Journal of Petrology Advance Access originally published online on September 24, 2004
Journal of Petrology 2005 46(1):191-217; doi:10.1093/petrology/egh069
Journal of Petrology vol. 46 issue 1 © Oxford University Press 2004; all rights reserved
The Grønnedal-Ika CarbonatiteSyenite Complex, South Greenland: Carbonatite Formation by Liquid Immiscibility
RALF HALAMA1,*,
TORSTEN VENNEMANN2,
WOLFGANG SIEBEL1 and
GREGOR MARKL1,
1 INSTITUT FÜR GEOWISSENSCHAFTEN, EBERHARD-KARLS-UNIVERSITÄT TÜBINGEN, WILHELMSTRASSE 56, D-72074,TÜBINGEN, GERMANY
2 INSTITUT DE MINÉRALOGIE ET GÉOCHIMIE, UNIVERSITÉ DE LAUSANNE, UNIL-BFSH2, CH-1015 LAUSANNE, SWITZERLAND
RECEIVED
NOVEMBER 6, 2003;
ACCEPTED
AUGUST 2, 2004
 |
ABSTRACT
|
|---|
The Grønnedal-Ika complex is dominated by layered nepheline
syenites which were intruded by a xenolithic syenite and a central
plug of calcite to calcitesiderite carbonatite. Aegirineaugite,
alkali feldspar and nepheline are the major mineral phases in
the syenites, along with rare calcite. Temperatures of 680910°C
and silica activities of 0·280·43 were
determined for the crystallization of the syenites on the basis
of mineral equilibria. Oxygen fugacities, estimated using titanomagnetite
compositions, were between 2 and 5 log units above the fayalitemagnetitequartz
buffer during the magmatic stage. Chondrite-normalized REE patterns
of magmatic calcite in both carbonatites and syenites are characterized
by REE enrichment (La
CNYb
CN = 1070). Calcite from
the carbonatites has higher Ba (

5490 ppm) and lower HREE concentrations
than calcite from the syenites (54106 ppm Ba). This is
consistent with the behavior of these elements during separation
of immiscible silicatecarbonate liquid pairs.
Nd(T =
1·30 Ga) values of clinopyroxenes from the syenites vary
between +1·8 and +2·8, and
Nd(T) values of whole-rock
carbonatites range from +2·4 to +2·8. Calcite
from the carbonatites has
18O values of 7·8 to 8·6
and
13C values of 3·9 to 4·6

.
18O
values of clinopyroxene separates from the nepheline syenites
range between 4·2 and 4·9

. The average oxygen
isotopic composition of the nepheline syenitic melt was calculated
based on known rockwater and mineralwater isotope
fractionation to be 5·7 ± 0·4

. Nd and CO
isotope compositions are typical for mantle-derived rocks and
do not indicate significant crustal assimilation for either
syenite or carbonatite magmas. The difference in
18O between
calculated syenitic melts and carbonatites, and the overlap
in
Nd values between carbonatites and syenites, are consistent
with derivation of the carbonatites from the syenites via liquid
immiscibility.
KEY WORDS: alkaline magmatism; carbonatite; Gardar Province; liquid immiscibility; nepheline syenite
 |
INTRODUCTION
|
|---|
Carbonatites typically occur in close association with alkaline
silicate igneous rocks, either in individual complexes or in
a regional association within particular magmatic provinces
(e.g. Harmer, 1999

). However, the genetic relationship between
carbonatites and associated silicate rocks is complex and not
completely understood (e.g. Bell, 1998

; Bell
et al., 1998

).
Proposed models for the generation of carbonatites include direct
melting of a carbonate-bearing mantle source (e.g. Wyllie &
Huang, 1976

; Dalton & Presnall, 1998

; Harmer & Gittins,
1998

; Moore & Wood, 1998

), derivation by immiscible separation
from a carbonated silicate melt (e.g. Koster van Groos &
Wyllie, 1973

; Freestone & Hamilton, 1980

; Kjarsgaard &
Hamilton 1988

; Brooker, 1998

; Kjarsgaard, 1998

) and crystal
fractionation of a carbonated alkali silicate melt (e.g. Lee
& Wyllie, 1994

; Korobeinikov
et al., 1998

; Veksler
et al.,
1998
a
). There is, however, still a lack of robust criteria that
can be used to distinguish carbonatitic melts directly derived
from the mantle from those produced by differentiation of a
parent silicate melt (Bell, 1998

).
This study explores geochemical and petrological data from the mid-Proterozoic Grønnedal-Ika nepheline syenitecarbonatite complex in the Gardar Province, South Greenland. The complex represents the most significant occurrence of carbonatite in the province. It is well suited to study the relationship between carbonatite and associated silicate rocks because the carbonatite occurs in close spatial association with nepheline syenitic rocks. Previous studies of Grønnedal-Ika investigated the petrography, whole-rock geochemistry and mineral chemistry of the complex (Emeleus, 1964
; Bedford, 1989
; Goodenough, 1997
; Pearce et al., 1997
). In this study, we focus on new O and Nd isotope data from minerals of the syenites combined with Nd isotope data from the carbonatite to address still open questions about liquid immiscibility. In situ trace-element data from minerals are used to evaluate the constraints given by experimental and theoretical carbonate liquidsilicate liquid partitioning data. Additionally, the petrological and geochemical characteristics of the syenites are compared with other syenitic complexes in the province in order to explore differences in terms of magma evolution that can be related to carbonatite formation.
 |
PREVIOUS WORK AND GEOLOGICAL SETTING
|
|---|
The first detailed account of the Grønnedal-Ika complex
was given by Callisen (1943)

