Journal of Petrology Advance Access originally published online on December 20, 2005
Journal of Petrology 2006 47(4):773-789; doi:10.1093/petrology/egi093
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Origin of Paleoproterozoic Komatiites at Jeesiörova, Kittilä Greenstone Complex, Finnish Lapland
1 ISOTOPE GEOCHEMISTRY LABORATORY, DEPARTMENT OF GEOLOGY, UNIVERSITY OF MARYLAND, COLLEGE PARK, MD 20742, USA
2 DEPARTMENT OF GEOSCIENCES, UNIVERSITY OF OULU, P.O. BOX 3000, FIN-90014 UNIVERSITY OF OULU, FINLAND
3 INSTITUTE FOR ROCK MAGNETISM, 291 SHEPHERD LABS, 100 UNION STREET S.E., MINNEAPOLIS, MN 55455-0128, USA
RECEIVED NOVEMBER 18, 2003; ACCEPTED NOVEMBER 16, 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGICAL SETTING AND ROCK...SAMPLESANALYTICAL TECHNIQUESRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Komatiites from the
2 Ga Jeesiörova area in Finnish Lapland have subchondritic Al2O3/TiO2 ratios like those in Al-depleted komatiites from Barberton, South Africa. They are distinct in that their Al abundances are higher than those of the Al-depleted rocks and similar to levels in Al-undepleted komatiites. Moderately incompatible elements such as Ti, Zr, Eu, and Gd are enriched. Neither majorite fractionation nor hydrous melting in a supra-subduction zone setting could have produced these komatiites. Their high concentrations of moderately incompatible elements may have resulted from contamination of their parental melt through interaction with metasomatic assemblages in the lithospheric mantle or enrichment of their mantle source in basaltic melt components. Re–Os isotope data for chromite from the Jeesiörova rocks yield an average initial 187Os/188Os of 0·1131 ± 0·0006 (2
), 
Os(I) = 0·1 ± 0·5. These data, coupled with an initial 
Nd of +4, indicate that melt parental to the komatiites interacted minimally with ancient lithospheric mantle. If their mantle source was enriched in a basaltic component, the combined Os–Nd isotopic data limit the enrichment process to within 200 Myr prior to the formation of the komatiites. Their Os–Nd isotopic composition is consistent with derivation from the contemporaneous convecting upper mantle. KEY WORDS: Finnish Lapland; Jeesiörova; komatiites; mantle geochemistry; petrogenesis; redox state; Re/Os isotopes; Ti enrichment
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND ROCK...SAMPLESANALYTICAL TECHNIQUESRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Komatiites have been divided into two broad chemical types, Al-undepleted (with near-chondritic Al2O3/TiO2 ratios of
22) and Al-depleted (subchondritic Al2O3/TiO2 ratios of typically 11), e.g. Nesbitt et al. (1979)
. Komatiites from Alexo and Pyke Hill in the Abitibi greenstone belt, Canada, are typical examples of the Al-undepleted type (also termed ‘Munro-type’; Pyke et al., 1973; Arndt, 1986), whereas those from the Komati and Mendon Formations in the Barberton Mountainland, South Africa, are examples of the Al-depleted type (also known as the ‘Barberton-type’; Nesbitt & Sun, 1976). The komatiitic rocks from the Jeesiörova area in the Central Lapland Greenstone Belt, northern Finland (Fig. 1) are distinct from these two types of komatiite in that they have subchondritic Al2O3/TiO2 ratios (typically 10–13), similar to those in Al-depleted rocks, yet compared with the latter they have significantly higher aluminum contents at given MgO levels (Hanski et al., 2001; Fig. 2a). However, the aluminum contents in the Jeesiörova rocks are similar to those of Al-undepleted komatiites at given MgO contents. In addition, the Jeesiörova rocks have significantly higher TiO2 contents at given MgO and Al2O3 contents than either Al-depleted or Al-undepleted komatiites from elsewhere (Fig. 2b). This distinctive combination of bulk-rock chemical characteristics for the Jeesiörova rocks prompted construction of a new classification scheme for komatiites using olivine-projected molecular Al2O3–TiO2 relations (Hanski et al., 2001). The Jeesiörova rocks, according to this scheme, plot in a distinct field of ‘Ti-enriched’ komatiites. Aluminum-depleted, Al-undepleted and/or Ti-enriched rock types can coexist in a single greenstone belt and, in some cases, within a single volcanic assemblage (e.g. Kidd–Munro assemblage in the Abitibi greenstone belt, Canada: Sproule et al., 2002). It is important to determine, therefore, whether the apparent Ti enrichment is a feature of the mantle source, and if so, what the implications of Ti enrichment are for the nature of the source and the petrogenetic processes involved in komatiite magma genesis.
