Journal of Petrology Volume 42 Number 12 Pages 2231-2257 2001
© Oxford University Press 2001
Phase Equilibrium Constraints on Intensive Crystallization Parameters of the Ilímaussaq Complex, South Greenland
GREGOR MARKL,*,
MICHAEL MARKS,
GREGOR SCHWINN and
HOLGER SOMMER
INSTITUT FÜR MINERALOGIE, PETROLOGIE UND GEOCHEMIE, EBERHARD-KARLS-UNIVERSITÄT, WILHELMSTRASSE 56, D-72074 TÜBINGEN, GERMANY
Received
June 26, 2000;
Revised typescript accepted
May 31, 2001
 |
ABSTRACT
|
|---|
The 1·13 Ga Ilímaussaq intrusive complex, South
Greenland, is composed of various types of alkali granite and
silica-undersaturated alkaline to agpaitic nepheline syenites
related to three subsequently intruded magma batches. Mineral
chemistry indicates continuous fractionation trends within each
rock type, but with distinct differences among them. The last,
peralkaline magma batch is the most fractionated in terms of
X
Femafic mineral, feldspar composition and mineral assemblage.
This indicates that an evolving magma chamber at depth discontinuously
released more highly fractionated alkaline melts. Fluid inclusions
in some sodalites record a pressure drop from 3·5 to
1 kbar indicating that crystallization started during magma
ascent and continued in the high-level magma chamber. On the
basis of phase equilibria and preliminary fluid inclusion data,
crystallization temperature drops from >1000°C (augite
syenite liquidus) to <500°C (lujavrite solidus) and silica
activity decreases from
0·8 to <0·3. An almost
pure methane fluid phase at high temperatures and an almost
pure aqueous fluid phase in the last crystallization stages
of the agpaitic rocks indicate a strong increase in water activity.
NaCl activity drops from 0·4 during magmatic sodalite
crystallization to <0·01 (3 wt % NaCl
equiv) in the
late magmatic aqueous fluids. Relative oxygen fugacity [
FMQ,
where FMQ is fayalite–magnetite–quartz)] depends
on silica and water activity via two solid–solid buffer
reactions. It decreases during fractionation in the augite syenite
from about FMQ – 1 to below FMQ – 4, but increases
in the peralkaline stage. The extreme peralkaline fractionation
trend appears to be governed by low water activity and low SiO
2 activity in the parental melt. Only then is methane a stable
fluid phase during most of the crystallization history, which
prevents early unmixing of an aqueous NaCl-bearing fluid phase.
KEY WORDS: Ilímaussaq; agpaite; intensive parameters; fractionation; peralkaline
|
INTRODUCTION
|
|---|
The Ilímaussaq intrusive complex in the mid-Proterozoic
Gardar rift province of South Greenland is for various reasons
one of the most famous alkaline plutonic complexes in the world:
- it is the type locality of agpaitic rocks, i.e. of rocks with a molar ratio of (Na + K)/Al
1·2 (Ussing, 1912
) and with complex Na–Ti–Zr silicates (Sørensen, 1997);
- it is a world-famous mineral locality with >200 species and is the type locality for sodalite, arfvedsonite, eudialyte and many other minerals.
- the kakortokite rock type shows one of the most spectacular examples of magmatic layering.
The field relations, whole-rock geochemistry, systematic mineralogy and economic potential for various metals such as Zr, Be and U (see review by Bailey et al., 1981; Sørensen, 2001) have been discussed in more than 100 contributions. Ussing (1912), Ferguson (1964), Engell (1973), Larsen (1976, 1977, 1981), Larsen & Sørensen (1987) and Sørensen & Larsen (1987) have presented a petrological framework based on field work, and on whole-rock geochemical and some mineral compositional data, which allows a qualitative understanding of the magmatic evolution of the Ilímaussaq complex. As most of the Ilímaussaq rocks are cumulates, whole-rock geochemistry has its limitations. The only quantitative work using silicate phase compositions that also covers the agpaitic rocks is by Larsen (1976, 1977), which provides important and detailed constraints on pyroxene and aenigmatite compositions, as well as on crystallization temperature, and which allows comparison with the work by, for example, Parsons (1972, 1981), Stephenson (1976), Powell (1978), Stephenson & Upton (1982) and Upton et al. (1985) on different, mostly saturated to slightly undersaturated, but not agpaitic Gardar rocks. Karup-Møller (1978) presented a thorough review of the temperature, fS2 and fO2 conditions of formation of ore minerals found in Ilímaussaq, which pertain also to their host rocks. Fluid inclusion studies by Petersilie & Sørensen (1970), Konnerup-Madsen et al. (1979, 1985, 1988) and Konnerup-Madsen & Rose-Hansen (1984) have mainly focused on the discovery of more complex hydrocarbons in agpaitic fluid inclusions, although aqueous, NaCl-bearing fluid inclusions have also been described. Such aqueous fluid inclusions from the alkali granite (not dealt with here) in conjunction with estimates of the former overburden (Poulsen, 1964) have been used to infer a pressure of intrusion of 1 kbar (Konnerup-Madsen & Rose-Hansen, 1984). Marks & Markl (2001) presented new data in a detailed phase equilibrium study on the augite syenite portion of the complex, which will be augmented by data for the agpaitic rocks from the present paper to allow derivation of quantitative petrological models.
