Journal of Petrology Volume 42 Number 5 Pages 855-876 2001
© Oxford University Press 2001
The Palaeoproterozoic Komatiite–Picrite Association of Finnish Lapland
E. HANSKI1,*,
H. HUHMA2,
P. RASTAS1 and
V. S. KAMENETSKY3
1GEOLOGICAL SURVEY OF FINLAND, PO BOX 77, FIN-96101 ROVANIEMI, FINLAND
2GEOLOGICAL SURVEY OF FINLAND, PO BOX 96, FIN-02151 ESPOO, FINLAND
3DEPARTMENT OF GEOLOGY, UNIVERSITY OF TASMANIA, GPO BOX 252-79, HOBART, TAS. 7001, AUSTRALIA
Received
January 25, 2000;
Revised typescript accepted
July 21, 2000
 |
ABSTRACT
|
|---|
The large range of chemical variation within intimately associated
highly magnesian volcanic rocks in the Palaeoproterozoic Central
Lapland Greenstone Belt prompted the construction of a new classification
scheme for MgO-rich volcanic rocks, based on an [Al
2O
3] vs [TiO
2]
diagram where the axes are the Al
2O
3 and TiO
2 contents (in mole
proportions) of the rocks projected from the olivine composition.
This diagram places the Lapland rocks in the fields of Ti-enriched
komatiites and picrites. Komatiitic rocks are depleted in both
light and heavy rare earth elements (LREE and HREE) relative
to middle REE (MREE) and possess relatively high TiO
2 even in
the most LREE-depleted varieties, whereas picritic rocks approach
geochemically Hawaiian picrites. Seven clinopyroxene and whole-rock
pairs analysed for Sm–Nd isotopes yield an average age
of 2056 ± 25 Ma for the komatiites. Uncontaminated komatiites
and picrites have similar positive
Nd values (+4) indicating
generation from a mantle source with a long-term depletion in
LREE relative to MREE. Geochemical characteristics of the komatiite–picrite
association, including REE and Nb/Y–Zr/Y systematics,
indicate chemical heterogeneities in the source region, which
seem to have been created by complex depletion and enrichment
processes shortly before or related to a dynamic melting process.
The high MgO contents of the rocks coupled with chemical similarity
between the Lapland and Hawaiian picrites supports a mantle
plume model for their genesis. Nevertheless, the geotectonic
evolution appears to have proceeded without significant regional
uplift shortly before volcanism.
KEY WORDS: Finland; komatiite; Nd isotopes; Palaeoproterozoic; picrite
|
INTRODUCTION
|
|---|
Komatiitic lavas with high MgO contents are often considered
to represent high degrees of mantle melting and consequently
to approach their mantle sources in composition (e.g. Bickle
et al., 1976
). However, as pointed out by Arndt (1986)
, they
can be a dirty window into the Archaean mantle because of some
complicating factors. These include the high temperature of
high-MgO liquids, which makes them susceptible to crustal contamination
during ascent or eruption (Huppert & Sparks, 1985
). Another
feature of komatiites—variable chemistry in terms of Al
2O
3/TiO
2 or heavy rare earth element (HREE) ratios—could have arisen
from garnet fractionation, and if this fractionation took place
during partial melting (e.g. Ohtani, 1990
), then the komatiite
composition may not reflect that of the source mantle. Observations
on ultra-depleted and enriched melt inclusions in early crystallized
minerals from recent primitive oceanic basalts have shown that
melts produced by partial melting of upwelling mantle undergoing
decompressional melting can be much more varied in their trace
element characteristics than the erupted lavas themselves (Gurenko
& Chaussidon, 1995
). This feature has been explained by
mixing of liquids produced by a dynamic melting process from
a homogeneous mantle source (Gurenko & Chaussidon, 1995
).
Dynamic melting has also been suggested to account for the chemical
variability of komatiites and picrites in Gorgona Island, Colombia
(Arndt
et al., 1997
).
Soon after the identification of the first komatiites in the late 1960s (Viljoen & Viljoen, 1969), it became apparent that there is a considerable heterogeneity within these rocks in terms of certain elements (e.g. Al/Ti, La/Sm, Gd/Yb; Nesbitt & Sun, 1976). Two principal chemical types were recognized: one having very high CaO/Al2O3 (
1·5) and occurring in the early Archaean Barberton Mountain Land, and the other possessing CaO/Al2O3 1·0 and occurring in late Archaean greenstone belts in various localities (Brooks & Hart, 1974). Later these two varieties were defined as Al-depleted and Al-undepleted komatiites by Nesbitt et al. (1979). As more geochemical and isotopic data have become available around the world, this simple picture of the distribution in time and space of these two komatiite types has turned out to be complex. Although the early Archaean komatiites are dominantly of the Al-depleted type, there are exceptions (e.g. Collerson et al., 1991; Puchtel et al., 1993a, 1993b; Lahaye et al., 1995). Also, late and middle Archaean greenstone belts contain numerous examples of komatiite and komatiitic basalt occurrences that display low Al2O3/TiO2 ratios deviating considerably from the chondritic value typical of Al-undepleted komatiites (e.g. Nesbitt et al., 1984; Cattell & Arndt, 1987; Schaefer & Morton, 1991; Xie et al., 1995; Fan & Kerrich, 1997; Tomlinson et al., 1999). Low Al2O3/TiO2 may be the consequence of depletion in Al, but not necessarily always exclusively so. In contrast, it seems that often it is due to an overabundance of TiO2 compared with normal Al-undepleted komatiites. Major element heterogeneity of Archaean primitive volcanic rocks cannot therefore be explained only in terms of depletion or non-depletion in aluminium, but there is another dimension of major element variation related to enrichment in TiO2, which is often coupled with enrichment in iron. Some of the Archaean rocks are sufficiently high in iron and titanium to allow them to be called ferropicrites (Stone et al., 1995; Francis, 1999) or iron-rich komatiites (Green & Schulz, 1977). Depletion in titanium in turn leads to rock compositions approaching recent boninites (Kerrich et al., 1998). The chemical heterogeneity within Archaean primitive volcanic rocks is even more pronounced when incompatible trace elements are concerned.