|
|
Here we compare and contrast the petrological, geochemical and isotopic characteristics of the Jeesiörova komatiites with those of the well-studied Munro and Barberton type komatiites in order to assess the origin of the Finnish rocks. We present electron microprobe data for chromites from the Finnish suite of rocks, and use them in conjunction with the Fe3+/
Fe ratios of selected chromite separates determined by Mössbauer spectroscopy in order to constrain the redox state of the parental magmas. Additionally, we present high-precision Re–Os isotope data for chromites in order to determine the initial Os isotopic composition of the mantle source of this chemically distinct type of komatiite. The initial Os isotopic composition, combined with previous estimates of the initial Nd isotopic composition of the same suite of rocks (Hanski et al., 2001), is used to evaluate the possible roles of (1) contamination by crustal materials or hypothesized metasomatized lithospheric mantle mineral assemblages rich in Ti and other moderately incompatible elements, and (2) recycling of oceanic crust in the generation of these rocks. Finally, we use previously published major and trace element data for the same suite of rocks (Hanski et al., 2001) to evaluate possible petrogenetic models. |
GEOLOGICAL SETTING AND ROCK TYPES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND ROCK...SAMPLESANALYTICAL TECHNIQUESRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Samples were collected from the Jeesiörova area, located in the southern part of the Kittilä Greenstone Complex, Central Lapland (Fig. 1a). Detailed discussions of the stratigraphic subdivisions, major rock types of the Kittilä Greenstone Complex, and their petrological and geochemical characteristics have been given by Lehtonen et al. (1998)
and Hanski et al. (2001). In brief, the Kittilä Greenstone Complex is part of an 400 km long belt, which runs through Finnish Lapland and extends to the Karasjok greenstone belt in northern Norway (Barnes & Often, 1990). The entire belt contains abundant, highly magnesian volcanic rocks that vary in chemical composition from light rare earth element (LREE)-depleted komatiites to LREE-enriched picrites (Hanski et al., 2001). A characteristic feature of the belt is the presence of ultramafic volcaniclastic rocks that vary from agglomerate-like deposits to fine-grained, laminated tuffs (Saverikko, 1985; Barnes & Often, 1990). Komatiitic rocks also occur as massive lavas, pillow lavas, and associated pillow breccias.
The Jeesiörova komatiites occur in a zone that is 15 km long and 4 km wide (Fig. 1a). They form part of the Savukoski Group together with underlying pelitic metasedimentary rocks. A fault zone, called the ‘Sirkka line’, separates the Savukoski Group rocks from younger (2·0 Ga) basaltic metavolcanic rocks and metasedimentary rocks of the Kittilä Group that crop out on the NE side of the Jeesiörova area. The clastic metasedimentary rocks of the Sodankylä Group are the oldest (2·3 Ga) rocks in the area, whereas the youngest supracrustal rocks are represented by the quartzites and conglomerates of the 1·88 Ga Kumpu Group, which lie unconformably on the rocks of the Savukoski and Kittilä Groups.
Drill cores through the lower contact of the komatiite-bearing lava succession in the Jeesiörova area show that the volcanic sequence was deposited on a 20 m thick unit of cherts and carbonate rocks that are underlain by black schists. The volcanism started with 20 m of tholeiitic lavas, followed by volcaniclastic komatiites (50 m) and finally by massive komatiitic lavas (>100 m). The total thickness of the volcanic rocks is difficult to estimate but may reach several hundreds of meters.