Ilímaussaq is one of the best-known peralkaline complexes and may serve as a type example to understand their petrogenesis. In agpaitic rocks, both whole-rock and mineral ratios of Mg/Fe, Ca/(Na + K) and K/Na approach nil, and they are characterized by extreme enrichment of elements such as Na, Zr, Cl and F, and in normally low-abundance elements such as Li, Be, Rb, Ga, REE, Nb, Ta, Hf, Zn, Sn, U and Th (Sørensen, 1997). On the basis of these and many other features, these intrusions are among the most differentiated of all magmatic rocks. Generally, peralkaline, agpaitic rocks are thought to be derived from highly fractionated alkali basaltic or nephelinitic melts (Sørensen, 1997), but in detail, their derivation, their fractionation trend and tectonic or geochemical prerequisites for their formation are still enigmatic. Because many peralkaline rocks show distinct enrichment in F and Cl, it has been argued that volatiles are responsible for the special fractionation trend that finally leads to agpaitic rocks (Gerassimovsky & Kuznetsova, 1967; Kogarko, 1974; Kogarko & Romanchev, 1977, 1982). However, in principle, pressure, temperature, the activities of silica, water and NaCl, oxygen fugacity, cooling rate or the tectonic setting could be equally responsible for the unusual agpaitic fractionation trend. As we still lack adequate quantitative estimates of intrinsic parameters for agpaitic melts, we lack the tools to fingerprint the influence of the various parameters on the crystallization behaviour of these melts. It is the aim of this study to present new phase compositional data, to discuss them quantitatively in the context of the published petrogenetic model for the Ilímaussaq intrusion (Larsen & Sørensen, 1987) and to evaluate the importance of the various parameters during the formation of agpaitic melts in general.
|
GEOLOGICAL SETTING
|
|---|
The Ilímaussaq intrusion is part of the 1·1–1·3
Ga Gardar failed rifting province of South Greenland (Upton
& Emeleus, 1987
; Fig.
1). During this period, 10 major plutonic
complexes of gabbroic and nepheline-bearing to quartz-saturated
granitoid rocks intruded a basement consisting of early Proterozoic
granites and gneisses (Emeleus & Upton, 1976
), which was
unconformably overlain by early Gardar basalts and sandstones
of the Eriksfjord formation (Poulsen, 1964
). The two intrusive
complexes of Grønnedal-Ika (Emeleus, 1964
) and Igaliko
(Upton & Fitton, 1985
) contain carbonatites, and many gabbroic
dykes all over the Gardar province (some thousands of square
kilometres) contain up to metre-scale anorthosite xenoliths
(Bridgwater, 1967
; Bridgwater & Harry, 1968
). The association
of basalts, gabbroic intrusions, anorthosite xenoliths and carbonatites
with the penecontemporaneous granitoid intrusive rocks has been
interpreted to reflect large-scale melting processes of asthenospheric
mantle. These melts (alkali basalts) were probably ponded at
the crust–mantle boundary and by fractionation gave rise
to massive-type anorthosites and later to alkaline or peralkaline
melts in the roof region of these magma chambers (see also Larsen
& Sørensen, 1987
).
According to Larsen & Sørensen (1987), the Ilímaussaq igneous complex was formed by successive intrusion of three melt batches at 3–4 km depth (1 kbar, Konnerup-Madsen & Rose-Hansen, 1984). The earliest of these three stages (stage I) is the augite syenite. It occurs both as thin shell with a thickness of a few hundred metres along the western, southern and southeastern margins of the intrusion and as a lid 150 m thick on the stratigraphic top of the intrusion (Fig. 1). The petrology and mineral chemistry of this early rock type has been discussed in detail by Larsen (1976, 1981) and by Marks & Markl (2001). The augite syenite lid was later intruded by an alkali granite (stage II), which will not be discussed further in this paper.
In stage III, nepheline- and sodalite-bearing syenites formed. Minor volumes of pulaskite, foyaite and the so-called ‘sodalite foyaite’ are interpreted to have crystallized in situ from the roof downwards, whereas the ‘naujaite’ represents a flotation cumulate of sodalite (Ussing, 1912; Ferguson, 1964). Mafic cumulates corresponding to this flotation cumulate are not exposed, but their existence is indicated by a strongly positive gravity anomaly beneath Ilímaussaq (Blundell, 1978; Forsberg & Rasmussen, 1978). At the top of the intrusion, the contacts between augite syenite, sodalite foyaite and naujaite appear somewhat gradational, whereas at other places, sharp cross-cutting relationships clearly demonstrate that the augite syenite is the oldest part of the intrusion. Chemically and texturally differing varieties of agpaitic nepheline syenites formed below the above-named roof rocks. The layered kakortokites crystallized at the lowermost visible stratigraphic position of the complex. The residual liquids from kakortokite crystallization formed a sequence of lujavrites, the largest masses of which occur between kakortokite and naujaite (Ferguson, 1964). Kakortokite contains xenoliths of all other rock types except for lujavrite, which also occurs as veins and irregular masses cutting through all other rock types. The rocks record extremely low viscosity of the lujavrite melt as evidenced by fluidal textures and by extreme variations in vein thickness and xenolith content over short distances.
Geochemically, typical monitors of fractionation such as Fe/Mg, Ca/(Na + K) or contents of incompatible elements indicate that later syenitic rock types in Ilímaussaq generally display higher degrees of fractionation than their precursors. The kakortokite as the only exception from this rule shows a lower Fe/Mg ratio, but on the other hand a more evolved mineral assemblage than its precursors. Eudialyte as a liquidus phase indicates that the lower Fe/Mg ratio is possibly related to contamination and does not disprove the general trend. The K/Rb ratios continuously decrease during fractionation from 500 in the augite syenite to 35 in the most evolved lujavrites, whereas Zr/Hf and Cl/Br continuously rise from 45 to 97 and from 170 to 15 000, respectively (Bailey et al., 2001). Similarly continuous trends are also exhibited by K/Cs, Sr/Ca, Li/Mg, Zn/Fe, Cs and Ga. Total rare earth element (REE) contents as well as LREE/HREE (light REE/heavy REE) ratios show a general increase with fractionation (Bailey et al., 2001). In contrast to these, more or less continously evolving fractionation indicators, elements such as Zr, Cl, F, Zn and W, show distinctive irregularities or maxima in some rock types, which are related to cumulus processes (Bailey et al., 2001). A good example for this is the unusually high content of Cl in naujaite (up to 2·9 wt %) compared with the earlier (<0·2 wt %) and later (maximum 0·3 wt %) Ilímaussaq rocks.
