Journal of Petrology Advance Access originally published online on November 11, 2004
Journal of Petrology 2005 46(2):377-406; doi:10.1093/petrology/egh081
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology vol. 46 issue 2 © Oxford University Press 2004; all rights reserved
Fenitizing Processes Induced by Ferrocarbonatite Magmatism at Swartbooisdrif, NW Namibia
1 INSTITUT FÜR ANGEWANDTE GEOWISSENSCHAFTEN, SEKRETARIAT BH1, TECHNISCHE UNIVERSITÄT BERLIN, ERNST-REUTER-PLATZ 1, D-10587 BERLIN, GERMANY
2 GEOCHEMISCHES INSTITUT, UNIVERSITÄT GÖTTINGEN, GOLDSCHMIDTSTR. 1, D-37077 GÖTTINGEN, GERMANY
3 MINERALOGISCHES INSTITUT, UNIVERSITÄT WÜRZBURG, AM HUBLAND, D-97074 WÜRZBURG, GERMANY
RECEIVED OCTOBER 29, 2002; ACCEPTED SEPTEMBER 20, 2004
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONThe southeastern margin of the anorthositic Kunene Intrusive Complex, NW Namibia, has been subsequently invaded by Mesoproterozoic syenite, nepheline syenite and ferrocarbonatite dykes along NE- and SE-trending faults. The first generation of carbonatite intrusions frequently contains fenitized anorthositic wall-rock fragments set in a ferrocarbonatite matrix; later, subordinate veins of massive ferrocarbonatite are almost xenolith-free and cut through the main carbonatite dykes. A mantle source for both carbonatite generations is constrained by their respective oxygen and carbon isotope compositions of ankerite (
18OSMOW 8·91–9·73
; 13CPDB –6·98 to –6·76). Na-rich fluids, released from the melt parental to the ferrocarbonatites, caused the fenitization of both the incorporated anorthosite xenoliths and the bordering anorthosite, syenite and nepheline syenite. This process is mainly characterized by the progressive transformation of Ca-rich plagioclase, K-feldspar and nepheline into albite and/or sodalite. The changing mineral modes indicate that the fenitizing fluids were sodium-rich and strongly Si-deficient solutions, which also contained significant amounts of Sr, Ba, Nb and the light rare earth elements. On the basis of mineral equilibria studies, it is possible to reconstruct the temperature conditions for carbonatite emplacement (c. 830 ± 200°C) and recrystallization (c. 480 ± 130°C), and for the metasomatic formation of sodalite (c. 700 ± 70°C). KEY WORDS: anorthosite; fenitization; ferrocarbonatite; sodalite; stable isotopes
|
INTRODUCTION |
|---|
The majority of the carbonatite complexes known from world-wide localities are predominantly composed of calcium- and magnesium-rich carbonatites. In these carbonatite centres late-stage ferrocarbonatite dykes may occur, which are characterized by strong enrichments of Fe, Mg, Ba, Sr, rare earth elements (REE), Th and U. Such ferrocarbonatites are commonly interpreted as the end-products of fractional crystallization and associated differentiation processes of the main Ca–Mg carbonatite body (Le Bas, 1977
, 1981, 1989, 1999; Gittins, 1989; Woolley & Kempe, 1989; Cooper & Reid, 1998). Magmatic natrocarbonatite is known only as the effusive product of the Oldoinyo Lengai volcano, Tanzania, which is also the only known active carbonatite volcano [see Bell & Keller (1995) for a review]. However, the close spatial association of Na- and/or K-rich fenites and carbonatites (Andersen, 1989; Platt & Woolley, 1990; Morogan, 1994; Hagni & Shivdasan, 2001; Williams-Jones & Palmer, 2002) suggests that carbonatites, which are now Ca–Mg–Fe carbonatites, were alkali-rich magmas when they intruded and subsequently expelled alkali-metals and volatiles during their differentiation and crystallization. This model has been confirmed by experimental studies (Watkinson & Wyllie, 1971; Freestone & Hamilton, 1980; Koster van Groos, 1990; Lee & Wyllie, 1994, 1997, 1998; Veksler & Keppler, 2000). Where carbonatitic centres are surrounded by metasomatic aureoles, the composition of the solidified carbonatite will differ significantly from that of the pristine carbonatite magma as a result of material losses and gains during differentiation, crystallization or post-magmatic re-equilibration. In this case, the carbonatites preserve only an incomplete memory of their magmatic history. The investigation of metasomatic mineral reactions at the magma to wall-rock interface and of fluid inclusions containing the fenitizing agents can thus provide important information about the geochemical nature and evolution of the fluids responsible.
The carbonatite dykes of Swartbooisdrif, NW Namibia, are an ideal study object as they are well exposed and have undergone only minor subsolidus alteration and weathering. Only few studies have been concerned with this carbonatite occurrence so far (e.g. Menge, 1986, 1996; Thompson et al., 2002), including our own work (von Seckendorff & Drüppel, 1999; von Seckendorff et al., 2000; this study), although it is host to a conspicuous sodalite deposit, which is confined to an area of c. 100 km2. The Swartbooisdrif carbonatites differ from other known carbonatite occurrences in (1) their exotic fenite mineral assemblage, (2) the uncommon anorthositic mineralogy of the wall rocks that are involved in the metasomatic processes and (3) the general lack of associated Ca–Mg carbonatites and subordinate, genetically related, felsic intrusives. The aim of this study is to characterize the magmatic evolution of the carbonatite intrusions and to elucidate the interrelated metasomatic processes, which led to the formation of the sodalite occurrences.
