilibrated to high-pressure assemblages (metamorphism). If they formed as magmatic cumulates or reaction products (e.g. high-Al cpx ± grt) near their equilibration depth, these magmatic silicates have re-equilibrated to lower-T subsolidus assemblages (as shown by garnet and rutile exsolution from clinopyroxene in Type II eclogites). If the sulfides were early phases in cumulates that have recrystallised to eclogite, we might expect them to be homogeneously distributed, rather than being found as sulfide inclusions enclosed only in the rims of garnet and pyroxene. Moreover, in the case of cognate immiscible sulfides the chemistry of the sulfides (e.g. Ni) and the chemistry of the associated silicate melt (e.g. Mg#) define positive correlations (e.g. Chamberlain, 1967; Papunen, 1970). More recently Sen et al. (2010) confirmed that this relationship also applies to cognate immiscible sulfides in mantle-derived rocks, However, this marker was tested in the case of Roberts Victor eclogites, but no correlation was observed. If the eclogites represent subducted oceanic crust, it is unlikely that sulfides and sulfur, respectively very mobile components and a very volatile element, would be preserved in the rocks through the “subduction factory” processes (e.g. dewatering, metamorphism, partial melting). Dale et al. (2009) have demonstrated the effects of these processes on the whole-rock Ni and S contents of the gabbros from the Zermatt–Saas ophiolite (Fig. 8-b). The unmetamorphosed gabbros show a good correlation between Ni and S, whereas the metamorphosed gabbros and basalts have lost their Ni contents. This depends strongly on the sulfide assemblages observed. Dale et al. (2009)

89

showed that the transition from gabbro to gabbroic eclogite resulted in a transition from a Po + Pn + Cp assemblage to Py ± Po ± Cp. They also show that basaltic eclogites carry Py + Cp assemblages while MORBs carry Po + Pn + Cp assemblages (Czamanske and Moore, 1977). Therefore even where S is not much affected, Ni is apparently removed during subduction. However, the Roberts Victor eclogites carry many Ni-rich phases (Po, Pn, smythite/violarite), and their presence is apparently inconsistent with a subduction event having affected this sulfide assemblage. Nevertheless, the Roberts Victor eclogites have Ni–S correlations similar to unmetamorphosed gabbros, which possibly can be seen as a cumulate feature. As mentioned before, the petrographic study shows that enclosed sulphides are only found in or on the outer rims of the silicates, which tends to indicate that the sulfides are not inherited from the protoliths of the eclogites, but have been secondarily introduced after the main stage of crystallisation of the present grt + cpx assemblage; they probably represent a metasomatic addition to the rocks. Their distribution along grain boundaries, or enclosed in the outer parts of garnet and pyroxene grains, suggests that they infiltrated the eclogites along existing grain boundaries, inducing recrystallisation so that some of the sulfide melt became trapped within the silicates. This metasomatic event must predate the kimberlite eruption, because interstitial polyphase sulfides are not associated with the secondary minerals that are produced by reaction between eclogite and the kimberlitic magma (e.g. amphiboles, spinel, secondary phlogopite). Furthermore, enclosed sulfides are surrounded by structures that typically are attributed to a physical response to explosive decompression, such as radiating cracks due to sulfide expansion and spongy cpx halos due to adiabatic melting (Pasteris, 1982; Dromgoole and Pasteris, 1986). This requires that the sulfides have been trapped prior to eruption. The secondary discrete pyrite grains, which are clearly linked to the host magma, support this interpretation by illustrating the difference in the type of sulfide that can be introduced from the kimberlite. In fact, this secondary pyrite may account for the strongly fractionated hydrothermal-like S/Se (up to 30,000) observed in Type II eclogites and kyanite-bearing eclogites, in which none or little Fe–Ni–Cu sulfides are present. Indeed, in situ LA-ICP-MS of this type of pyrite in both kimberlite and eclogites have also shown extremely fractionated S/Se N 90,000 (Gréau, 2011). The correlations between the whole-rock copper content, the number of sulfides present and some major-element compositions (Figs. 7, 8) imply that the introduction of the sulfides was linked to changes in the major-element composition of the rock. Moreover, the direct links between Cu and the mineral chemistry of the major silicate phases rule out the addition of copper from the kimberlite magma. Following the same reasoning than Gréau et al. (2011), it is significant that RV07-9, a phlogopite eclogite, is one of the most sulfide-rich samples, and also is one of the most enriched in incompatible elements (Fig. 10). Phlogopite is a common mineral in mantle rocks and its presence is typically attributed to metasomatic processes. In this case, the relationship between sulfide abundance and phlogopite enrichment suggests that sulfides, like phlogopite, have been introduced into the eclogites by a metasomatic process that also has modified the majorelement and trace-element compositions of the rocks. This is illustrated by the positive correlation between the MgO contents of the Type I eclogites and the number of sulfide grains per sample (Fig. 8c). Similarly, the broadly positive correlations between Se and various incompatible elements and between CuWR and ∑LREE N⁎ (Fig. 10) suggest that sulfides and LREE/HFSE enrichment are linked, consistent with a metasomatic addition of sulfide. This is also consistent with the fact that no sulfides have been observed in Type II eclogites and that the LREE enrichment of Type I eclogites have been shown to be the result