Contamination by basalt and quartz-rich roof and sidewall rocks that occur as xenoliths in the Ilímaussaq intrusive rocks was locally shown by Ferguson (1964). On the basis of Sm–Nd data and assimilation–fractional crystallization (AFC) calculations, Stevenson et al. (1997) showed that the earlier Ilímaussaq rocks (augite syenite and especially the alkali granite) reflect more crustal contamination than the later agpaitic rocks.
|
PETROGRAPHY AND MINERAL CHEMISTRY
|
|---|
Petrography and mineral assemblages
Below, we briefly describe the rock types dealt with in the
present paper. The mineral assemblages are listed in Table
1,
and mineral abbreviations in Table
2. Detailed descriptions
have been provided in many publications (e.g. Ferguson, 1964
).
The early magmatic and intercumulus assemblage is in some cases
partially overprinted by a later magmatic assemblage and in
most places by a post-magmatic or hydrothermal assemblage whose
development is still related to the same magmatic event but
occurs at temperatures below the solidus.
Augite syenite
The augite syenite shows a xenomorphic texture with grain size varying between 2 and 20 mm. The main liquidus mineral phases are strongly exsolved perthitic alkali feldspar, olivine, clinopyroxene and Fe–Ti oxides (Fig. 2a). In some samples, olivine is resorbed and overgrown by clinopyroxene. Most samples contain interstitial nepheline, one is quartz bearing, and some contain neither nepheline nor quartz. Biotite and amphibole may form rims around olivine, clinopyroxene or Fe–Ti oxides and are interpreted to be of near-solidus (or even subsolidus?) origin. Amphibole in most samples varies from ferropargasite to hastingsite and ferroedenite in composition. Nepheline is replaced by subsolidus assemblages consisting of analcime, sodalite, albite, muscovite and hydrogrossular (Fig. 2c and d).
View larger version (229K):
[in this window]
[in a new window]
|
Fig. 2. Observed mineral textures in the various rock types. (a) Granular xenomorphic texture in augite syenite with olivine, clinopyroxene, Fe–Ti oxides and apatite in a matrix of alkali feldspar. (b) Euhedral sodalite crystals in exsolved alkali feldspar in naujaite. (Note the small albite inclusion.) (c) and (d) in some samples of augite syenite, nepheline is replaced by subsolidus assemblages with analcime, sodalite, albite, hydrogrossular and muscovite. (e) Olivine and clinopyroxene as inclusions in alkali feldspar from sodalite foyaite. (Note the later interstitial amphiboles.) (f) Fluidal texture in lujavrite. Large and euhedral nepheline is enclosed in a mixture of albite, microcline, aegirine pyroxene and amphibole, which are strongly aligned. (g) Black kakortokite with amphibole and feldspar as cumulus phases and nepheline as a later interstitial mineral. (h) Euhedral eudialyte and feldspar crystals together with interstitial amphibole in red kakortokite. (i) Core of augitic pyroxene in sodalite foyaite, which is resorbed and overgrown by fluorite, aegirine pyroxene and arfvedsonite. (j) Amphibole rimmed by aegirine (sodalite foyaite). (k) Symplectic intergrowth of sodalite and analcime replacing primary sodalite and nepheline in sodalite foyaite.
|
|
Sodalite foyaite
This rock type is typically medium to coarse grained with grain sizes up to 20 mm. Early magmatic phases are exsolved euhedral perthitic alkali feldpar, nepheline, sodalite, olivine and resorbed relics of augite (Fig. 2e and i). Subsequent sector-zoned Na-rich clinopyroxene is followed by aenigmatite, fluorite, rare eudialyte and zoned ferrorichteritic, katophoritic or nyböitic amphibole, which again is partly rimmed by almost pure aegirine (Fig. 2j). In rare cases biotite overgrows early olivine. Analcime appears to occur as a late liquidus phase, but most analcime forms together with secondary sodalite by subsolidus replacement of primary sodalite and nepheline (see Fig. 2k).
Naujaite
Euhedral sodalite and nepheline are cumulate phases in this rock type (Fig. 2b). Their grain size reaches 1 cm. Magmatic fluorite occurs as inclusions in sodalite. Alkali feldspar occurs as large tabular crystals with frequent inclusions of sodalite and is interpreted to have crystallized as the first of the interstitial minerals, which additionally comprise aegirine, arfvedsonitic to katophoritic amphibole, eudialyte and again nepheline. Subsolidus natrolite and analcime replace sodalite and feldspar, and pectolite replaces pyroxene.
Kakortokite
The kakortokite occurs as a white, a red and a black variety. The differences are caused by variations of the modal content of arfvedsonitic amphibole (black), eudialyte (red) and feldspar (white). The white and red kakortokite contain eudialyte and feldspar as early, mostly euhedral liquidus phases (Fig. 2h), whereas the black variety shows amphibole and in places pyroxene as early phases (Fig. 2g). The black and white varieties show some gravitative alignment of the tabular minerals (feldspar, amphibole). Interstitial minerals in all three varieties are sodalite and fluorite, whereas nepheline, amphibole (katophoritic to arfvedsonitic) and pyroxene are late magmatic phases depending on the type of kakortokite. The white kakortokite bears interstitial analcime, and small prismatic albite crystals in addition to aegirine pyroxene and analcime formed during hydrothermal replacement of nepheline, sodalite and early alkali feldspar.