In this paper we present data for Palaeoproterozoic MgO-rich volcanic rocks from northern Finland, which indicate that chemical heterogeneity among primitive magmas was not less significant in the early Proterozoic than in the Archaean. As shown in Fig. 1, the Finnish occurrences are part of a longer belt extending to northern Norway. A characteristic feature of komatiites in this belt is enrichment in titanium and other high field strength elements (HFSE), which led Barnes & Often (1990) to call these rocks Ti-rich komatiites. The belt has also become well known because of the presence of large amounts of fragmental komatiites classified as pyroclastic or volcaniclastic (Saverikko, 1985; Barnes & Often, 1990; Räsänen, 1996), which are not so common in most Archaean komatiite successions.
View larger version (52K):
[in this window]
[in a new window]
|
Fig. 1. Occurrence of the Palaeoproterozoic komatiite–picrite association in northern Finland and Norway. GB, greenstone belt; CLGB, Central Lapland Greenstone Belt.
|
|
When considering the Lapland rocks together with other Palaeoproterozoic primitive volcanic rocks in the Fennoscandian Shield, such as the Pechenga ferropicrites and Onega picrites (Hanski & Smolkin, 1989; Hanski, 1992), there seems to be a more or less complete chemical variation in both major and trace elements. This observation raises the question of the classification of these primitive rocks and gave us an impetus to construct a new classification scheme for such volcanic rocks based on their Al2O3 and TiO2 contents. Another objective of this study was to provide new Sm–Nd isotope data for komatiitic and picritic rocks from Lapland to better constrain their age, which has been contentious until now. To this end, we have measured Nd isotopic compositions for whole rocks and primary clinopyroxenes from the Jeesiörova and Peuramaa areas at the southern edge of the Kittilä greenstone belt, where the rocks are relatively well preserved. We also provide new geochemical data and discuss the petrogenesis of these volcanic rocks with special reference to mantle plume models.
|
GEOLOGICAL BACKGROUND
|
|---|
In contrast to most Precambrian terranes, most widespread komatiitic
metavolcanic rocks in Finnish Lapland are early Proterozoic
rather than Archaean in age. Archaean komatiites, found in eastern
Lapland, have suffered from pervasive tectono-metamorphic modifications
and have therefore attracted limited attention from petrologists.
Instead, the Proterozoic komatiites have served as targets for
several studies in northern Finland (e.g. Saverikko, 1985
; Räsänen
et al., 1989
; Lehtonen
et al., 1998
) and in northern Norway
(Henriksen, 1983
; Krill
et al., 1985
; Barnes & Often, 1990
;
Davidsen, 1994
). These komatiites are part of the Central Lapland
Greenstone Belt, which extends from the eastern border of Finland
in the Salla area through the Sodankylä and Kittilä
areas and to the northern border between Finland and Norway
(Fig.
1). The most recent summary of the geology of the central
part of the belt has been reported by Lehtonen
et al. (1998)
.
The early Proterozoic komatiites principally occur at two stratigraphic horizons, referred to here as the lower and upper komatiites. They are separated by an assessed time span of 400 my (Lehtonen et al., 1998). These two occurrences clearly differ in their associated rock types, environment of eruption and geochemical characteristics. The lower komatiites, belonging to the Onkamo Group, have a stratigraphic position close to or at the bottom of the Proterozoic supracrustal sequence and have been most thoroughly studied in the type area at Möykkelmä, where they form part of a komatiite–tholeiite sequence of 250 m thickness extruded directly on an Archaean granitoid basement (Räsänen et al., 1989). The geochemical characteristics of these komatiitic rocks, including high light REE (LREE)/HREE and negative Ta anomalies, suggest a considerable amount of Archaean upper-crustal contamination. There exist no radiometric dates for the lower komatiites and associated rocks of the Onkamo Group. However, these rocks can be correlated with dated komatiitic rocks from the Vetreny Belt in Russian Karelia, on the basis of their similar stratigraphic position, lithology and geochemical characteristics. Recently, Puchtel et al. (1997) have obtained Sm–Nd isochron ages of 2449 ± 35 Ma and 2410 ± 34 Ma for Vetreny Belt komatiites.
The upper komatiites are associated with phyllites and black schists of the Savukoski Group deposited on a sequence of cratonic quartzites of the Sodankylä Group (Lehtonen et al., 1998). These komatiites occur together with picritic metavolcanic rocks and are therefore described below as the komatiite–picrite association. Komatiites and picrites display similar structures and textures and therefore we can distinguish them only on the basis of chemical composition, particularly the TiO2 content, which is clearly higher in picrites. Figure 1 depicts the areal distribution of this rock association in the Central Lapland Greenstone Belt and its continuation in northern Norway, totalling 400 km in length. Komatiitic varieties are abundant and well exposed in the Sattasvaara and Kummitsoiva areas, and have previously been described by Saverikko (1983, 1985) and Räsänen (1996), whereas picritic varieties are well developed in the Sotkaselkä area (Lehtonen et al., 1998) (see Fig. 1). Samples for this study were obtained from two localities, Jeesiörova and Peuramaa, situated in the most southwestern part of the komatiite belt (Figs 1 and 2).