, who established the intrusive
character of the alkaline rocks. Emeleus (1964)

reconstructed
the history of the complex based on detailed mapping and petrographic
observations. Later, Gill (1972
a
, 1972
b)

and Bedford (1989)
obtained numerous geochemical data on minerals and whole rocks
from the complex and associated phonolite dykes. C, O and Sr
and a few Nd whole-rock isotopic data were obtained by Goodenough
(1997)

, Pearce
et al. (1997)

and Coulson
et al. (2003)

.
The Grønnedal-Ika complex (Fig. 1a) is 8 km x 3 km in exposed dimension and consists predominantly of layered nepheline syenites, which were intruded by a xenolithic porphyritic syenite and a plug of carbonatite (Emeleus, 1964
). It was emplaced into the border zone between the Julianehåb batholith (18501800 Ma; Garde et al., 2002
), which is part of the Ketilidian mobile belt to the south and the Archean block to the north (Fig. 1b). The country rocks comprise quartzo-feldspathic gneisses, occasional amphibolites and rare calc-silicates (Bedford, 1989
). The complex was dated at 1299 ± 17 Ma by RbSr (Blaxland et al., 1978
). This age is consistent with field observations indicating that it pre-dates emplacement of a suite of olivine dolerite dykes (Brown Dikes) for which an UPb age of 1280 ± 5 Ma has been determined (Upton et al., 2003
). The layered and laminated syenites are considered to have formed mainly by consolidated by bottom accumulation and adhesion to steep-sided cooling surfaces (Emeleus, 1964
). The original structure of the complex comprises two series of layered syenitesthe Lower Series and the Upper Series, which are separated by a raft of gneiss (Emeleus, 1964
). Several distinct intrusive phases can be distinguished, but gradational relationships between different rock types within intrusive units are common (Callisen, 1943
; Emeleus, 1964
). For a detailed description of the syenitic rocks from Grønnedal-Ika, the reader is referred to the previous studies of Emeleus (1964)
and Bedford (1989)
. For the purpose of this study, the Upper and Lower Series syenites are combined as the Layered Syenites (LS). At one locality in the Lower Series syenites, some 20 gabbroic xenoliths, roughly 530 cm in diameter, were found. Porphyritic microsyenites (PMS) occur as dykes and sheets, and cut across earlier units of the complex. A xenolithic porphyritic syenite (XPS) is the youngest of the intrusive syenites. It contains abundant xenoliths of earlier syenites, trachytes, gneisses and amphibolites. Finally, a central plug of calciocarbonatite was emplaced into the syenites (Emeleus, 1964
; Pearce et al., 1997
). There are three major and several smaller outcrops of carbonatite that are thought to have belonged to one central plug (Emeleus, 1964
). The carbonatite is xenolithic with fragments of syenites and other rocks. It consists essentially of varying amounts of calcite, siderite and magnetite; calcite is dominant. Towards the centre of the carbonatite plug, the amount of siderite increases. Large amounts of magnetite occur where mafic dykes cut the siderite-rich part of the carbonatite (Emeleus, 1964
). The syenites in close contact with the carbonatite are altered and impregnated by calcite. Sodalite and sodalite ± calcite veins occur in various units of the complex. There is no field evidence for a significant time-gap between the emplacement of the syenites and the carbonatite. The last stage of magmatic activity, at about 1280 Ma, is represented by the intrusion of a variety of dykes, including lamprophyres, porphyritic dolerites and several olivine dolerites up to a few tens of meters wide. Trachytic, phonolitic and other alkaline dykes were emplaced during renewed magmatic activity during later Gardar (
12501150 Ma) times. Intense faulting, responsible for the elongated outline of the complex, took place largely following, but in part contemporaneous with, the intrusion of these dikes (Emeleus, 1964
).
Mainly on the basis of whole-rock REE and other trace-element
data, Bedford (1989)

proposed an origin for the Grønnedal-Ika
carbonatites via liquid immiscibility from a nepheline syenitic
liquid. He envisaged that the XPS represents the conjugate silicate
liquid to the carbonatite. However, Pearce
et al. (1997)

concluded
that the syenites and carbonatites are unrelated because of
differences in their C and O isotopic compositions. They presented
data suggesting that the carbonatites are, instead, linked to
lamprophyre dikes. The spatial association of carbonatites and
lamprophyres is a common feature in the Gardar Province, and
a general genetic link between them has been proposed previously
(Coulson
et al., 2003