Fig. 10. Correlations between common metasomatic tracers and sulfide addition. a Selenium versus (La/Sm)N⁎ (⁎: reconstructed from LA-ICP-MS analysis of the silicate assemblage). b Selenium versus Zr*. c Selenium versus Srcpx. d Selenium versus Pbcpx. e Selenium versus δ18OGrt f ΣLREEN⁎ versus Cu (XRF whole-rocks). g ΣLREEN⁎ versus Cucpx. h N*sulfide per sample versus Cucpx, showing how the number of sulfide grains broadly increases as copper is leached out from the silicates (here the cpx).

90

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

of a metasomatic event (Gréau et al., 2011; Huang et al., 2012). However, some care needs to be taken in the case of Se, as it has been shown here that Se may have been slightly mobile during the supergene alteration of the sulfides. However, preliminary in-situ analyses of the non-weathered/pristine sulfides from RV07-11 show superchondritic S/Se ratios (6000–10,000; Gréau, 2011) similar to those characterising metasomatic sulfides precipitated from volatile-rich carbonatitic small melt fractions (Lorand et al., 2003, 2004; Alard et al., 2011) but different from the Se- poor sulfides found in the associated kimberlite (S/Se N 90,000; Gréau, 2011). The absence of sulfide in Type II eclogites confirms that these eclogites did not undergo such a metasomatic event, which seems consistent with their petrographic features: extremely fresh silicate phases, equilibrated microstructures, and the absence of “primary” phlogopite or melt pockets in garnet (Gréau et al., 2011). Therefore, the original compositions of the Type II eclogites are more likely to have been preserved. This is consistent with the low MgO, Se, Te, (La/Sm)N⁎, ZrWR* and high CaO contents of Type II eclogites compared to Type I. 7.3. Sulfide mineralogy and nature of the metasomatic agent Two types of sulfide formation might be invoked to explain the origin of the sulfides: (a) metasomatic S-rich (SO2 or H2S) fluids, which precipitated sulfides by leaching Fe, Ni and Cu from the silicate matrix; and/or (b) introduced immiscible sulfide melts. Ballhaus and Stumpfl (1986) described the formation of immiscible sulfide melts and the precipitation of sulfides due to the interaction of an H2S fluid with the silicate minerals. Similarly, Libaudé and Sabatier (1980) and Peregoedova et al. (2006) have shown that mss can be produced by reaction of SO2 and silicate phases. These sulfidation reactions can be represented as: H2 SðfluidÞ þ FeO; NiO; CuOðsilicateÞ →Fe−Ni−Cu Sulfide þ H2 OðfluidÞ

ð1Þ

or SO2ðfluidÞ þ FeO; NiO; CuOðsilicateÞ →Fe−Ni−Cu Sulfide þ 3=2 O2 :

ð2Þ

In the Roberts Victor eclogite suite, the reconstructed compositions of Type II eclogites have higher copper contents than the Type I samples (Fig. 8-e), and the Fe contents of garnet and Ni contents of clinopyroxene are also generally higher in Type II eclogites than in Type I (Gréau et al., 2011). Therefore, the differences in the Fe–Ni–Cu contents of the silicates in Type I and Type II eclogites might be explained by precipitation of sulfide, after an introduced S-rich fluid has stripped Fe, Ni and Cu from the silicate assemblages, as described in the equations above. This mechanism is supported by the negative correlation between the Cu contents of the cpx and the LREE contents of the reconstructed whole-rock (Fig. 10-g), showing that as the degree of metasomatism increases copper is progressively leached away from the cpx. However, nothing permits to clearly decipher between a SO2 and H2S-rich fluid. Nevertheless, one could possibly assume that a reduced H2S fluid might be responsible, because the sulfides present in these eclogites are similar to those found in eclogitic diamonds from Roberts Victor indicating that the condition might have been reducing. Moreover, Lorand and Gregoire (2006) attributed the sulfides in some phlogopite-rich peridotite xenoliths from the Bultfontein kimberlite (Kaapvaal craton) to the first kind of sulfidation process (1). This study (Lorand and Gregoire, 2006) and Ballhaus and Stumpfl (1986) also described sulfides associated with hydrous silicate phases (e.g. biotite, phlogopite), which they link to the liberation of H2O during the reaction. While it is not a general feature, some of the sulfides in Robert Victor Type I eclogites are closely associated with a hydrous sheet silicate (Fig. 2-c) or phlogopite (Fig. 2-a). Nevertheless, although the contents of these metals in the primary silicates decrease from Type II to Type I, the whole-rock Type I