View larger version (48K):
[in this window]
[in a new window]
|
Fig. 2. Geological map of the southwestern part of the Kittilä greenstone complex [simplified after Lehtonen et al. (1998)].
|
|
After recognition of their komatiitic affinity, the ultramafic volcanic rocks in Central Lapland were regarded by some workers as Archaean in age (Gaál et al., 1978; Saverikko, 1983, 1985). This opinion has been questioned over the last decade, on the basis of detailed field geological studies resulting in the consensus that most komatiites in Finnish Lapland belong to Palaeoproterozoic schist belts deposited unconformably on the Archaean basement (e.g. Lehtonen et al., 1998). However, owing to the lack of any direct age determination on the komatiitic rocks in Finnish Lapland, their precise age has remained contentious. The only available dating of komatiitic rocks is that performed by Krill et al. (1985) using the Sm–Nd method on rocks from the Karasjok greenstone belt in Finnmark, northern Norway. They obtained an isochron age of 2085 ± 85 Ma, but this result has to be regarded with caution as the real eruption time (Hanski & Smolkin, 1989; Puchtel et al., 1998), because it relies on a geochemically heterogeneous population of whole-rock samples collected from several locations along the greenstone belt. Two other options have been presented. According to the stratigraphic scheme by Silvennoinen (1985), komatiite-bearing formations are assigned to the Lapponian Supergroup, meaning that they are pre-Jatulian and hence must have an age of >2·2 Ga. Recently, Puchtel et al. (1998) showed that the Onega plateau basalts in Russian Karelia possess an age of >1·98 Ga, similar to that obtained for the Pechenga ferropicrites (Hanski et al., 1990), and put forward a model in which the primitive volcanic rocks in these two areas together with komatiitic rocks in Central Lapland are genetically related to magmatism of the same starting mantle plume upwelling beneath the Fennoscandian Shield.
|
THE KOMATIITE–PICRITE ASSOCIATION
|
|---|
As mentioned above, komatiites and picrites show similar volcanic
structures in the field and cannot be distinguished from each
other without geochemical analyses, and therefore some uncertainty
exists about their areal distribution. Nevertheless, it is evident
that komatiitic varieties are more abundant than picritic ones,
although there are areas dominated by picritic rocks. As a whole,
there seems to be no systematic distribution of different types
within the komatiite belt. The most common eruption style of
both komatiites and picrites as subaqueous volcaniclastic deposits
and pillow lavas suggests that no large lateral magma movements
through magma channels on the surface took place and hence both
the magma types were extruded from vents located relatively
close to each other. The age relationship between the komatiitic
and picritic rocks is also unclear, although we know that in
some areas the first volcanic rocks erupted on phyllites are
komatiitic in composition and picritic dykes have been observed
to cut komatiitic tuffs. Hence, at least some of the komatiitic
rocks are older than the picritic ones.
The field characteristics of the Lapland primitive volcanic rocks have been described in the literature (Saverikko, 1983, 1985; Barnes & Often, 1990; Davidsen, 1994) and therefore need only brief mention here. Unsorted, agglomerate-like deposits are typical, with the size of rounded or angular fragments ranging from 1 to 10 cm. These deposits can be tens of metres thick and they are interlayered with fine-grained laminar tuffs. Massive lavas are common, and pillow lavas and associated pillow breccias are locally abundant. They alternate with volcaniclastic units reaching hundreds of metres in thickness and several kilometres in length. Differentiated lava flows with peridotitic and pyroxenitic cumulates in the lower part and gabbroic rocks in the upper part are not common but are found in the Sattasvaara area. Very rare olivine spinifex-textured rocks are found in these layered flows (Räsänen, 1996).
Even though the komatiitic and picritic metavolcanic rocks have often delicately preserved their original textures, the primary silicate mineralogy is commonly replaced by regional metamorphic minerals, which formed under greenschist to epidote amphibolite facies. The komatiitic and picritic lavas are usually tremolite–chlorite rocks with various amounts of serpentine, and basaltic komatiites contain actinolitic amphibole with small amount of oligoclase plagioclase. The magmatic olivine is always altered to serpentine or chlorite, and pyroxene usually to amphibole, but relics of magmatic pyroxene are occasionally still preserved.
|
THE JEESIÖROVA AND PEURAMAA AREAS
|
|---|
The Jeesiörova area is located 25 km east of the village
of Kittilä. There exists a lens-like, NW-trending, 3 km
x 12 km area of dominantly komatiitic rocks partly overlain
by younger coarse-clastic quartzites and conglomerates of the
Kumpu Group (Fig.
2). We have drilled through the lower contact
of the komatiite-bearing formation. The drill-core shows that
the black schists at the base are overlain by
10 m of dolomite
and a similar interval of chert. The latter in turn is overlain
by tholeiitic lavas 22 m thick. These are followed by 50 m of
fragmental komatiites and >90 m of massive, amygdale-bearing
komatiites.
In general, the komatiitic formation is elsewhere poorly exposed, except in the central part of the area where several outcrops of komatiitic rocks occur in the form of amygdaloidal massive flows, pillow lavas and tuffs. Some of the pillows display variolitic structures. Ultramafic lavas are often porphyritic with abundant olivine phenocrysts. There also exist some outcrops of tholeiitic basalt lavas but they seem to make up only minor intercalations within the komatiitic sequence. Microscopic investigations have revealed that some of the outcrops are composed of ultramafic rock with a cumulate texture with poikilitic clinopyroxene, indicating that these represent lower parts of differentiated lava flows or sills. Also, gabbroic and doleritic rocks are encountered whose geochemistry (low LREE/HREE, for example) is consistent with derivation from a komatiitic parental magma. A few porphyritic lavas have been found that possess chemical compositions transitional between komatiite and picrite.
Although it has limited exposure, the Jeesiörova area was chosen for this study because of its mineralogical preservation, which is the best so far encountered in the komatiite belt in northern Finland, despite the closeness to a long fault zone, the ‘Sirkka line’, on the southern edge of the Kittilä greenstone complex (Fig. 2). In this area, olivine has been replaced by secondary minerals, but clinopyroxene occurs as a poikilitic intercumulus mineral in olivine cumulates, prismatic grains in gabbroic differentiates and as needle-like crystals in pillowed and massive lavas. In addition, ultramafic lavas contain euhedral chromite grains, which were liquidus phases together with olivine during magma emplacement. These chromites are often translucent and brown in colour, and contain small melt inclusions, demonstrating that they are very fresh.
