The final stages of kimberlite petrogenesis

CHEMGE-18104; No of Pages 15 Chemical Geology xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

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

The final stages of kimberlite petrogenesis: Petrography, mineral chemistry, melt inclusions and Sr-C-O isotope geochemistry of the Bultfontein kimberlite (Kimberley, South Africa) Andrea Giuliani a,b,c,⁎, Ashton Soltys a, David Phillips a, Vadim S. Kamenetsky d, Roland Maas e, Karsten Goemann f, Jon D. Woodhead e, Russell N. Drysdale g, William L. Griffin b a

KiDs (Kimberlites and Diamonds), School of Earth Sciences, The University of Melbourne, Parkville, 3010, Victoria, Australia ARC Centre of Excellence for Core to Crust Fluid Systems and GEMOC, Department of Earth and Planetary Sciences, Macquarie University, North Ryde, 2019, NSW, Australia c Department of Earth Sciences, VU Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands d School of Physical Sciences, University of Tasmania, Hobart, 7001, Tasmania, Australia e Melbourne Isotope Geochemistry, School of Earth Sciences, The University of Melbourne, Parkville, 3010, Victoria, Australia f Central Science Laboratory, University of Tasmania, Hobart, 7001, Tasmania, Australia g School of Geography, The University of Melbourne, Parkville, 3010, Victoria, Australia b

a r t i c l e

i n f o

Article history: Received 21 June 2016 Received in revised form 4 October 2016 Accepted 6 October 2016 Available online xxxx Keywords: Kimberlite Carbonates Serpentine Melt/fluid inclusions Hydrothermal modification Melt differentiation

a b s t r a c t The petrogenesis of kimberlites is commonly obscured by interaction with hydrothermal fluids, including deuteric (late-magmatic) and/or groundwater components. To provide new constraints on the modification of kimberlite rocks during fluid interaction and the fractionation of kimberlite magmas during crystallisation, we have undertaken a detailed petrographic and geochemical study of a hypabyssal sample (BK) from the Bultfontein kimberlite (Kimberley, South Africa). Sample BK consists of abundant macrocrysts (N1 mm) and (micro-) phenocrysts of olivine and lesser phlogopite, smaller grains of apatite, serpentinised monticellite, spinel, perovskite, phlogopite and ilmenite in a matrix of calcite, serpentine and dolomite. As in kimberlites worldwide, BK olivine grains consist of cores with variable Mg/Fe ratios, overgrown by rims that host inclusions of groundmass phases (spinel, perovskite, phlogopite) and have constant Mg/Fe, but variable Ni, Mn and Ca concentrations. Primary multiphase inclusions in the outer rims of olivine and in Fe-Ti-rich (‘MUM’) spinel are dominated by dolomite, calcite and alkali carbonates with lesser silicate and oxide minerals. Secondary inclusions in olivine host an assemblage of Na-K carbonates and chlorides. The primary inclusions are interpreted as crystallised alkali-Si-bearing Ca-Mg-rich carbonate melts, whereas secondary inclusions host Na-K-rich C-O-H-Cl fluids. In situ Sr-isotope analyses of groundmass calcite and perovskite reveal similar 87Sr/86Sr ratios to perovskite in the Bultfontein and the other Kimberley kimberlites, i.e. magmatic values. The δ18O composition of the BK bulk carbonate fraction is above the mantle range, whereas the δ13C values are similar to those of mantle-derived magmas. The occurrence of different generations of serpentine and occasional groundmass calcite with high 87 Sr/86Sr, and elevated bulk carbonate δ18O values indicate that the kimberlite was overprinted by hydrothermal fluids, which probably included a significant groundwater component. Before this alteration the groundmass included calcite, monticellite, apatite and minor dolomite, phlogopite, spinel, perovskite and ilmenite. Inclusions of groundmass minerals in olivine rims and phlogopite phenocrysts show that olivine and phlogopite also belong to the magmatic assemblage. We therefore suggest that the crystallised kimberlite was produced by an alkali-bearing, phosphorus-rich, silica-dolomitic melt. The alkali-Si-bearing Ca-Mg-rich carbonate compositions of primary melt inclusions in the outer rims of olivine and in spinel grains with evolved compositions (MUM spinel) support formation of these melts after fractionation of abundant olivine, and probably other phases (e.g., ilmenite and chromite). Finally, the similarity between secondary inclusions in kimberlite olivine of this and other worldwide kimberlites and secondary inclusions in minerals of carbonatitic, mafic and felsic magmatic rocks, suggests trapping of residual Na-K-rich C-O-H-Cl fluids after groundmass crystallisation. These residual fluids may have persisted in pore spaces within the largely crystalline BK groundmass and subsequently mixed with larger volumes of external fluids, which triggered serpentine formation and localised carbonate recrystallisation. © 2016 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: Department of Earth Sciences, VU Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands. E-mail addresses: [email protected], [email protected] (A. Giuliani).

http://dx.doi.org/10.1016/j.chemgeo.2016.10.011 0009-2541/© 2016 Elsevier B.V. All rights reserved.

Please cite this article as: Giuliani, A., et al., The final stages of kimberlite petrogenesis: Petrography, mineral chemistry, melt inclusions and Sr-C-O isotope geochemistry of th..., Chem. Geol. (2016), http://dx.doi.org/10.1016/j.chemgeo.2016.10.011

2

A. Giuliani et al. / Chemical Geology xxx (2016) xxx–xxx

1. Introduction Kimberlites are enigmatic volcanic rocks which have been emplaced in continental areas betweenat least 2 Ga (Kiviets et al., 1998; Graham et al., 2004; Gurney et al., 2010; Smith et al., 2012; Tappe et al., 2014; Smart et al., 2016), and the present day (~ 9–12 ka; Brown et al., 2012). They are produced by carbonate-rich melts derived from low-degree melting of carbonated peridotites (e.g., Dalton and Presnall, 1998; Brey et al., 2008) or carbonated eclogites (Nowell et al., 2004; Paton et al., 2009). The radiogenic isotope composition of archetypal kimberlites (i.e. unradiogenic Sr, mildly radiogenic Nd and Hf) resembles that of ocean island basalts (e.g., Smith, 1983; Nowell et al., 2004; Tappe et al., 2013) and requires that kimberlite melts derive from the asthenosphere or deep lithospheric mantle. A sub-lithospheric origin of kimberlites is supported by the occurrence of rare ultra-deep xenoliths and diamonds in some kimberlites (e.g., Sautter et al., 1991; Pearson et al., 2014). The composition of kimberlite melts is poorly defined because: i) kimberlites interact with wall rocks while ascending through the lithosphere (e.g., Hunter and Taylor, 1982; Luth, 2009; Kamenetsky et al., 2014a; Kamenetsky and Yaxley, 2015; Soltys et al., 2016); ii) During ascent and emplacement kimberlite magmas exsolve C-O-H fluids and degas (e.g., Sparks et al., 2006; Nowicki et al., 2008; Russell et al., 2012), losing part of their volatile content (mainly CO2 and H2O; e.g., le Roex et al., 2003; Becker and le Roex, 2006; Kjarsgaard et al., 2009); iii) The composition of kimberlites may also be modified by crystal fractionation and flow differentiation (e.g., Dawson and Hawthorne, 1973; Apter et al., 1984; Nielsen and Sand, 2008; Willcox et al., 2015); iv) Finally, during and after the late stages of crystallisation, kimberlite rocks are modified by C-O-H fluids that commonly produce abundant serpentine, brucite, variable amounts of carbonates (mainly calcite and dolomite) and lesser amounts of low-temperature minerals such as sulfates, sulfides and chlorite (e.g., Clement, 1982; Mitchell, 1986, 2008). The low-temperature modification of kimberlites is perhaps the least resolved of these processes. In particular, the origin of serpentine is a hotly debated issue. Serpentine is a major groundmass constituent in the large majority of kimberlites, including the freshest examples, and commonly replaces olivine, monticellite, carbonates and other minerals (e.g., Skinner and Clement, 1979; Mitchell, 1986, 2013; Stripp et al., 2006). Mitchell (1986, 2008, 2013) has long argued that serpentine crystallises from deuteric (i.e. late-stage magmatic) fluids, whereas Sparks and co-workers (Sparks et al., 2006, 2009; Stripp et al., 2006; Buse et al., 2010; Brooker et al., 2011; Sparks, 2013) have presented petrographic, geochemical and experimental results supporting an essentially secondary (i.e. post-magmatic) origin for serpentine. Giuliani et al. (2014a) reviewed the available O isotope data for kimberlite serpentine and concluded that serpentine is probably generated by hydrothermal fluids that include abundant heated groundwater and lesser deuteric components. The origin of serpentine has profound implications for interpreting the composition of kimberlites, and therefore their parental melts (e.g., affecting bulk H2O/CO2, Si/Ca and Mg/Ca ratios; Sparks et al., 2009), because serpentine is a major constituent and the principal host of H2O in kimberlite rocks. Another major conundrum is the abundance of alkali-rich phases in kimberlite rocks. Alkali-rich carbonates, phosphates, halides and sulfates are major constituents of primary melt inclusions in spinel (Kamenetsky et al., 2013; Abersteiner et al., this issue) and secondary fluid/melt inclusions in olivine (Kamenetsky et al., 2009, 2014b; Mernagh et al., 2011) of kimberlites worldwide. However, these phases have only been identified as significant rock constituents in the Udachnaya-East kimberlite (Siberia; Kamenetsky et al., 2004), while alkali carbonates occur in the groundmass of a kimberlite dyke from Ontario, Canada (Watkinson and Chao, 1973; Cooper and Gittins, 1974) and in an altered kimberlite breccia from Wajrakarur (India; Parhasarathy et al., 2002). Whether or not the Udachnaya-East

kimberlite represents a mantle-derived kimberlite melt, uncontaminated by crustal material, remains unclear (see Kopylova et al., 2013 vs Kamenetsky et al., 2014b). Likewise, the significance of secondary fluid/melt inclusions in olivine is debated because alkali-rich C-O-H fluids are produced after extensive fractionation of a variety of magmas (Veksler and Lentz, 2006; Hanley et al., 2008 and references therein). On the other hand, the occurrence of Na and K-rich phases in primary melt inclusions hosted by kimberlite spinel suggests that the concentrations of alkalis in kimberlite melts (e.g., K2O = 0.8 ± 0.5 wt.%; Na2O = 0.16 ± 0.14 wt.%; Becker and le Roex, 2006) may be generally underestimated. The purpose of this work is to provide new constraints on the latestage evolution of kimberlite magmas, including modification of kimberlite rocks due to interaction with hydrothermal fluids. We have undertaken a detailed petrographic and geochemical study of a hypabyssal sample from the Bultfontein kimberlite (Kimberley, South Africa). We report the texture, mineralogy, mineral chemistry, melt/ fluid inclusion composition, Sr isotope systematics of calcite and perovskite, and bulk carbonate C-O isotope composition of this sample. This study is the first to document the composition of texturally primary melt inclusions in the magmatic rims of kimberlite olivine. Comparison between groundmass mineralogy and melt-inclusion compositions suggests that monticellite, and to a lesser extent calcite and dolomite, were more abundant before hydrothermal overprinting; and that the kimberlite melt evolved from alkali-bearing, silica-carbonate through alkali-Sibearing Ca-Mg-rich carbonate to Na-K-rich C-O-H-Cl compositions, via fractional crystallisation. 2. Geological setting The Bultfontein kimberlite is part of the Kimberley cluster, which includes the De Beers, Dutoitspan, Wesselton, Kimberley and probably Kamfersdam kimberlites, along with a number of smaller pipes and sill systems (e.g., Wesselton Floors, Benfontein) (Field et al., 2008). The cluster is located in the SW part of the Kaapvaal craton (Fig. 1). The host lithologies include Karoo sedimentary rocks, mainly shales of the Dwyka formation, intruded by Karoo dolerite sills at ~181–185 Ma (Jourdan et al., 2007; Svensen et al., 2012). The Karoo rocks overlie andesitic to basaltic lavas intercalated with quartzite layers of the ~2.7 Ga (Poujol et al., 2003) Ventersdorp supergroup and felsic gneisses of the Archean basement. The Bultfontein kimberlite has a simple internal geometry consisting of a regularly shaped diatreme that tapers into a dyke ~ 850–900 m below the present surface (Clement, 1982; Field et al., 2008). The pipe has a simpler geology than the other Kimberley kimberlites because it only hosts 3 units, which are the volumetrically dominant tuffisitic (possibly pyroclastic) unit B1, the kimberlite breccia B2 which envelops unit B1, and the hypabyssal kimberlite B3, which represents an enlargement of the lower dyke (Clement, 1982; Field et al., 2008). Different dating techniques (i.e. U/Pb zircon, U/Pb perovskite, Rb/Sr phlogopite, 40Ar/39Ar phlogopite) have constrained the emplacement ages of the Kimberley kimberlites to between 81 and 90 Ma (Allsopp and Barrett, 1975; Davis, 1977; Fitch and Miller, 1983; Batumike et al., 2008; Griffin et al., 2014). The most precise age estimate for the Bultfontein kimberlite was provided by Kramers et al. (1983), who obtained a Rb/Sr age of 84.0 ± 0.9 Ma for phlogopite grains in metasomatised mantle xenoliths. The Kimberley kimberlites are classified as Group I or archetypal kimberlites based on their petrography, mineral chemistry and Sr-Nd-Hf isotopic composition (Clement, 1982; Smith, 1983; Shee, 1984, 1985; Mitchell, 1986; Nowell et al., 2004; Becker and le Roex, 2006; Woodhead et al., 2009; Griffin et al., 2014). 3. Petrography Sample BK is a 5 × 3 × 2 cm offcut from a larger sample of hypabyssal kimberlite in the De Beers Group collection. The sample was originally collected at the Boshof Road Dumps, which host waste from historic