samples show much higher Cu and Ni contents than the Type II eclogites. This suggests the net addition of those elements to the rock through the introduction of ‘external’ immiscible sulfide melts. This would also explain why most of the sulfides enclosed in garnet and pyroxene have rounded shapes, implying that the sulfides existed as homogeneous liquid phases when the eclogites recrystallised in the subcontinental lithospheric mantle. Although these two ways of adding sulfides are different, they are not necessarily mutually exclusive. Indeed, evidences supporting both mechanisms are available in most eclogites (i.e. round sulfide blebs, cpx leaching, net Ni–Cu addition). One can possibly imagine that the interaction with a highly reactive sulfur-rich fluid happens in an early stage of the metasomatic event, when pre-metasomatic fluids are percolating up front of the main melt; melt that would subsequently interact with the eclogites and introduce an immiscible sulfide liquid. Metasomatic addition of sulfide is commonly described in ultramafic rocks (Alard et al., 2000; Lorand and Alard, 2001; Griffin et al., 2002; Lorand et al., 2003; Luguet et al., 2003; Griffin et al., 2004; Lorand and Gregoire, 2006; Alard et al., 2011). In these studies, the reconstituted bulk sulfide compositions correspond to the mss1 and/or the mss2 fields. Reconstruction of the bulk sulfide composition from the low-temperature assemblage allows us to estimate the composition of the original Fe–Ni–Cu sulfide and to assess its origin. The Roberts Victor bulk sulfide compositions fall into two groups. The first group ("fresh sulfides") corresponds to the mss1 field and is similar to sulfide inclusions in Roberts Victor diamonds (Fig. 4.3 and Fig. 4.5; Deines and Harris, 1995). The second group has not been described previously and is less relevant as it is the direct result of the supergene alteration process described above. It shows compositions intermediate between a “pyrite” end-member and smythite. These compositions are basically Fe-depleted compared to the first group of sulfides and to sulfide inclusions in eclogitic diamonds, and are Ni-depleted compared to peridotitic sulfides (Fig. 6) (Dromgoole and Pasteris, 1986; Szabo and Bodnar, 1995; Alard et al., 2000; Lorand and Gregoire, 2006) and the sulfide inclusions in diamonds of the peridotitic paragenesis (Deines and Harris, 1995; Bulanova et al., 1996; Ruzicka et al., 1998). Diamond-inclusion sulfides and in particular those ascribed to the eclogitic paragenesis (e.g.Deines and Harris, 1995; Bulanova et al., 1996; Ruzicka et al., 1998; Thomasot et al., 2009) show clear relationship to the most pristine sulfides in Roberts Victor eclogites. The common sulfide assemblage in eclogitic diamonds consists of pyrrhotite + pentlandite ± chalcopyrite, which corresponds to the fresh assemblage observed in RV07-11. Diamond-inclusion sulfides do not show the “cleaved” structure, and no Ni-rich pyrite or “smythite/violarite” has been reported as inclusions in diamond. This confirms that the “cleaved” feature is a late and superficial feature and that RV07-11 must be regarded as the most pristine sample of our suite in terms of sulfide assemblage. Sulfides are the most common inclusions in diamonds (Sharp, 1966; Sobolev, 1977; Harris, 1992; Sobolev et al., 1999) and they have been linked to the formation of diamond (Haggerty, 1986; Shushkanova and Litvin, 2008). The observations and data presented here strongly suggest that the sulfides in Roberts Victor eclogites are the products of a metasomatic process. In Type II eclogites, sulfides, diamonds and petrographic evidence of metasomatic processes are absent. On a worldwide scale, all diamond-bearing eclogites belong to Type I (as defined by contents of Na in garnet and K in clinopyroxene) and silicate inclusions in diamonds of the eclogitic paragenesis also belong to Type I. Recently, several studies have ascribed the growth of diamonds to metasomatic melts/fluids (Weiss et al., 2009; references therein), which is consistent with the apparent origin of the sulfides in Roberts Victor eclogites and the petrographic features of these Type I eclogites. Thus, there is a very strong coupling between metasomatism, sulfides and diamonds in Type I eclogites. The absence of diamonds, sulfides and metasomatic characteristics in Type II eclogites from Roberts Victor reinforces