Lujavrite
Many varieties of this rock type show flow textures. Euhedral nepheline, eudialyte, sodalite and clinopyroxene are enclosed in a mixture of albite, microcline, texturally later aegirine pyroxene, and amphibole, which are mostly aligned (Fig. 2f). Black lujavrite is distinguished from a green variety depending on the modal amount of either arfvedsonite (black) or aegirine (green). In some of the samples, rare minerals such as ussingite (NaAlSi3O8.NaOH), naujakasite, steenstrupine, villiaumite occur, in some special varieties even as major constituents. Some of the so-called lujavrites are in fact fenitized host rocks [such as a murmanite-bearing variety described by, for example, Ferguson (1964)], and these are not dealt with in the present paper. Rarely, the lujavritic melt developed silicate–silicate immiscibility features described in detail elsewhere (Markl, 2001).
Mineral chemistry
Analytical techniques
Microprobe wavelength-dispersive spectometry (WDX) analyses were performed on a Cameca SX100 system at the University of Freiburg, Germany. Natural standards supplied by Cameca were used for calibration. Analytical conditions were 15 kV and 20 nA, with counting times of 20 s for all elements on the peak and 10 s on the background. Raw data were processed by procedures described by Pouchou & Pichoir (1984). Uncertainties for major elements are estimated to be about ±1% (relative), for minor elements about ±5% (relative), and detection limits are 0·01 wt % depending on the specific element. The composition of exsolved microperthitic feldspar was reintegrated using the scanning mode and a window of about 20 µm x 30 µm size in cases where the exsolutions were of the order of a few microns only. For coarsely exsolved feldspars, point analyses in conjunction with image analysis techniques were used. Electron microprobe analyses of minerals used in the calculations below are reported in Tables 3–6. Data for Fe–Ti oxides have been reported in detail by Marks & Markl (2001) and are therefore not repeated here. The mineral chemistry of micas and amphiboles will be the topic of a separate paper and will therefore not be discussed here.
Olivine
Olivine occurs in the augite syenite and rarely in the sodalite foyaite. In both rock types, olivine is unzoned. Some representative analyses are shown in Table 3. Composition ranges from Fa71Tp2 to Fa94Tp4 in the augite syenite and shows a good correlation between Fe and Mn contents (Fig. 3c). In contrast, the sodalite foyaite olivine composition is almost constant between Fa94Tp6 and Fa97Tp3 (Fig. 3a). CaO in both rock types varies between 0·1 and 1·8 wt %. In the sodalite foyaite, Ca shows a strong correlation with Fa content, whereas in the augite syenite it shows wide scatter without any visible systematics (Fig. 3b). Ca-rich exsolution lamellae consisting of clinopyroxene–magnetite intergrowths in some olivines from augite syenite have been described by Markl et al. (2001).
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 3. (a) Fo–Fa–Tp diagram showing the Ilímaussaq olivine compositions. (b) and (c) correlation diagrams for Ca and Mn with Fa content. Here and in the following diagrams, p.f.u. stands for per formula unit.
|
|
Pyroxene
Larsen (1976) presented electron microprobe data of clinopyroxenes from various Ilímaussaq rock types which are now augmented by our new data (Table 4). Clinopyroxene in augite syenite is a chemically zoned subcalcic augite (Marks & Markl, 2001) with Quad components typically >90% (Fig. 4). The range of XFe (after Lindsley, 1983) in all samples is between 0·34 and 0·87. Fs content increases towards the rim of a single crystal, whereas Wo component stays approximately constant between 42 and 49 mol %, depending on the sample. Acmite contents range from 2 to 15 mol %, and only in one sample (GM1223) acmite content reaches 27 mol % (Fig. 4a).
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 4. (a) Pyroxene composition for the various rock types in the triangle acmite–jadeite–Quad. (b) Correlation diagram between Na and Fe3+ in Ilímaussaq pyroxenes.
|
|
In all other rock types, pyroxene Na and Fe3+ contents are significantly higher than in augite syenite whereas Ca and Fe2+ contents decrease strongly (Fig. 4a and b). Cores of resorbed pyroxene crystals in sodalite foyaite have augitic compositions very similar to those from the augite syenite (Fig. 4a and b). During reaction with the sodalite foyaitic melt, an aegirine pyroxene and fluorite replace the original augite (Fig. 2i). Later pyroxenes in sodalite foyaite are sector-zoned aegirine augites (Shearer & Larsen, 1994) with acmite contents up to 90 mol %. As shown in Fig. 5, zonation is mainly caused by discontinuous decrease of Fe2+, Ca, Mn and Zr, whereas Na and Fe3+ increase. Aegirine rims around amphibole (Fig. 2j) are relatively rich in jadeite component (up to 8 mol %).
View larger version (62K):
[in this window]
[in a new window]
|
Fig. 5. Chemical variation along a profile in a sector-zoned clinopyroxene crystal from sodalite foyaite.
|
|
In naujaite, the pyroxene composition is almost pure aegirine with <8 mol % of Quad and <6 mol % of jadeite component. In general, the acmite component increases towards the rim of a crystal.
Pyroxenes in all types of kakortokite show the same zonation and the same exchange of Quad against acmite component as those in the sodalite foyaite, but acmite content reaches only 75 mol %. Besides their higher Quad component, they have also a lower Fe/Mg ratio than naujaite and sodalite foyaite pyroxenes.