|
GEOLOGICAL FRAMEWORK |
|---|
The Swartbooisdrif ferrocarbonatites were emplaced near the southwestern margin of the anorthositic Kunene Intrusive Complex (KIC). The KIC, a massif-type anorthosite complex, is exposed in SW Angola and NW Namibia as a north–south elongated body of about 350 km in length and 30–80 km in width. It was emplaced at the southern margin of the Congo craton. An internal biotite–plagioclase whole-rock Rb–Sr isochron age of 1347 ± 13 Ma has been determined for one anorthosite sample from the Angolan part of the complex (Mayer et al., 2000
). This age has been constrained by Mayer et al. (2004) by a concordant U–Pb age of 1371 ± 2·5 Ma using zircons from a late-stage mangerite vein, inferred to be cogenetic with the KIC. These ages are in accordance with an almost concordant U–Pb zircon age of 1385 ± 25 Ma for an anorthosite sample of the Namibian part of the KIC (Drüppel et al., 2000). The anorthosite massif intruded the Palaeo- to Mesoproterozoic Epupa Complex (EC), which, in its main parts, consists of ortho- and paragneisses, metamorphosed under upper amphibolite-facies and subordinately under ultrahigh-T, granulite-facies conditions (Brandt et al., 2003). Close to the southernmost margin of the KIC, c. 7 km NW of the settlement of Swartbooisdrif, the anorthosite massif is transsected by numerous SE- and ENE-trending shear zones, which were subsequently intruded by syenite, quartz syenite and younger ferrocarbonatite dykes (Fig. 1). With U–Pb zircon ages of 1385 ± 5 Ma and 1335 ± 2 Ma (Littmann et al., in preparation) the syenite dykes of the Swartbooisdrif area appear to be genetically linked to the KIC. The ferrocarbonatites at Swartbooisdrif, however, are distinctly younger than both the anorthosites and the syenites as is evidenced by pyrochlore U–Pb ages of c. 1140–1120 Ma (Littmann et al., in preparation). The carbonatite dykes are host to significant amounts of sodalite, which are mined in an area of c. 100 km2.
|
Another alkaline suite is exposed within the basement gneisses of the EC in the Epembe area, about 30 km SW of Swartbooisdrif. This suite mainly comprises small plugs of nepheline syenite (Ferguson et al., 1975
; Menge, 1986), which yielded radiometric U–Pb zircon ages of 1216 ± 2·4 and 1213 ± 2·5 Ma (Littmann et al., 2000). Also assigned to the same magmatic event are east–west-striking calciocarbonatite, lamprophyre and subordinate ferrocarbonatite dykes (Ferguson et al., 1975).
In much of the earlier literature, the Swartbooisdrif and Epembe suites are grouped together (e.g. Ferguson et al., 1975; Menge, 1986; von Seckendorff et al., 2000), forming the Epembe–Swartbooisdrif Alkaline Province (ESAP; Menge, 1986). In addition, intrusive bodies and dykes of syenite and calciocarbonatite of the Lupongola Complex of Angola, 
20 km NW of Swartbooisdrif, have been considered part of the ESAP (Menge, 1996; Alberti et al., 1999; Thompson et al., 2002). With respect to their radiometric ages, however, at least the Swartbooisdrif and the Epembe carbonatite centres appear to represent independent rock suites.
|
GEOLOGY OF THE SODALITE MINING AREA |
|---|
The sodalite mining area near Swartbooisdrif (Fig. 1) is dominated by whitish, heavily altered anorthosites and leucogabbros of the KIC, the ‘white anorthosite’, which were subsequently intruded by small bodies of weakly altered, dark anorthosites, leucotroctolites, leucogabbros and leuconorites, the ‘dark anorthosite’ (Drüppel et al., 2001
). Numerous SE- and ENE-trending syenite and younger nepheline syenite and ferrocarbonatite dykes were subsequently emplaced along fractures in the anorthositic rocks of the KIC. Individual dykes may be interrupted over short distances and frequently change their strike and dip direction on a local scale. Felsic intrusives are mainly represented by brownish, strongly altered syenites, which form dykes 3–5 m wide. Where the syenite dykes are intruded by the younger carbonatites, the syenite is heavily strained and exhibits a pink colour. Exposed contacts between syenite and the younger carbonatites are curvilinear but tectonic, as syenitic fragments in all stages of detachment can be observed in the bordering carbonatite. At one locality a nepheline syenite dyke was observed in direct contact with a large carbonatite dyke. The nepheline syenite is exposed over just a short distance of about 6·5 m, has a width of 30 cm to 3 m and crosscuts an older syenite. At a distance from the carbonatite, the nepheline syenite is characterized by a homogeneous texture, with medium-grained nepheline and K-feldspar being the major constituents. When approaching the carbonatite dyke, the texture of the nepheline syenite becomes progressively inhomogeneous, with fine-grained K-feldspar-rich zones alternating with irregular pegmatoidal zones, mainly composed of nepheline crystals of up to 15 cm in length. At the direct contact with the carbonatite, nepheline exhibits a pale blue tint resulting from its partial replacement by sodalite.
A major swarm of carbonatite dykes and veins occurs in the area, dissecting the syenite dykes and the anorthosite body. Two main periods of carbonatite emplacement have been recognized.