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

the conclusion that this group has not been affected by such metasomatism. 8. Conclusions Petrographic and geochemical studies of Roberts Victor Type I eclogites indicate that the enclosed and interstitial sulfide populations have the same origin. However, they are not cogenetic with the eclogite and are attributed to a metasomatic event imposed on pre-existing eclogites near the base of the lithosphere. This event was characterised by the introduction of S-rich fluids and/or immiscible sulfide melts that precipitated sulfides in the eclogites. The general occurrence of sulfides in the rims of silicate mineral grains implies that this metasomatic event was accompanied by some recrystallisation of the rock. Correlations between whole-rock compositions, silicate mineral chemistry and sulfide abundance indicate that this metasomatic event also modified the chemistry of the Type I eclogites. The absence of sulfides, diamonds or other metasomatic products in Type II eclogites indicates that this group of eclogites did not experience the metasomatic event that produced the Type I eclogites. Attempts to relate the compositions of Type I eclogites to basaltic rocks, subduction and partial melting processes must consider this strong metasomatic overprint, which has significantly altered the original major-element, trace-element and isotopic compositions (Gréau et al., 2011; Huang et al., 2012). The presence of two distinct sulfide assemblages, Po + Pn + Cp and Ni-Py+“smythite/violarite” + Cp, is attributed to supergene weathering. Supergene weathering was also ultimately responsible for sulfur loss, which is marked by the replacement of sulfide grains by iron (oxy)hydroxides. Acknowledgments We are grateful to Peter Weiland for his generous technical assistance in the use of the analytical instruments at GEMOC. Norman Pearson also assisted with helpful discussions on the interpretation and instrumental aspects. This manuscript benefited greatly from the constructive comments provided by the Editor Laurie Reisberg and by the thoughtful reviews of Jean-Pierre Lorand and an anonymous reviewer. This project was supported by Macquarie University international postgraduate scholarships and ARC (Australian Research Council) Discovery and Linkage Grants (O'Reilly, Griffin). The analytical data were obtained using instrumentation funded by DEST Systemic Infrastructure Grants, ARC LIEF, NCRIS, industry partners and Macquarie University. This is contribution #889 in the GEMOC ARC National Key Centre (www.gemoc.mq.edu.au) and #324 from the CCFS ARC Centre of Excellence (www.ccfs.mq.edu.au). Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.chemgeo.2013.06.015. References Alard, O., Griffin, W.L., Lorand, J.-P., Jackson, S.E., O'Reilly, S.Y., 2000. Non-chondritic distribution of the highly siderophile elements in mantle sulphides. Nature 407, 891–894. Alard, O., Lorand, J.-P., Reisberg, L., Bodinier, J.-L., Dautria, J.-M., O'Reilly, S.Y., 2011. Volatile-rich metasomatism in Montferrier xenoliths (Southern France): implications fro the abundances of chalcophile and highly siderophile elements in the subcontinental mantle. Journal of Petrology 52 (10), 2009–2045. Anderson, D.L., 1982. Isotopic evolution of the mantle: a model. Earth and Planetary Science Letters 57, 13–24. Ballhaus, C.G., Stumpfl, E.F., 1986. Sulfide and pltainum mineralization in the Merensky Reef: evidence from hydrous silicates and fluid inclusions. Contributions to Mineralogy and Petrology 94, 193–204. Berg, G.W., 1968. Secondary alteration in eclogites from kimberlite pipes. American Mineralogist 53, 1336–1346.