). The data presented in this study are
used to re-evaluate the controversy, and a particular focus
is given to the XPS as a critical unit of the complex.
 |
ANALYTICAL METHODS
|
|---|
Electron microprobe analysis
Mineral compositions were determined using a JEOL 8900 electron
microprobe (EMP) at the Institut für Geowissenschaften,
Universität Tübingen, Germany. An internal


z correction
of the raw data was applied (Armstrong, 1991

). Both natural
and synthetic standards were used for major and minor elements.
Measuring times were 16 and 30 s on the peak positions for major
and minor elements, respectively. The emission current was 15
nA and the acceleration voltage 15 kV. For nepheline analyses,
a beam diameter of 5 µm was used to avoid errors resulting
from diffusion of Na. Feldspars with microperthitic exsolution
were measured with a 20 µm wide beam and an average of
24 analyses was taken as representative for the respective
grain. The bulk compositions of oxy-exsolved titanomagnetite
grains were reconstructed by combining image-processed (NIH
Image software) back-scattered electron images of the exsolved
grains and point analyses of exsolved ilmenite and magnetite
according to the method described byMarks & Markl (2001)

.
Laser ICPMS in situ trace-element measurements
In situ laser ablation inductively coupled plasmamass spectrometry (LA-ICPMS) analyses on
150 µm thick polished sections were performed at the EU Large-Scale Geochemical Facility (University of Bristol) using a VG Elemental PlasmaQuad 3 + S-Option ICPMS system equipped with a 266 nm NdYAG laser (VG MicroProbe II). The laser beam diameter at the sample surface was approximately 20 µm. Analyses of calcite were carried out in a beam rastering mode, as ablation of calcite was rapid. Details of the data acquisition have been described elsewhere (Halama et al., 2002
). The precision of trace-element concentrations, based on repeated analyses of NIST 610 and 612 glass standards, is approximately ±5% for element concentrations >10 ppm and ±10% for concentrations <10 ppm. Typical detection limits for most trace elements in this study were 0·11 ppm, except for Sc, Co and Zn, with detection limits in the range of 15 ppm. For the REE, detection limits were generally <0·5 ppm.
Oxygen and carbon isotope analysis
The oxygen isotope composition of hand-picked clinopyroxene separates was measured using a method similar to that described by Sharp (1990)
and Rumble & Hoering (1994)
. Hand-picked clinopyroxene (12 mg) was loaded onto a Pt-sample holder. The sample chamber was pumped to a vacuum of about 106 mbar and pre-fluorinated overnight. Samples were then heated with a CO2 laser in an atmosphere of 50 mbar of pure F2. Excess F2 was separated from O2 by conversion to Cl2 using KCl held at 150°C and the extracted O2 was collected on a molecular sieve (13X). Oxygen isotopic compositions were measured on a Finnigan MAT 252 mass spectrometer at the Institut für Geowissenschaften, Universität Tübingen. The results are reported as the
deviation from Vienna Standard Mean Ocean Water (V-SMOW) in the standard
-notation. Replicate analyses of the NBS-28 quartz standard yielded an average precision of ±0·1
(1
) for
18O values. In each run, standards were analyzed at the beginning and the end of the sample set. A correction, generally <0·3
, was then applied to the data, equal to the average difference between the mean measured value and the accepted value for the standard of 9·64
.
The carbon and oxygen isotope compositions of the carbonates were measured in automated mode using a GasBench from ThermoFinnigan connected directly to a Finnigan MAT 252 mass spectrometer. The method has been described in detail by Spötl & Vennemann (2003)
. About 200400 µg of sample material was recovered as powder by micro-drilling into the calcite. CO2 was obtained by reaction of the sample with several drops of 100% orthophosphoric acid and was carried via a He stream over traps to remove water vapor and a gas chromatograph-column to separate CO2 from possible interfering gases before being passed into the mass spectrometer for isotopic analysis. The C and O isotope compositions are reported in the standard
-notation, relative to PDB and V-SMOW, respectively; standard analytical errors are ±0·1
(1
) for both
13C and
18O.
Nd isotope analysis
For Nd isotope analyses, 1520 mg of hand-picked clinopyroxene from the same separates used for the O analysis, and 58 and 40 mg of whole-rock powder of carbonatite and basement samples, respectively, were used. Details of the analytical procedures have been described elsewhere (Halama et al., 2003
). 143Nd/144Nd ratios were normalized for mass fractionation to 146Nd/144Nd = 0·7219. The average 143Nd/144Nd ratio obtained for the Ames Nd-standard (Roddick et al., 1992
) was 0·512119 ± 10 (1
, n = 42) during the course of this study, and the La Jolla Nd standard gave 0·511831 ± 30 (1
, n = 12). The total procedural blank (chemistry and loading) was <100 pg for Nd.
 |
PETROGRAPHY AND MINERAL CHEMISTRY
|
|---|
Syenites
The major mafic silicate in the Layered Syenites is euhedral
to subhedral or poikilitic clinopyroxene (
Fig. 2ac).
Clinopyroxenes are typically cumulus phases and commonly rimmed
by intercumulus biotite. Most clinopyroxenes can be classified
as aegirineaugites (Morimoto
et al., 1988

), but diopsidic
pyroxene and aegirine occur occasionally. Typical clinopyroxene
compositions are summarized in
Table 1 and compositional trends
are shown in
Fig. 3ac. In some samples, clinopyroxene
shows a weak, patchy zonation with limited compositional variability.
Other samples contain clinopyroxenes with a marked growth zonation,
with a compositional range in a single crystal as large as that
in the entire sample suite. Compositions vary between Di
56Hed
34Aeg
10 and Di
3Hed
2 Aeg
95. In comparison with compositional trends from
other alkaline complexes of the Gardar Province (
Fig. 3d), the
extreme enrichment in Fe
2+ at low values of Na, as seen in Ilímaussaq
(Larsen, 1976

) and Igdlerfigssalik (Powell, 1978

), is lacking
in Grønnedal. Clinopyroxene compositions in a gabbroic
xenolith show a small range of Di
6559Hed
3036Aeg
5.