(1) The predominant carbonatite bodies, the ‘carbonatitic breccia’, are up to 80 m in width and frequently contain variable amounts of angular to subrounded fragments of fenitized wall-rock anorthosite and syenite, ranging continuously from 8 m down to several millimetres, set in an ankerite-rich carbonatite matrix (Fig. 2a). Most of the xenoliths are pervasively fenitized and partially broken into small particles, creating an interfragmental groundmass. A common feature of the carbonatitic breccia is a banded or streaky appearance caused by an alternation of ankerite-, magnetite- and silicate-rich layers oriented sub-parallel to the dyke walls. Flow-banding and impersistent magmatic folding on a millimetre to metre scale are common structures of this rock type (Fig. 2b), implying that the carbonatite was emplaced as a melt. Locally, the carbonatitic breccia may also contain massive carbonatite zones, almost entirely composed of medium- to coarse-grained ankerite and minor magnetite. The occurrence of fragmented rafts of the layered carbonatitic breccia, surrounded by younger massive ferrocarbonatite, supports evidence for repeated injections of carbonatite melt. Major phases in the inhomogeneous carbonatitic breccia are ankerite (35–85 vol. %), albite (10–65 vol. %) and magnetite (15–25 vol. %), although biotite and sodalite are locally observed in significant concentrations of >25 vol. %. Cancrinite, muscovite, apatite, calcite, ilmenite, pyrochlore and sulphides are minor constituents, which rarely exceed 5 vol. %. The occurrence of sodalite is confined to regions where several larger carbonatite dykes intersect at acute angles and the modal abundance of anorthositic xenoliths increases. Variable amounts of sodalite (10–90 vol. %) occur as conspicuous deep blue lenses, layers and breccias in several of the large carbonatite dykes, but are also present in narrow metasomatic aureoles of up to 1 m in width within the bordering dark anorthosite.
|
(2) The main carbonatite body is transsected by small veins and stringers of a second generation of almost silicate-free ferrocarbonatite, which are up to 1·5 m in width. These late-stage ferrocarbonatite veins have been recently investigated in some detail by Thompson et al. (2002)
. |
ANALYTICAL METHODS |
|---|
Electron microprobe analyses were performed on a CAMECA SX50 with three wavelength-dispersive spectrometers at the Institute of Mineralogy, University of Würzburg. For silicate, apatite and pyrochlore analyses the excitation voltage and beam current were set at 15 kV and 15 nA, with 1, 8 and 5 µm spot diameter, respectively. The bulk compositions of feldspars were determined with a defocused electron beam of 20 µm size. Carbonates were analysed at 12 kV, 15 nA with 8 µm spot sizes. For the calibration of Nb in pyrochlore pure Nb metal was used as a standard, whereas LaPO4 and CePO4 were chosen as calibration standards for La and Ce in pyrochlore and carbonates. Where possible, profiles of 10–60 points were analysed for each phase.
Major and trace elements were determined on fused glass discs using a Philips PW 1480 X-ray fluorescence (XRF) spectrometer at the Institute of Mineralogy, Würzburg. Matrix effects were corrected for automatically by the Philips X40 software. The relative analytical error for major and trace elements is 1% and 1–8%, respectively. The loss on ignition (LOI) was determined by heating 1 g of dried sample powder for 4 h at 1100°C and reweighing the sample after cooling. CO2 was determined volumetrically by reacting 1–3 g of the sample powder with 10 ml HCl (16–18%). The FeO contents were analysed using a ZEISS PMD 2 spectral photometer. FeO contents of carbonatite samples with elevated CO2 contents of >13 wt % were not detectable by this method, because of a strong chemical reaction of the sample powders with the HF–H2SO4 (SEROLAB) mixture.
Concentrations of the REE and Y in the anorthosites, nepheline syenite and REE-poor samples of the carbonatites were determined by inductively coupled plasma atomic emission spectrometry (ICP-AES) at the GeoForschungsZentrum Potsdam, Germany. Sample preparation followed the method of Zuleger & Erzinger (1988). Details on ICP equipment, operating conditions, background wavelength corrections and background equivalent concentrations, as well as the precision of the method, have been given iby Zuleger & Erzinger (1988). The concentrations of the REE and Y in two REE-rich carbonatite samples were analysed by inductively coupled plasma mass spectrometry (ICP-MS) using a Perkin-Elmer/SCIEX Elan 5000 quadrupole ICP mass spectrometer at the GFZ Potsdam. The samples were dissolved in PTFE digestion vessels with 1:1 HF–HClO4 for 16 h at 180°C. After cooling, samples were evaporated at 180°C for c. 4 h, dissolved with 5 ml HCl (10 mol/l) and heated again at 180°C to incipient dryness. The residues were taken up in 5 ml HCl (10 mol/l) and were heated in the closed vessels at 130°C for 12 h. After this procedure the samples were dried again and the sample cakes redissolved in 2 ml HCl (10 mol/l) and 10 ml Milli-Q water. The solutions were filled up to 50 ml with Milli-Q water. Before analysis, Ru and Re were added to aliquots of the solutions as internal standards to compensate for drift correction (Doherty, 1989), and the mixtures were diluted. A dilution factor of 50 000 was used for the measurements. Interference corrections consider the interference of MO+, MOH+ and MCl+ (Dulski, 1994). Calibration was carried out using multi-element solutions containing 10 mg/ml Rb, Sr, Y, Zr, Cs, Ba, all REE, Hf, Pb, Th and U. The precision ranged between 1 and 5% relative.
Trace element and REE analyses of sodalite in thin sections were performed with a 266 nm Nd:YAG laser (New Wave Research Inc., Merchantek Products) connected to a quadrupole ICP-MS system [Agilent 7500i; plasma power 1320 W; carrier gas flow 1·11 l/min (Ar); plasma gas flow 14·9 l/min (Ar); auxiliary gas flow 0·9 l/min (Ar)] at the Institute of Mineralogy, Würzburg. Laser parameters used were a frequency of 10 Hz and an energy setting of 50% (0·9 mJ). Ablation patterns were 600 µm long lines, ablated with a scan speed of 10 µm/s. The diameter of the sample pit created by the laser is 50 µm. Data acquisition was done in Time Resolved Analysis mode with measurements of the instrument background (20 s) and of sodalite (60 s). The certified reference material NIST 612 with the values of Pearce et al. (1997) was used as an external standard, whereas Si in sodalite (derived from the electron microprobe analyses) was used as internal standard. Raw counts for each element were solely corrected by subtracting the background counts and processed using the software GLITTER (Version 3·0; On-line Interactive Data Reduction for the LA-ICPMS, Macquarie Research Ltd., 2000).