91

Bodinier, J.L., Guiraud, M., Fabries, J., Dostal, J., Dupuy, C., 1987. Petrogenesis of layered pyroxenites from the Lherz, Freychinede and Prades ultramafic bodies. (Ariege, French Pyrénées) Geochimica et Cosmochimica Acta 51, 279–290. Bulanova, G.P., Griffin, W.L., Ryan, C.G., Shestakova, O.Y., Barnes, S.-J., 1996. Trace elements in sulfide inclusions from Yakutian diamonds. Contributions to Mineralogy and Petrology 124, 111–125. Caporuscio, F.A., Smyth, J.R., 1990. Trace element crystal chemistry of mantle eclogites. Contributions to Mineralogy and Petrology 105, 550–561. Chamberlain, J.A., 1967. Sulfides in the Muskox intrusion. Canadian Journal of Earth Sciences 4 (1), 105–153. Chinner, G.A., Cornell, D.H., 1974. Evidence of kimberlite–grospydite reaction. Contributions to Mineralogy and Petrology (45), 153–160. Craig, J.R., 1973. Pyrite–pentlandite assemblages and other low temperature relations in the Fe–Ni–S system. American Journal of Science 273 (A), 496–510. Czamanske, G., Desborough, G.A., 1968. Primary Fe–Ni–Cu sulfides in eclogite nodules in South African kimberlite. Geological Society of America Abstract 1967, 41. Czamanske, G.K., Moore, J.G., 1977. Composition and phase chemistry of sulfide globules in basalt from the Mid-Atlantic Ridge rift valley near 37oN lat. Geological Society of America Bulletin 88, 587–599. Dale, C.W., Burton, K.W., Pearson, D.G., Gannoun, A., Alard, O., Argles, T.W., Parkinson, I.J., 2009. Highly siderophile element behaviour accompanying subduction of oceanic crust: whole rock and mineral-scale insights from a high-pressure terrain. Geochimica et Cosmochimica Acta 73, 1394–1416. Deines, P., Harris, J.W., 1995. Sulfide inclusion chemistry and carbon isotopes of African diamonds. Geochimica et Cosmochimica Acta 59 (15), 3173–3188. Desborough, G.A., Czamanske, G., 1973. Sulfides in eclogite nodules from a kimberlite pipe, South Africa, with comments on violarite stoichiometry. American Mineralogist 58, 195–202. Donaldson, C.H., 1978. Petrology of the uppermost upper mantle deduced from spinellherzolite and harzburgite nodules at Calton Hill, Derbyshire. Contributions to Mineralogy and Petrology 65 (4), 363–377. Dromgoole, E.L., Pasteris, J.D., 1986. Interpretation of the sulfide assemblages in a suite of xenoliths from Kilbourne Hole, New Mexico. Geological Society of AmericaSpecial Paper 215. Mantle Metasomatism and Alkaline Magmatism, pp. 25–46. Frick, C., 1973. The sulphides in griquaite and garnet–peridotite xenoliths in kimberlite. Contributions to Mineralogy and Petrology 39 (1), 1–16. Furukawa, Y., Barnes, H.L., 1996. Reactions forming smythite, Fe9S11. Geochimica et Cosmochimica Acta 60 (19), 3581–3591. Garlick, G.D., MacGregor, I.D., Vogel, D.E., 1971. Oxygen isotope ratios in eclogites from kimberlites. Science 172 (3987), 1025–1971. Gréau, Y., 2011. Siderophile and chalcophile elements in Roberts Victor eclogites: insights on the generation of mantle heterogeneity. (PhD Thesis) Macquarie University. Gréau, Y., Huang, J.