Journal of Petrology | Volume 44 | Number 1 | Pages 3-38 | 2003
© Oxford University Press 2003
Alkali Picrites Formed by Melting of Old Metasomatized Lithospheric Mantle: Manîtdlat Member, Vaigat Formation, Palaeocene of West Greenland
1GEOLOGICAL SURVEY OF DENMARK AND GREENLAND, ØSTER VOLDGADE 10, DK-1350 COPENHAGEN K, DENMARK
2GEOLOGICAL MUSEUM, ØSTER VOLDGADE 5–7, DK-1350 COPENHAGEN K, DENMARK
3MINERALOGISK–GEOLOGISK MUSEUM, SARS GATE 1, N-0562 OSLO, NORWAY
4GEOLOGICAL INSTITUTE, UNIVERSITY OF COPENHAGEN, ØSTER VOLDGADE 10, DK-1350 COPENHAGEN K, DENMARK
5DANISH LITHOSPHERE CENTRE, ØSTER VOLDGADE 10, DK-1350 COPENHAGEN K, DENMARK
RECEIVED December 5, 2001; ACCEPTED July 1, 2002
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGICAL SETTINGANALYTICAL TECHNIQUESPETROGRAPHY AND MINERALOGYWHOLE-ROCK GEOCHEMISTRYDISCUSSIONCONCLUSIONSAPPENDIX: CALCULATION OF MELTING...SUPPLEMENTARY DATAREFERENCES |
|---|
Alkaline picrites and basalts constitute 20–200 m of lava flows and hyaloclastites in the middle part of an
2 km thick succession of tholeiitic picrites and basalts formed during continental rifting of West Greenland around 60 Ma. The alkaline rocks, found only in northern Disko, have phenocrysts of olivine + chromite ± clinopyroxene; lava flows contain abundant groundmass clinopyroxene and plagioclase, whereas pillow breccias contain abundant fresh alkali basaltic glass. Six compositional types are present; all are strongly but variably enriched in incompatible trace elements [Ba, U, Nb, Ta, light rare earth elements (LREE)], yet their major elements, with relatively high SiO2 and Al2O3 and low Na2O, do not suggest an origin by small degrees of mantle melting. The isotope compositions are unusual, with negative 
Nd and mostly negative Sr (below the mantle array), high 206Pb/204Pb (below the Northern Hemisphere Reference Line), and mostly negative 
Os. The most likely source for the alkaline magmas is old metasomatized lithospheric mantle in which melting was induced by the passing hot, asthenosphere-derived, tholeiitic magmas. Simple mass-balance calculations suggest that the melting assemblages consisted of 60% pargasitic amphibole, 26–30% clinopyroxene, 9% olivine and 1% apatite. Mica in the source is required for only the least enriched magma type. For the most enriched magmas small amounts of Ba–U–Nb–Sr–LREE-rich oxides (lindsleyite and hawthorneite) are required in the melting assemblage and dominate the Pb isotope compositions. The various magma types and the partly complementary relation between them suggest that the lithospheric mantle had an ordered structure, possibly with old metasomatic zones formed by successive trapping of elements in migrating fluids. KEY WORDS: alkali picrite; amphibole melting; Greenland; lithosphere melting; metasomatism
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL TECHNIQUESPETROGRAPHY AND MINERALOGYWHOLE-ROCK GEOCHEMISTRYDISCUSSIONCONCLUSIONSAPPENDIX: CALCULATION OF MELTING...SUPPLEMENTARY DATAREFERENCES |
|---|
Alkaline extrusive rocks are a volumetrically insignificant component in most large igneous provinces (LIPs). LIPs are overwhelmingly tholeiitic in character and are considered to be formed by relatively high degrees of melting, mainly of the asthenosphere, and usually in the presence of a mantle plume (e.g. Saunders et al., 1997). The alkaline magmas are generally considered to be products of lower degrees of melting; however, their mantle sources, whether asthenospheric, lithospheric, or both, are contentious. Alkaline extrusives have often been emplaced late in the formation of the LIPs, e.g. the meimechites capping the Siberian basalt plateau (Arndt et al., 1995, 1998), and the nephelinites and basanites capping the East Greenland basalt plateau and nunataks (Brooks et al., 1979; Brown et al., 1996; Bernstein et al., 2000). Both the Siberian and the East Greenland alkaline extrusives have been interpreted to be of asthenospheric origin (Brooks et al., 1979; Arndt et al., 1995, 1998; Brown et al., 1996; Bernstein et al., 2000) whereas in other LIPs, such as the Deccan and Yemen, lithospheric components are thought to be involved (e.g. Mahoney et al., 1985; Baker et al., 1997). Probably, the alkaline rocks in LIPs are polygenetic.
During Palaeogene rifting and continental break-up in the North Atlantic, large volumes of flood basalts were extruded on the continental margins of both West and East Greenland. The magmas are thought to be generated mainly by melting within the impacting Iceland mantle plume (e.g. review by Saunders et al., 1997). In contrast to East Greenland, the West Greenland flood basalts do not include any nephelinites but terminate with transitional to mildly alkaline basalts (Clarke & Pedersen, 1976; Larsen, 1977). The voluminous main succession is uniformly tholeiitic except within one limited interval in the middle part, which, in addition to tholeiitic picrites, contains three close-lying levels with alkali picrites and alkali basalts with distinctive and highly unusual geochemical and isotopic characteristics. This paper explores the petrogenesis of the alkaline melts and the nature of their unusual mantle sources.