The clinopyroxenes in lujavrites have acmite contents >80 mol %. Four types occurring in different samples can be distinguished both texturally and chemically (Fig. 6a–d). Early, relatively large tabular crystals have the lowest jadeite contents (5 mol %). Partially aligned prismatic crystals, fully aligned prismatic crystals and anhedral aggregates together with ussingite have successively higher jadeite contents up to 19 mol % (Fig. 4a).
View larger version (217K):
[in this window]
[in a new window]
|
Fig. 6. (a)–(d) Textures and corresponding phase diagrams in lujavrites relating various textures to various conditions of formation. (See text for discussion.)
|
|
The Zr contents in the various Ilímaussaq pyroxenes change significantly (Larsen, 1976) and systematically, and are comparable with those of the pyroxenes described by Jones & Peckett (1980) from the Igaliko complex in the Gardar province. As shown in Fig. 7, Zr in augite syenite is low and increases towards the outermost rims of the crystals (Fig. 7a) up to 0·9 wt % ZrO2, and in sample GM 1223 even up to 1·5 wt % (Fig. 7b). In sodalite foyaite (Fig. 7c), pyroxenes have constant Zr contents that are higher than in the naujaite pyroxenes (Fig. 7d), where Zr is also almost constant. Absolute amounts reach 1·1 and 0·6 wt % ZrO2, respectively. Zr in kakortokites (Fig. 7e) is comparable with that in the sodalite foyaite (1·3 wt % ZrO2), although the contents decrease slightly towards the rim. In lujavrite pyroxenes, Zr contents are generally low (<0·5 wt % ZrO2) and constant, but some pyroxenes with extreme Zr-rich cores (up to 2 wt % ZrO2) were observed (Fig. 7f). This change in zoning textures is interpreted to record the early increase and later decrease of Zr content in the melt. The decrease is mainly governed by the crystallization of the Zr silicate eudialyte and may monitor its appearance on the liquidus. A plot of Ca vs Zr (Fig. 8) reveals this systematic, but discontinuous change of Zr content with rock type and Ca content. The discontinuity may be related to the geochemical discontinuity of the three magma batches or to the lack of samples from the pulaskite and foyaite unit and from the unexposed part below the kakortokite.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 7. Zr contents of zoned pyroxene crystals in the various rock types: (a) and (b) augite syenite; (c) sodalite foyaite; (d) naujaite; (e) kakortokite; (f) lujavrite. Profile widths are 2·5 mm, 2·0 mm, 1·5 mm, 2·0 mm, 0·6 mm and 0·1 mm, respectively.
|
|
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 8. Ca vs Zr in Ilímaussaq pyroxenes. (Note the early increase and later decrease of Zr during fractionation.) The black arrow shows the fractionation trend. Also shown are published pyroxene analyses from Ilímaussaq according to Larsen (1976).
|
|
Alkali feldspar
In the augite syenite, sodalite foyaite, naujaite and kakortokite, alkali feldspar shows perthitic exsolution of albite in K-feldspar (Fig. 9a and b). Recalculated magmatic compositions in sodalite foyaite (Ab28–42An0–1) and kakortokite (Ab30–45An0–1) are very similar and Ca free, whereas the augite syenite (Ab40An6Or54–Ab55An12Or33) shows strongly ternary compositions (see Fig. 9d and Table 5). Texturally late, but euhedral feldspar crystals in kakortokite are pure albite. Textures indicate that the lujavrites crystallized two separate feldspars of almost albite and K-feldspar end-member composition (Fig. 9c).
Nepheline
Nepheline compositions range from Ne67Ks17Qtz16 to Ne74Ks11Qtz15 in the augite syenite (Table 5). Nephelines in sodalite foyaite, naujaite and kakortokite show a wide range in SiO2 contents. They are higher in Ks component than those in the other rocks (Fig. 10). Nepheline in lujavrite is significantly poorer in SiO2. The CaO content in all rock types is invariably <1 wt %. The spread in excess silica (Fig. 10) is related to patchy compositional differences within single grains and between grains, but not to a continuous growth zonation. We interpret the patchy variations as retrograde effects related to microfractures and hydrothermal fluids.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 10. Ne–Ks–SiO2 triangle showing the nepheline compositions from Ilímaussaq rocks. Liquidus lines are after Hamilton (1961).
|
|
Sodalite
Primary sodalite in sodalite foyaite, naujaite and kakortokite, and post-magmatic sodalite observed in some augite syenite samples (Fig. 2b–d) are almost pure sodalite end members with maximum SO3 contents of 0·5 wt % (see Fig. 11 and Table 6). Subsolidus sodalite in symplectites with analcime (Fig. 2k) in sodalite foyaite and kakortokite have SO3 contents up to 1·3 wt %. Sodalite in the various lujavrite samples covers a wide continuous compositional range with SO3 contents between 0·9 and 3·8 wt % (solid solutions with nosean). The villiaumite (NaF)-bearing sample GM1395 shows the largest variations and the highest SO3 contents.