View larger version (182K):
[in this window]
[in a new window]
|
Fig. 2. Photomicrographs (plane-polarized light) of mineral textures from the Grønnedal-Ika rocks. Sample numbers appear in the lower right corner. Ne, nepheline; fsp, alkali feldspar; cpx, clinopyroxene; m-cpx, matrix clinopyroxene; bt, biotite; amph, amphibole; ap, apatite; cal, calcite; ccn, cancrinite. (a) Typical texture of the layered nepheline syenites with an accumulation of clinopyroxene, partly surrounded by biotite and with associated apatite; (b) large, poikilitic amphibole; (c) interstitial calcite among biotite, clinopyroxene and alkali feldspar; the boundaries of the cancrinite in the upper left corner are have been outlined to increase its visibility; (d) and (e) clinopyroxene phenocrysts in the XPS showing discontinuous zonation from core to rim, together with small matrix clinopyroxenes; alkali feldspar and nepheline occur as phenocrysts and within the matrix; (f) calcite grains in calcite carbonatite.
|
|
View this table:
[in this window]
[in a new window]
|
Table 1: Representative single spot electron microprobe analyses of clinopyroxene from the Grønnedal-Ika syenites
|
|
In the XPS, clinopyroxene phenocrysts reach up to 2 mm in length
and show a marked discontinuous zoning, whereas matrix clinopyroxene
grains are considerably smaller (<200 µm) and optically
unzoned (
Figs 2d and e, and
4a and b). The clinopyroxene phenocrysts
in the XPS are characterized by a relatively Mg-rich core, overgrown
by a distinctly more Na- and Fe
3+-rich rim with a compositional
gap between (
Fig. 3c). The most primitive core compositions
in the XPS reach values of Di
57Hed
29Aeg
14, whereas the most
evolved rims are Di
7Hed
22Aeg
71. Matrix clinopyroxene compositions
in the XPS overlap with the more evolved phenocryst rim compositions.

View larger version (148K):
[in this window]
[in a new window]
|
Fig. 4. Back-scattered electron images of minerals in the Grønnedal-Ika syenites. Abbreviations as for Fig. 2, and hem, hematite; ab, albite; or, orthoclase. (a) and (b) Zoning in clinopyroxene phenocrysts from the xenolithic porphyritic syenite; (c) and (d) primary magmatic titanomagnetites showing trellis-type oxy-exsolution into magnetite (light) and ilmenitepyrophanitehematiteSS (dark lamellae); (e) late- to post-magmatic hematite associated with biotite; (f) perthitic alkali feldspar and cancrinite.
|
|
Olivine is completely absent from the samples studied, but scarce
relics of Fe-rich olivine have been observed (Emeleus, personal
communication). Amphiboles are also scarce (Bedford, 1989

),
but they occasionally occur as poikilitic or subhedral intercumulus
grains. Only one sample (GM 1531) was investigated in which
poikilitic amphibole with a katophorite composition is the dominant
mafic silicate phase (
Fig. 2b). Biotite, usually as intercumulus
grains (
Fig. 2a), occurs frequently and is often associated
with hematite. All syenites from the Grønnedal-Ika complex
are nepheline-bearing (Emeleus, 1964

), and nepheline usually
occurs as euhedral prisms (
Fig. 2ac). Nepheline is chemically
unzoned and its composition is fairly homogeneous within individual
samples and also within the whole sample suite, including the
PMS and the XPS (
Table 2). There are also no significant compositional
differences between phenocrysts and matrix nepheline. The compositional
variability spans the range Ne
7368Ks
2015Qz
716.
The FeO and CaO contents of nepheline in all rock types are
invariably <1 and <0·1 wt %, respectively. Nepheline
is often replaced by the alteration products gieseckite
(fibrous, micaceous aggregates) and cancrinite (
Fig. 2c). Cancrinite
also occurs as individual grains, partly with poikilitic character
suggestive of an intercumulus origin.
The abundant platy crystals of alkali feldspar frequently show
perthitic exsolution into almost pure albite and orthoclase.
Alkali feldspar phenocrysts of up to 1 cm in length occur in
both the PMS and the XPS (
Fig. 2d). Some samples contain feldspar
with exsolution on a µm scale, so that the original bulk
composition could be obtained using a defocused beam (
Table 2).
These bulk compositions are fairly homogeneous throughout
the different rock types and vary in the range An
01Ab
3954 Or
4660.
Primary magmatic titanomagnetite is relatively rare and characterized by a trellis-type oxy-exsolution texture [terminology after Buddington & Lindsley (1964
)] of magnetite with lamellae of the hematiteilmenitepyrophanite solid solution series (Fig. 4c and d). The compositional variation in the lamellae (Ilm5Pyr93Hem2Ilm56Pyr23Hem21) is in contrast to the constant composition of the magnetite (Mag9899Usp12). Reintegrated oxide compositions vary from Mag61Usp39 to Mag72Usp28 (Table 3). This range reflects the chemical variability in the lamellae and is considered to cover fairly well the overall variation in the samples, as the lamellae are volumetrically of minor importance (1030 vol. %). Except for MnO contents of up to 11·3 wt %, concentrations of minor elements in the reintegrated oxides are low (Al2O3
0·12 wt %, MgO
0·03 wt %, ZnO
0·24 wt %). Late- to post-magmatic hematite associated with biotite has almost a pure hematite end-member composition (Table 3, Fig. 4e).
Minor phases present include short prismatic apatite, zircon,
blue sodalite, analcite, calcite and fluorite. Calcite appears
to be a primary magmatic or late magmatic interstitial phase
and not an alteration product (
Fig. 2c). Sodalite can occur
as small, interstitial grains, partly associated with cancrinite
(Bedford, 1989