|
GEOLOGICAL SETTING |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL TECHNIQUESPETROGRAPHY AND MINERALOGYWHOLE-ROCK GEOCHEMISTRYDISCUSSIONCONCLUSIONSAPPENDIX: CALCULATION OF MELTING...SUPPLEMENTARY DATAREFERENCES |
|---|
The West Greenland basalts form part of the Nuussuaq Basin, a fault-controlled extensional basin at the continental margin where Precambrian basement is covered by Cretaceous–Palaeocene sediments overlain by volcanic rocks (Fig. 1; Chalmers et al., 1999). The major part of the onshore volcanic succession was erupted during a short time period around 60 Ma (Storey et al., 1998). The lower part of the 2–4 km thick volcanic succession is dominated by highly magnesian picritic rocks (the Vaigat Formation), whereas the upper part consists of more evolved, plagioclase-phyric basalts of the Maligât Formation (Clarke & Pedersen, 1976). The magnesian magmas of the Vaigat Formation were generated at high temperatures and very high production rates in the asthenosphere and passed swiftly through the lithosphere (Gill et al., 1992; Holm et al., 1993; Larsen & Pedersen, 2000; Pedersen et al., 2002). Most magmas escaped contamination, although a number of discrete crustal contamination episodes led to the formation of subordinate units of siliceous basalts and magnesian andesites (Pedersen, 1985a; Pedersen et al., 1996; Lightfoot et al., 1997). The volcanism of the Vaigat Formation occurred in three main cycles which formed the three main stratigraphic members; from older to younger these are the Anaanaa, Naujánguit and Ordlingassoq Members. The interval with alkaline rocks is the upper Naujánguit Member to lower Ordlingassoq Member, and the main part of the alkaline rocks forms one stratigraphic unit, which is formalized as the Manîtdlat Member (Pedersen, 1985b). The geographical distribution of these rocks is limited to northern Disko (Fig. 1).
|
|
Alkaline volcanic rock units
Alkaline rocks occur at three close-lying stratigraphic levels (Fig. 1). They are divided into a number of types as shown in Table 1. The total estimated volume of the alkaline rocks is around 30 km3; with possible extensions to the west and north the original volume may have been up to 50 km3, about 0·05% of the original volume of onshore basalts.
|
|
|
The oldest alkaline unit is a volcanic neck with a few associated lava flows of alkali basalt in the Stordal area (Fig. 1). These rocks (Stordal type) are of very limited extent and volume and are interbedded within picritic lava flows of the uppermost Naujánguit Member.
The second alkaline unit is an up to 20 m thick series of olivine-rich alkali picrite flows within the lowermost part of the Ordlingassoq Member. This unit (Type 0) can be followed over 20 km along the north coast of Disko and was most probably produced from one volcanic centre. It is overlain by 50–60 m of tholeiitic picrites, which are in turn overlain by the rocks of the Manîtdlat Member.
The third alkaline unit is the Manîtdlat Member, which forms a purplish brown marker horizon within the grey tholeiitic picrites of the Ordlingassoq Member (Fig. 2). The Manîtdlat Member is found within a 30 km by 20 km area in northern Disko (Fig. 1). It is on average about 50 m thick and represents eruptions from several volcanic centres, most probably fissure eruptions because all the lava flows are of pahoehoe type and no traces of volcanic edifices or explosive activity have been found. No eruption sites are known but they must be local. The volcanic rocks were produced from at least four alkaline centres and two tholeiitic centres, which interfinger laterally. The alkaline rocks are picrites and alkali basalts with four element enrichment patterns (Types 1a, 1b, 2 and 3; see Table 1). The youngest rocks are clinopyroxene-phyric basalts (ankaramites, Type 3), probably erupted from a volcanic centre in the southern part of the area. A dyke cutting the whole Manîtdlat Member is tholeiitic but somewhat enriched. Overlying the Manîtdlat Member are tholeiitic picrites of the Ordlingassoq Member.
Most of the volcanic rocks were erupted subaerially. However, the volcanic front was prograding laterally towards the east and SE into a volcanic-dammed lake (Pedersen et al., 1998), so that the thin lava flows pass laterally into thick hyaloclastite breccia deposits (Fig. 2). Consequently, all the magma types discussed here except the Stordal type exist in hyaloclastite facies with extremely fresh glassy rocks; in contrast, many of the subaerial lava flows are affected by zeolite-facies metamorphism. Stordal type glass is found in a chilled neck contact. Mantle xenoliths have not been found despite dedicated search.
In this paper, the alkaline units will be collectively referred to as the Manîtdlat Member.
|
ANALYTICAL TECHNIQUES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL TECHNIQUESPETROGRAPHY AND MINERALOGYWHOLE-ROCK GEOCHEMISTRYDISCUSSIONCONCLUSIONSAPPENDIX: CALCULATION OF MELTING...SUPPLEMENTARY DATAREFERENCES |
|---|
Microprobe analyses of phenocrysts, glass inclusions and matrix glasses were made on a JEOL Superprobe at the University of Copenhagen, using a combination of wavelength-dispersive (WDS) and energy-dispersive (EDS) detection systems. Normal operating conditions were 15 kV acceleration voltage, 15 nA beam current, 20 s total counting time for WDS and 60 s live time for EDS analyses. High-precision analyses of Ca, Cr, Ni and Ti in olivine and Ni, Ti and V in chromite were made using 100 nA beam current and 40 s total counting time (WDS); the lower limits of detection for these conditions are 15–30 ppm (Pedersen, 1985a). Glasses were analysed with 15 nA beam current and extended total counting times (WDS) for Ti (60 s), K (60 s) and P (120 s), yielding reproducibilities of 0·03 wt % TiO2 and 0·02 wt % K2O and P2O5 (2
on 11 repeat analyses on glass standards). Sixty whole-rock samples from the alkaline units and 15 samples of the contemporaneous tholeiitic picrites were analysed for major and trace elements by X-ray fluorescence spectrometry (XRF). Major elements were determined on fused glass discs at the Geological Survey of Denmark and Greenland, as described by Kystol & Larsen (1999). FeO was determined by titration. Trace elements were determined on pressed powder pellets at the Geological Institute, University of Copenhagen, using a Philips PW 1400 spectrometer and standard analytical methods with USGS reference materials for calibration.