The bedrock in this part of the Jeesiörova area (Fig. 1b) is dominated by olivine-phyric komatiitic rocks. As a result of limited outcrops, individual lava flows cannot be delineated. There are, however, zones of cumulate-textured komatiites that probably represent the lower parts of thick lava flows. Komatiitic rocks also occur as pillow lavas with the pillow size varying from 0·2 to 1 m. Various mafic rocks are interbedded with the ultramafic rocks (Fig. 1b). The mafic rocks include LREE-depleted tuffs and dolerites. There are also mafic amygdaloidal lava flows, marked as tholeiites in Fig. 1b, which are moderately LREE enriched and probably not genetically related to the associated komatiites.
|
SAMPLES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND ROCK...SAMPLESANALYTICAL TECHNIQUESRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The samples that form the basis of this study are all komatiites collected from outcrops (sample locations are shown in Fig. 1b), and are well characterized in terms of their major and trace element, and Sm–Nd isotope compositions (Hanski et al., 2001
). Whole-rock samples were first crushed into millimeter-sized pieces. The freshest unaltered pieces were ground in a steel pan. Different aliquots of the same whole-rock powder were used for major and trace element and isotopic analysis of each sample.
Mineral separates (chromite, clinopyroxene, and sulfide) were obtained using standard crushing, milling, concentration table, and heavy liquid techniques. Separation of large quantities of chromite was possible because of their relatively large grain size and well-preserved nature (low magnetic susceptibility). After removing the most magnetic fraction with a hand magnet, the chromite separates were divided into different fractions, depending on the degree of alteration (presence of magnetite rim) and using different current settings in a FrantzTM magnetic separator. Likewise, for sulfide separates, we obtained a fraction dominated by pentlandite, whereas the other fractions were mixtures containing pyrrhotite, pentlandite, and chalcopyrite.
|
ANALYTICAL TECHNIQUES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND ROCK...SAMPLESANALYTICAL TECHNIQUESRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Details of the whole-rock chemical compositions of the komatiites have been reported elsewhere (Hanski et al., 2001
). Whole-rock X-ray fluorescence (XRF) data for major and minor elements were obtained using a Phillip PW 1480 spectrometer at the Geological Survey of Finland (GSF) and a Phillips PW 1410 spectrometer at the University of Tasmania (UT). The concentrations of REE and other trace elements were determined by inductively coupled plasma mass spectrometry (ICP-MS) using a Perkin-Elmer Sciex Elan 5000 instrument at the GSF and an HP 4500 instrument at the UT. The whole-rock major and trace element data of Hanski et al. (2001) are supplemented by two additional samples (10C-PPR and 314-LVP). The complete dataset is included in Electronic Appendix 1 (http://www.petrology.oxfordjournals.org).
Electron microprobe
The chemical compositions of chromites from selected samples were determined using a Cameca Camebax SX50 microprobe at the GSF (Table 1). The analytical conditions included an accelerating potential of 25 kV, a sample current of 47 nA, and a beam diameter of 1 µm.
|
Mössbauer analyses
Mössbauer spectra were collected for five selected chromite separates at the Institute for Rock Magnetism at the University of Minesota, Minneapolis. The chromite separates were crushed and
20–30 mg of powder was dispersed in powdered sugar in a 12 mm diameter sample holder. The spectra were collected at room temperature using a constant acceleration transmission spectrometer with a 57Co in Rh matrix source and an activity of 5 mC. The spectra were fitted with four doublets and two sextets using a least-squares fitting routine. The hyperfine parameters as well as peak area and width were allowed to float with no bounds enforced. Three of the samples, 12-PPR, 10C-PPR and 6.2-EJH, have significant amounts of a magnetically ordered phase with the hyperfine parameters consistent with magnetite. Of the four doublets that were fitted for all of the samples, three can be assigned to Fe2+ with isomer shifts above 0·63 mm/s and quadrupole splitting above 1·2 mm/s, and one can be assigned to Fe3+ with isomer shift ranging from 0·16 to 0·45 mm/s and quadrupole splitting between 0·53 and 0·86 mm/s. Although some ambiguity exists because of the three overlapping Fe2+ doublets, the overall area assignment to Fe2+ is not affected excepting only the relation between the Fe2+ doublets. The Fe2+/Fe3+ ratios were calculated from the paramagnetic doublets as defined above. In order to calculate the Fe3+/Fe ratios of paramagnetic iron, we excluded the iron in the magnetite. Assuming stoichiometry, magnetite can be expressed as Fe3+[Fe2+Fe3+]O4 with the tetrahedral site containing only Fe3+ and the octahedral site containing equal contents of Fe2+ and Fe3+. The ferric/ferrous ratios of selected chromite separates obtained through analyses of Mössbauer spectra are included in Table 1.