), or as larger aggregates in almost pure sodalite
veins. The chemical characteristics of the minor minerals relevant
to this study are briefly summarized based on the work of Bedford
(1989)

. Sodalite contains around 8 wt % Cl but only minor amounts
(<0·3 wt %) of S. Cancrinite (
Fig. 4f), with the ideal
structural formula (Sirbescu & Jenkins, 1999

) Na
6Ca
2[Al
6Si
6O
24](CO
3)
2 +
nH
2O, also has low (<0·15 wt %) concentrations of
S and shows some exchange of Ca for Na. Calcite in the syenites
is almost pure CaCO
3.
Carbonatites
The Grønnedal-Ika carbonatites can be classified as calcite carbonatite and calcitesiderite carbonatite (Woolley & Kempe, 1989
) or, on the basis of whole-rock geochemical data with a maximum of 23·1 wt % Fe2O3 (Bedford, 1989
), as calciocarbonatite and ferruginous calciocarbonatite (Gittins & Harmer, 1997
). The carbonatites consist essentially of varying amounts of calcite, siderite and magnetite. They are poor in UThNbREE minerals compared with other carbonatites of the world. Magnetite is exclusively secondary after original siderite as a result of decarbonation and oxidation in the vicinity of the dolerite dykes (Bedford, 1989
). In places, anhedral to subhedral calcite crystals form almost 100% of the rock (Fig. 2f). Individual calcite crystals are 50500 µm in diameter and commonly show prominent twinning. Siderite occurs as euhedral rhombs or as minute grains and is commonly rusty-brown because of surface weathering. In the siderite-bearing samples, calcite grains tend to have a cloudy appearance. Minor quantities of apatite, sphalerite, pyrite, strontianite, monazite, pyrochlore, feldspar and alkali amphibole are also present in some samples (Emeleus, 1964
; Bedford, 1989
; Bondam, 1992
).
 |
GEOCHEMISTRY
|
|---|
Trace elements
Calcite
Chondrite-normalized REE patterns of calcites are shown in
Fig. 5a.
Three different patterns can be distinguished in the calcites
from the carbonatites. One samplea pure calcite carbonatite
(GM 1502)contains exclusively calcite characterized by
a steep REE pattern with high LREE abundances (type A). There
is a slight positive Eu anomaly present (Eu/Eu
* = 1·25)
and the degree of LREE enrichment relative to the HREE expressed
as La
CN/Yb
CN is

69. The overall REE enrichment of this sample
is within the lower range of whole-rock analyses from the Grønnedal
carbonatite and typical of carbonatites in general (Woolley
& Kempe, 1989

). Two distinctly different types of REE patterns
occur in calcites from sample GM 1498, where type A patterns
are absent. The sample is a calcitesiderite carbonatite
in which siderite is characterized by brown oxidation rims.
Both REE patterns show a strong depletion in LREE relative to
the type A pattern (
Fig. 5a). The main characteristic of the
type B REE pattern is a convex-upward shape. In contrast, type
C calcite has about one order of magnitude lower normalized
LREE and MREE abundances and exhibits an almost continuous increase
from very low LREE abundances towards the HREE. There are no
textural differences between types B and C. Interstitial calcite
from two syenitic samples (Layered Syenites, samples GR 44 and
GR 63) shows LREE-enriched patterns similar to type A, but with
slight negative Eu anomalies (Eu/Eu
* = 0·66) and a flatter
slope (La
CN/Yb
CN = 10·526·3). In a similar
fashion, the REE pattern of the XPS whole-rock matrix is flatter
than that of the whole-rock carbonatites (
Fig. 5a). In addition
to the REE, Sr, Ba, Pb, Y and Ga also occur in significant amounts
in the calcites, whereas all calcites have very low concentrations
of the HFSE (e.g. Nb, Ta, Zr, Hf), mostly below the detection
limits (
Table 4). Type A calcite (GM 1502) is characterized
by a relative enrichment of Ba, Sr, Eu and Ga compared with
the calcites from the syenites (
Fig. 6a). Type B and C calcites
show marked enrichments in Pb and Sr compared with the neighboring
elements. These features and the low REE contents of type B
and C calcite are not entirely clear. They are consistent with
postmagmatic loss of these elements by fluid-controlled recrystallization
and remobilization of carbonate phases (Bau, 1991