-X., Griffin, W.L., Renac, C., Alard, O., O'Reilly, S.Y., 2011. Type I eclogites from Roberts Victor kimberlites: products of extensive mantle metasomatism. Geochimica et Cosmochimica Acta 75, 6927–6954. Griffin, W.L., O'Reilly, S.Y., 2007. Cratonic lithospheric mantle: is anything subducted? Episodes 30 (1), 43–53. Griffin, W.L., Spetsius, Z.V., Pearson, N.J., O'Reilly, S.Y., 2002. In situ Re–Os analysis of sulfide inclusions in kimberlitic olivine: new constraints on depletion events in the Siberian lithospheric mantle. Geochemistry, Geophysics, Geosystems 3 (11), 1069. http://dx.doi.org/10.1029/2001GC000287. Griffin, W.L., Graham, S., O'Reilly, S.Y., Pearson, N.J., 2004. Lithosphere evolution beneath the Kaapvaal Craton: Re–Os systematics of sulfides in mantle-derived peridotites. Chemical Geology 208, 89–118. Griffin, W.L., Powell, W.J., Pearson, N.J., O'Reilly, S.Y., 2008. GLITTER: data reduction software for laser ablation ICP-MS. Laser Ablation-ICP-MS in the Earth Sciences. Mineralogical Association of Canada Short Course Series 40, 204–207. Gros, M., Lorand, J.P., Bezos, A., 2005. Determination of total sulfur contents in the International rock reference material SY-2 and other mafic and ultramafic rocks using an improved scheme of combustion/iodometric titration. Geostandards and Geoanalytical Research 29 (1), 123–130. Guo, J., Griffin, W.L., O'Reilly, S.Y., 1999. Geochemistry and origin of sulphide minerals in mantle xenoliths: Quilin, Southeastern China. Journal of Petrology 40 (7), 1125–1149. Haggerty, S.E., 1986. Diamond genesis in a multiply-constrained model. Nature 320, 34–37. Handler, M.R., Bennett, V.C., Dreibus, G., 1999. Evidence from correlated Ir/Os and Cu/S for late-stage Os mobility in peridotite xenoliths: Implications for Re–Os systematics. Geology 27 (1), 75–78. Harris, J.W., 1992. Diamond geology. In: Field, J.E. (Ed.), The Properties of Natural and Synthetic Diamond. Academic Press, London, pp. 345–393. Harte, B., Gurney, J.J., 1975. Evolution of clinopyroxene and garnet in an eclogite nodule from the Roberts Victor kimberlite pipe, South Africa. Physics and Chemistry of the Earth 9, 367–387. Hatton, C.J., 1978. The Geochemistry and Origin of Xenoliths from the Roberts Victor Mine. (PhD Thesis) University of Cape Town. Hatton, C.J., Gurney, J.J., 1977. Igneous fractionation trends in Roberts Victor eclogites. 2nd International Kimberlite Conference, AGU Santa Fe. Hattori, K.H., Arai, S., Clarke, D.B., 2002. Selenium, tellurium, arsenic and antimony contents of primary mantle sulfides. The Canadian Mineralogist 4, 637–650. Huang, J.-X., Gréau, Y., Griffin, W.L., O'Reilly, S.Y., 2010. Metasomatic hide and seek: origins of the Roberts Victor eclogites, South Africa. 20th General Meeting of the International Mineralogical Association, Budapest, Mineralogica-Petrographica. Huang, J.-X., Gréau, Y., Griffin, W.L., O'Reilly, S.Y., Pearson, N.J., 2012. Multi-stage origin of Roberts Victor eclogites: progressive metasomatism and its isotopic effects. Lithos (142–143), 161–181.