Re–Os isotope analyses
The chemical separation techniques for Re–Os analysis employed in this study follow previously published work (Shirey & Walker, 1995; Cohen & Waters, 1996) and have been described in detail elsewhere (Gangopadhyay & Walker, 2003; Gangopadhyay, 2004; Gangopadhyay et al., 2005). Approximately 3 g of whole-rock powder was dissolved in reverse aqua regia, and frozen and equilibrated with spikes in sealed PyrexTM Carius tubes. The tubes containing whole-rock powder were heated at 240°C for at least 24 h, whereas those with mineral separates (chromite, clinopyroxene and sulfide) were heated for at least 48 h to facilitate complete digestion. Osmium was separated by solvent extraction into carbon tetrachloride and finally transferred into concentrated HBr (Cohen & Waters, 1996). The final purification for Os was accomplished via micro-distillation. Rhenium was recovered from aqua regia via anion exchange column chemistry (Morgan & Walker, 1989).
The isotopic compositions of Re and Os were obtained using negative thermal ionization mass spectrometry (Creaser et al., 1991; Völkening et al., 1991). The mass spectrometric procedures followed in this study have been discussed by Walker et al. (1994) and Morgan et al. (1995). Samples with high Os abundances (whole-rock, chromite and sulfide) were analyzed with Faraday cups in the static mode (using a Sector 54 mass spectrometer), whereas those with relatively low abundances of Os (e.g. clinopyroxene) and/or Re (e.g. chromite) were analyzed with an electron multiplier using a 12'' NBS mass spectrometer.
The total ranges in analytical blanks for Re and Os were 1·2–6·4 pg (excluding one outlier with 12·7 pg of Re) and 1·3–6·9 pg (n = 6), respectively, and are negligible compared with their concentrations in most of the whole-rock samples. The blank contribution, however, was substantial (typically between 2 and 10%) for some of the Re analyses of chromites. The isotopic compositions of the blanks were natural for Re and 187Os/188Os varied between 0·135 and 0·223 (excepting one outlier with 0·319). All data were corrected for blanks and the isotopic and concentration data in Table 2 represent the corrected values. The uncertainties in blanks are reflected in respective uncertainties for isotopic and concentration data in Table 2. The external reproducibilities on standard analyses for Os and Re are typically better than ±0·1 and ±0·3%, respectively.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND ROCK...SAMPLESANALYTICAL TECHNIQUESRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Whole-rock geochemical characteristics
The Jeesiörova komatiites have subchondritic Al2O3/TiO2 ratios and characteristic chondrite-nomalized (CN) heavy rare earth element (HREE)-depleted patterns (Gd/YbCN >1), similar to those typical of Al-depleted komatiites (e.g. Lahaye et al., 1995
). Additionally, the Jeesiörova rocks have high abundances of high field strength elements (HFSE; e.g. Ti, Zr) and moderately incompatible REE (e.g. Nd, Sm, Eu, Gd) relative to those at a given MgO content in both Al-depleted and Al-undepleted komatiites (Fig. 2b; also see Hanski et al., 2001). In contrast to high concentrations of Ti and Zr, virtually all Jeesiörova samples show negative Nb anomalies and characteristic depletions in LREE (Fig. 3).
|
Re–Os concentrations and isotopic compositions
Re–Os concentrations and isotopic compositions for the whole-rocks and mineral separates (chromite, clinopyroxene and sulfide) are reported in Table 2. The concentrations of Re and Os, and Re/Os ratios for all the sample types are plotted in Fig. 4a and b. The abundances of both Os and Re vary over two orders of magnitude (Fig. 4a). The Os concentrations are highest in chromite and lowest in clinopyroxene separates.