The Peuramaa area is located 30 km NW of Jeesiörova. The lavas in this area cover a smaller area than those at Jeesiörova and are mostly basaltic (Fig. 2). These occur as massive lavas, tuffs and pillow lavas. Cogenetic mafic dykes also occur. Picritic rocks are interbedded with the mafic lavas and occur either as massive, porphyritic lavas or lapilli tuffs. As is shown below, geochemical data demonstrate that all the rocks in the Peuramaa area form a co-magmatic suite. The primary minerals are not so well preserved here as at Jeesiörova. All the primary silicates are destroyed, but fresh chromites have been observed in picritic rocks.
|
GEOCHEMISTRY
|
|---|
Whole-rock major and minor element abundances were determined
by X-ray fluorescence analysis (XRF) using a Phillips PW 1480
spectrometer at the Geological Survey of Finland and a Phillips
PW 1410 spectrometer at the University of Tasmania, Australia.
Rare earth elements and other trace elements were mostly analysed
by inductively coupled plasma mass spectrometry (ICP-MS) using
a Perkin–Elmer Sciex Elan 5000 instrument at the Geological
Survey of Finland and an HP 4500 instrument at the University
of Tasmania. Some samples were also analysed previously for
REE by instrumental neutron activation analysis (INAA) in connection
with the Lapland Volcanite Project (Lehtonen
et al., 1998
).
Representative whole-rock major and trace element analyses are
reported in Tables
1 and
2.
Regardless of the relatively well-preserved nature of the rocks at Jeesiörova and Peuramaa, we concur with Barnes & Often (1990) that whole-rock analyses of rocks in the komatiite belt are not good indicators of the original contents of mobile elements such as Na, K, Rb, Ba and Sr, and we therefore do not discuss these elements further. Instead, the coherent behaviour of many more immobile elements, including HFSE and REE, makes them valuable in characterizing the rocks and the processes involved in their genesis.
The total range of the measured MgO contents of komatiitic rocks varies from 31% to <10%. Because of the porphyritic nature of many of these rocks, as well as the presence of clear cumulate textures in some samples, enrichment of intratelluric olivine and chromite has evidently taken place. Massive lavas contain normally 22–24 wt % MgO whereas in pillow lavas MgO can be as high as 27 wt % (calculated on a volatile-free basis) but is most commonly between 14 and 18 wt %. Therefore, strictly speaking, many of the Jeesiörova magnesian lavas are komatiitic basalts rather than komatiites. It seems that the high MgO attained by volcaniclastic (19·2–24·8%) and pillow lavas (24·7–26·1%) in the Karasjok greenstone belt are not usual in our study area. Abundant (intratelluric) chromite grains often produce high Cr concentrations of 4000–5000 ppm. The Al2O3/TiO2 ratio in komatiitic rocks is typically around 13, thus falling well below the chondritic value of 22. This has been shown to result from a relatively high TiO2 content (Barnes & Often, 1990; Lehtonen et al., 1998), which in our sample suite is 0·6 wt % at an MgO content of 24 wt %, and 0·8 wt % at MgO of 15 wt % (Fig. 3). CaO/Al2O3 is most often close to 1·0 although there is some variation, most probably as a result of mobility of CaO.
Picritic tuffs at Peuramaa most often have MgO contents close to 17–18 wt %, which may be representative of the magma composition. Picritic lavas have a lower MgO content (11–15 wt %). Chromium and nickel behave coherently with MgO and do not distinguish komatiitic rocks from picritic. Compared with komatiitic rocks, the picrites have a much higher TiO2 at the same MgO content and consequently their Al2O3/TiO2 ratio is very low, in the range of 5·0–6·5. The Jeesiörova komatiites possess Ti/V close to the value observed in Archaean komatiites (16, Nesbitt & Sun, 1976), whereas the Peuramaa picrites have considerably higher values of 30. The difference is caused by the higher TiO2 in picrites, as the V content of these two magma types is comparable. The most striking difference between the komatiitic and picritic suites is the elevated HFSE concentrations in the latter; this feature is not restricted to titanium but occurs also for Zr and particularly Nb and Ta. Niobium contents in picritic rocks reach 10 ppm, whereas in komatiitic rocks they remain below 1 ppm. This is reflected in high Nb/Y ratios in picrites as shown in a Zr/Y vs Nb/Y plot in Fig. 4. Fitton et al. (1997) utilized this diagram to distinguish plume-related Icelandic basalts and picrites from N-MORB (normal mid-ocean ridge basalts). Picrites from our study area fall totally within the Icelandic field, whereas the komatiitic rocks plot at the lower edge of the N-MORB field.
The difference between komatiites and picrites is also pronounced when considering their REE characteristics. In Fig. 5, chondrite-normalized REE profiles are shown for various rock types from the Jeesiörova area. Komatiitic lavas have hump-shaped patterns with very low (La/Sm)CN of 0·2–0·5 in pillow lavas and slightly higher value of 0·6 in massive lavas, whereas (Gd/Yb)CN varies between 1·3 and 1·8 (Fig. 5a and b). Arndt et al. (1989) have observed similar REE patterns in Brazilian komatiites in the Crixas greenstone belt and suggested that they are at least in part due to mobility of HREE. We have analysed a few primary clinopyroxenes for REE, and their HREE patterns closely resemble the corresponding whole-rock patterns (Fig. 5f) and demonstrate that the REE characteristics of the rocks are magmatic features. Some komatiitic samples display negative europium anomalies, most probably as a result of secondary alteration (see Sun & Nesbitt, 1978). Doleritic and gabbroic rocks follow the same middle REE (MREE)-enriched pattern, indicating that they are fractionation products of a similar parental magma to that of the magnesian volcanic rocks (Fig. 5d). Figure 5c displays analyses of a tholeiitic lava and two ultramafic lavas, which have transitional composition between komatiite and picrite. Mafic tuffs are present as two chemically distinct varieties, one with an LREE-depleted REE pattern and the other with enrichment in LREE (Fig. 5e). Because of the presence of chromite phenocrysts, the latter type is peculiar in having exceptionally high Cr of 1500–1750 ppm for its MgO content of 7–8 wt %.