; Bau &
Möller, 1992

), for which brown rims around siderite might
be an indication, or they could be related to the contemporaneous
siderite crystallization itself.
View this table:
[in this window]
[in a new window]
|
Table 4: Average trace-element compositions of clinopyroxene and calcite in Grønnedal-Ika rocks determined by laser ablation ICPMS
|
|
Clinopyroxene
A detailed study of the REE and trace-element abundances in
clinopyroxene and amphibole in the Grønnedal-Ika syenites
has been made by Marks
et al. (2004
a)

. In this study, we concentrate
on data from clinopyroxene phenocrysts in the XPS to evaluate
whether features of the trace-element chemistry indicative of
silicatecarbonate liquid immiscibility can be detected.
Typical trace-element concentrations in clinopyroxene cores
and rims are summarized in
Table 4. Chondrite-normalized REE
patterns (
Fig. 5b) indicate distinct differences between core
and rim compositions. The core patterns are characterized by
a smooth increase in normalized abundances from La to Nd and
a smooth decrease from Nd to Tm or Yb, with a slight increase
towards Lu. La
CN/Yb
CN ratios in the cores range from 4·7
to 6·9 and Eu anomalies are slightly positive, with values
between 1·02 and 1·37. In the Na-rich rims, REE
patterns are sinusoidal with generally lower LREE and higher
HREE abundances than in the cores (
Fig. 5b). La
CN/Yb
CN ratios
range between 0·32 and 0·75. Eu anomalies in the
rims are not significantly different from those in the cores
and vary from 0·99 to 1·24.
Trace-element concentrations of XPS clinopyroxene cores and rims normalized to primitive mantle values (McDonough & Sun, 1995
) are shown in Fig. 6b. A principal feature of all the patterns are strong negative Ba, Pb and Ti spikes and depletions in Sc, V and Co. Negative Sr spikes are present in all patterns except for two phenocryst cores. These two cores are also characterized by the lowest Zr and Hf values. All rim compositions have strong positive Zr, Hf and Sn, but negative Li anomalies. In comparison with the core compositions, rims are also enriched in Zn, but depleted in Rb, Co and Sc. All rim patterns are very similar to those of aegirineaugites from the LS, both in enrichment level and shape (Fig. 6b). Aegirines from another silica-undersaturated syenitic complex of the Gardar Province, Ilímaussaq (Marks et al., 2004a
) have much lower absolute trace-element concentrations, but the shape of their patterns has a striking similarity to that of the phenocryst rims, except for Li and Pb.
Ratios of isovalent trace elements with a similar geochemical behavior (Bau, 1996
) were investigated in order to identify characteristic differences between the growth of clinopyroxene cores and rims (Fig. 7). Zr/Hf ratios in the cores vary between 31·5 and 42·1, and scatter around the primitive mantle value of 37·1 (McDonough & Sun, 1995
). In the rims, however, Zr/Hf ratios are higher than in the cores, with values ranging from 45·7 to 58·8. In contrast, Y/Ho ratios in individual phenocrysts are slightly higher in the cores (Y/Ho = 19·622·7) compared with the rims (Y/Ho = 16·820·5). Nb/Ta ratios scatter widely (Nb/Ta = 4·216·3) and there is no consistent relation between core and rim compositions.
Carbon and oxygen isotope analyses
Goodenough (1997)

, Pearce
et al. (1997)

, Coulson
et al. (2003)
and Taubald
et al. (2004)

have reported C and O isotopic data
from Grønnedal-Ika, but data on silicates from the syenites
were lacking so far. Recent oxygen isotope studies have shown
that individual coarse-grained minerals are less sensitive to
low-
T alteration than whole-rock samples (e.g. Vroon
et al.,
2001

). Extensive studies of mineral separates from ocean island
basalts (Eiler
et al., 1997

) and oceanic-arc lavas (Eiler
et al., 2000
a
) demonstrated that analyses of mineral separates
consistently span a narrower range than comparable whole-rock
data. At Grønnedal-Ika, whole-rock oxygen and carbon
isotope data were used as an argument against liquid immiscibility
between syenites and carbonatites because of their distinct
differences in
18O
V-SMOW and
13C
PDB (Pearce
et al., 1997

). To
evaluate this argument in the view of mineral data, the oxygen
isotope compositions of fresh magmatic clinopyroxene and amphibole
in the syenites were measured, augmented by C and O isotopic
data of microdrilled calcite (
Table 5).
The oxygen isotopic composition of clinopyroxene from the syenites
varies from 4·2 to 4·9

and that of amphibole (
18O
= 4·7

) is within this range. The average
18O value of
clinopyroxene from the layered syenites is 4·6

±
0·2

(1

). There is no significant isotopic compositional
difference between the layered syenites and the late syenites.
It was not possible to analyze core and rim compositions in
the clinopyroxenes from the XPS separately, but, from petrographic
observations, the proportion of rim composition in the clinopyroxene
population should be relatively small compared with the core
composition.
18O values of clinopyroxene from the XPS are in
the lower range of the entire dataset. Clinopyroxene from a
gabbroic xenolith (sample GR 08) is slightly, but significantly,
higher in
18O than from the syenites, with
18O = 5·1