A subset of 13 alkaline and four tholeiitic samples, mainly fresh pillow breccias, was analysed for trace elements by inductively coupled plasma mass spectrometry (ICP-MS) and for Sr, Nd and Pb isotopes. The ICP-MS analyses were performed on a Perkin–Elmer SCIEX Elan 6000 at the University of Durham, using methods described by Turner et al. (1999). Reproducibility, based on replicate digestion of samples, varied from 1·5% to 3% for most analyses.
Sr, Nd and Pb isotope ratios were determined on unleached samples on a VG354 instrument at the Mineralogisk–Geologisk Museum, Oslo, using methods described by Griffin et al. (1988). Average values for repeated standard analyses during the analytical period were 87Sr/86Sr = 0·71023 ± 3 (2 SE) for NBS987 and 143Nd/144Nd = 0·511112 ± 5 (2 SE) for the J&M standard batch no. S819093A. The Pb standard NBS981 gave 206Pb/204Pb = 16·897 ± 0·005, 207Pb/204Pb = 15·434 ± 0·005 and 208Pb/204Pb = 36·540 ± 0·015 (2 SE).
Seven samples were analysed for Os isotopes at Geocentre Copenhagen. Samples were spiked with an 188Os- and 187Re-enriched solution and digested in inversed (14N HNO3:10N HCl = 3:1) aqua regia in carius tubes at 230°C for 1 week (Shirey & Walker, 1995). Os was distilled from aqua regia directly into 8N HBr (Nägler & Frei, 1997) and purified following Roy-Barman & Allègre (1994). Os isotope analyses were performed on a VG Sector 54 solid-source negative thermal ionization mass spectrometer, using a multi-collector static routine and single multiplier peak jump mode for small Os beams. Re was purified using the liquid extraction method of Cohen & Waters (1996) and the concentrations were measured by multiple-collector (MC)-ICP-MS on an Axiom instrument, using Ir-doped sample solutions for controlling mass fractionation of Re through monitoring the 190Ir/194Ir ratio. Procedural blanks for Re were <30 pg and for Os <3 pg.
Mössbauer analyses were performed on handpicked glass chips at the Royal Veterinary and Agricultural University, Copenhagen. The spectra were obtained at 295 and 80 kV, using a constant acceleration spectrometer, and were fitted using three Fe2+ doublets and one Fe3+ doublet.
Analytical results are presented in Tables 2–6. The complete dataset is available for downloading from the Journal of Petrology web site at http://www.petrology.oupjournals.org (Electronic Appendix A).
|
|
|
|
|
|
|
PETROGRAPHY AND MINERALOGY |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL TECHNIQUESPETROGRAPHY AND MINERALOGYWHOLE-ROCK GEOCHEMISTRYDISCUSSIONCONCLUSIONSAPPENDIX: CALCULATION OF MELTING...SUPPLEMENTARY DATAREFERENCES |
|---|
The alkaline rocks have simple mineralogies with olivine and chromite as the earliest phenocryst phases, just as in the contemporaneous tholeiites (Larsen & Pedersen, 2000). In contrast to the tholeiites, clinopyroxene was the next phase to crystallize, followed by plagioclase. The glassy rocks are 1–3 cm thick pillow rims with phenocrysts (up to 1–2 mm) and microphenocrysts embedded in 70–80 vol. % clear, pristine, pale yellow glass, which has yielded well-defined 39Ar–40Ar ages of 60·3 ± 1·0 Ma and 60·7 ± 0·5 Ma (Storey et al., 1998). In the glassy rocks the amount of clinopyroxene increases gradually with decreasing Mg content of the rocks, from none or just a few tiny microphenocrysts in the glasses of the most Mg-rich rocks (Type 0), through increasingly frequent microphenocrysts, often in clusters, in Type 1 and 2 glasses, to up to 1 mm phenocrysts in the glasses of Type 3 where clinopyroxene is the dominant phenocryst phase. Plagioclase is absent from the glassy rocks of Type 0, and it forms microlites in the Type 1 and 2 glasses, and sparse microphenocrysts in the Type 3 glasses. The Stordal rocks are aphyric to olivine–clinopyroxene microphyric.
In the crystalline lava samples the groundmass consists of clinopyroxene, plagioclase, olivine, Fe–Ti oxides and apatite in intersertal, intergranular or subophitic textures. No primary mica or amphibole has been found. The alkaline and calcic character is reflected in very high modal proportions of clinopyroxene (17–40% normative di), with crystals often showing hourglass zoning and purplish colours. Late-stage segregation veins contain purple, zoned, prismatic clinopyroxene crystals, zoned plagioclases often heavily zeolitized, semi-skeletal magnetite and ilmenite and frequent apatite crystals, all embedded in a matrix of fine-grained zeolite–smectite aggregates. In contrast to the fresh glassy pillow breccias, the crystalline samples are affected by zeolite-facies metamorphism, and interstices are filled with green and brown smectites and colourless fine-grained aggregates of zeolites and Ca-hydrosilicates. These interstices probably include the breakdown products of nepheline. Vesicles are filled with massive zeolites. Sulphides form secondary pyrite in alteration zones, and primary sulphide liquid drops infrequently preserved as tiny 1–5 µm spherules in glass inclusions in olivine phenocrysts in the Mg-rich Types 0 and 1a.
Olivine
Olivine comprises several textural types similar to those from the tholeiitic rocks described by Larsen & Pedersen (2000). Most olivines are clear euhedral to subhedral to skeletal phenocrysts; some have inclusion-filled zones and healed cracks, and others have cores speckled with numerous small inclusions of oxides and sometimes glass. Some olivines are obviously xenocrystic, with anhedral and serrated outlines.