92

Y. Gréau et al. / Chemical Geology 354 (2013) 73–92

Jacob, D., Jagoutz, E., Lowry, D., Mattey, D., Kudrjavtseva, G., 1994. Diamondiferous eclogites from Siberia: remnants of Archean oceanic crust. Geochimica et Cosmochimica Acta 58 (23), 5191–5207. Jacob, D.E., Bizimis, M., Salters, V., 2002. Lu–Hf isotopic systematics of subducted ancient oceanic crust: Roberts Victor eclogites. Geochimica et Cosmochimica Acta 66 (15A), A360-A360. Jacob, D.E., Bizimis, M., Salters, V.J.M., 2005. Lu–Hf and geochemical systematics of recycled ancient oceanic crust: evidence from Roberts Victor eclogites. Contributions to Mineralogy and Petrology 148 (6), 707–720. Jacob, D.E., Viljoen, K.S., Grassineau, N.V., 2009. Eclogite xenoliths from Kimberley, SouthAfrica — a case study of mantle metasomatism in eclogites. Lithos 112s (2), 1002–1013. Jin, Y., Taylor, L.A., 1988. Metasomatic partial melting of clinopyroxene in spinel lherzolite xenoliths in alkali basalts from eastern China. Geological Society of America Abstracts with Programs 20 (4), 273. König, S., Luguet, A., Lorand, J.P., Wombacher, F., Lissner, M., 2012. Selenium and tellurium systematics of the Earth's mantle from high precision analyses of yltradepleted orogenic peridotites. Geochimica et Cosmochimica Acta 86, 354–366. Kramers, J.D., 1979. Lead, uranium, strontium, potassium and rubidium in inclusionbearing diamonds and mantle-derived xenoliths from Southern Africa. Earth and Planetary Science Letters EPSL 42, 58–70. Krupp, R.E., 1994. Phase relations and phase transformations between the lowtemperature iron sulfides mackinawite, greigite and smythite. European Journal of Mineralogy (6), 265–278. Kullerud, G., Yund, R.A., Moh, G.H., 1969. Phase relation in the Cu-Fe-Ni, Cu-Ni-S and Fe-Ni-S systems. Economic Geology 4, 323–343. Libaudé, J., Sabatier, G., 1980. Contribution a l'étude de la sulfuration des olivines nickeliferes en phase vapeur. Mémoires du Bureau de Recherches Géologique et Minières 242–253. Lorand, J.-P., 1987. Caracteres mineralogiques et chimiques generaux des microphases du systeme Cu–Fe–Ni–S dans les roches du manteau superieur: exemples d'heterogeneites en domaine sub-continental. Bulletin de la Societe Geologiqu de France 4, 643–656. Lorand, J.P., 1989a. Abundance and distribution of Cu–Fe–Ni sulfides, sulfur, copper and platinum-group elements in orogenic-type spinel lherzolite massifs of Ariege (northeastern Pyrenees, France). Earth and Planetary Science Letters 93, 50–64. Lorand, J.P., 1989b. Sulfide petrology of spinel and garnet pyroxenite layers from mantle-derived spinel lherzolite massifs of Ariege, northeastern Pyrenees, France. Journal of Petrology 30, 987–1015. Lorand, J.P., 1989c. Mineralogy and chemistry of Cu–Fe–Ni sulfides in orogenic-type spinel peridotite bodies for Ariege (Northeastern Pyrenees, France). Contributions to Mineralogy and Petrology 103, 335–345. Lorand, J.P., 1990. Are spinel lherzolite xenoliths representative of the abundance of sulfur in the upper mantle? Geochimica et Cosmochimica Acta 54, 1487–1492. Lorand, J.-P., Alard, O., 2001. Platinum-group element abundances in the upper mantle: new constraints from in situ and whole-rock analyses of Massif Central xenoliths (France). Geochimica et Cosmochimica Acta 65, 2789–2806. Lorand, J.P., Alard, O., 2010. Determination of selenium and tellurium concentrations in Pyrenean peridotites (Ariege, France): new insight into S/Se/Te systematics of the upper in mantle samples. Chemical Geology 278 (1–2), 120–130. Lorand, J.P., Alard, O., 2011. Pyrite tracks assimilation of crustal sulfur in Pyrenean peridotites. Mineralogy and Petrology 101 (1–2), 115–128. Lorand, J.-P., Gregoire, M., 2006. Petrogenesis of base metal sulphide assemblages of some peridotites from the Kaapvaal craton (South Africa). Contributions to Mineralogy and Petrology 151, 521–534. Lorand, J.-P., Alard, O., Luguet, A., Keays, R.R., 2003. Sulfur and selenium systematics of the subcontinental lithospheric mantle: inferences from the Massif Central xenolith suite (France). Geochimica et Cosmochimica Acta 67 (21), 4137–4151. Lorand, J.-P., Delpech, G., Grégoire, M., Moine, B., O'Reilly, S.Y., Cottin, J.-Y., 2004. Platinum-group elements and the multistage metasomatic history of Kerguelen lithospheric mantle (South Indian Ocean). Chemical Geology 208, 195–215. Lorand, J.P., Chevrier, V., Sautter, V., 2005. Sulfide mineralogy and redox conditions in some shergottites. Meteoritics and Planetary Science 40 (8), 1257–1272. Lorand, J.-P., Luguet, A., Alard, O., 2008. Platinum-group elements: a new set of key tracers for the Earth's interior. Elements 4 (4), 247–252. Luguet, A., Lorand, J.-P., Seyler, M., 2003. Sulfide petrology and highly siderophile element geochemistry of abyssal peridotites: a coupled study of samples from the Kane Fracture Zone (45° W 23°20 N, MARK Area, Atlantic Ocean). Geochimica et Cosmochimica Acta 67 (8), 1553–1570. MacGregor, I.D., Carter, J.L., 1970. The chemistry of clinopyroxenes and garnets of eclogite and peridotite xenoliths from the Roberts Victor mine, South Africa. Physics of the Earth and Planetary Interiors 3, 391–397. MacGregor, I.D., Manton, W.I., 1986. Roberts Victor eclogites: ancient oceanic crust. Journal of Geophysical Research 91 (B14), 14063–14079. Manton, W.I., Tatsumoto, M., 1971. Some Pb and Sr isotopic measurements on eclogites from the Roberts Victor Mine, South Africa. Earth and Planetary Science Letters 10 (2), 217–226. McCandless, T.E., Gurney, J.J., 1989. Sodium in garnet and potassium in clinopyroxene: criteria for classifying mantle eclogites. In: Ross, J., Jaques, A.L., Ferguson, J., et al. (Eds.), Kimberlites and Related Rocks. Perth, Special Publication — Geological Society of Australia, 14, pp. 827–832. McDonough, W.F., Sun, S.S., 1995. The composition of the Earth. Chemical Geology 120 (3–4), 223–253. Meyer, H.O.A., Boctor, N.Z., 1975. Sulfide-oxide minerals in eclogite from Stockdale Kimberlite, Kansas. Contributions to Mineralogy and Petrology 52–57. Nickel, E.H., 1972. Nickeliferous smythite from Canadian occurrences. Canadian Mineralogist 11, 514–519.