3

Fig. 1. Schematic map of southern Africa showing the location of major diamond mines in archetypal (or Group I) kimberlites, including those in Kimberley, and orangeites (modified from Field et al., 2008; Giuliani et al., 2015). Some of the major structural units (after Thomas et al., 1993) are also shown. SCLM: sub-continental lithospheric mantle.

mining of the Bultfontein kimberlite (Robey J.V.A., personal communication). Therefore, while the exact location within the pipe cannot be constrained, it is likely that this sample derives from unit B3, which is the only hypabyssal unit in the Bultfontein pipe, or from a dyke crosscutting the other units. The petrography of sample BK was characterised by optical microscopy and SEM (scanning electron microscopy) analysis (see Methods in Supplementary Material). Sample BK has a porphyritic texture (Fig. 2a, b) with macrocrysts (~15 vol.%) and euhedral/subhedral phenocrysts (~30 vol.% including micro-phenocrysts, i.e. b0.5 mm) of olivine, lesser phlogopite (~ 5 vol.%), and scarce ilmenite, orthopyroxene and garnet macrocrysts (b 1 vol.%) in a carbonate and serpentine-rich groundmass. ‘Macrocrysts’ are loosely defined as grains larger than 0.5–1.0 mm with anhedral shapes and/or deformation features indicative of derivation from pre-existing rocks (e.g., Clement et al., 1984); while ‘phenocrysts’ generally include a xenocrystic core (see Giuliani et al., 2016, and Section 4.1). Olivine macrocrysts are more fractured and serpentinised than the phenocrysts, with serpentine replacement limited to the outermost rims of phenocrysts (Fig. 2c). The size of olivine phenocrysts (and micro-phenocrysts) ranges from ~1.0 to 0.1 mm with no preferential size distribution (i.e. no dominant size population). Phlogopite macrocrysts and phenocrysts commonly show concentric zoning with one or more rims/layers around a xenocryst or antecryst (i.e. produced by earlier magmatic activity) core (Giuliani et al., 2016). Apart from olivine micro-phenocrysts, the groundmass includes abundant, ≤ 20 μm grains of apatite (Figs. 2e, f and 3c), complexly zoned (0.1–0.5 mm) euhedral to subhedral grains of phlogopite (Fig. 2c; see also Giuliani et al., 2016), smaller (b0.1 mm) crystals of MUM spinel (solid solution of magnesian ulvospinel, ulvospinel and magnetite; Mitchell, 1986), perovskite (Fig. 2d, e) and minor ilmenite (ilmenite-geikielite solid solution), barite and Ni ± Fe sulfides in an interstitial matrix of calcite, serpentine and lesser dolomite. The relative proportions of calcite, dolomite and serpentine vary on a cm scale such that either calcite, serpentine or, less commonly, dolomite is the dominant matrix mineral (Fig. 3a, b). The estimated dolomite/calcite ratio in the groundmass is 1/4 by volume. Matrix calcite, dolomite and serpentine exhibit complex textural relationships. In the calcite-rich zones dolomite occurs interstitial to calcite while serpentine is only present as pseudomorphic replacement of monticellite. In the dolomite-rich zones dolomite is intergrown with minor amounts of calcite and rare serpentine (Fig. 3c). In the serpentine-rich areas calcite appears resorbed whereas euhedral dolomite is occasionally observed (Fig. 3d). Serpentine after olivine appears to locally replace the adjacent

groundmass and, importantly, also replaces dolomite as shown in Fig. 3e. The BK groundmass includes 10–30 μm, euhedral grains pseudomorphed by serpentine (Fig. 3f) sometimes intergrown with carbonates; occasional fresh remnants suggest that this phase was monticellite. The oxide minerals sometimes occur in clusters, comprising MUM spinel and perovskite grains with ilmenite inclusions in spinel and/or perovskite (Fig. 3g). MUM spinels can have a core with TIMAC (titanian magnesian aluminous chromite; Mitchell, 1986) composition, which may encase a xenocrystic core of Cr spinel (Fig. 3h). FE-SEM observations have shown that sub- to μm-sized grains of Na or Ca carbonate and strontianite (SrCO3) may occur within interstitial serpentine (Fig. 3i) and calcite, respectively. Inclusions of strontianite and Sr-Ca carbonates were previously reported in the Ti-rich ‘magmatic’ rims of phlogopite macrocrysts in sample BK (Giuliani et al., 2016). Some apatite grains in the groundmass of sample BK have occasional inclusions of dolomite (Fig. 3j). The other prominent features of the BK groundmass are calcite ± dolomite (Fig. 3b) and less abundant serpentine-calcite segregations. The pure carbonate segregations usually have rounded shapes (Fig. 2g), are up to 0.5 mm in diameter and consist of a few large calcite crystals with minor intergrown dolomite and inclusions of oxide minerals and phlogopite near the contact between the segregations and the groundmass. The serpentine-calcite segregations occupy interstitial spaces between micro-phenocrysts, and include serpentine in the core and inclusion-free calcite grains overgrowing groundmass calcite along the segregation rims. 4. Mineral chemistry The major- and trace-element compositions of minerals in sample BK were determined by electron microprobe and laser ablation (LA) ICPMS analyses, respectively (see Methods in Supplementary material), respectively. The results are summarised below with extended data-sets provided in Supplementary Tables S1 to S7. The composition of BK matrix phlogopite was described previously by Giuliani et al. (2016) and exhibits a typical kimberlitic trend of increasing Al and Ba with decreasing Fe and Ti (Mitchell, 1986, 1995; Reguir et al., 2009). 4.1. Olivine SEM observations show that olivine macrocrysts, phenocrysts and micro-phenocrysts in sample BK are zoned (Fig. 3k, l) with


4


(a)

5 mm

(b)

ol

(c)

(d)

core phl rim

ol ol

ol P

ilm ol

2 mm

ol

(e)

srp

300 µm

300 µm

ol M

ol

ol

phl

(f)

(g)

spl ap

cc pvk

spl

phl

cc

ap

cc

srp

100 µm

150 µm

(h)

200 µm

(i)

s.f.i.

(j) ol

spl

ol

p.f.i.

p.f.i.

spl s.f.i. phl 80 µm

100 µm

80 µm

Fig. 2. Thin section scan (a) and transmitted-light optical photomicrographs (b–j) of kimberlite sample BK. (a) Porphyritic texture with large (N1 mm) transparent macrocrysts of olivine and brown macrocrysts of phlogopite. (b) Transmitted light photomicrograph showing abundant macrocrysts and phenocrysts of olivine (ol) and a single ilmenite (ilm) macrocryst. (c, d) Macrocrysts (M) and micro-phenocrysts (P) of olivine, zoned phenocrysts and micro-phenocrysts of phlogopite (phl), and abundant oxide minerals (black and dark brown grains) in a fine-grained carbonate-rich groundmass. (e) Apatite (ap), spinel (spl) and perovskite (pvk) grains in calcite (cc) groundmass. (f) Apatite in serpentine (srp) groundmass. (g) Carbonate segregation. (h) Inclusions of spinel and phlogopite in the rim of an olivine phenocryst (grain mount). (i) Inclusions of spinel and swarms of secondary fluid inclusions (s.f.i.) in olivine phenocryst. (j) Primary fluid/melt inclusions (p.f.i.) in the rim of an olivine phenocryst. (To view color photomicrographs, the reader is referred to the web vesion of this article).

very rare exceptions, which is similar to the observations of Boyd and Clement (1977) for the De Beers kimberlite. Zoning in olivine is defined by cores showing higher and, in fewer cases, lower Mg# (atomic Mg/(Mg + Fe); 85.5–93.6) than the rims (Fig. 4; Supplementary Table S1). Most core analyses fall in the range of peridotite xenoliths entrained by kimberlite magmas in the Kimberley area (e.g., Erlank et al., 1987) and other cratons (Menzies et al., 2004; Sobolev et al., 2009; Fig. 4a); this is consistent with previous studies of kimberlite olivine (e.g., Kamenetsky et al., 2008; Nielsen and Sand, 2008; Brett et al., 2009; Bussweiler et al., 2015; Giuliani and Foley, 2016). Rare cores are significantly enriched in Fe (Mg# ~ 85–86) and these resemble the olivine in Fe-rich dunite xenoliths from the Kimberley kimberlites (Dawson et al., 1981; Rehfeldt et al., 2007).

The olivine rims show a very restricted range of Mg# (88.8 ± 0.2 (1sd); n = 29), but variable Ni, Ca and Mn contents (Fig. 4), which appears to be a characteristic feature of olivine in fresh kimberlite rocks (Kamenetsky et al., 2008; Brett et al., 2009; Arndt et al., 2010; Pilbeam et al., 2013; Bussweiler et al., 2015). The Mg# of olivine rims in BK are comparable with the composition of rims in the De Beers (88–90; Boyd and Clement, 1977; Moore, 2012) and Wesselton kimberlites (87–88; Shee et al., 1994). 4.2. Calcite Calcite grains have high Sr concentrations (SrO of 0.66–1.54 wt.%; Supplementary Fig. S1 and Table S2), comparable to the results for the De Beers kimberlite (Exley and Jones, 1983). Conversely, BK calcite



5

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)

(k)

(l)

Fig. 3. Back-scattered electron (BSE) SEM (a–f, h, k, l) and FE-SEM (g, i, j) images showing details of kimberlite sample BK. (a) serpentine (srp)-rich zone traversing calcite (Cc)-rich zone in BK groundmass. (b) Calcite ± dolomite segregation (Cc ± Dol), partly serpentinised olivine (ol), phlogopite (phl) and spinel (spl) in BK groundmass. (c) Dolomite (dol)-rich matrix; note the abundance of apatite. (d) Resorbed calcite rimmed by euhedral dolomite in serpentine matrix. (e) Pseudomorphic serpentine after olivine and serpentine replacing dolomite. (f) Pseudomorphic replacement of monticellite (mtc) by serpentine and serpentine + calcite in calcite-rich matrix. (g) Aggregate of oxide minerals which comprise perovskite (pvk), MUM spinel (mum) and minor ilmenite (ilm); note inclusion of olivine in MUM spinel. (h) Zoned spinel grain in BK groundmass with a core of xenocrystic Cr spinel (crsp) embedded in TIMAC spinel (timac), which is sequentially rimmed by MUM spinel; MUM spinel hosts inclusions of zoned perovskite and is discontinuously rimmed by magnetite (not shown). (i) Inclusions of Na-rich carbonate (NaC) in groundmass serpentine. (j) Inclusion of dolomite in groundmass apatite. (k) Zoned olivine macrocryst with perovskite inclusion in the rim zone; note the irregular outline of the core-rim boundary. (l) Two cores of distinct composition in an olivine phenocryst (grain mount).