.
In the studied suite of samples, the oxygen isotope composition of calcite ranges from 7·8 to 8·6
18O and carbon isotope composition varies between 3·9 and 4·7
13C. Our data overlap with the data from the literature. They are shown in Fig. 8, together with the field of primary, mantle-derived carbonatites as proposed by Taylor et al. (1967)
. The C and O isotopic compositions of the Grønnedal-Ika carbonates are within the upper half of the primary carbonatite field of Taylor et al. (1967)
.
Nd isotope analyses
Nd isotopic analyses of mineral separates and whole-rock samples
are presented in
Table 6.
Nd(T) values were calculated for T
= 1300 Ma based on the RbSr isochron age of the complex
(Blaxland
et al., 1978

). They range from +1·8 to +2·8
for the syenites and from +2·4 to +2·8 for the
carbonatites. The average
Nd(T) of the carbonatites (2·59
± 0·20) is slightly higher, but not significantly
different from the average of the syenites (2·36 ±
0·40). Thus, the Nd isotopic composition of the Grønnedal-Ika
complex appears to be relatively homogeneous. However,
Nd(T)
is slightly lower compared with whole-rock syenite data from
Grønnedal-Ika (
Nd = 3·23·9; Goodenough,
1997

) and a value for the Gardar mantle source (
Nd > 4) defined
by Andersen (1997)

. Plotted in a SmNd isochron diagram
(
Fig. 9), all analyses of the mafic minerals and the carbonatites
scatter closely around the 1300 Ma reference line. Clinopyroxene
from a gabbroic xenolith (sample GR 08) yielded a distinctly
lower
Nd(T) value than those typical for the syenites [
Nd(T)
= 1·7]. A whole-rock sample of relatively unaltered
basement gneiss has
Nd(T) = 13·3.

View larger version (21K):
[in this window]
[in a new window]
|
Fig. 9. SmNd isochron diagram for clinopyroxene and amphibole mineral separates from syenites and whole-rock carbonatites from Grønnedal-Ika. The 1300 Ma reference line is based on the RbSr age of 1299 ± 17 Ma (Blaxland et al., 1978 ). Errors (±0·5% for 147Sm/144Nd and ±0·002% for 143Nd/144Nd) are not larger than the symbols.
|
|
 |
CALCULATION OF PHYSICO-CHEMICAL PARAMETERS
|
|---|
Temperature, silica activity and oxygen fugacity in the nepheline syenites
The quantification of intensive crystallization parameters can
be used to distinguish separate phases in the evolution of the
complex or individual units. Furthermore, the determination
of equilibration temperatures is important for obtaining accurate
estimates of the oxygen isotope composition of the melt from
the mineral data, because mineralmelt isotopic exchange
is temperature dependent. The quantification of silica activities
and oxygen fugacities is relevant for comparison with other
Gardar complexes to evaluate the possible influence of these
parameters on the origin of the carbonatite.
Equilibria among Ne, Ab and Jd components in nepheline, alkali feldspar and clinopyroxene solid solutions, respectively, provide means to estimate temperatures and silica activities in nepheline syenites (Markl et al., 2001
). These equilibria are believed to reflect the conditions during formation of the last mineral in a specific assemblage (Markl et al., 2001
). In the absence of a reliable pressure estimate, 1 kbar was used for all calculations. This value is an estimate from the Ilímaussaq intrusion further south in the Gardar Province, based on fluid inclusion studies (Konnerup-Madsen & Rose-Hansen, 1984
). For the nepheline syenites of Grønnedal-Ika, the following three equilibria are relevant:
 | (1) |
 | (2) |
 | (3) |
Following the approach of Markl et al. (2001)
, we assume equilibration of the magmatic mineral assemblage is achieved as long as melt remained in contact with the minerals and in the absence of growth zonation (feldspars and nephelines). Zoned clinopyroxene crystals record various stages of crystallization, and the increase in Na+ and Fe3+ in clinopyroxene rims and along cracks may reflect oxidation by residual melt or fluid phases (Marks et al., 2003
). Therefore, estimates of magmatic T and aSiO2 conditions were made using clinopyroxene compositions that presumably reflect early crystallization, i.e. the most diopside-rich compositions of the matrix clinopyroxenes. Activities of jadeite in clinopyroxene (aJd) were calculated after Holland (1990)
at temperatures of 800, 900 and 1000°C. For feldspars, the complete range of compositions measured was used to determine the activity of albite (aAb) after Fuhrman & Lindsley (1988)
, again at 800, 900 and 1000°C. Nepheline activities (aNeph) were calculated using a mixing-on-sites model and do not show a large variation, either between or within samples. Therefore, an average aNeph for each sample was used. Input parameters and results are summarized in Table 7. Calculations using minimum and maximum values for aJd and aAb resulted in four positions of the same invariant point that define parallelogram-shaped fields in TaSiO2 space (Fig. 10). These fields are considered to represent the equilibration conditions of the assemblage.
View this table:
[in this window]
[in a new window]
|
Table 7: Activities used in calculations for jadeitealbitenepheline equilibria and resulting temperatures and silica activities of syenitic rocks from Grønnedal-Ika
|
|
For the LS and the PMS, equilibrium conditions vary from 680
to 850°C and
aSiO
2 values of 0·280·41
(
Table 7). Temperatures agree well with those obtained by nepheline
geothermometry after Hamilton (1961)

, which yields temperatures
between 700 and 850°C. Only a few nepheline analyses poor
in SiO
2 indicate temperatures as low as 500°C, which suggests
prolonged re-equilibration (Powell, 1978