|
The very high concentrations of Os, coupled with low Re concentrations in chromite, result in very low Re/Os ratios (Fig. 4b). The Re–Os isotope data for the chromites do not yield a precise isochron as a result of the very limited spread in their 187Re/188Os ratios (
0·01–0·07: Table 2) and relatively large uncertainties in Re concentrations. Because of their low Re/Os ratios and minimal correction of 187Os/188Os for age, chromites were used to precisely determine the initial 187Os/188Os for these rocks. The average initial 187Os/188Os ratio obtained for the chromites (0·1131 ± 0·0006, 2) corresponds to a Os of 0·1 ± 0·5 at the 2·056 Ga age of crystallization (Fig. 5a). This precisely chondritic average initial Os isotopic composition is indistinguishable from the projected Os isotopic composition of the contemporaneous convecting upper mantle (Snow & Reisberg, 1995; Walker et al., 1996, 2002; Brandon et al., 2000).
|
The concentrations of Os in the whole-rocks are typically high, varying between
1·3 and 6·2 ppb (Fig. 4b). Similarly, the whole-rock Re concentrations are high (mostly above 0·5 ppb) relative to komatiites from elsewhere (e.g. Gangopadhyay & Walker, 2003; Gangopadhyay et al., 2005). The whole-rock samples mostly plot above a chondritic reference isochron (Fig. 5b) and display a large variation in their respective calculated initial Os values (Table 2).
The sulfide separates have the highest Re concentrations and most of them have very high Re/Os ratios (Fig. 4b). The sulfide separates give generally consistent Os model ages of 1·8 Ga. This model age is 250 Myr younger than the Sm–Nd age of these rocks (2056 ± 25 Myr: Hanski et al., 2001). This suggests that the primary Re–Os systematics of the sulfides were disturbed 250 Ma after their crystallization because of secondary processes such as hydrothermal alteration and/or metamorphic resetting. As a result, the sulfides give anomalous initial Os values for the crystallization age of the rocks (Table 2). It is not uncommon for the Re–Os system to display open-system behavior in sulfide phases associated with komatiites (e.g. Luck & Allègre, 1984; Shirey & Walker, 1995; Walker et al., 1997).
Clinopyroxene separates have both relatively high Os (>0·5 ppb to 1 ppb, excepting one separate, 17-EJH) and Re concentrations (up to 1·4 ppb). Clinopyroxene separates have highly variable initial Os values (ranging between +6 and +80). Most of the whole-rock and clinopyroxene separates in these rocks give generally consistent initial Nd isotopic compositions (Hanski et al., 2001), whereas the Re–Os isotope systematics of these samples indicate large-scale open-system behavior. This may be due to the presence of secondary sulfide inclusions within clinopyroxene or sulfide impurities in our clinopyroxene separates.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND ROCK...SAMPLESANALYTICAL TECHNIQUESRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Are the Jeesiörova komatiites Al depleted?
The low Al2O3/TiO2 ratios of the Jeesiörova komatiites could result from depletion in Al or enrichment in Ti relative to the Al-undepleted Munro-type komatiites. Several aspects of the Jeesiörova rocks suggest that the parental liquids were not Al depleted. First, the bulk aluminum contents at given MgO levels of the Jeesiörova rocks are significantly higher than in the Al-depleted Barberton rocks, and similar to those in Al-undepleted Munro rocks (Fig. 2a). Second, the chromites in the Jeesiörova rocks have distinctly higher Al contents than those in the Al-depleted komatiites from elsewhere (Barnes & Roeder, 2001
; Fig. 6a). The chromites in the Finnish rocks are, on the other hand, chemically similar to chromites from the Al-undepleted komatiites from Gorgona Island (Echeverria, 1980) and Pyke Hill in the Munro Township (Arndt et al., 1977; Fig. 6a). The similar aluminum contents of the chromites in both the Jeesiörova rocks and those in Al-undepleted komatiites from Pyke Hill, Canada (Arndt et al., 1977), Vetreny, Russia (Puchtel et al., 1996) and Belingwe, Zimbabwe (Zhou & Kerrich, 1992) are also apparent in Fig. 6b. Combined, these characteristics suggest that aluminum contents in the magmas parental to the Finnish rocks were more similar to those of Al-undepleted, rather than Al-depleted rocks.
|
As noted, the Jeesiörova komatiites have high concentrations of moderately incompatible elements, but are depleted in both more incompatible (LREE) and less incompatible elements (HREE). As a result, these rocks show characteristic convex-up patterns in mantle-normalized trace element diagrams (Fig. 3). In addition, they have subchondritic Al2O3/TiO2 ratios similar to those commonly observed in Al-depleted komatiites (Nesbitt & Sun, 1976
; Herzberg, 1995; Lahaye et al., 1995).