Stable isotope ratios are reported for hand-picked mineral separates of feldspar, nepheline, magnetite and ankerite. Silicates and oxides were treated with ClF3 according to the method described by Borthwick & Harmon (1982). Carbonate samples were treated with 100% phosphoric acid. CO2 was measured mass spectrometrically with a Finnigan MAT 251 at the Geochemical Institute, University of Göttingen. All 18O values are given relative to SMOW, 13C values relative to PDB. The reproducibility of both values is better than ±0·2.
|
PETROGRAPHY |
|---|
Fenitized anorthosite
The petrography and mineral chemistry of the anorthositic rocks of the KIC has been discussed in detail by Drüppel et al. (2001)
and Drüppel (2003). In summary, both the white and the dark anorthosite suite display a primary magmatic assemblage of plagioclase (An37–53 and An43–75, respectively) ± olivine (XMg 0·54–0·65) ± orthopyroxene (XMg 0·53–0·71) ± clinopyroxene (XMg 0·66–0·76) + ilmenite + magnetite ± biotite (XMg 0·39–0·67) ± amphibole (XMg 0·46–0·73; Fig. 3a).
|
As the contact with the carbonatitic breccia dykes is approached the magmatic plagioclase of the dark anorthosite suite becomes progressively altered to sodium-rich plagioclase (An1–17; Fig. 3b) with or without sericite. In the same direction, the modal abundance of carbonate, pyrite and chalcopyrite increases; clinopyroxene and orthopyroxene are replaced, to an increasing extent, by biotite–epidote–calcite assemblages and/or fine-grained intergrowths of bluish magnesio-riebeckite (XMg 0·75–0·82) and calcite. Fluorite and calcite may fill cracks in the rock. At the direct contact with the carbonatite dykes, the anorthosites exhibit a banded appearance as a result of alternation of layers rich in albite (An0–3), ankerite and biotite (XMg 0·26–0·39) (Fig. 3b). The replacement of albite by sodalite and/or sodalite–ankerite intergrowths along both its margins and the albite twin lamellae (Fig. 3c) is responsible for the colour change of the fenitized rocks from grey to purplish blue. The absence of anhydrous Fe–Mg silicates and the progressive transformation of intermediate plagioclase into pure albite and sodalite provide evidence for the hydrous and alkali-rich nature of the fenitizing fluid. The most likely source of the fenitizing solutions is the Swartbooisdrif carbonatites, because fenitization occurs only at the contacts of large carbonatite dykes and along fractures and joints within a few metres of the carbonatites, whereas anorthosites bordering syenites are not altered.
Fenitized syenite
Unaltered syenite is characterized by a massive appearance. The main minerals present are coarse- to medium-grained K-feldspar (Or91–98) containing albite exsolution lamellae (An1–4) and plagioclase (An0–14; Fig. 3d). Late magmatic clinopyroxene (XMg 0·45–0·51) and/or hastingsite (XMg 0·01–0·06) form discrete grains in the interstices between feldspar crystals, whereas subordinate and texturally late quartz has been observed in only one of 28 samples investigated. Common accessories are titanite, epidote, clinozoisite, Fe–Ti oxides and zircon.
The main constituent of the syenites intruded by carbonatite is mostly twinned and subordinately untwinned granular albite (An0–5) associated with muscovite (Fig. 3e), both presumably formed at the expense of former K-feldspar. Fragmented relics of chessboard albite are preserved locally. Fine-grained and granular ankerite forms part of the matrix. Heavily strained and partially recrystallized biotite (XMg 0·26–0·39) as well as anhedral ilmenite and magnetite grains occur as patches or irregular streaks. Biotite has presumably formed at the expense of magmatic hastingsite and/or clinopyroxene. Sodalite as well as minor analcite, both replacing secondary albite, have been observed in only one sample. Fissures in the rock are invaded by ankerite. In the vicinity of carbonatite, hastingsite, clinopyroxene or K-feldspar relics have never been observed in syenites, suggesting that they were pervasively altered by Na-rich hydrous fluids.
Fenitized nepheline syenite
This rock type is characterized by a streaky appearance caused by changes in the grain sizes of the nepheline and K-feldspar from medium to coarse grained. Rare unaffected parts of the nepheline syenite show euhedral nepheline and subhedral tabular perthitic orthoclase as the cumulus phases of the nepheline syenite. Minor post-cumulus biotite (XMg 0·50–0·56) is the dominant mineral among the mafic silicates. It occurs as fine-grained anhedral intercumulus grains intimately associated with late-magmatic ilmenite.
When reaching the ferrocarbonatite contacts, the grain size of the nepheline increases, suggesting that carbonatite emplacement may have triggered the nepheline growth. Muscovite and cancrinite are localized in patches, marginally replacing individual nepheline grains. Granular ankerite fills cracks in the rock. Both at its grain margins and along cracks nepheline is altered to sodalite (Fig. 3f), suggesting an interaction of the nepheline syenite with NaCl-rich, Si-undersaturated fluids. The fact that the conversion of nepheline into sodalite is restricted to the contacts between nepheline syenite and carbonatite makes the carbonatite magma the most likely source of fenitization.
Carbonatitic breccia
Because the composition and internal structure of the carbonatitic breccia change on a millimetre scale, only a simplified description of this rock type can be given. A detailed description of the sulphide and oxide mineralogy and textures of this rock-type has been presented by von Seckendorff & Drüppel (1999), von Seckendorff et al. (2000) and Drüppel (2003).