has BaO concentrations (0.06–0.18 wt.%) lower than De Beers groundmass carbonates (0.24–0.44 wt.%), but higher compared to secondary calcite in late-stage veins in the De Beers kimberlite (b0.05 wt.%; Exley and Jones, 1983). Relatively fresh sills in the ‘Water Tunnels’ of the Wesselton mine contain carbonates with SrO and BaO concentrations of ~ 0.4–0.9 wt.% and ~ 0.09–0.20 wt.%, respectively White et al., 2012); similar to those of BK calcite. There is a strong correlation between the concentrations of different REE (e.g., La and Ce) and between REE and Y contents in BK calcite (Supplementary Fig. S1). Variable REE concentrations denote small-scale variations in the composition of the kimberlite melt, probably resulting

from crystallisation of perovskite and apatite, which are the major REE hosts in kimberlite rocks. 4.3. Apatite Apatite is a major groundmass constituent in sample BK and in the Kimberley kimberlites generally (Pasteris, 1983; Clement et al., 1986; Shee et al., 1994; Ogilvie-Harris et al., 2009; White et al., 2012). BK apatite shows uniform composition, intermediate between fluoro- and hydroxyl-apatite with minor impurities, mainly SiO2 (1.8–2.4 wt.%, with one value of 1.0 wt.%; Supplementary Table S3). Fluorine and chlorine


6


(a)

0.6

NiO (wt.%)

0.5

(a)

0.92

Fe3+/(Fe 3++Al+Cr) 0.88

0.4 0.84 0.3 0.80

core

0.2

rims

0.1

0.76

unzoned

Mg#

Fe2+/(Fe 2++Mg)

(b) MnO (wt.%)

0.23

(b)

TiO2 (wt.%) 22

0.20 0.17

20

0.14 18 0.11 16

0.08

Mg#

Fe2+/(Fe 2++Mg)

(c)

CaO (wt.%)

0.30

5.0 0.25

(c)

Al2O3 (wt.%)

4.5

0.20

4.0

0.15 0.10

3.5

Mg#

0.05

3.0

Fe2+/(Fe 2++Mg)

0.00 85

86

87

88

89

90

91

92

93

94

Fig. 4. Mg# vs NiO (a), MnO (b) and CaO (c) variation diagrams for olivine grains in kimberlite sample BK. Grey field: olivine in Kimberley dunites (Dawson et al., 1981); thick black line: Mg# variation of olivine in metasomatised peridotites from the Kimberley kimberlites (Erlank et al., 1987); purple field: olivine in peridotites from Siberian kimberlites (Sobolev et al., 2009). (To view colour figures, the reader is referred to the web version of this article).

concentrations vary between 1.5 and 1.9 wt.% and 0.04–0.07 wt.%, respectively, and the OH content of the average BK apatite composition is 1.75 wt.% (estimated following Ketcham, 2015). This composition is typical of apatite in kimberlites worldwide (Mitchell, 1986).

2.5 0.58

0.60

0.62

0.64

0.66

0.68

Fig. 5. Atomic Fe2+/(Fe2+ + Mg) vs atomic Fe3+/(Fe3+ + Al + Cr) (a), TiO2 (b) and Al2O3 (c) variation charts for MUM spinel grains in kimberlite sample BK. The arrow indicates the core to rim compositional trend.

poorly developed here. Magnetite also forms occasional isolated grains in the BK groundmass. The small size (b10 μm) of spinel inclusions in olivine rims did not allow quantitative EMP determinations, but semiquantitative EDS analyses show that most correspond to TIMAC spinel.

4.5. Perovskite 4.4. Spinel Some groundmass grains (e.g., Fig. 3h) preserve the entire spectrum of compositions occurring in sample BK. These grains consist of a small (b 10 μm) core of xenocrystic Cr-spinel, a ≤20 μm wide overgrowth of TIMAC (i.e. kimberlitic chromite) spinel (one EMP analysis: ~ 50 wt.% Cr2O3; Supplementary Table S4) and a rim of MUM spinel. MUM spinel shows a progressive enrichment in Fe2 + and Fe3 + (FeOtotal = 55–66 wt.%), and a decrease in TiO2 (15–22 wt.%) at constant Al2O3 concentrations (3.2–3.7 wt.%, one outlier of 5.3 wt.%) towards the rim (Fig. 5). These compositions are similar to those of MUM grains in the Wesselton kimberlite (Shee, 1984; White et al., 2012), but with marginally lower MgO and Al2O3; and plot in the field of worldwide kimberlite spinels (Mitchell, 1986; Roeder and Schulze, 2008). A discontinuous layer of magnetite sometimes rims MUM spinel, although the characteristic atoll shape of kimberlitic spinel elsewhere (e.g., Mitchell, 1986) is

Perovskite grains in the Bultfontein kimberlite exhibit zoning from BSE-bright cores to BSE-darker rims with a sharp boundary between them (Fig. 3h). Individual analyses of perovskite represent one of the two zones or, more frequently, a mixture due to the small size of the perovskite grains (mainly 20–30 μm; Figs. 2e and 3g). Perovskite major (Supplementary Table S5) and trace element results (Supplementary Table S6) show significant variations (Fig. 6), which are consistent with zoning from a core enriched in LREE, Th, Ta, Pb and to a lesser extent Na, to a rim depleted in these elements and marginally richer in Ca. This is the ‘normal zoning’ trend seen in kimberlites worldwide (e.g., Chakhmouradian and Mitchell, 2000; Chakhmouradian et al., 2013; Castillo-Oliver et al., 2016) including the Dutoitspan kimberlite (Ogilvie-Harris et al., 2009) and is due to progressive melt depletion in incompatible elements (e.g., LREE, Th, Ta) with perovskite crystallisation.



0.045

(a)

Na (apfu)

(b)

0.029

Ce (apfu)

0.027

0.040

7

0.025 0.035

0.023

0.030

0.021 0.019

0.025

0.017

Ca (apfu)

(d) Nb (apfu)

0.012

Fe (apfu)

0.045

Ca (apfu)

(c) 0.011

0.040

0.010

0.035

0.009 0.008

0.030

0.007

Ca (apfu)

Ca (apfu)

0.025

0.89

0.90

0.91

0.92

7500

0.93

0.006 0.89

(e)

25000

Na (ppm)

7000

0.90

0.91

0.92

0.93

(f)

Ce (ppm)

24000 23000

6500

22000

6000

21000 5500

20000

5000

19000

4500

Th (ppm)

(g)

Th (ppm)

(h)

55

Pb (ppm)

50

Ta (ppm)

1700

18000

45 1500

40

1300

35

1100

30 25

900

20 700

15

Th (ppm)

(i)

Th (ppm)

(j)

10 450

U (ppm)

Mn (ppm)

280

400

260

350

240

300

220

250

Th (ppm) 200

Th (ppm) 200

0

2000

4000

6000

8000

10000

0

2000

4000

6000

8000

10000

Fig. 6. Major and trace element co-variation diagrams for perovskite grains in kimberlite sample BK. Major element concentrations are expressed as atoms per formula unit (apfu). The arrows indicate core to rim compositional trends.

4.6. Serpentine Fifty-nine EMP analyses of serpentine reveal large compositional variations (Fig. 7 and Supplementary Table S7) with SiO2 concentrations and Mg# between 41 and 50 wt.% and 82–89, respectively. Calcium, Na and K abundances in serpentine are usually very low (e.g., Deer et al., 1992). However, analyses of BK serpentine show CaO, Na2O and K2O

concentrations up to 3.3, 5.8 and 1.1 wt.%, respectively. Serpentine analyses with high concentrations of CaO, Na2O and/or K2O at low SiO2 abundances are attributed to contributions from carbonate minerals, which is consistent with the occurrence of μm-sized carbonate inclusions in serpentine grains (Fig. 3i). Serpentine analyses with minimal or no contribution from carbonate inclusions display a weak positive correlation between K and Si (Fig. 7g) and, to a lesser extent, K and Al


8


(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Fig. 7. Major element co-variation diagrams for serpentine in kimberlite sample BK. Circles and square symbols represent analyses with minimal and significant contributions from included carbonates, respectively. ‘gmass’: analyses of groundmass serpentine; ‘after Mtc’: serpentine replacing monticellite; ‘after Oliv’: serpentine replacing olivine; ‘other’: serpentine replacing calcite and showing unclear textural relationships. (To view colour figures, the reader is referred to the web version of this article).

contents (Fig. 7h), which could be due to limited solid solution between serpentine and a smectite component. In contrast, Ca and Na do not show any correlation with Si (Fig. 7e, f) and Al concentrations (not shown). Donaldson and Reid (1982) measured Na2O and K2O concentrations up to 0.4 and 0.1 wt.%, respectively for serpentine in kimberlite dykes from the De Beers pipe. Significantly higher Na2O (up to 3.3. wt.%) and K2O contents (up to 2.2 wt.%) are found in saponite and chloritesmectite grains in the Wesselton ‘water tunnel’ sills (White et al., 2012) and Wesselton tuffisitic units (Mitchell et al., 2009), respectively. However, the high Al2O3 contents of saponite (17–21 wt.%) and chlorite-smectite (6–9 wt.%) are distinct from the composition of serpentine in this study (Al2O3 ≤ 0.7 wt.%).

Different textural varieties of serpentine in sample BK show distinct compositions. Groundmass serpentine (n = 29) has the highest Mg# values (≥85.5, except for two analyses), and TiO2, Al2O3 and Cr2O3 concentrations comparable to pseudomorphic serpentine replacing monticellite (n = 5). Serpentine after olivine (n = 19) is distinguished by lower Mg# values (≤85.1), TiO2 and Al2O3 but higher Cr2O3 contents than the other two textural types (Fig. 7). There is a good positive correlation between Mg# values and SiO2 concentrations for serpentine grains after olivine (once the analyses with carbonate contribution are removed; i.e. blue circles in Fig. 7a). Two analyses of serpentine replacing calcite are in the range of pseudomorphic serpentine after olivine (Table S7).



((Na,K)2Ca(CO3)2), magnesite], phlogopite, oxide minerals (perovskite, ilmenite, MUM spinel), phosphates [apatite and bradleyite (Na3Mg(PO4)(CO3)), chlorides (sylvite and sylvite-halite solid solution) and arcanite (i.e. K2SO4). We interpret these inclusions to be the crystallisation products of an alkali-Si-bearing Ca-Mg-rich carbonate melt (see Section 8.1). Secondary inclusions in olivine occur along trails and elongated swarms (Fig. 2i), which probably represent healed fractures. These inclusions have rounded shapes and are mostly b5 μm in size. Inclusion decrepitation in the SEM vacuum reveals daughter crystals of halite, sylvite and their solid-solution mixtures in a matrix of Na ± K carbonate (Fig. 8f, g). SEM-EDS analyses also show minor Ca, S and F in the bulk composition of these inclusions. Secondary inclusions in olivine therefore host a Na-K-rich C-O-H-Cl fluid. Mernagh et al. (2011) documented the occurrence of similar daughter Na-K-Ca carbonates and minor sulfates in secondary fluid inclusions in Wesselton olivine.