Previous studies of Al-depleted Barberton-type komatiites (e.g. Lahaye et al., 1995) have documented negative correlations between HREE depletions (e.g. Gd/YbCN >1) and HFSE anomalies [e.g. Zr/Zr* = Zr/10(log Sm + log Nd)/2: McCuaig et al., 1994]. These features have been explained by separation of the high-pressure form of garnet (majorite garnet at >14 GPa) in the deep mantle (Yurimoto & Ohtani, 1992; Xie & Kerrich, 1994; Lahaye et al., 1995). This interpretation was based on the close correspondence between the variations in HFSE and HREE concentrations in Al-depleted rocks, consistent with calculated majorite fractionation trends using experimentally determined partition coefficients (Kato et al., 1988; Yurimoto & Ohtani, 1992). The depletions in HREE and variations in HFSE vs REE in the Jeesiörova rocks are plotted in Fig. 7a and b. Calculated majorite fractionation trends using published partition coefficients (Yurimoto & Ohtani, 1992) do not correspond to the variations of the respective elements in the Jeesiörova rocks. Nor are the Re and Al concentrations in the Jeesiörova komatiites consistent with majorite fractionation. A previous experimental study has suggested that Re is compatible in garnet during mantle melting (DRegarnet/silicate melt = 2·7: Righter & Hauri, 1998). If generally true for majorite garnet, it is likely that majorite fractionation in the mantle source of the Jeesiörova komatiites would have led to significant depletion of Re. Instead, the Re concentrations in virtually all the Jeesiörova samples (0·5–0·9 ppb: Table 2) are higher than in Al-undepleted komatiites (e.g. 0·3–0·4 ppb in Alexo komatiites: Gangopadhyay & Walker, 2003). Similarly, Al, which is a major constituent of majorite garnet, is likely to be depleted in komatiitic melts as a result of majorite fractionation during their generation. The Al2O3 contents of the Jeesiörova komatiites, on the other hand, are similar to those at given MgO contents of Al-undepleted rather than Al-depleted komatiites (Fig. 2a).
|
Collectively, the distribution patterns of moderately incompatible elements along with Re and Al concentrations in the Jeesiörova komatiites suggest that either (1) the D values considered for the majorite fractionation model are not appropriate for the Finnish rocks, (2) the choice of primitive mantle as a source composition is not valid, (3) majorite fractionation was not involved in the generation of these rocks and rather than being depleted in HREE, the rocks are enriched in middle REE (MREE), or (4) another petrogenetic process intervened. The first two possibilities are considered unlikely because the D values and/or model starting mantle parameters would have to be considerably different from those used in previous studies (e.g. Lahaye et al., 1995
) in order to account for the abundances of the respective elements in the Jeesiörova komatiites. The last two possibilities are discussed below.
In a different model, some of the petrological and geochemical characteristics of Barberton-type komatiites were explained in terms of hydrous melting in supra-subduction zone settings (e.g. Parman et al., 1997, 2003; Grove et al., 1999). For example, the negative correlation between Ti/Zr and La/Sm ratios in the chilled margins of some komatiites and komatiitic basalts from Barberton resembles that observed in Phanerozoic boninites (Parman et al., 2003). This feature was interpreted by Parman et al. (2003) as a result of mixing between a silicate melt with high La/Sm ratio and a hydrous fluid with high Ti/Zr ratio. This model is also relevant to consider here because the Jeesiörova komatiites show characteristic negative Nb anomalies (Fig. 3), similar to some Phanerozoic boninites (Hickey-Vargas, 1989) and also typically crustally contaminated tholeiites and komatiites. However, on a primitive mantle (PM)-normalized diagram of La/Sm vs Ti/Zr, the Jeesiörova rocks do not show any evidence for mixing between hydrous fluids and silicate melts compositionally similar to those that were suggested for the Barberton rocks (Fig. 7c; also see Parman et al., 2003). Additionally, as can be seen on a primitive mantle-normalized diagram of (Ti/V)PM vs (Ti/Sc)PM, the Finnish rocks, unlike boninites (Hickey & Frey, 1982), are enriched rather than depleted in Ti relative to other transitional elements, such as V and Sc (Fig. 7d). Finally, the convex-up REE pattern of the Jeesiörova komatiites (Fig. 3), resulting from high concentrations of MREE relative to both LREE and HREE, is in contrast to the characteristic U-shaped REE profile of boninites (Arndt, 2003, and references therein).