Subhedral to anhedral and partially recrystallized granular ankerite (5–40 vol. %) is the main constituent of the cementing ferrocarbonatite matrix (Fig. 4a), whereas subordinate anhedral calcite fills the interstices between ankerite grains. Optically zoned apatite has been observed in two sodalite-bearing and two sodalite-free rock samples. The euhedral to subhedral apatite prisms frequently contain inclusions of albite, magnetite and rare biotite. Apatite is commonly intimately associated with subhedral biotite (XMg 0·37–0·57) and a first, mostly euhedral and inclusion-free generation of magnetite grains of up to 0·5 mm in diameter (Fig. 4b), but may also occur as isolated grains, dispersed in the carbonatite matrix. Apatite, biotite and magnetite appear to have co-crystallized early in the magmatic history of the carbonatite, although biotite and magnetite have overgrown apatite locally. At their margins, both biotite and apatite are partially recrystallized to granular mosaics, supporting evidence that shearing outlasted the carbonatite emplacement and crystallization. Common accessories are fine-grained, unzoned pyrochlore crystals up to 100 µm in diameter, pyrite and chalcopyrite. Rare albite xenocrysts are dispersed in the carbonate matrix. In clear contrast to the anhedral and fragmented albite in silicate-rich zones, these albites are oval or subrounded in shape, suggesting that they underwent a marginal solution by the carbonatite magma.
|
Relic fenitized wall-rock material may be enriched in layers dominated by fragmented and very fine-grained pure albite with undulose extinction. In addition, wall-rock fragments form irregular bounded zones mainly composed of twinned to untwinned granular albite pseudomorphing the former xenolith boundaries and are wrapped by carbonate-rich layers. Sodalite occurs in different textural positions: (1) as monomineralic layers where sodalite replaces very fine-grained fragmented albite; (2) as a replacement product of brecciated and albite-rich wall-rock xenoliths, surrounded by narrow reaction rims of subhedral biotite or small euhedral magnetite grains. Sodalite of both textures may be accompanied by ankerite (Fig. 4c). As evidenced by biotite–sodalite inclusion relationships, the replacement of albite by sodalite started during the end-stages of biotite formation. Along cracks sodalite is altered to an as yet unknown, fibrous, Si-free Na–Al phase (Drüppel, 2003
). A second generation of anhedral, poikilitic magnetite grains up to 2 cm in diameter, which frequently contain inclusions of albite, sodalite, apatite, biotite and ankerite, occurs as matrix-overgrowths. Subordinate muscovite and cancrinite replace albite, whereby previously formed sodalite is preserved. Cancrinite may also be present as reaction rims around albite- and sodalite-rich patches, separating carbonate-rich from silicate-rich zones. Late-stage calcite occurs in most of the samples, where it overgrows all previously mentioned minerals. Dolomite, haematite, ilmenite, rutile, epidote, chlorite and various sulphides are common accessories. In two REE-rich samples of the carbonatitic breccia, carbocernaite [general formula (Ca,Na)(Sr,Ce,Ba)(CO3)2; up to 15 vol. %] is concentrated in large brown carbonate aggregates, most probably pseudomorphing the shape of a suspected subhedral precursor carbonate (presumably calcite or ankerite). In these samples, the Ca–Sr–REE carbonate is associated with subordinate subhedral barite and anhedral strontianite grains up to 40 µm in width, both localized between ankerite, carbocernaite and calcite grains; chalcopyrite–pyrite assemblages are replaced by late millerite and violarite. The extreme REE enrichment in the two samples may result from a strong fractionation of the carbonatite melt; however, the mineralization style, i.e. the co-crystallization of carbocernaite, strontianite and barite, rather suggests a secondary REE enrichment of these rocks by the late injection of highly fractionated, Sr-, Ba- and REE-enriched fluids.
Ferrocarbonatite
Major phases of the massive ferrocarbonatites are subhedral to anhedral ankerite grains with curved grain boundaries and subordinate subhedral magnetite grains that frequently display albite inclusions. Minor calcite fills the interstices between ankerite grains whereas rutile occurs as rims around magnetite. In one sample optically zoned pyrochlore euhedra, containing inclusions of pure albite (An0), have been observed in the vicinity of magnetite grains (Fig. 4d). The contacts between the ferrocarbonatite veins and the older carbonatitic breccia are locally marked by trails of magnetite grains within the ferrocarbonatite. In two samples, cancrinite rims up to 2 mm are developed between the Si-poor ferrocarbonatite and the carbonatitic breccia, pointing to a minor degree of metasomatic exchange between the two rock units. Rounded fragments of sodalite and mostly untwinned albite (An0–1) with undulose extinction, both presumably derived from the fenitized wall-rock anorthosite or the carbonatitic breccia, may form subordinate constituents of this rock type. Thompson et al. (2002) additionally described the presence of quartz within the ferrocarbonatite veins, incorporated from the fenitized wall-rock anorthosite. In clear contrast to those workers, we never observed quartz in a total of 112 samples investigated, comprising both the carbonatitic breccia and the younger ferrocarbonatite veins. The lack of quartz in our rock samples agrees well with the general absence of quartz in the anorthosites of the KIC (Drüppel et al., 2001) and their fenitization products.