View larger version (38K):
[in this window]
[in a new window]
|
Fig. 5. Chondrite-normalized REE patterns for various rock types from the Jeesiörova area. In (f), one example of REE for a whole-rock and clinopyroxene pair is shown.
|
|
Chondrite-normalized REE patterns for rocks from the Peuramaa area are presented in Fig. 6. These include analyses of mafic pillow lavas, massive lavas, tuffs and dykes, and picritic massive lavas and tuffs. All display similar sloping, LREE-enriched patterns with a total range of (La/Sm)CN of 1·4–1·7 and (La/Yb)CN of 4·4–5·4. This feature coupled with similar ratios of other incompatible elements, such as Ti/V, Zr/Y and La/Nb, suggests that the rocks at Peuramaa form a co-magmatic suite produced by fractional crystallization from a similar parental magma.
To emphasize the huge spread in the abundances of LREE, a combination of selected chondrite-normalized REE patterns of komatiitic to picritic rocks from the Jeesiörova and Peuramaa areas is displayed in Fig. 7. In the whole suite, La concentrations vary from <2 times chondritic to 40 times chondritic and all the rocks show fractionated HREE patterns. Along with increasing LREE/MREE, there is a concomitant increase in HFSE concentrations.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 7. Representative chondrite-normalized REE patterns for rocks of the komatiite–picrite association. Shown are MgO contents (wt %) of the samples calculated as volatile free.
|
|
|
CLASSIFICATION OF HIGH-MgO VOLCANIC ROCKS
|
|---|
On the basis of the above discussion it is evident that there
is a relatively large variation in the composition of magnesian
volcanic rocks from both the Archaean and Proterozoic greenstone
belts, which has led to them being described using terms such
as Ti-rich komatiites, iron-rich komatiites, Al-depleted komatiites
or Al-undepleted komatiites (Green & Schulz, 1977
; Nesbitt
et al., 1979
; Barnes & Often, 1990
). On the other hand,
younger magnesian volcanic rocks are commonly classified as
picrites, with good examples being those from Hawaii, Iceland,
Baffin Bay, the Deccan Trap and the Karoo. A closer chemical
comparison reveals that there is an overlap in composition of
rocks termed picrites and komatiitic basalts or komatiites.
A characteristic feature of many Phanerozoic picritic magmas
is an elevated TiO
2 content compared with Archaean komatiites,
but no well-established definitions exist in the literature
that could be used to classify a magnesian volcanic rock either
as picrite or komatiite. Textural criteria are not unequivocal,
as spinifex textures have also been recognized in picritic rocks
(Hanski & Smolkin, 1995
).
The revised IUGS classification for high-Mg and picritic rocks recommends that volcanic rocks with MgO between 12 and 18 wt %, SiO2 between 30 and 52 wt % and (Na2O + K2O) < 3 wt % are called picrites (Le Bas, 2000). This scheme has some weaknesses when applied to our study area or to Precambrian greenstone belts in general: it is based on mobile elements (Si, Na, K) and totally eliminates the terms komatiitic basalt or basaltic komatiite. Furthermore, it cannot distinguish between fractionated products (MgO < 18 wt %) of magmas ranging from various types of komatiites to meimechite.
As the magnesian rock suites have often undergone olivine fractionation or olivine accumulation, a strict value of TiO2, for example, is not a good criterion. Nor is the Al2O3/TiO2 ratio, because both these components vary. We have developed a new [Al2O3] vs [TiO2] diagram (Fig. 8a) in which [Al2O3] and [TiO2] are contents of these components in mole proportions (normalized to unity) as projected from the olivine composition. They are calculated using the following equations: [Al2O3] = Al2O3/(2/3 -MgO -FeO), [TiO2] = TiO2/(2/3 -MgO -FeO). For simplicity, all iron is taken as ferrous. The rationale of the projection scheme has been graphically explained by Hanski (1992). All rock or liquid compositions that lie on an olivine control line have constant [Al2O3] and [TiO2] values and therefore a fixed position on the diagram. The diagram exhibits fields for boninites, Al-undepleted komatiites, Al-depleted komatiites, Ti-enriched komatiites, Ti-enriched and Al-depleted komatiites, picrites, Al-depleted picrites and meimechites. Shown in Fig. 8b are lines with constant Al2O3/TiO2 ratios as expressed in wt %. The boundaries between the separate fields in Fig. 8 could be drawn in many different ways. We have chosen the boundaries nonparallel to the axes of the diagram mainly for two reasons. First, komatiites and boninites are most effectively distinguished by a line projecting to the origin and therefore having a constant Al2O3/TiO2 ratio. The lines between the overlying fields separating komatiites, Ti-enriched komatiites and picrites were drawn parallel to the komatiite–boninite boundary. Second, the reason for an oblique instead of vertical boundary between the Al-depleted and Al-undepleted rock types is the observation that many natural compositional trends of picritic rocks exhibit a slight decrease in [Al2O3] with increasing [TiO2]. As a result of the method of construction of the diagram, there is some loss of information, as the diagram does not distinguish between komatiites and komatiitic basalts, for example. This has to be done on the basis of MgO content. Thus the fields of the diagram represent rock series. Moreover, additional attributes can be used for rocks with some additional peculiarities, such as a high iron content (ferropicrites, ferrokomatiites).