The olivines span the compositional range mg-number 92·3–77·4 (Table 2) with a compositional gap around mg-number 90. The most magnesian olivines (mg-number >90) are found in the most magnesian rocks and there is a crude correlation between the olivine compositional range within a sample and the bulk-rock MgO contents, as also found in the tholeiitic rocks (Larsen & Pedersen, 2000). All the olivines, including those with mg-number >90, have glass inclusions and high contents of CaO and Cr2O3, indicating a magmatic origin. Possible mantle xenocrysts would have very low contents of CaO and Cr2O3 (Larsen & Pedersen, 2000) and have not been found.
The minor elements MnO, CaO, Cr2O3 and NiO, measured with high precision, show distinct differences between olivines in alkaline and tholeiitic rocks (Fig. 3). First, olivines in the alkaline rocks show a far greater scatter than those in the tholeiites. Second, the main olivine populations in the alkaline rocks have distinctly higher contents of CaO (and MnO, not shown), similar or higher Cr2O3, and lower NiO than the tholeiitic olivines. Within a single sample, olivine crystals with widely different minor-element contents and zoning patterns may exist side by side. A few olivine crystals, often the larger ones, have minor-element contents similar to those of the tholeiitic olivines. This is particularly evident for some olivines with low CaO. The zoning patterns in some individual crystals are also shown in Fig. 3, and the significance of the data is discussed below.
|
Oxides
The primary magmatic oxides are brown, semi-transparent chromites, which occur as small euhedral crystals enclosed in olivine and, in Type 0 picrites only, also within the matrix glass. Some olivine cores are speckled with tiny oxide inclusions ranging from dust-size particles to greenish brown bleb-like grains and opaque vermicular grains. Rare opaque oxide xenocrysts, fringed with clinopyroxene crystals, occur free in the matrix glass.
The compositional variation of the oxides is illustrated in Fig. 4 and Table 2. The primary magmatic chromites have cr-numbers [atomic 100Cr/(Cr + Al)] between 40 and 60, generally lower than in the tholeiitic rocks (Fig. 4a). The dyke chromites have high cr-numbers. There is a good correlation between the mg-number of the chromites and the enclosing olivines (not shown); chromites with mg-number 75–80 are enclosed within olivines with mg-number 91–92·5, and the iron-rich olivine xenocrysts contain correspondingly iron-rich chromites.
The primary magmatic chromites show differences in the iron oxidation states between the magma types (Fig. 4b; Table 2). The Type 1a chromites have the highest Fe2+/
Fe whereas those in Type 1b have the lowest. With progressive crystallization (decreasing mg-number) the chromites become more reduced (higher Fe2+/Fe), in contrast to the tholeiitic chromites, which become more oxidized (lower Fe2+/Fe).
The opaque oxide xenocrysts are ferroan magnesioferrites, which have thin rims that are more reduced and more magnesian than the centres. They are extremely low in both Cr and Ti (Table 2, numbers 16 and 17).
The vermicular and bleb-like oxides in some olivine crystals range from ferroan magnesioferrite in olivine cores through greenish bleb-like Al-rich spinel sensu stricto to chromian spinel approaching normal magmatic compositions in the olivine rims (Fig. 4c). These oxides are not magmatic but are solid-state high-temperature oxidation and re-equilibration products, as discussed below.
Clinopyroxene and plagioclase
The clinopyroxenes show complicated oscillatory zoning, which is a feature very commonly found in alkaline rocks. The microphenocrysts generally correspond to the outermost 2–3 zones of the larger phenocrysts. Compositionally, however, all clinopyroxene phenocrysts span a relatively narrow range, En42–48 Fs4–10 Wo42–52, and they are thus classical diopsides (Table 2). There is little or no difference between the clinopyroxenes from the various chemical rock types, except that those of Type 2 tend to have slightly higher Ti and slightly lower Wo.
Plagioclase microphenocrysts with slight normal zoning (in Type 3) and microlites (in Types 1–3) span the compositional range An87–71 Or0·7–2·3. There are no differences between the various rock types.
Primary sulphides
Tiny globules of sulphide preserved within glass inclusions in olivine have chemical compositions close to Fe–Ni monosulphide with small amounts of Cu. Most globules are too small for ‘clean’ microprobe analyses, but energy spectra show about equal amounts of Fe and Ni in sulphide in olivine with mg-number 92, and successively decreasing Ni in sulphides in olivines with mg-number 88–85. Similar globules are also present in the tholeiitic rocks. The globules seem to be too large to be exsolved from the trapped liquid and are considered to be trapped as liquid sulphide together with the silicate melts.
Matrix glasses and glass inclusions
Matrix glasses were analysed in all compositional types. Glass inclusions in olivine phenocrysts, representing melts at earlier stages of crystallization, were analysed in Mg-rich samples of Types 0, 1 and 2. The glass inclusions were not homogenized and have lost olivine as a result of post-entrapment crystallization, but as long as there are no other daughter minerals the incompatible element ratios of the trapped melts should be unaffected.