|
MINERAL CHEMISTRY OF SELECTED MINERALS |
|---|
Carbonate minerals
Ankerite
Carbonates with XMg at or near the ankerite–dolomite transition were observed in two samples of the carbonatitic breccia, where they occur (1) as euhedral, early crystallized carbonates (XMg 0·61–0·77) and (2) as inclusions (XMg 0·79–0·84) in early crystallized magnetite (Table 1). Anhedral matrix carbonates of the main carbonatite body display distinctly lower Mg contents, which slightly decrease in the sequence of sodalite-poor carbonatitic breccia (XMg 0·56–0·74) to sodalite-rich carbonatitic breccia (XMg 0·43–0·64) to REE-rich carbonatitic breccia (XMg 0·25–0·61). All analysed ankerites show minor oscillatory to patchy zoning with the Mg contents commonly decreasing from core towards the rim, probably reflecting the extensive, synchronous biotite formation. In the same direction both MnO and SrO increase from values of 0·6 to 2·5 wt % and 0·1 to 0·8 wt %, respectively. The Ce, La and Na contents of ankerite are near or below the detection limit.
|
Ankerite (XMg 0·43–0·64) from the ferrocarbonatite veins has Mg amounts comparable with those of matrix ankerite of the carbonatitic breccia (Table 1), but, at the same time, distinctly higher MnO contents of 2·3–3·4 wt %, suggesting its crystallization from a more fractionated melt. Most of the ankerites preserve nearly unzoned Fe-rich cores (XMg 0·43–0·49) and display strongly reversed rims (XMg 0·55–0·64) where the Mg content increases abruptly. This increase is most probably related to the onset of magnetite crystallization.
Carbonates in the fenitized anorthosites and syenites are ankerites with a patchy zonation and Mg/Fe ratios (XMg 0·48–0·64) in the range of the main carbonatite body, indicating that these rocks were invaded by the ferrocarbonatite melt.
Calcite
Calcite filling the interstices between ankerite grains of the carbonatitic breccia has high Sr/Mn ratios (0·2–0·3 wt % MnO; 0·7–0·8 wt % SrO), whereas secondary calcite that overgrows the granulated matrix displays comparably high MnO contents of 0·7–1·6 wt % and corresponding SrO contents of 0·3–1·3 wt % (Table 1). With MnO contents of 1·2–1·8 wt % and SrO contents of 0·5–0·6 wt %, calcite in the fenitized syenite has the lowest Sr/Mn of all calcites analysed.
Carbocernaite and calcian strontianite
Carbocernaite (Ca,Na)(Sr,Ce,La)(CO3)2 has been analysed in two REE-rich samples of the carbonatitic breccia (Table 1). It has been distinguished from the chemically similar mineral burbankite by X-ray diffraction and its stoichiometry. The analysed carbocernaite grains have a rather variable composition with the CaO and SrO contents in the range of 14·3–18·1 wt % and 13·6–18·6 wt %, respectively. Decreases of Ca and Sr are accompanied by increases of Na, Ce and La. The composition of the analysed carbocernaite ranges between (Ca1·01Na0·09Sr0·58Ba0·06Ce0·15La0·11)(CO3)2 and (Ca0·97Na0·10Sr0·50Ba0·03Ce0·23La0·17)(CO3)2. The chemistry of the carbocernaite is similar to that obtained by Wall et al. (1993) for carbocernaite exsolution lamellae in calcite in a REE-rich carbonatite dyke from Rajasthan, India. The analysed strontianite crystals are characterized by high CaO contents of up to 18·3 wt %.
Feldspars
Representative analyses of feldspars are given in Table 2. In the unaltered syenite, alkali-feldspar (reintegrated composition Ab43·3An1·9Or54·8–Ab88·3An3·1Or8·6) exhibits a strong perthitic exsolution, with a potassium feldspar host (Or91–98) containing albite (Ab95–99) exsolution lamellae. Plagioclase in the syenite may be subdivided into two types, one displaying an inverse zonation with An contents increasing slightly from the core (An9–11) towards the rim (An11–14) and a second type of unzoned plagioclase crystals (An13–18). The alkali-feldspar of the syenite is transformed into almost pure albite (An0–1) via the reaction
 | (1) |
 | (2) |
|
Nepheline
The analysed compositions of magmatic nepheline are in the range of Ne68·4Ks19·1Qtz12·5 to Ne73·6Ks18·6Qtz7·8 (Table 3). In most cases the grains display a continuous growth zonation, with SiO2 decreasing towards the rim. When approaching the outermost margins or cracks nepheline is significantly richer in the Ne component (Ne73·1Ks21·4Qtz5·5–Ne80·3Ks11·2Qtz8·5). These chemical variations may result from an interaction of the nepheline syenite with sodium-rich fluids, which presumably caused the subsolidus alteration of nepheline into sodalite. A similar subsolidus origin of sodalite has been proposed by Finch (1991)
for sodalite in nepheline syenites of the Igaliko complex, South Greenland. The stability of sodalite under hydrothermal conditions has been investigated experimentally by Kostel'nikov & Zhornyak (1995), who found that the stability of sodalite with respect to nepheline increases with increasing NaCl concentrations of the fluid.
|
Sodalite
The metasomatic sodalite, formed (1) in the fenitized wall-rock anorthosite, (2) in fenitized wall-rock xenoliths incorporated by the carbonatites and (3) as a replacement product of nepheline in the nepheline syenite, is the almost pure sodalite end-member with SO3 contents below the detection limit (Table 4). The transformation of plagioclase in the anorthosites and nepheline in the nepheline syenite into sodalite necessitates the introduction of large amounts of NaCl, whereas Si will be released, as is evidenced by the end-member reactions
 | (3) |
 | (4) |
|
For the analysis of the trace element contents of sodalite, in situ LA-ICP-MS analyses were performed, as measurements of hand-picked sodalite separates gave mostly unreasonable and highly variable values of the respective elements, resulting from a variable but generally high amount of mineral and fluid inclusions contained in the sodalite. Trace element contents of sodalite are generally low, but unexpectedly high values of As (422–657 ppm) have been recorded (Table 4). Remarkably, the REE patterns of sodalite are similar to those obtained for the anorthosite samples (Fig. 5), even though sodalite contains lower absolute values of the respective elements. All REE patterns are characterized by an enrichment of the light REE (LREE) and a weak positive Eu anomaly, although the Ca contents of sodalite are below the detection limit. This fact is taken as evidence that the sodalite, which is abundant in the carbonatitic breccia, was formed at the expense of plagioclase in the incorporated anorthosite xenoliths.