5. Melt/fluid inclusions 5.1. Inclusions in olivine The compositions of inclusions in olivine and spinel of sample BK were investigated by field emission (FE) SEM (see Methods in Supplementary material). Texturally primary (i.e. not associated with healed fractures) inclusions in olivine phenocrysts and micro-phenocrysts are commonly concentrated in the rim zones near the grain boundaries. Olivine rims host single-phase inclusions of chromite and less common perovskite, MUM spinel, phlogopite and ilmenite (Figs. 2h, i and 3k). Spinel, ilmenite and perovskite inclusions in olivine rims have also been described from the Wesselton kimberlite (Shee, 1984). Rare unzoned olivine grains host abundant disseminated needles of rutile, which are also observed in olivine from the Wesselton (Shee, 1984) and De Beers pipes (Clement et al., 1986). The only inclusion type we observed in olivine cores is low-Ti, Cr-spinel, i.e. the spinel typical of mantle peridotites (e.g., Schulze, 2001). Primary multiphase inclusions are uncommon and occur as isolated inclusions or in small clusters close to the grain boundaries (Figs. 2j and 8a). These inclusions are usually ovoid in shape with irregular edges, up to 15 μm across, and are only partially filled when observed with the SEM, due to plucking and/or inclusion decrepitation. They host an assortment of phases (Fig. 8a–e), which include (in order of decreasing abundance): carbonates [dolomite, calcite, zemkorite

(a)

primary

5.2. Inclusions in MUM spinel MUM spinel grains in sample BK host single-phase inclusions of perovskite, ilmenite, olivine, rare baddeleyite (ZrO2) and primary multiphase inclusions in the rim zones. They have rounded to ovoid or more complex shapes and their size is commonly b 5 μm. The mineralogical content varies between inclusions and is dominated by carbonates [calcite, dolomite, zemkorite, shortite (Na2Ca2(CO3)3) and

(b)

primary

mgs

zmk

9

primary

(c)

brdl ap

dol

KNaCl

pvk ap

hal

dol

olivine

olivine

olivine 25 µm

5 µm

primary

(d)

5 µm

primary

(e)

dol

secondary

phl hal

arcan

sylv

zmk olivine

5 µm

secondary

(g)

hal olivine

5 µm

olivine

5 µm

primary

(h)

primary CaSi

wth

NaC olivine

1 µm

(i)

Na-cc

cc

hal

5 µm

NaKCl

zmk

brdl

sylv

(f)

dol

zmk

ap shrt mum

1 µm

CaF

mum

Fig. 8. Back-scattered electron (BSE) SEM (a) and secondary electron (SE) FE-SEM (b–i) images of primary (a–e) and secondary inclusions (f, g) in olivine and primary inclusions in MUM spinel (h, i). (a) Primary, multiphase inclusion with apatite (ap), perovskite (pvk) and zemkorite (zmk) in olivine rim; this is probably a composite solid + (crystallised) melt inclusion. (b– e) Primary, multiphase inclusion of crystallised fluid/melt in the rims of olivine crystals; included daughter minerals comprise dolomite (dol), magnesite (mgs), halite (hal), bradleyite (brdl), apatite, K-Na halide (KNaCl), phlogopite (phl), zemkorite, sylvite (sylv) and arcanite (arcan; see text for chemical formulas of the unusual phases). (f, g) Decrepitated, secondary fluid inclusions situated along inclusion trails and swarms in olivine; daughter minerals comprise Na (±K) carbonate, halite (hal), sylvte and halite-sylvite solid solution (NaKCl). (h, i) Primary, multiphase inclusion of crystallised fluid/melt in MUM spinel (mum); the daughter mineral assemblages include calcite (cc), zemkorite, apatite, witherite (wth), Na-bearing calcite (Na-cc), shortite (shrt), a Ca silicate (CaSi, possibly cuspidine) and fluorite (CaF2).


10


witherite (BaCO3)] with minor apatite and one instance of a Ca silicate (perhaps cuspidine) and Na-bearing fluorite (Fig. 8h, i). As per the primary multiphase inclusions in olivine, we interpret the primary inclusions in MUM spinel to represent the crystallisation product of an alkali-Si-bearing Ca-Mg-rich carbonate melt.

is 0.70454 ± 0.00019 (95% confidence; n = 15/15; MSWD = 0.61; Fig. 9 and Fig. S3), which is within the range of previous measurements of perovskite in the Kimberley kimberlites (0.70429–0.70461; Woodhead et al., 2009). 7. Bulk carbonate C-O isotopes

6. Sr isotope geochemistry 6.1. Calcite Sr isotopes Inclusion-free groundmass grains and segregations of calcite were selected for in-situ Sr isotope analysis by laser ablation (LA) multi-collector (MC) ICPMS ICPMS (see Methods in Supplementary material). The 87Sr/86Sr ratio of calcite in sample BK varies between 0.70412 ± 0.00022 (2se) and 0.70508 ± 0.00016 (Supplementary Table S8 and Fig. S2). 87Rb/86Sr values of calcite are very low (≤ 0.018, but ≤ 0.006 for 35/39 analyses) with no correlation between measured 87Sr/86Sr and 87Rb/86Sr values (Supplementary Fig. S2). 87Sr/86Sr values at 84 Ma (emplacement age of the Bultfontein kimberlite; Kramers et al., 1983) thus are identical to the measured values. The weighted mean calcite 87Sr/86Sr(84Ma) value is 0.70432 ± 0.00005 (95% confidence; n = 35/39; MSWD = 5.1; Fig. 9 and Fig. S2). This ratio is marginally lower than that of groundmass calcite and dolomite in a carbonaterich dyke from the De Beers kimberlite (0.7046–0.7051; Exley and Jones, 1983).

Bulk carbonate C-O isotope analyses of two aliquots of sample BK are compared with previous analyses of bulk carbonates from the Kimberley kimberlites in Fig. 10 (see Methods in Supplementary material for details of the analytical procedure). The δ13C and δ18O values of the Bultfontein sample are − 4.3 and − 5.2‰ (compared to VPDB), and 11.7 and 10.5‰ (compared to VSMOW) respectively. The δ13C values of the bulk BK carbonates are within the mantle range defined by most carbonatites and kimberlites from worldwide localities (e.g., Taylor et al., 1967; Deines, 1989; Wilson et al., 2007; Giuliani et al., 2014a). The δ18O values extend towards heavier isotopic compositions, which is a common feature of bulk carbonate analyses of kimberlites (Giuliani et al., 2014a). The BK carbonate O isotope composition is within uncertainty of the average values for carbonates in hypabyssal kimberlites from Wesselton (11.5 ± 2.7‰) and the Benfontein Sills (11.1 ± 3.4‰; Sheppard and Dawson, 1975; Kobelski et al., 1979; Kirkley et al., 1989). In contrast, the δ13C composition is marginally higher than that of Wesselton (−6.4 ± 0.8‰), but similar to Benfontein carbonates (−5.2 ± 0.5‰) (Fig. 10), which might indicate isotopically different sources for these kimberlites.

6.2. Perovskite Sr isotopes 8. Discussion Perovskite grains in sample BK are b 50 μm in size and most grains are in the 20–30 μm size range. Therefore, for MC-LA-ICPMS analyses of perovskite (see Methods for details) we employed a 26 μm beam size, which resulted in elevated analytical uncertainties (2se) of between 0.0004 and 0.0015 87Sr/86Sr (Supplementary Table S9). The measured 87Sr/86Sr values range from 0.7038 ± 0.0012 (2se) to 0.7051 ± 0.0007 and show no correlation with 87Rb/86Sr ratios (0.001–0.042; Supplementary Fig. S3). The perovskite 87Sr/86Sr(84Ma) weighted mean

Bultfontein kimberlite sample BK shows textural and mineralogical features that are common to many hypabyssal kimberlites worldwide (e.g., Mitchell, 1986, 2008; Skinner, 1989) including other Kimberley kimberlites (e.g., Pasteris, 1980; Clement, 1982; Shee, 1985). These include abundant olivine and lesser phlogopite macrocrysts and (micro-)phenocrysts in a groundmass dominated by carbonates and serpentine with apatite and lesser spinel, perovskite and phlogopite.

Fig. 9. Strontium isotope composition of calcite and perovskite in kimberlite sample BK reported as 87Sr/86Sr at 84 Ma (emplacement age of the Bultfontein kimberlite; Kramers et al., 1983) compared to initial 87Sr/86Sr ratios of perovskite in the Kimberley and southern African Jurassic-Cretaceous kimberlites, southern African transitional kimberlites and orangeites (Woodhead et al., 2009); initial 87Sr/86Sr values of magmatic and secondary carbonates in a kimberlite dyke from the De Beers pipe (Exley and Jones, 1983); Sr isotope compositions at 84 Ma of wall rocks in the underground sections of the Kimberley diamond mines (Barrett and Berg, 1975); and present-day isotopic ratios of groundwaters in the Kimberley diamond mines (Barrett and Berg, 1975). Dashed lines for ‘Kimberley country rocks’ indicate that the 87Sr/86Sr values extend to values above 0.8 for some pegmatites (Barrett and Berg, 1975).



4

6

8

10

12

14

16

18

20

δ18 O (‰)

-1 -3 -5 -7 -9

Kimberley area Benfontein

-11 δ13C -13

(‰)

Wesselton hypabbysal Bultfontein, BK

Fig. 10. δ13CVPDB - δ13CVSMOW composition of bulk carbonate analyses for sample BK from the Bultfontein kimberlite. Bulk carbonate analyses for the other Kimberley kimberlites, including hypabyssal samples from Wesselton and the Benfontein sills, are reported for comparison (Deines and Gold, 1973; Sheppard and Dawson, 1975; Kobelski et al., 1979; Kirkley et al., 1989). Carbonate mantle range (box with dotted outline) is after Taylor et al. (1967), Deines (1989), Wilson et al. (2007) and Giuliani et al. (2014a).

Clement (1982), Shee (1985) and le Roex et al. (2003) estimated that hypabyssal macrocrystic kimberlites in the Kimberley area contain between ~10–30 vol.% of olivine macrocrysts; ~15 vol.% of sample BK consists of olivine macrocrysts. Given the above petrographic features, it is likely that sample BK came from the only hypabyssal unit (B3) of the Bultfontein pipe or from a dyke that intruded the pipe during or after emplacement of the main kimberlite units. Hypabyssal kimberlite units in the Kimberley pipes display variable mineralogical composition within and across different units. This is exemplified by samples from the De beers pipe where the groundmass may be dominated by calcite, monticellite or phlogopite (above 50 vol.% of each; Clement et al., 1986). Calcite, serpentine and apatiterich kimberlites, such as that reported in this study, are widespread in the root zones of the Kimberley-area pipes (Clement, 1982). In addition, the major element characteristics of olivine (Fig. 4), spinel (Fig. 5) and perovskite (Fig. 6), the Sr isotope compositions of calcite and perovskite (Fig. 9), the carbonate C-O isotopic signature (Fig. 10) and the Nd isotope systematics of leachates of sample BK (see Appendix I and Table S10 in Supplementary material) are consistent with results from the other Kimberley kimberlites. Therefore, the conclusions drawn from this study are considered to be applicable to most of the Kimberley kimberlites, and more broadly to carbonate-serpentine kimberlites in southern Africa and elsewhere. In the following sections we employ the compositions of primary and secondary inclusions in olivine and spinel of the Bultfontein kimberlite to discuss how kimberlite melts that emplace non-explosively (i.e. as hypabyssal bodies) may evolve during fractional crystallisation. We then address the origin of carbonate minerals and serpentine, i.e. the most abundant groundmass constituents of sample BK, and present a model that summarises the late petrogenetic history of this kimberlite. 8.1. Constraints on the evolution of kimberlite melts from the compositions of melt/fluid inclusions Primary alkali-Si-bearing Ca-Mg-rich carbonate melt/fluid inclusions similar to those in BK olivine and spinel have been identified previously in spinel grains of the Koala (Canada; Kamenetsky et al., 2013) and Venetia (South Africa; Abersteiner et al., this issue) kimberlites, and ilmenite and olivine grains of Bultfontein polymict breccia xenoliths (Giuliani et al., 2012, 2013, 2014b), which are regarded as failed kimberlite intrusions at mantle depths (e.g., Lawless et al., 1979; Pokhilenko, 2009; Giuliani et al., 2014b). However, it is noteworthy that the melt/ fluid inclusions documented in this study are confined to the outer rims of olivine and to grains of MUM spinel. This spinel composition is produced after chromite formation, late in the crystallisation sequence