The alkaline matrix glasses (Table 3) are homogeneous and relatively fractionated, with 6·4–7·9 wt % MgO and elevated contents of the incompatible elements Ti, Na, K and P relative to the bulk rocks. In a K2O–TiO2–P2O5 triangular diagram (Fig. 5), the matrix glasses of the various alkaline types have well-defined K2O–TiO2–P2O5 ratios (Fig. 5a), with Types 0, 1a and 1b being closely similar. The tholeiitic glasses with their very low K2O and P2O5 contents plot close to the TiO2 apex. The inclusion glasses (Fig. 5b) show considerable scatter, but the main trend goes from the matrix glasses towards the TiO2 apex. A plot of CaO/Al2O3 versus the composition of the olivine (Fig. 5c) shows that low-CaO olivines have glass inclusions with low CaO/Al2O3; these are the same as those with high TiO2/P2O5 in Fig. 5b.
|
The glass inclusions in the high-CaO olivines have CaO/Al2O3 ratios of 0·9–1·2. Except for Type 0 there is a general tendency of declining CaO/Al2O3 with increasing crystallization (decreasing olivine mg-numbers) to low values in the matrix glasses, distinctly lower than in the bulk rocks. This is ascribed to the formation of the abundant clinopyroxene microphenocrysts in the matrix glasses in these rocks.
Crystallization temperatures and oxidation states of alkaline vs tholeiitic melts
During quenching the last olivine rims that formed in contact with the alkaline matrix glasses have mg-numbers varying from 83·5 in the Type 0 picrites to 79·6 in the Type 3 ankaramites. When the oxidation state of iron in the glass is derived by assuming an olivine–melt Fe–Mg distribution coefficient at 1 atm of 0·30 (Roeder & Emslie, 1970), quench temperatures can be calculated after Ford et al. (1983) and range from 1190°C in the Type 0 picrites to 1150°C for the ankaramites. In most cases the measured and calculated olivine compositions are very similar, indicating equilibrium between olivine rims and glass. In comparison, the quench temperatures for the tholeiites are in the range 1210–1180°C for rocks without plagioclase phenocrysts (Larsen & Pedersen, 2000).
The complicated oxides indicate early intratelluric events of oxidation of the olivine phenocrysts, leading to oxidation-exsolution of magnesioferrite (Khisina et al., 1995), and subsequent solid-state re-equilibration at low oxygen fugacities, leading to formation of Fe3+-poor, Al-rich spinel and then to Fe3+-poor Cr-spinel (Fig. 4c). During both primary chromite crystallization and re-equilibration, the oxides from the alkaline rocks show progressively decreasing oxidation states, in contrast to the evolution trend in the chromites from the tholeiitic rocks (Fig. 4b). Mössbauer analysis of matrix glasses (Table 3) also shows higher Fe2+/Fe in the alkaline than in the tholeiitic glasses. Thus, there is circumstantial evidence that the alkaline melts were significantly more reduced than the tholeiitic melts.
|
WHOLE-ROCK GEOCHEMISTRY |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL TECHNIQUESPETROGRAPHY AND MINERALOGYWHOLE-ROCK GEOCHEMISTRYDISCUSSIONCONCLUSIONSAPPENDIX: CALCULATION OF MELTING...SUPPLEMENTARY DATAREFERENCES |
|---|
Major elements
Representative XRF analyses are shown in Table 4, and plots of the major elements vs MgO are shown in Fig. 6. The alkaline rocks contain 7–23 wt % MgO, and all have much higher contents of P2O5 and K2O than the contemporaneous tholeiites. They form a number of compositional groups (types, presented in Table 1), which are particularly evident in the P2O5 and TiO2 diagrams. Most of the alkaline rocks have low SiO2 and high CaO compared with the tholeiites. Types 0 and 3 show the greatest relative enrichment in CaO and P2O5 but no enrichment in TiO2. In contrast, Type 2 rocks have relatively low CaO and P2O5 but the greatest enrichment in TiO2 and K2O. The alkaline rocks have only slightly lower Al2O3 than the tholeiites, whereas the levels of FeO* are the same or slightly lower. Perhaps surprisingly, the alkaline rocks are not enriched in Na2O; indeed, many have lower Na2O contents than the tholeiites. Although Na2O is somewhat scattered as a result of secondary alteration, low Na2O is a primary feature of the magmas and is also seen in the analyses of the fresh matrix glasses (Table 3 and Fig. 6). The tholeiitic glasses have the same Na2O contents as the glasses of Types 0 and 2; the highest Na2O contents are seen in the glasses of Types 1 and 3 that were quenched after significant clinopyroxene crystallization. The K2O diagram shows considerable scatter, much of which is due to secondary redistribution of K in the lava samples, which, as described above, are often altered whereas the pillow breccias are fresh. The data from the pillow breccias alone strongly suggest that the Type 2 melts were generated with higher K2O than the other types. The dyke has tholeiitic abundances of most of the major elements but has slightly elevated contents of P2O5 and K2O.
Three Stordal samples and four Type 1a samples have increased SiO2 and decreased CaO relative to other rocks of the same type; together with other elemental fingerprints this suggests these samples are crustally contaminated, as discussed below.
For each rock type, the compositional variation seen in Fig. 6 is dominantly caused by olivine fractionation and accumulation; the clinopyroxene-phyric Type 3 rocks also show evidence of clinopyroxene fractionation, or perhaps accumulation, in the changed slope of the CaO trend. CaO/Al2O3 is not changed by olivine fractionation or accumulation, and the different levels of this ratio in the various magma types (Fig. 7) may be features of the primary magmas. Types 0, 1b and 3 have the highest CaO/Al2O3 ratios, up to 1·35, caused by both high CaO and low Al2O3 (Fig. 6). The sloping CaO/Al2O3 trend in the Type 3 rocks is produced by clinopyroxene fractionation or accumulation. Type 2 and Stordal have CaO/Al2O3 ratios corresponding to the tholeiites.