|
Apatite
Apatites in both the sodalite-free and the sodalite-rich carbonatitic breccia are strontian fluorapatites with F contents increasing from an average value of 2·80 wt % (1·3–4·8 wt %) in sodalite-rich samples to 3·64 wt % (2·2–5·1 wt %) in sodalite-free samples, with the values partly even exceeding two atoms per formula unit (Table 5). In the same sequence of samples the concentrations of Na2O (0·33–1·09 wt %) and BaO (0–0·14 wt %) increase. The SrO contents of the apatites are in the range of 1·06–2·46 wt %, with the highest SrO concentrations obtained for an apatite of a sodalite-free carbonatitic breccia (Fig. 6). Most of the investigated apatites exhibit only minor amounts of Cl, which, however, may reach values of up to 0·31 wt % in some apatites. Even though the fluorapatites are optically zoned no chemical zonation has been detected with respect to the elements analysed by electron microprobe. High-resolution synchrotron micro-XRF analyses of apatite, however, revealed oscillatory zoning patterns with respect to the REE, whereas Fe and Y increase towards the outer margins (Drüppel, 2003
; K. Drüppel & K. Rickers, unpublished data, 2003). The 
REE of apatite ranges between 0·9 and 3·1 wt %. The fluorapatites have relatively high La contents of 0·2–0·7 wt % and high, positively correlated (La/Nd)cn and (La/Yb)cn ratios increasing from 1·5 to 6·4 and from 33 to 240, respectively. All fluorapatites display straight REE patterns and weakly positive to distinctly negative (Eu/Eu*)cn anomalies of 0·3–1·3 (Drüppel et al., 2002; Drüppel, 2003). Most apatites involve a substitution mechanism of the ‘belovite scheme’ (Rønsbo, 1989): Ca2+ + Ca2+ = Na+ + REE3+. The high REE, (La/Nd)cn and (La/Yb)cn ratios of the analysed apatites as well as their straight REE patterns agree well with the findings of Bühn et al. (2001) for fluorapatites that crystallized from carbonatite melts.
|
|
Biotite
Representative analyses of biotite in the anorthosite and the carbonatitic breccia are given in Table 6. End-member calculations define the analysed biotites as meroxenes and lepidomelanes. All biotites are characterized by minor oscillatory to concentric zoning, involving a core preference for Al and Fe, whereas the rim trends to higher Si and Mg. Biotite displays a considerable variation of XMg between both rock types and individual samples. Based on major mineralogical differences between the respective samples, biotite in the carbonatitic breccias can be subdivided into the following three groups: (1) metasomatically formed biotite in assemblage with albite and/or sodalite in fenitized wall-rock xenoliths (Bt1); (2) magmatic biotite in assemblage with apatite in sodalite-bearing samples (Bt2); (3) magmatic biotite in assemblage with apatite in sodalite-free samples (Bt3). The XMg of secondary Bt1 is highly variable but clusters around 0·35 (0·26–0·60), whereas magmatic biotite in assemblages (2) and (3) exhibits higher and more restricted average values of 0·49 (0·44–0·57) and 0·44 (0·37–0·52), respectively.
|
The highest but most variable Cl contents of up to 0·24 wt % occur in Bt1 of the fenitized wall-rock xenoliths. In contrast, magmatic Bt2 and Bt3 in assemblage with apatite contain Cl amounts at or below the detection limit (up to 0·11 wt %) corresponding to the higher XMg values (Table 6). The results are in agreement with experimental studies of Munoz & Swenson (1981)
and Volfinger et al. (1985), who suggested a preferential incorporation of Cl into Fe-rich biotites, although the predicted trend of ‘Mg–Cl avoidance’ is not very pronounced in the rock samples investigated. In particular, Bt1 in sample Ku-98-59b has Cl contents that are much more variable than the corresponding XMg. Therefore, it appears that Cl incorporation in the metasomatically formed Bt1 is not solely governed by Mg–Cl avoidance, but additionally results from the interaction of Bt1 with metasomatic fluids. The generally low chlorine contents in the analysed micas, especially in Bt2 and Bt3 of presumed magmatic origin, as well as those of the analysed apatites, are consistent with those found in biotite and apatite of carbonatite complexes elsewhere. According to Gittins (1989), chlorine preferentially goes into Cl- and alkali-rich fenitizing fluids, which are separated when a carbonatite magma reaches water saturation. Distinct trends are also displayed by the Si/Al ratios of the biotite subgroups (Fig. 7). The majority of Bt1 exhibits a scattered distribution but plots within the respective compositional fields of biotite of the bordering anorthosite and syenite. In clear contrast, Bt2 and Bt3, which both crystallized from the carbonatite melt, display a distinct trend extending from high Al values towards Al-deficient compositions compared with ideal Si + Al = 4 behaviour, suggesting that the carbonatite melt contained only minor amounts of Al that were consumed during biotite formation.
|
Pyrochlore
Medium-grained and optically zoned pyrochlore is restricted to the late ferrocarbonatite veins; the grain sizes of unzoned pyrochlore crystals in the carbonatitic breccia range from fine grained to very fine grained. Electron microprobe analyses reveal high SrO and Na2O contents of 0·96–2·24 wt % and 6·54–7·98 wt %, respectively (Table 7). The fine-scale oscillatory zonation of pyrochlore, observed microscopically, is also displayed in its mineral chemistry. All analysed grains show a concentric zonation pattern with a core preference for Ca and Ti, whereas Fe, Ba, Ce and Sr are concentrated in the rim.