11

of the kimberlite groundmass (e.g., Shee, 1984; Mitchell, 2008), from melts that have already undergone extensive fractionation (Pasteris, 1983; Shee, 1984; Jones and Wyllie, 1985; Mitchell, 1986; Roeder and Schulze, 2008). We therefore suggest that these alkali-Si-bearing CaMg-rich carbonate fluids/melts represent a fractionated kimberlite melt, or an exsolved fluid, trapped after crystallisation of olivine and, probably, other phases (e.g., chromite, ilmenite). Similar melt compositions in chromite grains of the Koala and Venetia kimberlites, and olivine and ilmenite grains of Bultfontein polymict breccia xenoliths, might indicate that also these inclusions represent residual or exsolved fluids rather than parental melts/fluids. This interpretation agrees with the occurrence of primary CO2-rich fluid inclusions in apatite of carbonatite rocks, which represent examples of fluids associated with the carbonatite magma rather than the parental melt (Costanzo et al., 2006). Secondary fluid inclusions in BK olivine are dominated by daughter Na ± K carbonates and Na-K chlorides, and, therefore, probably represent a Na-K-rich C-O-H-Cl fluid. We suggest that these inclusions trapped the remaining fluid(s) after crystallisation of the kimberlite groundmass. Comparison between the compositions of primary melt/ fluid inclusions in BK olivine and spinel with secondary inclusions in olivine reveals a progressive increase in alkalis and Cl in the residual kimberlite fluids. Veksler and Lentz (2006, and references therein) and Hanley et al. (2008, and references therein) documented a similar evolution of residual fluids in carbonatitic, ultramafic, mafic and felsic magmatic rocks towards alkali-Cl-rich C-O-H compositions. Smith et al. (2004) have reported the only detailed study documenting the alteration of country rocks by kimberlitic fluids in the Murowa-Sese cluster (Zimbabwe). They noted the formation of carbonates and alkali-rich amphiboles and pyroxenes in granite wall rocks, but overlooked the high Na/K of the metasomatising fluids compared to the low Na/K of associated kimberlites. We suggest that the Na-K-rich CO-H-Cl fluids trapped as secondary inclusions in olivine of the Bultfontein and other kimberlites (Kamenetsky et al., 2009, 2014b; Mernagh et al., 2011) represent examples of kimberlite fluids, residual after groundmass crystallisation. We speculate that these fluids could be released into the kimberlite wall rocks, causing alteration (see Smith et al., 2004). However, in a study of kimberlite interaction with gneissic wall rocks in the De Beers pipe, Ferguson et al. (1973) identified enrichment in Ca, Mg, Sr, CO2 and H2O, whereas the alkalis exhibited erratic behaviour. Therefore, other factors appear to influence the composition of wall rocks that interacted with kimberlite-related fluids (see Section 9). 8.2. The origin of carbonates There is broad consensus that calcite is a major primary magmatic mineral in kimberlites (Mitchell, 1986). However: i) secondary calcite veins typically traverse kimberlite rocks and carbonatisation (i.e. the replacement of magmatic phase by secondary carbonates) is common in kimberlites; ii) the O isotope compositions of bulk kimberlite carbonates indicate common hydrothermal overprinting of magmatic carbonates (e.g., Wilson et al., 2007; Giuliani et al., 2014a); iii) detailed Sr isotope work on calcite from the Jos kimberlite (Canada) has revealed the coexistence of multiple carbonate generations of various derivation (Malarkey et al., 2010). Furthermore, Kamenetsky et al. (2004, 2008, 2009, 2014b) have challenged the view that texturally primary calcite is entirely magmatic and proposed that calcite could largely derive from recrystallization of alkali-rich carbonates. Therefore, there remain major uncertainties regarding the ultimate origin of carbonates in kimberlite rocks. The elevated Sr concentrations of BK calcite (SrO = 0.66–1.54 wt.%) are typical of ‘magmatic’ carbonates in kimberlites (Armstrong et al., 2004). This is consistent with the Sr isotopic signature of the majority of calcite grains, which is similar to that of perovskite from the Kimberley kimberlites (Fig. 9). However, other observations show that the


12


magmatic carbonates in sample BK were partially overprinted after crystallisation. i. Some monticellite grains that are completely pseudomorphed by serpentine appear to be entirely embedded in calcite (Fig. 3f). Therefore, unless the H2O-rich fluids that interacted with monticellite (and produced calcite) were deuteric, some primary magmatic calcite had to recrystallise in the presence of external (or mixed) fluids. ii. The 87Sr/86Sr analyses of some groundmass grains (and segregations) of calcite extend towards more radiogenic values (up to 0.70508 ± 0.00016; Supplementary Fig. S2 and Table S8) compared to most calcite analyses (0.70432 ± 0.00005; 95% confidence; n = 35/39). 87Sr/86Sr values higher than magmatic ratios for carbonates in a dyke associated with the De Beers pipe were previously interpreted to reflect crystallisation from late-stage hydrothermal fluids equilibrated with country rocks (Exley and Jones, 1983). This is consistent with the highly radiogenic Sr isotopic composition of the Kimberley country rocks and associated groundwaters (Fig. 9; Barrett and Berg, 1975). Brookins (1967) and Donnelly et al. (2012) attributed anomalously high and low Sr isotopic values for carbonates in kimberlites from Kansas (US) and Kuruman (South Africa) to equilibration with crustal fluids or contamination by country rocks with very radiogenic and unradiogenic 87Sr/86Sr ratios, respectively. iii. The C-O isotope systematics of BK carbonates (i.e. δ18O higher than mantle values but 13C/12C typical of mantle-derived carbonate-rich magmas; Fig. 10) is typical of kimberlite carbonates partially overprinted by hydrothermal fluids, which may or may not include a significant deuteric component (Giuliani et al., 2014a).

Therefore, petrographic observations and Sr-C-O isotope arguments suggest that some calcite probably recrystallised and/or formed during the hydrothermal overprinting of the Bultfontein kimberlite. Dolomite is one of the most abundant daughter minerals in primary melt inclusions in olivine rims (Fig. 8c–e) and forms single-phase inclusions in apatite (Fig. 3j). This contrasts with the relative paucity of dolomite in sample BK compared to calcite. Dolomite occurs as minor constituents in carbonate segregations (Fig. 3b), as rare interstitial grains in the calcite-rich zones, rare euhedral grains in the serpentinerich zones (Fig. 3d) and is only found as major matrix mineral in occasional dolomite-rich zones of the groundmass (Fig. 3c). Furthermore, some magmatic dolomite (i.e. that closely associated with calcite) was replaced during serpentinisation (Fig. 3e); whereas dolomite in serpentine-rich zones is probably largely of hydrothermal origin (see Section 8.3). It is therefore difficult to constrain the amount of dolomite that crystallised directly from the kimberlite melt. The abundance of dolomite in primary inclusions in olivine and apatite might suggest that dolomite was possibly more abundant in the magmatic assemblage of sample BK before serpentinisation. This would be in line with abundant experimental evidence that carbonate melts of mantle derivation, such as the possible kimberlite precursor (e.g., Kamenetsky et al., 2008, 2014b; Patterson et al., 2009; Giuliani et al., 2012; Russell et al., 2012), are broadly 'dolomitic' regardless of the starting material employed in the experimental runs (e.g., Wallace and Green, 1988; Thibault et al., 1992; Dalton and Presnall, 1998; Brey et al., 2008). Daughter crystals of alkali-rich carbonates (and, to a lesser extent, alkali-rich phosphates, halides and sulfates) occur in primary melt inclusions in olivine rims and MUM spinel grains, but are absent in the kimberlite groundmass, except for occasional μm-sized grains of Na ± K carbonate included in serpentine (Fig. 3i). We argue below (Section 8.3) that serpentine was produced by mixed deuteric-external fluids. Therefore, the carbonate grains included in serpentine could be magmatic or have been introduced from an external crustal source. A magmatic (or mixed) origin seems more likely because it is consistent with the occurrence of alkali-rich daughter crystals in primary inclusions in

olivine and MUM spinel (Fig. 8). Alkali-rich carbonates, phosphates, sulfates and halides are highly soluble in water, and therefore it is likely that the alkali and alkali-earth elements may have been extensively mobilised during hydrothermal modification of these rocks (see le Roex et al., 2003). This suggests higher, though unconstrained, concentrations of Na and K in the original kimberlite magmas than those measured in bulk rock analyses and estimated in reconstructed melt compositions (e.g., le Roex et al., 2003; Becker and le Roex, 2006; Kjarsgaard et al., 2009). 8.3. The origin of serpentine Serpentinisation commonly affects kimberlites worldwide (Mitchell, 1986). Mitchell (1986, 2013) has documented three differential textural types of serpentine in hypabyssal kimberlite rocks, which include: 1) retrograde serpentine replacing olivine along veins and, then, margins; 2) serpophitic (i.e. cryptocrystalline) serpentine in segregations and groundmass, including serpentine after groundmass phases such as calcite, apatite and monticellite; 3) prograde serpentine replacing retrograde serpentine. The origin of the serpentinising fluid(s) is contentious. It has long been assumed that deuteric (i.e. late-stage magmatic) fluids crystallised interstitial serpentine in the kimberlite groundmass, and to some extent replaced olivine and other earlier magmatic minerals (e.g., Clement, 1982; Mitchell, 1986, 2008, 2013; Kopylova et al., 2007). More recent studies have attributed serpentine formation to post-magmatic alteration of olivine and other igneous phases, including silicate glass (Skinner and Marsh, 2004; Willcox et al., 2015), or to direct precipitation from hydrothermal fluids that include a significant or dominant groundwater component (Stripp et al., 2006; Sparks et al., 2006, 2009; Sparks, 2013; Giuliani et al., 2014a; Afanasyev et al., 2014). In addition, experimental data indicate that the H2O concentrations of bulk kimberlite rocks would exceed the solubility of water in silicate and silicate-carbonate melts at the emplacement conditions of kimberlite magmas (Sparks et al., 2009; Brooker et al., 2011; Moussallam et al., 2016). Therefore, there is an apparent discrepancy between early petrographic studies, which supported a largely deuteric (i.e. magmatic) origin for serpentine (Clement, 1982; Mitchell, 1986) based on its widespread occurrence as major kimberlite-forming constituent, and geochemical and experimental constraints, which require that most H2O in kimberlitic serpentine be of external derivation (Sparks et al., 2009; Brooker et al., 2011; Giuliani et al., 2014a; Moussallam et al., 2016). Below we employ our new petrographic and serpentine compositional data to provide an updated model on serpentine formation in hypabyssal kimberlites. In sample BK, serpentine replaces olivine (Fig. 3e), monticellite (Fig. 3f), calcite (Fig. 3d) and minor phlogopite, and occurs both as a major matrix phase (Fig. 3a) and in segregations with calcite. Cross-cutting relationships indicate that olivine serpentinisation occurred after the formation of serpentine-rich zones in the groundmass, which is consistent with a study of serpentine in the Ham kimberlite, Canada (Jago and Mitchell, 1985). Two distinct episodes of serpentinisation in sample BK are supported by serpentine compositions. Groundmass serpentine (i.e. the serpophitic serpentine of Mitchell, 1986, 2013) and serpentine after monticellite are richer in Mg, Al and Ti, but poorer in Fe and Cr than serpentine after olivine (the retrograde serpentine of Mitchell, 1986, 2013). Jago and Mitchell (1985) documented similar lower Mg# values and Al2O3 concentrations in serpentine replacing olivine compared to serpophitic serpentine in the Ham kimberlite. The elevated Mg# of serpophitic groundmass serpentine in sample BK suggests contributions from evolved magmatic fluids, because extremely high Mg/Fe ratios are typical of late-stage kimberlite melts, which produce high-Mg olivine rinds in kimberlites worldwide (Kamenetsky et al., 2008; Bussweiler et al., 2015; Howarth and Taylor, 2016). The Mg enrichment of these fluids is confirmed by coeval precipitation of dolomite and high-Mg serpentine in the BK groundmass (e.g., Fig. 3d).