|
All samples except for three evolved ones contain <3 wt % total alkalis, and according to the IUGS classification (Le Bas, 2000) the rocks are simply picrites and basalts. The alkaline character is better reflected in the CIPW norms, particularly of the matrix glasses (Fig. 8). None of the Type 2 and Stordal bulk rocks are ne normative; Type 1 is mixed, and all rocks of Type 0 and Type 3 are ne normative although the maximum ne is only 5·3%. However, all matrix glasses except Stordal and the dyke are ne normative, with maximum ne = 9·8–10·2% in Type 3. The three high-Si Stordal lavas are slightly Q normative (not plotted). Figure 8 also shows the high contents of normative diopside, the Type 3 ankaramites attaining a maximum of 40% di. The Type 3 glasses have somewhat decreased di because of clinopyroxene fractionation, whereas the Type 0 glasses with 39% di were quenched just before clinopyroxene saturation was reached.
|
Trace elements
Figure 9 shows a range of trace elements plotted against MgO. The incompatible elements Ba, Sr, Nb and La show different levels of enrichment relative to the tholeiites, and Type 1 clearly splits up in two groups (a and b). Types 0, 1b and 3 are the most strongly enriched in Ba, Sr and La, Type 1a is intermediate, and Type 2 and Stordal are the least enriched. Types 0, 1 and 3 have extremely high Nb/Zr ratios in the range 0·5–0·7. In contrast, Type 2 is relatively more enriched in Zr and has lower Nb/Zr ratios of 0·3–0·4, but still higher than the tholeiitic values of <0·14. The dyke is close to the tholeiitic values for many elements but is clearly enriched in all the incompatible elements (see also Fig. 11). It has a Nb/Zr ratio of 0·19.
|
The compatible elements Ni and Cr show the effects of olivine and chromite control; however, the alkaline rocks have distinctly lower Ni contents than tholeiites with similar MgO. The diagrams for V and Sc show a maximum at 10–11 wt % MgO, indicating clinopyroxene fractionation in magmas with <10 wt % MgO, in accordance with the major-element variations.
A wider range of trace elements was obtained by ICP-MS analysis of two samples of each of the alkaline types and one of the dyke (Table 5). The two samples from each alkaline type gave closely similar results, except for the Stordal samples, of which one (264165) is a high-Si variety. These data are presented in Figs 10 and 11 as rare earth element (REE) and multi-element diagrams with comparisons.
|
Two element enrichment patterns can be clearly distinguished from the multi-element diagrams. One pattern is common to the rocks of Types 0, 1a, 1b and 3. It shows extreme enrichment in Ba, U, Nb–Ta and light to middle REE (LREE to MREE), deep troughs for Rb, Th, K, Pb and Ti, and lesser troughs for Sr, P and Zr–Hf. In comparison, melilitites, nephelinites and meimechites show similar levels of enrichment for many elements but have much smoother spectra, with moderate K troughs being the most distinctive (Rogers et al., 1992; Arndt et al., 1995, 1998; Hegner et al., 1995; Wilson et al., 1995; Bernstein et al., 2000; Späth et al., 2001). Enriched picrites from Nuanetsi have incompatible element concentrations up to 200 times primitive mantle; however, their spectra show no K anomalies and have distinct Nb–Ta troughs (Ellam & Cox, 1989, 1991). Carbonatites and allegedly carbonatite-metasomatized mantle xenoliths show variable degrees of enrichment and have spiky patterns with similarities to those of the Manîtdlat Member, showing deep troughs for Rb, K, Zr and Ti, and strong relative enrichment in Ba, Nb and LREE (Nelson et al., 1988; O’Reilly & Griffin, 1988; Yaxley et al., 1991; Larsen & Rex, 1992; Le Roex & Lanyon, 1998; Coltorti et al., 2000). The Manîtdlat Member rocks have low to very low Th/U (down to 1·2 in Type 3), whereas most carbonatites have high Th/U although the ratios are very variable.
A different element enrichment pattern is seen in the Type 2 rocks. The incompatible-element enrichment is less extreme, Rb is enriched, Ba much less so, the troughs for Th, K, Pb and Sr are very small, and there is a large Nb–Ta peak and lesser peaks for Zr–Hf and Ti. This pattern is similar to those of some basanites and alkali basalts from eastern Australia (O’Reilly & Zhang, 1995; Zhang et al., 2001), and remarkably similar to patterns of amphibole from some mantle xenoliths (Moine et al., 2001).
The Stordal type has an enrichment pattern most similar to those of Types 0, 1 and 3, but with smaller troughs for Rb, Th and K. The high-Si Stordal sample has higher Rb and Pb, and lower U, Nb, Ta and REE than the other Stordal sample.
The dyke is the least enriched of the alkaline rocks. Its incompatible element contents all lie between those of the tholeiites and the other alkaline rocks, and the trace-element pattern, with low Rb and high Nb/La, mostly resembles that of Type 1a.
The uniqueness of the Manîtdlat Member magmas may be illustrated by comparison with incompatible element ratios in other strongly enriched rocks (Fig. 12). In melilitites, nephelinites and meimechites, K/Ba, Rb/Sr, Ba/Nb and La/Nb ratios vary over about one order of magnitude. In the Manîtdlat Member rocks these ratios vary over about two orders of magnitude, and whereas Type 1a and Stordal often plot together with other rocks, Type 2 lies at one extreme end with higher K/Ba and Rb/Sr and lower La/Nb and Ba/Nb than most other rocks, and Types 0, 1b, and 3 lie at the opposite extreme with lower K/Ba and Rb/Sr and higher La/Nb and Ba/Nb than almost all other rocks. The significance of the spread in Fig. 12 and the good correlation within the Manîtdlat Member is discussed below.
|
Isotopes
The alkaline rocks of the Manîtdlat Member have very unusual isotope compositions (Table 5). The Sr–Nd–Pb results for the two samples analysed of each type are mutually consistent, and the results for the tholeiitic picrites of the Ordlingassoq Member, analysed simultaneously with the alkaline rocks, are in complete agreement with earlier results by Holm et al. (1993), Lightfoot et al. (1997) and Graham et al. (1998).