|
|
GEOCHEMISTRY |
|---|
Major and trace element compositions of anorthosites, syenites and carbonatites are presented in Table 8.
|
Anorthosite
Major and trace element data for the anorthositic rock suites display a consistent correlation pattern, involving an increase of MnO, MgO, FeOt and TiO2 and decreasing SiO2, Al2O3 and CaO from anorthosite to troctolite and leucogabbronorite. Because of the strong variability of the chemical composition of the anorthositic rocks, fenitized anorthosite samples mostly plot within the compositional range of unaltered anorthosites in selected Harker diagrams (Fig. 8).
|
Syenite
The data for samples of unaltered syenites form coherent arrays in major and trace element variation diagrams (Fig. 8), involving a negative correlation of SiO2 with TiO2, Al2O3, MnO, MgO, CaO, Na2O and Sr, whereas K2O, Rb and Y correlate positively with SiO2. Samples of weakly fenitized syenite trend to higher Na2O, Zn, Sr, Nb, CO2 and H2O and lower SiO2, K2O and Ba, but do not display a clear correlation for the respective elements.
Nepheline syenite
The weakly fenitized nepheline syenite analysed has an alkali-rich, peraluminous and silica-undersaturated composition, thus differing significantly from the older syenites. The Al2O3, Na2O, K2O and Rb contents are higher than those of the fenitized syenites, whereas SiO2, CaO, FeOt, Sr, Ba, CO2 and H2O are considerably lower (Fig. 8). With a value of 43 ppm, the 
REE of the fenitized nepheline syenite is extremely low. The sample is characterized by a straight REE pattern and comparably low (La/Nd)cn and (La/Yb)cn ratios of 1·2 and 3·4, respectively (Fig. 9c). A distinctly negative (Eu/Eu*)cn anomaly of 0·5 suggests plagioclase fractionation of the magma parental to the nepheline syenite.
|
Carbonatitic breccia and ferrocarbonatite veins
Compared with the anorthosites, the 18 samples of the carbonatitic breccia (excluding the REE-rich samples) generally contain significantly lower absolute amounts of SiO2 (22–45 wt %) but, nevertheless, mostly display similar correlation patterns when plotted in selected Harker diagrams (Fig. 8). Hence anorthosite is assumed to be the dominant component of the wall-rock xenoliths. Increasing carbonatite/xenolith ratios are mirrored by increases of FeOt, CaO, P2O5, S, Sr, Nb, Zn, Ba, Th, U, the LREE and CO2, suggesting that the above-mentioned elements represent primary constituents of the carbonatite magma. Sodalite-rich samples of the carbonatitic breccia trend to high Al2O3 and Na2O, whereas the sodalite-poor and sodalite-free samples are generally higher in TiO2, CaO, MgO, FeOt and V, reflected by higher contents of biotite, magnetite and ankerite.
Because most of the wall-rock fragments incorporated and subsequently fenitized by the carbonatite are almost entirely composed of albite and/or sodalite (95–100 vol. %), the totals of Na2O were subtracted from the analyses, together with corresponding amounts of SiO2 and Al2O3 (calculated for the petrographically observed albite:sodalite ratios of each single sample, i.e. 0·7–1·0 in Sdl-poor and 0·45–0·55 in Sdl-rich samples), to obtain an approximation of the composition of the carbonatite matrix (Table 9). Samples containing metasomatically formed biotite and/or muscovite were excluded from the calculation. Recalculated SiO2 and Al2O3 contents display ranges of 2–10 wt % and 0·5–6 wt %, respectively, with the highest values being obtained for samples that also contain significant amounts of K2O (1–5 wt %), mirrored by the abundance of orthomagmatic biotite. Values for FeOt, CaO, MgO, MnO, TiO2 and P2O5 are similar to those of ferrocarbonatites world-wide [compiled by Woolley & Kempe (1989)]. Following the IUGS classification (Le Maitre, 1989), which is adopted for this study, the carbonatite matrix of the carbonatitic breccia is ferrocarbonatitic in nature (Fig. 9a and b).
|
REE-rich samples of the carbonatitic breccia exhibit similar low SiO2 (4–11 wt %) and Al2O3 contents (2–6 wt %), both positively correlated with the Na2O content (6–8 wt %), pointing to a minor degree of wall-rock contamination. The amounts of FeOt and MgO are too low to classify the investigated samples as ferrocarbonatites, whereas extremely high values were obtained for CaO (15–16 wt %), Sr (3·8–8·9 wt %), Ba (0·8–2·2 wt %), La (2·2–5·4 wt %), Ce (3·3–7·9 wt %) and Nd (1·1–3·0 wt %), reflected in the high modal amounts of secondary Ca–Sr–REE carbonates, strontianite and barite.
Compared with the carbonatitic breccia, the late ferrocarbonatite veins are characterized by low SiO2 contents (0–1·4 wt %), because of the low potential of these small-scale veins to incorporate wall-rock material. Values of FeOt, MgO and CaO are similar to the mean values of ferrocarbonatites world-wide, as summarized by Woolley & Kempe (1989). The MnO contents are high (2·33–2·43 wt %), whereas the amounts of K2O, TiO2, P2O5 and S are near the detection limit. The late ferrocarbonatite veins exhibit Sr, La, Ce and Nd contents in the ppm range, although Sr contents may be as high as 0·5 wt %. According to the classification of both Le Maitre (1989) and Gittins & Harme