Nickel, E.H., Ross, J.R., Thornber, M.R., 1974. The supergene alteration of pyrrhotite– pentlandite ore at Kambalda, Western Australia. Economic Geology 69, 93–107. O'Neill, H.S.C., Dingwell, D.B., Borisov, A., Spettel, B., H.P., 1995. Experimental petrochemistry of some highly siderophile elements at high temperatures, and some implications for core formation and the mantle's early history. Chemical Geology 120, 255–273. Ongley, J.S., Basu, A.R., Kyser, T.K., 1987. Oxygen isotopes in coexisting garnets, clinopyroxenes and phlogopites of Roberts Victor eclogites: implications for petrogenesis and mantle metasomatism. Earth and Planetary Science Letters 83, 80–84. Ozima, M., Saito, K., 1975. 40Ar–39Ar of the omphacite, garnet and whole rock of eclogite from the Roberts-Victor Mine. Eos, Transactions of the American Geophysical Union 56 (6), 472. Papunen, H., 1970. Sulphide mineralogy of the Kotalahti and Hitura nickel-copper ores, Finland. Annales Academiae Scientiarum Fennicae, Series A III 109. Pasteris, J.D., 1982. Evidence of potassium metasomatism in mantle xenoliths. Eos, Transactions of the American Geophysical Union 63, 462. Pearson, N.J., O'Reilly, S.Y., Griffin, W.L., 1995. The crust–mantle boundary beneath cratons and craton margins: a transect across the south-west margin of the Kaapvaal craton. Lithos 36 (3–4), 257–287. Peregoedova, A., Barnes, S.J., Baker, D.R., 2006. An experimental study of mass transfer of platinum-group elements, gold, nickel and copper in sulfur-dominated vapor at magmatic temperatures. Chemical Geology 235 (1–2), 59–75. Pouchou, J.L., Pichoir, F., 1984. A new model for quantitative X-ray microanalysis. Part 1: application to the analysis of homogeneous samples. 5, 13–38. Recherche Aerospatiale 5, 13–38. Ruzicka, A., Riciputi, L.R., Taylor, L.A., Snyder, G.A., Greenwood, J., Keller, R.A., Bulanova, G.P., Milledge, H.J., 1998. Petrogenesis of mantle derived sulfide inclusions in Yakutian diamonds: chemical and isotopic disequelibrium during quenching from high temperatures. 7th International Kimberlite Conference. Sautter, V., Harte, B., 1988. Diffusion gradients in a eclogite xenolith from the Roberts Victor kimberlite pipe: 1. Mechanism and evolution of garnet exsolution in Al2O3-rich clinopyroxene. Journal of Petroleum 29, 1325–1352. Schulze, D.J., Helmstaedt, H.H., 1988. Coesite–sanidine eclogites from kimberlite; products of mantle fractionation or subduction? Journal of Geology 96 (4), 435–443. Sen, I.S., Bizimis, M., Sen, G., 2010. Geochemistry of sulfides in Hawaiian garnet pyroxenite xenoliths: implications for highly siderophile elements in the oceanic mantle. Chemical Geology 273 (3–4), 180–192. Sharp, W.E., 1966. Pyrrhotite: a common inclusion in South African diamonds. Nature (211), 402–403. Shushkanova, A.V., Litvin, Y.A., 2008. Diamond nucleation and growth in sulfide-carbon melts: an experimental study at 6.0–7.1 GPa. European Journal of Mineralogy (20), 349–355. Smart, K.A., Heaman, L.M., Chacko, T., Simonetti, A., Kopylova, M., Mah, D., Daniels, D., 2009. The origin of high-MgO diamond eclogites from the Jericho Kimberlite, Canada. Earth and Planetary Science Letters 284 (3–4), 527–537. Smith, C.B., Allsopp, H.L., Kramers, J.D., Hutchinson, G., Roddick, J.C., 1985. Emplacement ages of Jurassic–Cretaceous South African kimberlites by the Rb–Sr method on phlogopite and whole-rock samples. Transactions of the Geological Society of South Africa 88 (2), 249–266. Smyth, J.R., Caporuscio, F.A., McCormick, T.C., 1989. Mantle eclogites: evidence of igneous fractionation in the mantle. Earth and Planetary Science Letters 93, 133–141. Sobolev, N.V., 1977. Deep-seated Inclusions in Kimberlites and the Problem of the Composition of the Upper Mantle. A.G.U., Washington. Sobolev, N.V., Yefimova, E.S., Koptil, V.I., 1999. Mineral inclusions in diamonds in the norteast of the Yakutian diamondiferous province. 7th International Kimberlite Conference, Cape Town. Szabo, C., Bodnar, R.J., 1995. Chemistry and origin of mantle sulfides in spinel peridotite xenoliths from alkaline basaltic lavas, Nograd-Gomor Volcanic Field, northern Hungary and southern Slovakia. Geochimica et Cosmochimica Acta 59 (19), 3917–3927. Taylor, L.A., Neal, C.R., 1989. Eclogites with oceanic crustal and mantle signatures from the Bellsbank Kimberlite, South Africa; part I, Mineralogy, petrography, and whole rock chemistry. Journal of Geology 97 (5), 551–567. Taylor, L.A., Williams, K.L., 1972. Smythite (Fe, Ni)9S11—a redefinition. American Mineralogist 57, 1571–1577. Thomasot, E., Cartigny, P., Harris, J.W., Lorand, J.P., Rollion-Bard, C., Chaussidon, M., 2009. Metasomatic diamond growth: a multi-isotope study (13C, 15N, 33S, 34S) of sulphide inclusions and their host diamonds from Jwaneng (Botswana). Earth and Planetary Science Letters (282), 79–90. Tsai, H.M., Shieh, Y., Meyer, H.O.A., 1979. Mineralogy and S (super 34) S (super 32) ratios of sulfides associated with kimberlite, xenoliths and diamonds. In: Boyd, F.R., Meyer, H.O.A. (Eds.), The Mantle Sample: Inclusions in Kimberlites and Other Volcanics. A.G.U., Washington. Viljoen, K.S., Robinson, D.N., Swash, P.M., Griffin, W.L., Otter, M.L., Ryan, C.G., Win, T.T., 1991. Diamond- and graphite-bearing peridotite xenoliths from the Roberts Victor kimberlite, South Africa. 5th International Kimberlite Conference, Brazil, Companhia de Pesquisa de Recursos Minerais. Watmuff, I.G., 1974. Supergene alteration of the Mt Windarra nickel sulphide ore deposit, Western Australia. Mineralium Deposita 9, 199–221. Weiss, Y., Kessel, R., Griffin, W.L., Kiflawi, I., Klein-BenDavid, O., Bell, D.R., Harris, J.W., Navon, O., 2009. A new model for the evolution of diamond-forming fluids: evidence from microinclusion-bearing diamonds from Kankan, Guinea. 9th International Kimberlite Conference, Frankfurt, Germany, Lithos. Wulff-Pedersen, E., Neumann, E.-R., Jensen, B.B., 1996. The upper mantle under La Palma, Canary Islands: formation of Si–K–Na-rich melt and its importance as a metasomatic agent. Contributions to Mineralogy and Petrology 125, 113–139.