Fluid inclusions
Fluid inclusion studies by Petersilie & Sørensen (1970) and Konnerup-Madsen et al. (1979, 1985, 1988) have shown that methane is the most important fluid phase in equilibrium with the various peralkaline melts. Higher hydrocarbons occur in small percentages, and water is almost completely absent. According to those workers, the alkali granite contains abundant primary aqueous, highly saline fluid inclusions, whereas aqueous inclusions of variable salinity are extremely rare in most of the agpaitic rocks. Preliminary results of Schwinn (1999) and Sommer (1999) indicate that the primary inclusions in naujaite, kakortokite and some lujavrite samples are almost pure methane hydrocarbon inclusions, whereas some lujavrites show primary aqueous, low-salinity (3 wt % NaCl equivalent) fluid inclusions. Secondary inclusions in the methane-bearing rocks consist of methane–water mixtures, of almost pure methane or of low-salinity aqueous fluids, whereas they comprise only aqueous, low-salinity inclusions in the investigated lujavrite samples.
|
PHASE EQUILIBRIA
|
|---|
In the different rock types, various mineral assemblages provide
constraints on intensive parameters. Phase equilibria involving
olivine, clinopyroxene, quartz, spinel
sensu lato and ilmenite
in the augite syenite allow the most detailed estimates concerning
temperature, silica activity and oxygen fugacity (see Larsen,
1976
, 1977
; Marks & Markl, 2001
). Additional information
on this rock type and most information on the other rock types
are related to equilibria in the Na–Al–Si–O–H–Cl
system involving albite, jadeite, nepheline, sodalite, analcime,
NaCl, SiO
2 and H
2O. The equilibria depend on pressure, temperature
and some of them also on SiO
2, H
2O and/or NaCl activity. All
five phases in equilibrium principally allow the calculation
of four of these variables. However, as the early magmatic assemblage
never involves analcime, only three parameters can be calculated
for the early magmatic stages. As discussed below, pressure
was fixed at 1 kbar, and hence, temperature, silica activity
and NaCl activity could be calculated using the reactions
and
For the calculation of end member component activities from mineral formulae, the solution models of Fuhrman & Lindsley (1988) for feldspar, Holland (1990) for clinopyroxene and Ghiorso (http://melts.geology.washington.edu) for nepheline were used. Sodalite was regarded as pure Cl end member in accordance with most of the non-lujavrite microprobe analyses. In the absence of growth zonation, we assume equilibration of the magmatic mineral assemblage as long as melt is in contact with the minerals. Accordingly, zoned clinopyroxene crystals record various stages of crystallization. For the calculation of the early magmatic crystallization conditions, we used clinopyroxene core compositions and the nepheline compositions with the highest amounts of excess silica [i.e. recording the highest temperatures (Hamilton, 1964), see Fig. 10].
For the late-magmatic assemblages, analcime-involving equilibria such as
or
were
used. In these calculations analcime was treated as a pure phase.
SiO
2 unit activity was referred to a standard state of a pure
SiO
2 modification at
P and
T; H
2O and NaCl unit activities refer
to pure water and halite at
P and
T.
In principle, clinopyroxene–olivine–Fe–Ti oxide equilibria and equilibria involving Na–Al silicates provide information on solidus and liquidus conditions, respectively. The former depend on fast-diffusing cations such as Fe and Mg and involve minerals known to re-equilibrate and exsolve during cooling. The latter, in contrast, can be assumed to behave as a closed system after their formation—even in the presence of a melt—as re-equilibration would involve coupled substitutions and reconfiguration of the crystal structure after re-equilibration. This is unlikely to happen and therefore these equilibria are believed to be robust and to reflect early magmatic conditions or, more precisely, conditions during formation of the last mineral in a specific assemblage, whereas the Fe–Mg-involving equilibria probably reflect last mutual contact with a melt and hence approach solidus conditions.
In some augite syenite samples, the primary magmatic nepheline is replaced by a mixture of hydrogrossular, analcime, sodalite and muscovite that is in mutual contact with both K-feldspar and plagioclase. In conjunction with the sodalite–nepheline–NaCl reaction (4), the reactions
and
can be used to estimate temperature and minimum
NaCl activity during this post-magmatic event. K-feldspar and
plagioclase activities were calculated after Fuhrman & Lindsley
(1988)
; end-member grossular was used in the calculations as
a proxy for the almost pure hydrogrossular end member found
in the samples, muscovite activity was calculated using an ideal
site mixing model; sodalite and analcime were regarded having
unit activities and a range of nepheline activities reflecting
the measured range of nepheline compositions were used in the
calculations.
Oxygen fugacity in the augite syenite was estimated by Larsen (1976) and by Marks & Markl (2001) based on Fe–Ti oxide stability. In the agpaitic rocks, the reaction
can, in principle, be used to estimate oxygen fugacity
in rocks with an amphibole–clinopyroxene–aenigmatite
assemblage. No thermodynamic data exist for these phases, but
the experiments of Ernst (1962)
allow an approximation of the
above equilibrium in the Mg- and Al-free system. Ernst’s
results were obtained in a quartz- and water-saturated system,
but we corrected them for reduced SiO
2 and H
2O activities by
graphically estimating a log
K value from his figures.
Equilibria involving olivine, clinopyroxene and Fe–Ti oxides were calculated using QUILF (Andersen et al., 1993), whereas all other calculations used the GEOCALC software of Berman et al. (1987) and Lieberman & Petrakakis (1990) with the database of Berman (1988). Thermodynamic data for sodalite, NaCl, analcime and nepheline, however, were inserted into this database, and were taken from Sharp et al. (1989; nepheline, NaCl and sodalite) and from the SUPCRT92 database (Johnson et al., 1992; analcime). Where feasible, only data from one source were used in one set of calculations to assure internal consistency.
|
CHANGES IN THE INTENSIVE PARAMETERS DURING CRYSTALLIZATION
|
|---|
Pressure
Pressure was set at 1 kbar in all calculations, for the following
reasons:
- fluid inclusions in the alkali granite (Konnerup-Madsen & Rose-Hansen, 1984) and in the lujavrite (Sommer, 1999) are consistent with a pressure of formation of 1 kbar;
- the overburden of the Eriksfjord formation on top of Ilímaussaq was estimated to 3–4 km, corresponding again to a pressure of 1 kbar within the intrusion (Konnerup-Madsen & Rose-Hansen, 1984);
- the various melts are assumed to have crystallized and/or equilibrated (if they brought phenocrysts with them) all at the same level.