The tholeiitic rocks have positive Nd and negative Sr (Fig. 13) and plot within the field for the Iceland mantle plume (e.g. Stecher et al., 1999). They have been interpreted by earlier workers (Holm et al., 1993; Lightfoot et al., 1997; Graham et al., 1998) as produced by melting of the asthenospheric mantle in the proto-Icelandic mantle plume, and the data fields for the Ordlingassoq Member shown in Figs 13 and 14 thus conceivably represent the local contemporaneous asthenospheric mantle.
|
|
The alkaline rocks (excepting the dyke) all have negative Nd, and all except Type 2 and the high-Si Stordal lava also have negative Sr (Fig. 13). Types 0, 1b and 3 have nearly identical Nd–Sr isotope compositions. Except for Type 2, all the alkaline rocks plot well below the oceanic mantle array in the Nd–Sr isotope diagram, in an area of the lower left quadrant occupied by very few uncontaminated mantle-derived rocks. None are known from the North Atlantic Igneous Province; the few we have noted are some potassic rocks from the Elkhead Mountains, Colorado (Leat et al., 1988; Thompson et al., 1989), some carbonatites and kimberlites from the Archangelsk region, NW Russia (Mahotkin et al., 2000), and some nephelinites from the Napak and Mt Elgon volcanoes in East Africa (Simonetti & Bell, 1994, 1995).
The Pb isotope compositions (Fig. 14) of the tholeiites of the Ordlingassoq Member cluster around the Northern Hemisphere Reference Line (NHRL) and fall within the Iceland field. In contrast, the alkaline rocks have high 206Pb/204Pb ratios and plot below the NHRL. The basement rocks in the area have low Pb isotope ratios (Fig. 14), and the high-Si Stordal sample is clearly displaced towards basement values. In terms of the vertical distance from NHRL as defined by Hart (1984), the Manîtdlat Member rocks have 
7/4 = -8 to -25, 8/4 = 0 for Type 2 and 8/4 = -100 to -257 for the other types, outside the range of all modern mid-ocean basalts (MORB) and ocean island basalts (OIB) (Thirlwall, 1997). We do not know of any other mantle-derived uncontaminated igneous rocks with Pb isotope ratios similar to those of the Manîtdlat Member. Many alkaline rocks have similarly high 206Pb/204Pb ratios, but for a given 206Pb/204Pb they all have higher 207Pb/204Pb and 208Pb/204Pb ratios, most of which lie close to or above the NHRL (e.g. Nelson et al., 1988; Simonetti & Bell, 1994, 1995; Hegner et al., 1995; Wilson et al., 1995; Kalt et al., 1997; Le Roex & Lanyon, 1998; Bell & Tilton, 2001; Späth et al., 2001, and compilation therein). Pb isotope ratios similar to those of the Manîtdlat Member are, however, reported from some highly metasomatized peridotite xenoliths with metasomatic oxides from South Africa (Hawkesworth et al., 1990) and in metasomatic oxide (mathiasite) from such xenoliths (Fig. 14; Griffin et al., 1999).
All the samples analysed for Os isotopes have low to very low 187Os/188Os ratios (0·1342–0·1067, Table 6) and all except the Type 1a sample have negative Os (Fig. 15). There is an inverse correlation between the Os isotope ratios and the amounts of Os present in the samples (Table 6). One sample of Type 0 has exceptionally high Os, 44 ppb. These data are as unusual as the other isotope data: 187Os/188Os ratios below 0·110 have previously not been reported from igneous rocks but only from peridotite xenoliths from old subcontinental lithospheric mantle (Pearson et al., 1995; Hanghøj et al., 2001).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL TECHNIQUESPETROGRAPHY AND MINERALOGYWHOLE-ROCK GEOCHEMISTRYDISCUSSIONCONCLUSIONSAPPENDIX: CALCULATION OF MELTING...SUPPLEMENTARY DATAREFERENCES |
|---|
Modification of the primary alkaline magmas
Identification of crustal contamination
The effects of crustal contamination on the tholeiitic volcanic rocks on Disko and Nuussuaq are well described (Pedersen, 1979, 1985a; Pedersen & Pedersen, 1987; Goodrich & Patchett, 1991; Pedersen et al., 1996; Lightfoot et al., 1997). Contamination with either Precambrian basement or Mesozoic sediments leads to increase in SiO2, K2O, Rb and Pb, and decrease in FeO*, CaO, TiO2 and Nb (but not appreciably Zr). Sr isotope ratios increase, and Nd and Pb isotope ratios decrease. Compositional shifts of this kind are seen in the three high-Si Stordal samples, one of which is analysed isotopically, and to a lesser extent in four Type 1a samples (Figs 6, 7, 9, 13 and 14). We therefore consider these samples to represent crustally contaminated alkaline magmas. Modelling of the crustal contamination process is outside the scope of the present paper.
On the other hand, the character of the alkaline rocks as such, with their strong enrichment in Ba, U, Nb, LREE and P, and low Sr isotope ratios and high Pb isotope ratios, cannot be explained in terms of contamination with any known crustal components in the region.
Evidence for mixing of tholeiitic magma into alkaline magma
The alkaline magmas were erupted within a regional zone of eruption centres that mainly produced tholeiitic magmas, and it is likely that the magmas utilized the same conduit systems. The conduits would be possible sites of mixing between alkaline and tholeiitic magma batches and their phenocrysts, of which the tholeiites contained only olivine + chromite. The heterogeneous olivine populations within single samples suggest that such mixing has taken place (Fig. 3). Low-CaO zoned olivines such as phenocryst 1 in Fig. 3 are explicable as crystals that originally formed in tholeiitic magmas and later were mixed into the alkaline magmas where they partly re-equilibrated and continued their growth. The ‘tholeiitic’ levels of Ca, Cr and Ni in the cores are preserved, whereas Mg and Fe are re-equilibrated to lower mg-numbers. The low CaO/Al2O3 in the glass inclusions in the low-CaO olivines (Fig. 5) confirms the tholeiitic character of their parental melts. These low-Ca olivines have mg-numbers of 82–92·3 (Fig. 3). Phenocrysts 2–4 in Fig. 3 are normal phenocrysts in the alkaline magma.