Oxygen-isotope analyses of serpophitic serpentine in hypabyssal kimberlites yield a range of values (Ham West (Canada), δ18O = −1.8 to +0.1‰ and +1.2 to +1.6‰ for 2 different units; Lac de Gras (Canada), δ18O = +2.0 to +3.0‰) (Mitchell, 2013). At the likely conditions of serpentine formation in kimberlites (200–400 °C), the O isotope fractionation between serpentine and water varies between −1 and +3‰ (Wenner and Taylor, 1971; Zheng, 1993), which translates to serpentine δ18O values of between +8 and +12‰, assuming δ18O ~+9‰ for the magmatic fluids (Giuliani et al., 2014a). The low and variable δ18O values of serpophitic serpentine are therefore inconsistent with formation solely from deuteric fluids and require input from external groundwaters at this stage of kimberlite evolution. We therefore propose that residual hydrous fluids of magmatic origin persisted in interstitial spaces within the largely crystalline BK groundmass and subsequently mixed with probably larger volumes of external fluids that progressively infiltrated the pipe (see also Sparks et al., 2009; Afanasyev et al., 2014). These mixed deuteric-external fluids crystallised serpophitic serpentine and dolomite partly at the expense of calcite, monticellite and phlogopite and partly filling voids and interstitial spaces (see also Mitchell and Putnis, 1988). Weak compositional trends from groundmass serpophitic serpentine to serpentine after olivine (Fig. 7a–d) might indicate that olivine was serpentinised by these same mixed fluids after the formation of serpophitic serpentine. Alternatively, these trends resulted from mixing between deuteric-external fluids and additional groundwaters. The mixing scenario is favoured, as this is consistent with the good linear correlation displayed by SiO2 vs Mg# values of serpentine grains replacing olivine (Fig. 7a), and the localised, bulk replacement of BK groundmass by retrograde serpentine. It is noteworthy that the Mg# of serpentine after olivine ranges between 82 and 85, i.e. below the value (Mg# = 88; Fig. 4) of olivine rims that are replaced by serpentine. Provided that brucite [Mg(OH)2] is not associated with serpentine replacing olivine in sample BK, we can infer an Fe enrichment for these late serpentinising fluids, unlike the Mg-rich composition of earlier deuteric fluids. 9. Conclusions: the final stages of kimberlite petrogenesis In summary, hydrothermal fluids consisting of deuteric and external components largely overprinted the Bultfontein kimberlite by triggering 1) replacement and recrystallization of calcite, 2) replacement of magmatic dolomite and crystallisation of hydrothermal dolomite, 3) removal of alkali carbonates (as well as alkali phosphates, alkali sulfates and halides), 4) formation of at least two generations of serpentine partly at the expense of calcite, dolomite, monticellite, olivine and phlogopite. We therefore propose that before serpentinisation, the BK groundmass consisted of calcite, monticellite, apatite and minor dolomite, phlogopite, spinel, perovskite and ilmenite, and coexisted with a residual alkali-rich C-O-H deuteric fluid. Some of these groundmass components (i.e. phlogopite, perovskite, spinel, ilmenite) occur as inclusions in olivine rims (Fig. 2h–j) and phlogopite (micro-) phenocrysts (Giuliani et al., 2016), and a single olivine inclusion was observed in a MUM spinel (Fig. 3g). Therefore, the olivine rims and additional phlogopite also belong to the magmatic assemblage. While a quantitative assessment of the composition of the melt parental to this assemblage is beyond the scope of this work, it is apparent that an alkali-bearing, P-rich Ca-Mg-rich silicate-carbonate melt is the most likely candidate. This is probably the composition of the melt that crystallised as the Bultfontein kimberlite, but might not reflect the kimberlite composition at source in the deep mantle. This silicatecarbonate composition contrasts with that of primary melt inclusions in BK olivine and spinel, which are more enriched in alkalis and CO2, but depleted in Si. This evidence, combined with the distribution of primary melt inclusions in the outer rims of olivine (i.e. after abundant olivine crystallisation) and in MUM spinel with evolved composition, suggests that these alkali-Si-bearing Ca-Mg-rich carbonate melts

13

represent either a fractionated product of the kimberlite magma or a fluid that exsolved after crystallisation of olivine, and probably ilmenite and chromite. Secondary fluid inclusions in olivine of this and other worldwide kimberlites (Kamenetsky et al., 2009, 2014b; Mernagh et al., 2011) are dominated by Na ± K-rich C-O-H-Cl compositions, which represent residual fluids during groundmass crystallisation. These alkaline fluids probably mix with groundwaters and, depending on the extent of dilution with external fluids, may occasionally become effective metasomatising agents of kimberlite wall rocks (e.g., Smith et al., 2004). In conclusion, we envisage a transition from magmatic to hydrothermal conditions in the late stages of groundmass crystallisation (during or after calcite formation at Bultfontein) when external fluids infiltrated the kimberlite pipe in response to the negative pressure gradient produced by the diatreme-forming event. Residual alkali-rich fluids trapped as secondary inclusions in olivine probably represent the best examples of kimberlite deuteric fluids before mixing with groundwaters occurs. A step forward in constraining the characteristics of kimberlite parental magmas will require determination of the amount of magmatic water in serpentine and assessment of the composition and abundance of (magmatic) carbonates before hydrothermal overprinting. The employment of bulk kimberlite analyses to estimate parental melt compositions remains unwarranted unless specific criteria are applied to ‘see through’ the hydrothermal modification of kimberlite rocks. Acknowledgments We would like to thank Graham Hutchinson and Alan Greig for support with the EMP and LA-ICPMS analyses at the University of Melbourne, and Simon Shee and Jock Robey for insightful discussions on the geology and petrology of the Kimberley kimberlites. We are also grateful to the De Beers Group for providing access to sample BK. This work benefitted from very constructive and thorough reviews by Hugh O'Brien, Troels Nielsen and an anonymous referee, and the careful editorial handling of Sebastian Tappe. AG acknowledges funding from the Australian Research Council through a DECRA award (DE-150100009) and from the Australian Academy of Sciences through a J. G. Russell Award. This is contribution 856 from the ARC Centre of Excellence for Core to Crust Fluid Systems (http://www.ccfs.mq.edu.au) and 1114 in the GEMOC Key Centre (http://www.gemoc.mq.edu.au). Appendix A. Supplementary data Supplementary data to this article can be found online at doi:10. 1016/j.chemgeo.2016.10.011. References Abersteiner, A., Giuliani, A., Kamenetsky, V.S., Phillips, D., Petrographic and melt inclusion constraints on the petrogenesis of a magmaclast from the Venetia kimberlite cluster, South Africa. Chem. Geol., http://dx.doi.org/10.1016/j.chemgeo.2016.08.029 (this issue). Afanasyev, A.A., Melnik, O., Porritt, L., Schumacher, J.C., Sparks, R.S.J., 2014. Hydrothermal alteration of kimberlite by convective flows of external water. Contrib. Mineral. Petrol. 168 (1), 1038–1055. Allsopp, H.L., Barrett, D.R., 1975. Rb-Sr age determinations on South African kimberlite pipes. Phys. Chem. Earth 9, 605–617. Apter, D.B., Harper, F.J., Wyatt, B.A., Smith, B.H.S., 1984. The geology of the Mayeng kimberlite sill complex, South Africa. In: Kornprobst, J. (Ed.), 3rd International Kimberlite Conference. Elsevier, Amsterdam, pp. 43–57. Armstrong, J.P., Wilson, M., Barnett, R.L., Nowicki, T., Kjarsgaard, B.A., 2004. Mineralogy of primary carbonate-bearing hypabyssal kimberlite, Lac de Gras, Slave Province, Northwest Territories, Canada. Lithos 76 (1–4), 415–433. Arndt, N.T., et al., 2010. Olivine, and the origin of kimberlite. J. Petrol. 51 (3), 573–602. Barrett, D.R., Berg, G.W., 1975. Complementary petrographic and strotium-isotope ratio studies of South African kimberlite. Phys. Chem. Earth 9, 619–635. Batumike, J.M., et al., 2008. LAM-ICPMS U-Pb dating of kimberlitic perovskite: Eocene-Oligocene kimberlites from the Kundelungu Plateau, D.R. Congo. Earth Planet. Sci. Lett. 267 (3–4), 609–619.


14


Becker, M., le Roex, A.P., 2006. Geochemistry of South African on- and off-craton, group I and group II kimberlites: petrogenesis and source region evolution. J. Petrol. 47 (4), 673–703. Boyd, F.R., Clement, C.R., 1977. Compositional zoning of olivines in kimberlites of the De Beers mine, Kimberley, South Africa. Carnegie Institute Washington Yearbook. vol. 76, pp. 485–493. Brey, G.P., Bulatov, V.K., Girnis, A.V., Lahaye, Y., 2008. Experimental melting of carbonated peridotite at 6-10 GPa. J. Petrol. 49 (4), 797–821. Brett, R.C., Russell, J.K., Moss, S., 2009. Origin of olivine in kimberlite: phenocryst or impostor? Lithos 112S (0), 201–212. Brooker, R., Sparks, R., Kavanagh, J., Field, M., 2011. The volatile content of hypabyssal kimberlite magmas: some constraints from experiments on natural rock compositions. Bull. Volcanol. 73 (8), 959–981. Brookins, D.G., 1967. The strontium geochemistry of carbonates in kimberlites and limestones from Riley County, Kansas. Earth Planet. Sci. Lett. 2 (3), 235–240. Brown, R., et al., 2012. Eruption of kimberlite magmas: physical volcanology, geomorphology and age of the youngest kimberlitic volcanoes known on earth (the Upper Pleistocene/Holocene Igwisi Hills volcanoes, Tanzania). Bull. Volcanol. 74 (7), 1621–1643. Buse, B., Schumacher, J., Sparks, R., Field, M., 2010. Growth of bultfonteinite and hydrogarnet in metasomatized basalt xenoliths in the B/K9 kimberlite, Damtshaa, Botswana: insights into hydrothermal metamorphism in kimberlite pipes. Contrib. Mineral. Petrol. 160 (4), 533–550. Bussweiler, Y., Foley, S.F., Prelević, D., Jacob, D.E., 2015. The olivine macrocryst problem: new insights from minor and trace element compositions of olivine from Lac de Gras kimberlites, Canada. Lithos 220–223, 238–252. Castillo-Oliver, M., et al., 2016. Trace-element geochemistry and U–Pb dating of perovskite in kimberlites of the Lunda Norte province (NE Angola): petrogenetic and tectonic implications. Chem. Geol. 426, 118–134. Chakhmouradian, A.R., Mitchell, R.H., 2000. Occurrence, alteration patterns and compositional variation of perovskite in kimberlites. Can. Mineral. 38 (4), 975–994. Chakhmouradian, A.R., Reguir, E.P., Kamenetsky, V.S., Sharygin, V.V., Golovin, A.V., 2013. Trace-element partitioning in perovskite: implications for the geochemistry of kimberlites and other mantle-derived undersaturated rocks. Chem. Geol. 353 (0), 112–131. Clement, C.R., 1982. A Comparative Geological Study of Some Major Kimberlite Pipes in the Northern Cape and Orange Free State. University of Cape Town (431 pp). Clement, C.R., Harris, J.W., Robinson, D.N., Hawthorne, J.B., 1986. The De Beers kimberlite pipe - a historic South African diamond mine. In: Anhaeusser, C.R., Maske, S. (Eds.), Mineral Deposits of Southern Africa. Geological Society of South Africa, Johannesburg, pp. 2193–2214. Clement, C.R., Skinner, E.M.W., Smith, B.H.S., 1984. Kimberlite redefined. The Journal of Geology 92 (2), 223–228. Cooper, A.F., Gittins, J., 1974. Shortite in kimberlite from the upper Canada gold mine, Ontario: a discussion. J. Geol. 82 (5), 667–669. Costanzo, A., Moore, K.R., Wall, F., Feely, M., 2006. Fluid inclusions in apatite from Jacupiranga calcite carbonatites: evidence for a fluid-stratified carbonatite magma chamber. Lithos 91, 208–228. Dalton, J.A., Presnall, D.C., 1998. The continuum of primary carbonatitic-kimberlitic melt compositions in equilibrium with lherzolite: data from the system CaO-MgO-Al2O3SiO2-CO2 at 6 GPa. J. Petrol. 39 (11–12), 1953–1964. Davis, G.L., 1977. The ages and uranium content of zircons from kimberlites and associated rocks. Carnegie Institute Washington Yearbook. vol. 76, pp. 631–635. Dawson, J.B., Hawthorne, J.B., 1973. Magmatic sedimentation and carbonatitic differentiation in kimberlite sills at Benfontein, South Africa. J. Geol. Soc. 129 (1), 61–85. Dawson, J.B., Hervig, R.L., Smith, J.V., 1981. Fertile iron-rich dunite xenoliths from the Bultfontein kimberlite, South Africa. Fortschr. Mineral. 59, 303–324. Deer, W.A., Howie, R.A., Zussman, J., 1992. An Introduction to the Rock-forming Minerals. Longman (696 pp). Deines, P., 1989. Stable isotope variations in carbonatites. In: Bell, K. (Ed.), Carbonatites: Genesis and Evolution. Unwin Hyman, London, pp. 301–359. Deines, P., Gold, D.P., 1973. The isotopic composition of carbonatite and kimberlite carbonates and their bearing on the isotopic composition of deep-seated carbon. Geochim. Cosmochim. Acta 37 (7), 1709–1733. Donaldson, C.H., Reid, A.M., 1982. Multiple intrusion of a kimberlite dyke. Trans. Geol. Soc. S. Afr. 85, 1–12. Donnelly, C.L., et al., 2012. In situ U-Pb dating and Sr-Nd isotopic analysis of perovskite: constraints on the age and petrogenesis of the Kuruman Kimberlite Province, Kaapvaal Craton, South Africa. J. Petrol. 53 (12), 2497–2522. Erlank, A.J., et al., 1987. Evidence for mantle metasomatism in peridotite nodules from the Kimberley Pipes, South Africa. In: Menzies, M.A., Hawkesworth, C.J. (Eds.), Mantle Metasomatism. Academic Press, London, pp. 221–311. Exley, R.A., Jones, A.P., 1983. 87Sr/86Sr in kimberlitic carbonates by ion microprobe: hydrothermal alteration, crustal contamination and relation to carbonatite. Contrib. Mineral. Petrol. 83 (3–4), 288–292. Ferguson, J., Danchin, R.V., Nixon, P.H., 1973. Fenitization associated to kimberlite magmas. In: Nixon, P.H. (Ed.), Lesotho Kimberlites. Cape and Transvaal Printers, Cape Town, pp. 207–213. Field, M., Stiefenhofer, J., Robey, J., Kurszlaukis, S., 2008. Kimberlite-hosted diamond deposits of southern Africa: a review. Ore Geol. Rev. 34 (1–2), 33–75. Fitch, F.J., Miller, J.A., 1983. K-Ar age of the east peripheral kimberlite at De Beers Mine, Kimberley, R.S.A. Geol. Mag. 120, 505–512. Giuliani, A., Foley, S.F., 2016. The Geochemical Complexity of Kimberlite Rocks and their Olivine Populations: a Comment on Cordier et al. (Journal of Petrology, 56, 1775– 1796, 2015). J. Petrol. 57, 921–926.