It is only in some naujaite samples that complications arise. As shown in Fig. 12, primary fluid inclusions in sodalite crystals in various naujaite samples crystallized at magmatic temperatures between 700 and 900°C at pressures between 1·5 and 3·5 kbar. Secondary methane-bearing inclusions along cracks indicate pressures of 1–1·5 kbar at temperatures of 500–700°C. The secondary inclusions are interpreted to reflect cracking of the phenocrysts during final emplacement. Aqueous inclusions in naujaite indicate 1 kbar at 400°C (for derivation of the various temperature estimates see below). We interpret this to result from sodalite crystallization within a rising magma emplaced and solidified at 1 kbar. This interpretation is consistent with the textural interpretation that sodalite is an early liquidus phase and that the naujaite is a sodalite flotation cumulate (Ferguson, 1964).
Temperature and silica activity
Information on temperature and silica activity of most Ilímaussaq rock types is compiled in Figs
9, 10, 13 and 14. Nepheline liquidus thermometry after Hamilton (1964) reveals significant differences among the various sample groups, although the spread in calculated temperatures is large (Fig. 10). Some analyses from the augite syenite plot on the 1068°C isotherm whereas the nepheline analyses lowest in SiO2 indicate temperatures as low as 500°C. The highest temperatures recorded by nepheline in a single rock type are regarded as liquidus or at least closest to liquidus temperatures and are broadly consistent with other temperature estimates. Accordingly, nepheline compositions indicate liquidus temperatures in excess of 1000°C in the nepheline-bearing portions of the augite syenite, slightly lower temperatures as high as 950°C in sodalite foyaite, kakortokite and naujaite, and liquidus temperatures of 750°C in lujavrite.
View larger version (13K):
[in this window]
[in a new window]
|
Fig. 13. Temperature–silica activity diagram showing the estimates for all rock types derived from QUILF calculations (Andersen et al., 1993) as well as from Ab–Jd–Ne equilibria.
|
|
View larger version (34K):
[in this window]
[in a new window]
|
Fig. 14. Temperature–silica activity diagrams showing equilibria among Ab, Ne and Jd for the agpaitic rocks of Ilímaussaq, except for the lujavrites, which are shown in Fig. 6. The activity of Ne was calculated from the Ne analysis corresponding to the highest temperature (see Fig. 10), the activity of jadeite in (a), (c) and (d) was calculated from the Jd-richest analysis of texturally early clinopyroxene, in (b) of texturally late aegirine, and the two activities of Ab correspond to the highest and lowest Ab contents in alkali feldspars of the respective rock type.
|
|
On the basis of feldspar compositions (Fig. 9), liquidus temperatures of the augite syenite were in excess of 950°C, whereas solidus temperatures deduced from mafic mineral equilibria reached as low as 600°C. Silica activity in these rocks decreases during fractional crystallization mainly of alkali feldspars from values of 0·8 or even unity (in rare cases where presumably sandstone contamination produced a quartz-saturated assemblage) to values of 0·4. These estimates were derived from QUILF calculations (Andersen et al., 1993) as well as from Ab–Jd–Ne equilibria. They are reported in detail by Marks & Markl (2001) and compiled in Fig. 13.
Equilibria among Ne, Ab and Jd (components in early magmatic nepheline solid solution, alkali feldspar and clinopyroxene) provide the only means to estimate temperatures or silica activities in the sodalite foyaite, naujaite and kakortokite (Fig. 14a–d). Equilibration temperatures between 800 and 700°C occur at SiO2 activities between 0·3 and 0·5. Late-stage aegirine rims around magmatic amphiboles (Fig. 14b) indicate temperatures of 550°C at silica activities around 0·25.
In the lujavrites, the texturally different types of clinopyroxene (Fig. 6a–d) record temperatures between 800°C and 350°C. Pyroxenes indicating temperatures of 450–500°C are found in textures that reflect solidus conditions (Fig. 6c), whereas the lowest temperatures are derived from pyroxenes that have probably grown during hydrothermal alteration (Fig. 6d). Silica activity drops from values of 0·6 at 800°C to 0·15 at 350°C; these values reflect magmatic and hydrothermal conditions, respectively.
The hydrothermal sodalite–hydrogrossular–analcime assemblage in some augite syenite samples (Fig. 2c and 2d) records 400°C at a fixed silica activity of 0·1 if calculated with the whole variation of mineral chemistry observed (Fig. 15a). In summary, the Ilímaussaq nepheline syenites record a crystallization interval between 1000 and 450–500°C. During fractional crystallization and subsequent hydrothermal activity, the silica activity continuously decreases from 0·8 to 0·1.
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 15. Temperature–log aNaCl diagrams showing phase equilibria used to estimate NaCl activity in augite syenite (a) and the agpaitic rocks (b and c). In augite syenite (a), the range of measured mineral compositions was used, resulting in a broad band rather than a single reaction line.
|
|
NaCl and water activity
NaCl activity can be calculated from both phase equilibria and fluid inclusion freezing point depression data. Sodalite–nepheline equilibrium buffers NaCl activity in almost the whole course of agpaite crystallization (sodalite is absent in relatively few kakortokite and lujavrite samples only). On the basis of this equilibrium, NaCl activity drops strongly during fractionation. For early magmatic temperatures of 800°C, phase equilibria among nepheline and sodalite reveal NaCl activities of 0·3–0·4, whereas NaCl activities between 0·01 and 0·1 are indicated for late-magmatic temperatures in all agpaitic rocks (Fig. 15b and c). In support of this, primary and secondary aqueous inclusions in lujavrites as detailed above show salinities in the 3 wt % NaCl equivalent range corresponding to NaCl activities of 0·01 if an ideal solution model is used. Because the hydrogrossular–analcime–sodalite assemblage in the augite syenite is consistent with these low salinities, NaCl activities drop from 0·4 to 0·01 in the course of crystallization and post-magmatic hydrothermal activity.