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 8. (a) [Al2O3] vs [TiO2] diagram for classifying high-MgO volcanic rocks. [Al2O3] and [TiO2] are Al2O3 and TiO2 projected from olivine composition and are calculated in mole proportions (normalized to unity) using the following equations: [Al2O3] = Al2O3/(2/3 -MgO -FeO) and [TiO2] = TiO2/(2/3 -MgO -FeO) (see Hanski, 1992). (b) Rocks of the komatiite–picrite association and komatiites from the Barberton Mountain Land, Munro Township and Gorgona Island plotted in the [Al2O3] vs [TiO2] diagram. List of literature references available from the first author upon request.
|
|
In Fig. 8b, analyses of the Kittilä komatiite–picrite association are compared with literature data on komatiites from Barberton, Munro Township and Gorgona Island in an [Al2O3] vs [TiO2] diagram. The Al-undepleted Munro Township and Gorgona Island komatiites straddle the line with Al2O3/TiO2 of 20. The majority of the Barberton komatiites, typical examples of Al-depleted komatiites, possess variable but very low [Al2O3] values compared with the Munro Township or Gorgona komatiites. Several of the Barberton samples, however, overlap with the Al-undepleted komatiites in terms of the Al2O3/TiO2 value. Some of them are also relatively high in [TiO2]. Compared with the Munro Township komatiites, the Jeesiörova komatiites display similar [Al2O3] values but are in most cases distinctly higher in [TiO2] and plot within the field of Ti-enriched komatiites. The Peuramaa picrites in turn have still higher [TiO2] and accordingly plot within the picrite field.
In Fig. 9, the Lapland komatiite–picrite association is compared with picrites from the North Atlantic volcanic province and Hawaii. The trend formed by the data on the Baffin Bay and Greenland picrites demonstrates the non-vertical slope of many natural suites. Barnes & Often (1990) pointed out the similarity between the Karasjok komatiite and Baffin Bay picrites. Figure 9 shows, however, that these two magmas differ somewhat with respect to their Al2O3 contents. The Peuramaa volcanic rocks mimic Hawaiian picrites not only in their Al–Ti systematics but also in their overall chemical composition including trace elements and therefore it is more appropriate to call them picrites than komatiitic basalts or Ti-rich komatiitic basalts.
View larger version (40K):
[in this window]
[in a new window]
|
Fig. 9. Rocks of the komatiite–picrite association compared with picrites from Baffin Bay, Greenland and Iceland in an [Al2O3] vs [TiO2] diagram. Also shown is a field for Hawaiian picrites. List of literature references available from the first author upon request.
|
|
|
NEODYMIUM ISOTOPE DATA AND THE AGE OF THE ROCKS
|
|---|
For isotopic analyses, clinopyroxene separates were made from
seven samples from Jeesiörova using standard separation
methods following purification by hand-picking. Six of the samples
represent komatiitic volcanic rocks and one of them is a doleritic
dyke. In addition to these mineral separates and their whole-rock
samples, whole-rock Nd isotopic analyses were carried out for
two komatiitic lavas and one gabbroic sample. Because no primary
silicates are preserved in the picritic suite at Peuramaa, we
had to resort to whole-rock isotope analyses, which were carried
out on two picrite samples. Nd isotopic compositions were analysed
on a VG SECTOR 54 mass spectrometer at the Geological Survey
of Finland. Measurements of
143Nd/
144Nd for the La Jolla standard
during 1995–1999 have yielded a ratio of 0·511851
± 0·000006 (SD,
n = 48). The analytical procedure
has been described by Peltonen
et al. (1996)
. The results are
shown in Table
3.
The pyroxene separates have relatively large range in 147Sm/144Nd (0·284–0·428) probably owing to their different modes of occurrence (poikilitic to prismatic) and likely different cooling conditions in which they crystallized. The 147Sm/144Nd values are always higher than those of the corresponding whole rocks (0·231–0·285). On the other hand, Sm and Nd concentrations in pyroxenes can be either lower or slightly higher than in the whole rocks. Picritic rocks display low 147Sm/144Nd (0·152, 0·156). As a whole, the dataset thus possesses a large spread in 147Sm/144Nd. When plotted in a 147Sm/144Nd vs 143Nd/144Nd diagram, the analyses form a linear trend (Fig. 10). There is, however, some scatter [mean square weighted deviation (MSWD) = 16] indicating heterogeneity in the initial ratios. Regression of the whole dataset of 25 analyses yields an age of 2081 ± 29 Ma, which is identical to that published by Krill et al. (1985) for the Karasjok komatiites (2085 ± 85 Ma).
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 10. 143Nd/144Nd vs 147Sm/144Nd isochron plot for komatiitic whole rocks and clinopyroxene separates from Jeesiörova and picritic whole rocks from Peuramaa. A 2060 Ma isochron with Nd of +4 is shown for reference.
|
|
There are several examples in the literature of Sm–Nd studies on mafic to ultramafic rocks that have resulted in erroneous isochron ages based on whole-rock samples (Cattel et al., 1984; Hegner et al., 1984; Chauvel et al., 1985, 1993; Gruau et al., 1987, 1990, 1992; Claoue-Long et al., 1988; Jahn & Ernst, 1990; Tourpin et al., 1991). The reasons include heterogeneities in the source, crustal contamination or mobility of REE as a result of post-crystallization alteration, particularly carbonatization. One of the aims of our study was to assess whether the age obtained by Krill et al. (1985) is realistic. They used whole rocks having both LREE-depleted komatiitic and LREE-enriched picritic compositions. Regressed separately, their analytical data for komatiites and picrites yield very different ages: 2180 Ma for the former and <1700 Ma for the latter. The age of 2085 Ma obtained using the whole dataset is valid only if the komatiitic and picritic rocks had a similar initial Nd value. Given the large difference in REE chemistry between komatiites and picrites, the Nd isotopic heterogeneity in the corresponding source regions is a possibility that evidently increases the uncertainty of the obtained whole-rock isochron age for the Karasjok metavolcanic rocks. Simple calculations show that if the Nd values for the individual komatiite samples were between +2·8 and +4·2 and those for picrite samples between +4·3 and +4·9, both datasets would plot on separate but parallel isochrons with an age of 2200 Ma. Thus a relatively small heterogeneity in the source region would produce easily an age error of >100 my.