Giuliani, A., et al., 2013. Oxide, sulphide and carbonate minerals in a mantle polymict breccia: metasomatism by proto-kimberlite magmas, and relationship to the kimberlite megacrystic suite. Chem. Geol. 353, 4–18. Giuliani, A., et al., 2012. Nature of alkali-carbonate fluids in the sub-continental lithospheric mantle. Geology 40 (11), 967–970. Giuliani, A., et al., 2014a. Stable isotope (C, O, S) compositions of volatile-rich minerals in kimberlites: a review. Chem. Geol. 374–375, 61–83. Giuliani, A., Phillips, D., Kamenetsky, V.S., Goemann, K., 2016. Constraints on kimberlite ascent mechanisms revealed by phlogopite compositions in kimberlites and mantle xenoliths. Lithos 240–243, 189–201. Giuliani, A., et al., 2014b. Petrogenesis of mantle polymict breccias: insights into mantle processes coeval with kimberlite magmatism. J. Petrol. 55 (4), 831–858. Giuliani, A., et al., 2015. Did diamond-bearing orangeites originate from MARID-veined peridotites in the lithospheric mantle? Nat. Commun. 6, 6837. Graham, S., Lambert, D., Shee, S., 2004. The petrogenesis of carbonatite, melnoite and kimberlite from the Eastern Goldfields Province, Yilgarn Craton. Lithos 76 (1–4), 519–533. Griffin, W.L., et al., 2014. Emplacement ages and sources of kimberlites and related rocks in southern Africa: U–Pb ages and Sr–Nd isotopes of groundmass perovskite. Contrib. Mineral. Petrol. 168 (1), 1032–1045. Gurney, J.J., Helmstaedt, H.H., Richardson, S.H., Shirey, S.B., 2010. Diamonds through time. Econ. Geol. 105 (3), 689–712. Hanley, J.J., Mungall, J.E., Pettke, T., Spooner, E.T.C., Bray, C.J., 2008. Fluid and halide melt inclusions of magmatic origin in the ultramafic and lower banded series, stillwater complex, Montana, USA. J. Petrol. 49, 1133–1160. Howarth, G.H., Taylor, L.A., 2016. Multi-stage kimberlite evolution tracked in zoned olivine from the Benfontein sill, South Africa. Lithos 262, 384–397. Hunter, R.H., Taylor, L.A., 1982. Instability of garnet from the mantle: glass as evidence of metasomatic melting. Geology 10 (12), 617–620. Jago, B.C., Mitchell, R.H., 1985. Mineralogy and petrology of the ham kimberlite, SOMERSET Island, Northwest Territories. Can. Mineral. 23, 619–634. Jones, A.P., Wyllie, P.J., 1985. Paragenetic trends of oxide minerals in carbonate-rich kimberlites, with new analyses from the Benfontein sill, South Africa. J. Petrol. 26 (1), 210–222. Jourdan, F., Féraud, G., Bertrand, H., Watkeys, M.K., 2007. From flood basalts to the inception of oceanization: Example from the 40Ar/39Ar high-resolution picture of the Karoo large igneous province. Geochem. Geophys. Geosyst. 8 (2), Q02002. Kamenetsky, M.B., et al., 2004. Kimberlite melts rich in alkali chlorides and carbonates: a potent metasomatic agent in the mantle. Geology 32 (10), 845–848. Kamenetsky, V.S., et al., 2014a. Chemical abrasion of zircon and ilmenite megacrysts in the Monastery kimberlite: implications for the composition of kimberlite melts. Chem. Geol. 383, 76–85. Kamenetsky, V.S., et al., 2014b. Towards a new model for kimberlite petrogenesis: evidence from unaltered kimberlites and mantle minerals. Earth Sci. Rev. 139, 145–167. Kamenetsky, V.S., Grutter, H., Kamenetsky, M.B., Goemann, K., 2013. Parental carbonatitic melt of the Koala kimberlite (Canada): Constraints from melt inclusions in olivine and Cr-spinel, and groundmass carbonate. Chem. Geol. 353, 96–111. Kamenetsky, V.S., et al., 2008. Olivine in the Udachnaya-East kimberlite (Yakutia, Russia): Types, compositions and origins. J. Petrol. 49 (4), 823–839. Kamenetsky, V.S., et al., 2009. How unique is the Udachnaya-East kimberlite? Comparison with kimberlites from the Slave Craton (Canada) and SW Greenland. Lithos 112S, 334–346. Kamenetsky, V.S., Yaxley, G.M., 2015. Carbonate–silicate liquid immiscibility in the mantle propels kimberlite magma ascent. Geochim. Cosmochim. Acta 158, 48–56. Ketcham, R.A., 2015. Calculation of stoichiometry from EMP data for apatite and other phases with mixing on monovalent anion sites. Am. Mineral. 100, 1620–1623. Kirkley, M.B., Smith, H.S., Gurney, J.J., 1989. Kimberlite carbonates - a carbon and oxygen stable isotope study. In: Glover, J.E., Harris, P.G. (Eds.), Kimberlites and Related Rocks Vol. 1: Their Composition, Occurrence, Origin and Emplacement. 4th International Kimberlite Conference. Geological Society of Australia, Perth, pp. 264–281. Kiviets, G.B., et al., 1998. In: Gurney, J.J., Gurney, J.L., Pascoe, M.D., Richardson, S.H. (Eds.), 40 Ar/39Ar dating of yimengite from the Turkey Well kimberlite, Australia: the oldest and the rarest. 7th International Kimberlite Conference, Extended Abstracts. Red Roof Design, Cape Town, pp. 432–434. Kjarsgaard, B.A., Pearson, D.G., Tappe, S., Nowell, G.M., Dowall, D.P., 2009. Geochemistry of hypabyssal kimberlites from Lac de Gras, Canada: comparisons to a global database and applications to the parent magma problem. Lithos 112S, 236–248. Kobelski, B.J., Gold, D.P., Deines, P., 1979. Variations in stable isotope compositions for carbon and oxygen in some South African and Lesothan kimberlites. In: Boyd, F.R., Meyer, H.O.A. (Eds.), The Mantle Sample. 2nd International Kimberlite Conference. American Geophysical Union, Washington, DC, pp. 252–271. Kopylova, M.G., Kostrovitsky, S.I., Egorov, K.N., 2013. Salts in southern Yakutian kimberlites and the problem of primary alkali kimberlite melts. Earth Sci. Rev. 119 (0), 1–16. Kopylova, M.G., Matveev, S., Raudsepp, M., 2007. Searching for parental kimberlite melt. Geochim. Cosmochim. Acta 71 (14), 3616–3629. Kramers, J.D., Roddick, J.C.M., Dawson, J.B., 1983. Trace element and isotope studies on veined, metasomatic and “MARID” xenoliths from Bultfontein, South Africa. Earth Planet. Sci. Lett. 65 (1), 90–106. Lawless, P.J., Gurney, J.J., Dawson, J.B., 1979. Polymict peridotites from the Bultfontein and De Beers mines, Kimberley, South Africa. In: Boyd, F.R., Meyer, H.O.A. (Eds.), The Mantle Sample. 2nd International Kimberlte Conference. American Geophysical Union, Washington, DC, pp. 145–155. le Roex, A.P., Bell, D.R., Davis, P., 2003. Petrogenesis of group I kimberlites from Kimberley, South Africa: evidence from bulk-rock geochemistry. J. Petrol. 44 (12), 2261–2286. Luth, R.W., 2009. The activity of silica in kimberlites, revisited. Contrib. Mineral. Petrol. 158 (2), 283–294.