Because of the evidence for mixing it is necessary to evaluate the extent to which this process has modified the composition of the original alkaline magmas. The main constraint on this comes from the near-constant Nd isotope ratios in Types 0, 1b and 3; the total variation in 143Nd/144Ndi for these rocks is only 0·00002 (0·51236–0·51238, Table 5). Because the Nd contents are strongly dominated by the alkaline component, moderate amounts of tholeiitic magma can be mixed into the alkaline magma before the Nd isotope ratio changes significantly. A Type 0 alkaline end-member magma may be mixed with up to 25% tholeiitic magma before the Nd isotope ratio increases by more than 0·00002 (mixing curve in Fig. 13), and 25% is therefore considered a maximum amount of in-mixed tholeiitic magma. We have no means of quantitatively constraining the amounts of tholeiitic magma further, but based on the relative scarcity of the tholeiitic olivine crystals we consider that the amount of tholeiite in the alkaline magmas was normally 10% or less. It is possible that the tholeiitic component is dominated by olivine crystals picked up in the mush zones in the conduit systems, with very little accompanying tholeiitic melt.
With an upper limit of 25% tholeiitic magma in the alkaline magmas, the major-element composition of the unknown pure alkaline end-member melt will not be very different from that of the erupted magmas. SiO2 in the end-member will be lower than in the erupted rocks by 0·5 wt % or less; CaO/Al2O3 will be higher but still <1·26. TiO2 and Na2O are invariably low. The incompatible trace elements in the alkaline end-member will be higher by a factor of 1·3 or less; ratios of more incompatible elements (e.g. Nb/La, Th/U) will be virtually unaffected, whereas ratios of more incompatible to less incompatible trace elements (e.g. Nb/Y, Ba/Ti) will be lowered by the tholeiitic component. With an upper limit of 10% in-mixed tholeiite the alkaline magmas are practically unchanged. In conclusion, there is undoubtedly a small tholeiitic component in the alkaline magmas, but the compositional influence of this is negligible for most elements.
The dyke: mixing of alkaline magma into tholeiitic magma
Whereas the composition of an alkaline magma is fairly robust against addition of minor amounts of tholeiitic magma, the opposite is not the case. This is illustrated by the investigated dyke that cuts the entire Manîtdlat Member succession. The dyke can be interpreted in terms of mixing of alkaline magma into tholeiitic. Its major-element composition is tholeiitic (Fig. 6), whereas its incompatible trace elements (including K and P) are intermediate between tholeiitic and alkaline values, higher than those in the tholeiites by a factor of 2–4 (Fig. 11). The Nd–Sr isotope composition of the dyke is also intermediate between that of tholeiitic and alkaline rocks (Fig. 13). The low Rb and normalized Ta/La>1 suggest a relation to Type 1a, and the dyke is actually situated in an area where the alkaline rocks are solely of Type 1a. Simple mixing calculations between tholeiite (sample 326783) and Type 1a (sample 326787) give consistent results for most trace elements and the Nd isotopes, suggesting that the dyke is a tholeiitic magma that contains around 15% alkaline component.
Primary alkaline magmas
The composition of the most magnesian, possibly primary, alkaline magmas may be estimated from the most magnesian cognate olivines present. The most magnesian high-CaO olivine has mg-number 90 (Fig. 5c), which corresponds to a melt calculated to have around 15 wt % MgO, somewhat dependent on the oxidation state. Thus, samples with >15 wt % MgO most probably contain accumulated olivines whereas samples with lower MgO may represent the erupted and more or less fractionated magmas. At 15 wt % MgO the tholeiitic melts had temperatures close to 1400°C (Larsen & Pedersen, 2000). The parental alkaline melts would have had lower temperatures, loosely estimated around 1300°C.
Mantle sources and melting processes
The rocks of the Manîtdlat Member have incompatible trace element concentrations that are enriched by up to 100–300 times primitive mantle for elements such as Ba, U, Nb and La (Fig. 11). Similar enrichment levels in basic igneous rocks are normally found in melilitites, nephelinites and meimechites, and also some kimberlites. These rock types are all strongly silica undersaturated, sometimes larnite normative, plagioclase free or plagioclase poor, and are considered to be formed by small degrees of melting of enriched, volatile-bearing mantle (e.g. Nelson et al., 1988; Wilson, 1989; Rogers et al., 1992; Taylor et al., 1994; Arndt et al., 1995, 1998; Hegner et al., 1995; Wilson et al., 1995; Mahotkin et al., 2000). In comparison with these rocks, those of the Manîtdlat Member are not highly undersaturated and have relatively high contents of SiO2 and Al2O3, and low TiO2, alkalis and P2O5 (Fig. 16). The effusive eruption style and the anhydrous mineralogy do not suggest that the volatile contents of the alkaline Manîtdlat Member magmas were higher than in the tholeiites.
|
Manîtdlat Member melts such as those of Types 0 and 1 have CaO/Al2O3 around 1·2 and, at the 15 wt % MgO level, 11·5 wt % Al2O3 and 1·0 wt % Na2O. According to the melting model of Herzberg & Zhang (1996), such melts cannot be produced near the solidus of an ordinary dry lherzolite at any pressure. Olivine fractionation from a deep near-solidus melt may produce enriched alkali picrites (Milholland & Presnall, 1998), but they will not be like the alkali-poor Manîtdlat