From the above discussion it follows that we regard it as more reliable to try to calculate the age using primary minerals and their whole rocks. The seven whole-rock–clinopyroxene pairs analysed from komatiites at Jeesiörova provide Sm–Nd ages ranging from 2019 ± 44 to 2118 ± 86 Ma with a weighted average of 2056 ± 25 Ma (1 SD). On the basis of this age, the initial Nd values range from +2 to +4 for individual samples, with the most LREE-depleted komatiites having the highest Nd. Using the same age, we obtained initial Nd value of +3·5 and + 3·7 for the two picritic samples from Peuramaa. The high Nd values of +4·0 and +3·7 obtained for the gabbroic and doleritic samples, respectively, corroborate the conclusion reached on the basis of geochemistry on the genetic relationship between these rocks and the komatiitic lavas.
These new isotopic results are compatible with the data reported by Krill et al. (1985) from Finnmark. The isotopically most depleted samples from both the picritic and komatiitic rocks display similar high Nd values of +4. As a whole, there is thus no correlation between the Nd isotopic composition and the major element composition of the volcanic rocks. However, when the whole-rock analyses of the komatiitic sub-population are considered, there seems to be a positive correlation between Nd and 147Sm/144Nd (Fig. 11). This means that Nd is the highest for the most LREE-depleted samples. The decrease in Nd with an increase in LREE/HREE may be a mantle source character or a consequence of crustal contamination.
The Sm–Nd isotopes show that the age obtained by Krill et al. (1985) is close to the new results we obtained using whole rocks and mineral separates from a single area. The reason for this is that komatiites and picrites of the Kittilä–Karasjok belt were derived from an isotopically homogeneous mantle source at least in terms of neodymium isotopes. Both Ti-enriched komatiites and picrites in Lapland were produced from a mantle source with a long-term depletion in LREE relative to MREE.
|
DISCUSSION
|
|---|
Neodymium isotopic evolution
One way to explain the observed variation of
Nd in the komatiitic
rocks is crustal contamination. Two possible sources of crustal
contamination can be envisaged. The komatiites may have digested
pelitic sediments of the Savukoski Group on which the komatiite–picrite
unit of the same group is lying or they may have interacted
with continental rocks deeper in the crust during their ascent
or in an intermediate magma chamber. We have previously analysed
pelitic metasediments for their Nd isotopic and trace element
compositions from
60 km east of Jeesiörova (Hanski
et al.,
1997
) and assume that the pelitic rocks underlying the komatiite
suite in the present study area possess similar elemental and
isotopic compositions. Mica schists and black schists have high
REE concentrations, are LREE enriched and possess low
Nd(2060
Ma) values ranging between -3·7 and -6·7. Even
minor amounts of contamination with such crustal material should
be readily detectable in terms of isotopic and trace element
variations in the komatiitic rocks. We use the isotopically
and geochemically most depleted komatiite sample as the uncontaminated
parental magma. Results of the mixing calculations demonstrate
that the observed range in
Nd and
147Sm/
144Nd can be accounted
for by assimilation of up to 4–6% of pelitic material.
In Fig. 12, calculated Nd isotopic evolutions of two samples, one komatiitic and the other picritic, are shown together with the corresponding REE profiles of these samples. It is seen that the evolution lines cross each other at a value that is very close to that of the proposed contemporaneous upper mantle as defined by Nägler & Kramers (1998). The contrast between the positive Nd values, indicating a long-term LREE depletion in the source, and the observed LREE-enriched character of the picritic rocks is similar to the situation with the Hawaiian tholeiitic picrites and alkali basalts and Pechenga ferropicrites, for example (Stille et al., 1986; Hanski et al., 1990). The possible explanations for these results are that such melts were derived by very low degrees of partial melting or by dynamic melting processes so that parent–daughter abundance ratios were changed markedly during partial melting or that the source experienced enrichment in incompatible elements by a ‘metasomatizing’ process, which must have occurred relatively shortly before partial melting to avoid the influence on the initial isotopic signature. Utilizing the most LREE-depleted komatiites and typical LREE-enriched picrites as representing the relative proportions of LREE in the mantle source before and after the enrichment event, it can be shown that every 50 my extension of the enrichment event back in time from the melting event at 2060 Ma would decrease the initial Nd(2060 Ma) value of the picritic rocks by one unit. The result would not be much different if allowance is made for some Sm–Nd fractionation during partial melting.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 12. Neodymium isotopic evolution of typical komatiitic and picritic samples with high and low f Sm/Nd values, respectively, compared with the evolution of depleted model upper mantle (DM) as defined by Nägler & Kramers (1998). Also shown are the chondrite-normalized REE profiles of these samples and an evolution line for a mantle source with f Sm/Nd = 0·85. f Sm/Nd is an enrichment factor relative to the chondritic standard reservoir (CHUR) and is calculated as (147Sm/144Nd)m/(147Sm/144Nd)CHUR -1.
|
|
Low LREE/MREE ratios in komatiites are in harmony with their positive Nd values. It is generally presumed that in the case of komatiites the Sm–Nd fractionation during partial melting is negligible. The most LREE-depleted komatiitic lava analysed for Sm–Nd has 147Sm/144Nd ratio of 0·275 (equivalent to f = 0·401 in Fig. 12). The mantle source with such isotopic characteristics would possess relatively low Nd values before 2·1 Ga, clearly off the depleted mantle source (DM). The Lapland komatiites have hump-like chondrite-normalized REE patterns with (Gd/Yb)CN varying between 1·3 and 1·8. Specifically, if the source had chondritic relative abundances of HREE before melting, then the heaviest REE were retained more preferably in the source compared with the lighter ones, most probably as a result of the influence of residual garnet. Model calculations using batch melting of garnet peridotite demonstrate that if the source was unfractionated with respect to HREE and these elements were fractionated during melting to the extent observed in the Lapland komatiites, and LREE could not escape from mutual fractionation during melting, then the mantle source would have had even higher Sm/Nd than that measured for the komatiites. In Fig. 12, a Nd isotopic evolution line is also shown for the source of the komatiite sample assuming that this source had originally chondritic Gd/Yb and produced the observed REE profile of the melt with (Gd/Yb)CN of 1·63. This corresponds to Sm–Nd fractionation from f Sm/Nd = 0·85 to f Sm/Nd = 0·40 upon melting. Provided that the depleted model mantle with positive Nd is the principal source for primitive magmas, the LREE depletion seen in the most depleted komatiites should have take place shortly before or during the melting event that produced the komatiites. A similar conclusion was reached above concerning a potential enrichment event related to the generation of the picrites, and hence there seems to have been a complicated fractionation process in the source mantle shortly before or at 2·06 Ga.