In addition to their whole-rock trace element characteristics, the redox state of the Jeesiörova rocks is also unlikely to support their generation in a highly oxidized arc environment. The redox state of the Jeesiörova rocks was estimated using Fe3+-number [Fe3+/(Fe3+ + Cr + Al)] and Fe3+/Fe ratios of selected chromites determined by electron microprobe and Mössbauer analysis, respectively (Table 1). The Fe3+-number for the chromites in the Jeesiörova rocks, and corresponding fO2 values, are compared with those in the 1 atm experiments of Murck & Campbell (1986) and Roeder & Reynolds (1991), which used natural komatiitic lavas as starting materials (Fig. 8). The entire range of Fe3+-number of chromites in the Jeesiörova rocks corresponds approximately to a range in fO2 from FMQ (fayalite–magnetite–quartz) to FMQ + 1·4. This range in fO2 is consistent with that obtained for mafic–ultramafic lavas by Fisk & Bence (1980), and Barnes (1986). It is, however, recognized that conventional electron microprobe analysis of chromites does not always yield accurate estimates of fO2, whereas the Fe3+/Fe ratios of chromites determined by Mössbauer analysis are potentially more accurate for the purpose of oxygen barometry (Wood & Virgo, 1989). For this reason, the Fe3+/Fe ratios of selected chromite separates from the Finnish rocks were determined by Mössbauer spectroscopy. The Fe3+/Fe ratios for five chromite separates vary within a limited range of 0·16–0·28 that corresponds to fO2 of –1·5 < 
FMQ < +0·7 (Parkinson & Arculus, 1999, and references therein). Thus, microprobe and Mössbauer analyses of chromites collectively yield fO2 for the Finnish rocks of approximately FMQ ± 1·5. For comparison, arc-related lavas have redox states that vary from FMQ to as high as FMQ + 6 [as summarized by Lee et al. (2003)]. In contrast, estimates for the oceanic mantle as sampled by mid-ocean ridge basalt (MORB) and abyssal peridotites are typically FMQ to FMQ – 2. Although the entire range of fO2 values obtained for the Jeesiörova rocks overlaps with the relatively rare, least oxidized arc rocks, it is lower than that of most arc-related rocks. Instead, the fO2 estimates for the Jeesiörova rocks are indistinguishable from those estimated for non-arc oceanic mantle (Lee et al., 2003). The range of our empirical estimates of fO2 for the Finnish rocks is also indistinguishable from those based on the whole-rock V systematics of the Al-undepleted komatiitic basalts from Fred's Flow in the Munro Township (Canil & Fedortchouk, 2001) and, in general, with those obtained for the Archean Al-undepleted komatiites (Canil, 1999). This suggests that the redox states during the generation of both the Al-undepleted Munro rocks and the Jeesiörova komatiites were not significantly different.
|
The initial Os isotopic composition [
Os(I) = +0·1 ± 0·5] of the mantle source of the Jeesiörova komatiites is also inconsistent with their generation in supra-subduction zone setting. Modern arc-related lavas and sub-arc mantle xenoliths commonly have variably 187Os-enriched isotopic compositions that are likely to be a result of transport of radiogenic 187Os via Cl-rich hydrous fluid into the mantle wedge (e.g. Brandon et al., 1996, 1999; Peslier et al., 2000; Lassiter & Luhr, 2001; Widom et al., 2003; Saha et al., 2005). The precisely chondritic initial Os isotopic composition of the Jeesiörova rocks, on the other hand, is indistinguishable from the projected Os isotopic evolution trajectory of the primitive upper mantle (PUM; Meisel et al., 2001), suggesting that these rocks sampled a mantle source that had no prior history of resolvable 187Os enrichment.
Are the Jeesiörova komatiites Ti enriched?