A. Giuliani et al. / Chemical Geology xxx (2016) xxx–xxx Malarkey, J., Pearson, D.G., Kjarsgaard, B.A., Davidson, J.P., Nowell, G.M., Ottley, C.J., Stammer, J., 2010. From source to crust: Tracing magmatic evolution in a kimberlite and a melilitite using microsample geochemistry. Earth Planet. Sci. Lett. 299, 80–90. Menzies, A., et al., 2004. Peridotitic mantle xenoliths from kimberlites on the Ekati Diamond Mine property, N.W.T., Canada: major element compositions and implications for the lithosphere beneath the central Slave craton. Lithos 77 (1–4), 395–412. Mernagh, T.P., Kamenetsky, V.S., Kamenetsky, M.B., 2011. A Raman microprobe study of melt inclusions in kimberlites from Siberia, Canada, SW Greenland and South Africa. Spectrochim. Acta A Mol. Biomol. Spectrosc. 80 (1), 82–87. Mitchell, R.H., 1986. Kimberlites: Mineralogy, Geochemistry and Petrology. Plenum Publishing Company, New York (442 pp). Mitchell, R.H., 1995. Kimberlites, Orangeites and Related Rocks. Plenum Press, New York (410 pp). Mitchell, R.H., 2008. Petrology of hypabyssal kimberlites: relevance to primary magma compositions. J. Volcanol. Geotherm. Res. 174 (1–3), 1–8. Mitchell, R.H., 2013. Oxygen isotope studies of serpentine in kimberlite. Proceedings of the 10th International Kimberlite Conference. Geological Society of India, Bangalore, pp. 1–12. Mitchell, R.H., Putnis, A., 1988. Polygonal serpentine in segregation-textured kimberlite. Can. Mineral. 26, 991–997. Mitchell, R.H., Skinner, E.M.W., Scott Smith, B.H., 2009. Tuffisitic kimberlites from the Wesselton Mine, South Africa: mineralogical characteristics relevant to their formation. Lithos 112S, 452–464. Moore, A.E., 2012. The case for a cognate, polybaric origin for kimberlitic olivines. Lithos 128–131, 1–10. Moussallam, Y., Morizet, Y., Gaillard, F., 2016. H2O–CO2 solubility in low SiO2-melts and the unique mode of kimberlite degassing and emplacement. Earth Planet. Sci. Lett. 447, 151–160. Nielsen, T.F.D., Sand, K.K., 2008. The Majuagaa kimberlite dike, Maniitsoq region, western Greenland: constraints on an Mg-rich silicocarbonatitic melt composition from groundmass mineralogy and bulk compositions. Can. Mineral. 46 (4), 1043–1061. Nowell, G.M., et al., 2004. Hf isotope systematics of kimberlites and their megacrysts: new constraints on their source regions. J. Petrol. 45 (8), 1583–1612. Nowicki, T., Porritt, L., Crawford, B., Kjarsgaard, B., 2008. Geochemical trends in kimberlites of the Ekati property, Northwest Territories, Canada: insights on volcanic and resedimentation processes. J. Volcanol. Geotherm. Res. 174 (1–3), 117–127. Ogilvie-Harris, R.C., Field, M., Sparks, R.S.J., Walter, M.J., 2009. Perovskite from the Dutoitspan kimberlite, Kimberley, South Africa: implications for magmatic processes. Mineral. Mag. 73 (6), 915–928. Parhasarathy, G., Chetty, T.R.K., Haggerty, S.E., 2002. Thermal stability and spectroscopic studies of zemkorite: a carbonate from the Venkatampalle kimberlite of southern India. Am. Mineral. 87, 1384–1389. Pasteris, J.D., 1980. Opaque Oxide Phases of the De Beers Pipe Kimberlite (Kimberley, South Africa). Yale University, New Haven (US). Pasteris, J.D., 1983. Spinel zonation in the De Beers Kimberlite, South Africa; possible role of phlogopite. Can. Mineral. 21 (1), 41–58. Paton, C., Hergt, J.M., Woodhead, J.D., Phillips, D., Shee, S.R., 2009. Identifying the asthenospheric component of kimberlite magmas from the Dharwar Craton, India. Lithos 112S, 296–310. Patterson, M., Francis, D., McCandless, T., 2009. Kimberlites: magmas or mixtures? Lithos 112S, 191–200. Pearson, D.G., et al., 2014. Hydrous mantle transition zone indicated by ringwoodite included within diamond. Nature 507 (7491), 221–224. Pilbeam, L.H., Nielsen, T.F.D., Waight, T.E., 2013. Digestion fractional crystallization (DFC): an important process in the genesis of kimberlites. evidence from olivine in the Majuagaa kimberlite, Southern West Greenland. J. Petrol. 54 (7), 1399–1425. Pokhilenko, N.P., 2009. Polymict breccia xenoliths: evidence for the complex character of kimberlite formation. Lithos 112S, 934–941. Poujol, M., Robb, L.J., Anhaeusser, C.R., Gericke, B., 2003. A review of the geochronological constraints on the evolution of the Kaapvaal Craton, South Africa. Precambrian Res. 127 (1–3), 181–213. Reguir, E.P., Chakhmouradian, A.R., Halden, N.M., Malkovets, V.G., Yang, P., 2009. Majorand trace-element compositional variation of phlogopite from kimberlites and carbonatites as a petrogenetic indicator. Lithos 112S, 372–384. Rehfeldt, T., Jacob, D.E., Carlson, R.W., Foley, S.F., 2007. Fe-rich dunite xenoliths from south African kimberlites: cumulates from karoo flood basalts. J. Petrol. 48 (7), 1387–1409. Roeder, P.L., Schulze, D.J., 2008. Crystallization of groundmass spinel in kimberlite. J. Petrol. 49 (8), 1473–1495. Russell, J.K., Porritt, L.A., Lavallee, Y., Dingwell, D.B., 2012. Kimberlite ascent by assimilation-fuelled buoyancy. Nature 481 (7381), 352–356. Sautter, V., Haggerty, S.E., Field, S., 1991. Ultradeep (N300 kilometers) ultramafic xenoliths: petrological evidence from the transition zone. Science 252 (5007), 827–830. Schulze, D.J., 2001. Origins of chromian and aluminous spinel macrocrysts from kimberlites in southern Africa. Can. Mineral. 39 (2), 361–376. Shee, S., 1984. In: Kornprobst, J. (Ed.), The oxide minerals of the Wesselton Mine kimberlite, Kimberley, South Africa. 3rd International Kimberlite Conference. Elsevier, Amsterdam, pp. 59–73. Shee, S., 1985. The Petrogenesis of the Wesselton Mine Kimberlites. University of Cape Town, Kimberley, Cape Province, R.S.A. (220 pp). Shee, S.R., Clement, C.R., Skinner, E.M.W., 1994. The petrology of the Wesselton kimberlite sills, Kimberley, Cape Province, South Africa. In: Meyer, H.O.A., Leonardos, O.H. (Eds.),

15

Kimberlites, Related Rocks and Mantle Xenoliths. 5th International Kimberlite Conference. CPRM Special Publication, Araxa, Brazil, pp. 98–114. Sheppard, S.M.F., Dawson, J.B., 1975. Hydrogen, carbon and oxygen isotope studies of megacryst and matrix minerals from lesothan and South African kimberlites. Phys. Chem. Earth 9 (0), 747–763. Skinner, E.M.W., 1989. Contrasting Group I and Group II kimberlite petrology: towards a genetic model for kimberlites. In: Glover, J.E., Harris, P.G. (Eds.), Kimberlites and Related Rocks. 4th International Kimberlite Conference. Geological Society of Australia, Perth, pp. 528–544. Skinner, E.M.W., Clement, C.R., 1979. Mineralogical classification of southern African kimberlites. In: Boyd, F.R., Meyer, H.O.A. (Eds.), The Mantle Sample. 2nd International Kimberlte Conference. American Geophysical Union, Washington, DC, pp. 129–139. Skinner, E.M.W., Marsh, J.S., 2004. Distinct kimberlite pipe classes with contrasting eruption processes. Lithos 76 (1‚Äì4), 183–200. Smart, K.A., Tappe, S., Stern, R.A., Webb, S.J., Ashwal, L.D., 2016. Early Archaean tectonics and mantle redox recorded in Witwatersrand diamonds. Nat. Geosci. 9 (3), 255–259. Smith, C.B., 1983. Pb, Sr and Nd isotopic evidence for sources of southern African Cretaceous kimberlites. Nature 304, 51–54. Smith, C.B., et al., 2004. Kimberlite metasomatism at Murowa and Sese pipes, Zimbabwe. Lithos 76 (1–4), 219–232. Smith, E.M., Kopylova, M.G., Nowell, G.M., Pearson, D.G., Ryder, J., 2012. Archean mantle fluids preserved in fibrous diamonds from Wawa, Superior craton. Geology 40 (12), 1071–1074. Sobolev, N.V., et al., 2009. Petrogenetic significance of minor elements in olivines from diamonds and peridotite xenoliths from kimberlites of Yakutia. Lithos 112 (Suppl. 2(0)), 701–713. Soltys, A., et al., 2016. In-situ assimilation of mantle minerals by kimberlitic magmas — direct evidence from a garnet wehrlite xenolith entrained in the Bultfontein kimberlite (Kimberley, South Africa). Lithos 256–257, 182–196. Sparks, R.S.J., 2013. Kimberlite Volcanism. Annu. Rev. Earth Planet. Sci. 41 (1), 497–528. Sparks, R.S.J., et al., 2006. Dynamical constraints on kimberlite volcanism. J. Volcanol. Geotherm. Res. 155 (1–2), 18–48. Sparks, R.S.J., et al., 2009. The nature of erupting kimberlite melts. Lithos 112S, 429–438. Stripp, G.R., Field, M., Schumacher, J.C., Sparks, R.S.J., Cressey, G., 2006. Post-emplacement serpentinization and related hydrothermal metamorphism in a kimberlite from Venetia, South Africa. J. Metamorph. Geol. 24 (6), 515–534. Svensen, H., Corfu, F., Polteau, S., Hammer, O., Planke, S., 2012. Rapid magma emplacement in the Karoo large Igneous Province. Earth Planet. Sci. Lett. 325–326, 1–9. Tappe, S., Kjarsgaard, B.A., Kurszlaukis, S., Nowell, G.M., Phillips, D., 2014. Petrology and Nd–Hf isotope geochemistry of the Neoproterozoic Amon kimberlite sills, Baffin Island (Canada): evidence for deep mantle magmatic activity linked to supercontinent cycles. J. Petrol. 55, 2003–2042. Tappe, S., Pearson, G.D., Kjarsgaard, B.A., Nowell, G., Dowall, D., 2013. Mantle transition zone input to kimberlite magmatism near a subduction zone: origin of anomalous Nd–Hf isotope systematics at Lac de Gras, Canada. Earth Planet. Sci. Lett. 371–372, 235–251. Taylor, H.P., Frechen, J., Degens, E.T., 1967. Oxygen and carbon isotope studies of carbonatites from the Laacher See District, West Germany and the Alno District, Sweden. Geochim. Cosmochim. Acta 31 (3), 407–430. Thibault, Y., Edgar, A.D., Lloyd, F.E., 1992. Experimental investigation of melts from a carbonated phlogopite lherzolite: implications for metasomatism in the continetal lithopsheric mantle. Am. Mineral. 77, 784–794. Thomas, R.J., von Veh, M.W., McCourt, S., 1993. The tectonic evolution of southern Africa: an overview. J. Afr. Earth Sci. 16, 5–24. Veksler, I.V., Lentz, D., 2006. Parental magmas of plutonic carbonatites, carbonate-silicate immiscibility and decarbonation reactions: evidence from melt and fluid inclusions. In: Webster, J.D. (Ed.), Mineralogical Association of Canada Short Courses. vol. 36, pp. 123–149. Wallace, M.E., Green, D.H., 1988. An experimental determination of primary carbonatite magma composition. Nature 335, 343–346. Watkinson, D.H., Chao, G.Y., 1973. Shortite in kimberlite from the Upper Canada gold mine, Ontario. J. Geol. 81 (2), 229–233. Wenner, D.B., Taylor Jr., H., 1971. Temperatures of serpentinization of ultramafic rocks based on O18/O16 fractionation between coexisting serpentine and magnetite. Contrib. Mineral. Petrol. 32, 165–185. White, J.L., et al., 2012. Kimberlite sills and dykes associated with the Wesselton kimberlite pipe, Kimberley, South Africa. S. Afr. J. Geol. 115 (1), 1–32. Willcox, A., et al., 2015. Petrology, geochemistry and low-temperature alteration of lavas and pyroclastic rocks of the kimberlitic Igwisi Hills volcanoes, Tanzania. Chem. Geol. 405, 82–101. Wilson, M.R., Kjarsgaard, B.A., Taylor, B., 2007. Stable isotope composition of magmatic and deuteric carbonate phases in hypabyssal kimberlite, Lac de Gras field, Northwest Territories, Canada. Chem. Geol. 242 (3‚Äì4), 435–454. Woodhead, J., Hergt, J., Phillips, D., Paton, C., 2009. African kimberlites revisited: in situ Srisotope analysis of groundmass perovskite. Lithos 112S, 311–317. Zheng, Y.-F., 1993. Calculation of oxygen isotope fractionation in hydroxyl-bearing silicates. Earth Planet. Sci. Lett. 120, 247–263.