The relation between picritic and komatiitic magmas
The association of LREE-depleted and LREE-enriched Archaean komatiites has been explained by a combination of crystal fractionation and crustal contamination (Cattell, 1987). Barnes & Often (1990) noted the Ti- and LREE-enriched nature and heterogeneity of the Norwegian Karasjok komatiites and examined several potential explanations for their unusual geochemistry compared with typical Archaean komatiites. They concluded that alteration, upper-crustal contamination or low-pressure crystal fractionation cannot be invoked in this connection. It suffices to note here that the same conclusion applies to the relation between komatiitic and picritic rocks of this study, and the reader is referred to Barnes & Often (1990) for detailed discussion on the processes mentioned above.
We are confronted with a difficult question: How much of the observed heterogeneity is related to fractionation processes during partial melting and how much to compositional differences in the mantle sources? The implication of the Nd isotope studies is that if there was significant variation in the mutual ratios of REE in the mantle, this was created very soon before the generation of the komatiite–picrite association. This association has two particular features, the large variation in LREE/MREE within the rock suite and concomitant depletion in both LREE and HREE in the komatiitic rocks, which any melting models should be able to explain if these fractionations are attributed to partial melting effects.
Various melting models have been called upon to account for chemical variations in oceanic basalts, particularly as a consequence of findings of ultradepleted melt inclusions in near-liquidus minerals. For example, Gurenko & Chaussidon (1995) have successfully replicated the observed range of incompatible elements in depleted and enriched melt inclusions in olivine from Icelandic tholeiites by using a critical (dynamic) melting model. This is an intermediate process between pure fractional and batch melting and assumes that a critical amount of melt is retained in the mantle residue. The same model was utilized by Arndt et al. (1997) to explain moderate to extreme trace element depletion in the Gorgona Island komatiitic lavas. Fractional melting depletes the residue in incompatible elements efficiently, and consequently incremental melts may differ considerably in composition from each other; this difference is also reflected in the composition of the pooled melts. Dynamic melting thus has a potential for creating both LREE-enriched and LREE-depleted melts from the same mantle source and is therefore an attractive hypothesis in our case.
To simulate polybaric, near-fractional melting we have carried out model calculations using incremental, non-modal batch melting (Shaw, 1970) with small steps of 1 kbar pressure release, as this approach allows changing melting modes to be taken into account easily. We consider decompressional melting of upwelling mantle with two potential temperatures leading to different pressures of solidus intersection, 28 and 40 kbar, and accordingly initiation of melting within the spinel and garnet stability fields, respectively. We performed calculations using six starting mantle compositions with flat or variously LREE- and/or HREE-depleted REE patterns, as shown in Fig. 13. Other input parameters in the model are given in the caption to Fig. 13.
Pooled melt compositions produced from various sources are compared with measured rock compositions in Fig. 13 in terms of primitive mantle normalized La/Nd and Nd/Yb. Melting in the spinel peridotite facies produces curved trends approximately parallel to the komatiite–picrite array (Fig. 13a), but only those mantle sources having high Nd/Yb already before melting (model compositions 4 and 6) are able to create melt compositions with HREE fractionation similar to that observed in the komatiites and picrites. It is immediately evident that the whole observed range of elemental ratios cannot be achieved by dynamic melting to any reasonable degree in these conditions. It is also questionable whether low degrees of melting can produce high-MgO liquids at the pressures of the spinel lherzolite facies.
As all the rocks in the study area display high Gd/Yb (i.e. a strong ‘garnet signature’), partial melting in the garnet peridotite facies could provide a relevant explanation. Irrespective to the source composition, the calculated decompression melting paths produced by low degrees of melting approximately parallel the komatiite–picrite array but are shifted to much higher (Nd/Yb)PM values (Fig. 13b). On the other hand, high-degree melts tend to form nearly horizontal trends with little variation in (La/Nd)PM crossing the komatiite–picrite array. The same problem occurs for melts produced by melting of garnet lherzolite and those from spinel lherzolite, namely the limited variation of La/Nd compared with the measured dataset. Computations demonstrate that mantle sources with low (La/Nd)PM (<1) obviously fail to recover REE characteristics of the picritic rocks. Instead, those characteristics are successfully reproduced by LREE-enriched model composition 2 in Fig. 13.
Figure 13a and b assumes continuous mixing of melt increments after their extraction from the source. Much larger compositional variation in pooled melts derived from a single mantle source can be attained if the early formed melts are isolated from subsequent melt increments in some stage of the melting process, either because of tapping of the magma chamber, in which melt aggregation takes place, or pooling of the later melt increments in a separate chamber. Figure 14 illustrates the effect of isolation of the first formed pooled melts on the subsequent compositional evolution of the melts. Two starting materials are used, an enriched source (EM, Fig. 14a) and a depleted source (DM, Fig. 14b), with the latter chosen to be in accord with a long-term LREE-depleted character of the source as indicated by the obtained Nd(2060 Ma) values. It is interesting to note that in this kind of melting process, both strongly LREE-enriched and LREE-depleted melts as well as melts with pronounced MREE humps can be generated from both the EM and DM sources. However, only the EM source allows melts to be generated that correspond in their REE characteristics to those of the Peuramaa picrites. The Jeesiörova komatiites can instead represent moderate- to high-degree, second-stage pooled melts from both sources.
TOP
ABSTRACT
INTRODUCTION
) rocks from Lapland plotted in a Nb/Y vs Zr/Y diagram showing the fields for Icelandic basalts and N-MORB after Fitton et al. (1997)