). Calculations with
a higher crystallization pressure of 2 kbar yield slightly higher
temperatures and silica activities, but this does not affect
the general conclusions of this study. For the XPS, calculations
were performed with clinopyroxene compositions from phenocryst
cores, phenocryst rims and matrix grains, as it is not clear
from petrographic observations which clinopyroxene equilibrated
with the nepheline and feldspar. Calculations with the most
Mg-rich matrix cpx compositions indicate temperatures of 710775°C
and silica activities of 0·280·36. Equilibria
with Mg-rich rim compositions yield similar but slightly elevated
values of
aSiO
2 (0·320·43) and temperatures
of 785910°C. For both matrix and phenocryst rim compositions,
calculated values of
T and
aSiO
2 overlap with the intensive
parameters determined for the Layered Syenites. However, equilibria
based on typical phenocryst core compositions (
aJd = 0·080·12)
yield temperatures of <550°C at silica activities <0·25.
It is very unlikely that the cores grew at such low temperatures,
and most probably the clinopyroxene core compositions were not
in equilibrium with nepheline and alkali feldspar during growth.
Thus, we believe that the clinopyroxene phenocryst cores of
the XPS crystallized during a distinct earlier event in the
magmatic history, possibly as the only crystallizing phase or
at higher
P.
The abundant occurrence of cancrinite (Fig. 4f) suggests a considerable build-up in CO2 in the late stages of the solidification of the magma, as cancrinite crystallization temperatures, derived from fluid inclusion data, range from 450 to 650°C at 1·11·5 kbar (Sobolev et al., 1974
). This enrichment of CO2 might be linked to the forcible emplacement of the carbonatite into the nephelinesyenites.
Oxygen fugacities in the nepheline syenites were estimated from the compositions of reintegrated titanomagnetites, and independent temperature determinations from NeAbJd equilibria. In Fig. 11, theoretical oxygen buffer curves (FMQ, HM) are plotted together with their displaced positions appropriate to the activities determined from mineral analyses in this study (FMQ*, HM*, FsMQ*). FMQ* marks the minimum oxygen fugacity in the nepheline syenites, as this is the position of the buffer curve recalculated with mineral compositions typical for the syenites (aSiO2 = 0·37, aMag = 0·43) if olivine with aFa = 1 were present. At high temperatures, FeTi oxides equilibrate readily with a fluid phase because of their fast reaction kinetics in hydrothermal experiments (Buddington & Lindsley, 1964
), and it can be inferred that this is true for melts as well. Then, for a given temperature and magnetiteulvöspinelSS composition, the fO2 of the fluidmelt can be read from the fO2T graph (Fig. 11). Because this fluidmelt phase, in turn, determines the fO2 of the rock (Buddington & Lindsley, 1964
), we can estimate the oxygen fugacities of the nepheline syenitic melts at their equilibration temperatures of 710910°C with the typical oxide composition Mag6070Usp3020. The latter are represented as projections of tie-lines connecting coexisting magnetiteulvöspinelSS and hematiteilmeniteSS pairs after Buddington & Lindsley (1964)
and they define a dark grey shaded field in Fig. 11. At temperatures appropriate for the syenites (700910°C) fO2 falls between +2 and +5 log units above the FMQ buffer (cross-hatched pattern in Fig. 11). This result agrees well with the position of the FsMQ* (ferrosilitemagnetitequartz) buffer at very low aFs values (
0·01) that corresponds to the virtually absent ferrosilite component in the aegirineaugites. The occurrence of presumably late- to post-magmatic hematite indicates that oxygen fugacities increased to values >1 log unit above the HM buffer during the later stages of magmatic evolution (light grey arrow).
Stable isotope composition of the Grønnedal-Ika magmas
Oxygen isotopes in nepheline syenites
Oxygen isotope data for mineral separates from Grønnedal-Ika
and other Gardar igneous rocks (Halama
et al., 2003

; Marks
et al., 2003

) are plotted in
Fig. 12a versus the respective
Nd(T)
values. Compared with the Eriksfjord Formation basalts, samples
from Grønnedal-Ika syenites have slightly lower
18O values
at slightly higher values of
Nd(T). Clinopyroxene from the gabbroic
xenolith overlaps with the EF basalt samples. In contrast, most
minerals from the silica-oversaturated rocks from the Puklen
intrusion tend towards lower
18O values and strongly negative
Nd(T) values, which were interpreted to reflect the influence
of open-system late-stage magmatic processes and assimilation
of crustal material (Marks
et al., 2003

). These factors do not
seem to have affected the Grønnedal-Ika rocks. However,
not only the oxygen isotopic composition of the melt, but also
temperature and chemical composition of the minerals influence
18O
mineral values (Zheng, 1993
a
, 1993
b
). Therefore, it is useful
to calculate
18O
melt from the mineral data. The calculation
of the oxygen isotopic composition of the original melts from
mineral data is relatively straightforward for basaltic rocks
(Kalamarides, 1986

), but less so for the Grønnedal samples
because of lower equilibration temperatures and the formation
of diopsideaegirine solid solutions. We therefore used
a different approach, which is briefly described below.
Based on the nephelinealbitejadeite equilibria,
we assume that the equilibration temperature (
Teq) for all nepheline
syenitic samples is