Maximum water activities were estimated for the augite syenite from the equilibrium
for the silica
activities calculated above (Fig.
16). Late-magmatic biotite
was assumed to have formed between 600 and 700°C to estimate
maximum water activities in this stage. Using an ideal site
mixing model for the biotite and the fayalite (two site mixing)
and the Fuhrman & Lindsley (1988)
model for feldspar, extremely
low water activities of
0·2 were calculated (Fig.
16).
The absence of biotite in the early crystallization stages is
interpreted to reflect even lower water activities. In constrast,
the late-stage agpaitic analcime assemblages indicate water
activities between 0·7 and unity at 400°C in the
augite syenite (Fig.
17a), between 0·4 and 0·8
at 500°C in the sodalite foyaite, naujaite and kakortokite
(Fig.
17b), and around unity at 500°C in the lujavrites
(Fig.
17c). These results are in agreement with fluid inclusion
data. Hence, contrary to NaCl activity, water activity significantly
increases during fractionation.
View larger version (32K):
[in this window]
[in a new window]
|
Fig. 17. (a) aH2O–aSiO2 and (b, c) temperature–aH2O diagrams with equilibria in the system Na–Al–Si–O–H–Cl used to calculate post-magmatic water activity in the various rock types: (a) for augite syenite; (b) for foyaite, naujaite and kakortokite; (c) for lujavrite.
|
|
Oxygen fugacity
Relative oxygen fugacities in the augite syenite decrease during fractional crystallization from about FMQ (fayalite–magnetite–quartz) – 1 to FMQ – 2, to below FMQ – 4, reaching in some samples the calculated stability of native iron, which was, however, not observed in any sample (Marks & Markl, 2001).
The agpaitic rocks lack both Fe–Ti-oxides and quartz, but the occurrence of arfvedsonitic amphibole, aegirine pyroxene and aenigmatite or olivine allows us to constrain the oxygen fugacity (see also Larsen, 1977) by reactions such as reaction (9). After steady decrease of fO2 in the augite syenite stage, oxygen fugacity rises approximately on or parallel to the ‘Arf–Acm’ buffer curve in Fig. 18, which represents the approximate, activity-corrected curve where arfvedsonite and aegirine coexist according to the experiments of Ernst (1962). A part of it is represented by reaction (9). The bold black curve in Fig. 18 is meant to represent combined deep-level (at high temperatures) and high-level (at lower temperatures) fractionation, and the grey curve labelled augite syenite refers to high-level fractionation only. Augite syenite and agpaitic rocks show different, but mostly subparallel evolution trends. As shown by the remarks at the top of Fig. 18, fO2 may be buffered by various solid-phase buffers or may not be buffered at all during the various stages of crystallization. Oxygen fugacity buffered by aegirine–arfvedsonite–aenigmatite assemblages increases during fractionation and cooling, probably to values of FMQ + 2 to FMQ + 4; this is close to or equal to the magnetite–haematite buffer (Fig. 18). After aenigmatite is no longer a liquidus phase, i.e. in the lujavrites, oxygen fugacity is no longer constrained by a solid–solid buffer reaction and fO2 can, in principle, increase or decrease. The transition from a methane-dominated to an aqueous fluid in the late lujavrite crystallization stages appears to occur at oxygen fugacities close to the magnetite–haematite buffer (Fig. 18).
|
DISCUSSION
|
|---|
The above calculations will be discussed based on the genetic
model of Larsen & Sørensen (1987)
, which assumes
fractionation at depth (crust–mantle boundary?) of the
melts parental to the Ilímaussaq rocks. Successive ‘leaking’
of this deep-level magma chamber is responsible for the three
magma batches rising to the high-level chamber, where further
low-pressure fractionation continues. The calculations and observations
reported above reveal the following important points:
- the earliest augite syenite melt at Ilímaussaq was reduced (FMQ – 1 to FMQ – 2), had methane as a stable fluid phase, very low water activities (<0·2), relatively low silica activities (0·8) and relatively high alkali/Al ratios [0·9 based on whole-rock analyses by Engell (1973); 0·83 after Bailey et al. (2001)].
- High-level fractionation of augite syenitic melts leads to drastically reduced SiO2 activities down to 0·4 and strong reduction down to FMQ – 4 and below (Marks & Markl, 2001).
- At the same time, deep-level fractionation also leads to decreasing SiO2 activities and increasing (Na + K)/Al ratios (however, much more slowly than during high-level fractionation, as a result of the much larger magma reservoir involved) thereby giving rise to the highly alkaline and finally agpaitic rocks in the next syenitic magma batch. Increasing alkali/Al ratios require that plagioclase or spinel has to fractionate until the alkali/Al ratio exceeds unity. The anorthosite xenoliths found all over the Gardar province may be the witness of such deep-seated processes. After this state is reached, crystallization of alkali feldspar and nepheline further increases the alkali/Al ratio (it should be noted that crystallizing a Na:Al = 1:1 phase such as albite from a
TOP
ABSTRACT
INTRODUCTION