The TiO2 contents of a significant portion of the chromites in the Jeesiörova rocks are higher than those in chromites from Al-undepleted komatiites with similar aluminum contents reported from Pyke Hill, Canada (Arndt et al., 1977; Fig. 6b), Vetreny, Russia (Puchtel et al., 1996) and Belingwe, Zimbabwe (Zhou & Kerrich, 1992), suggesting that the Jeesiörova parental magmas had higher Ti concentrations than those of most Al-undepleted komatiites. In most Jeesiörova whole-rock komatiites, however, mantle-normalized concentrations of Ti are similar to, or only slightly higher than, those for other moderately incompatible elements (e.g. MREE), which, in turn, appear to be enriched in comparison with less or more incompatible elements (e.g. HREE and LREE, respectively). For comparison, the entire range of Ti/Ti* in the Al-undepleted and Al-depleted komatiites is 0·6–1·1, and that in the Jeesiörova rocks is 0·9–1·3 (Fig. 9a). As noted above, the Jeesiörova komatiites have higher concentrations of Ti than in Munro- or Barberton-type rocks (Fig. 2b), yet significant enrichments in Ti relative to similarly incompatible elements (e.g. MREE) are found only in a limited number of Jeesiörova samples (Ti/Ti* > 1·1: Fig. 9a). Elevated abundances of Ti, similar to those in the Jeesiörova rocks, are also found in komatiites from the presumed extension of the Lapland complex in the Karasjok greenstone belt, northern Norway (Barnes & Often, 1990), mafic–ultramafic lavas and sills (MgO 14–30 wt %) from the Onega plateau in the SE Baltic Shield (Puchtel et al., 1998), and picritic rocks from Baffin Bay, West Greenland (Clarke, 1970; Francis, 1985).
|
High concentrations of Ti and other moderately incompatible elements as a feature of the mantle source
Based on the major and trace element characteristics of the Ti-rich komatiites from the Karasjok greenstone belt, northern Norway, Barnes & Often (1990)
concluded that the Ti-rich nature of the Norwegian rocks is unlikely to be a result of (1) post-magmatic alteration, (2) crustal contamination or (3) lower degrees of melting. These possibilities for the Jeesiörova rocks are considered briefly below.
First, it is possible that elevated concentrations of some HFSE, such as Ti and Zr, could result from the loss of other elements or addition of HFSE during some stage of alteration. However, Ti and Zr are among the least mobile elements during alteration of komatiites (e.g. Arndt, 1994), and there are strong negative correlations on plots of MgO vs TiO2 (see Hanski et al., 2001) and Zr in the Jeesiörova rocks. This strongly suggests that Ti abundances were not significantly affected during secondary alteration. Second, the Jeesiörova parental magmas are unlikely to have suffered a large degree of contamination by old continental crustal materials. The LREE-depleted nature and 143Nd-enriched initial isotopic composition of the Jeesiörova rocks (Nd +4: Hanski et al., 2001) is consistent with their derivation from a source similar to the contemporaneous depleted upper mantle (DePaolo, 1981), suggesting that these komatiites were not significantly contaminated by typical continental crustal materials. Importantly, Ti contents in most crustal rocks are significantly lower than in the Jeesiörova komatiites, so the high concentrations of Ti are not a result of crustal contamination. Third, the high relative concentrations of Ti and moderately incompatible elements in the Jeesiörova komatiites are probably not due solely to lower degrees of partial melting of the mantle source than for the Munro- or Barberton-type rocks. This is because the incompatibility of Ti during mantle melting is intermediate between that of Eu and Gd (MREE), and significantly lower than that of the LREE (Sun & McDonough, 1989). Accordingly, for the case of modal batch melting, the rate of increase in LREE/MREE (e.g. Ce/Sm) in the partial melt as a function of degree of melting is likely to be higher than that for Ti/MREE (e.g. Ti/Ti*; see Rollinson, 1998). For comparison, high Ti contents in picrites from both Hawaii and Finnish Lapland, which are presumably products of lower degrees of partial melting than Munro-type komatiites, are accompanied by corresponding increases in Ce/Sm ratios (Norman & Garcia, 1999; Hanski et al., 2001; Fig. 9b). No such correlation is observed for the Jeesiörova komatiites, and, instead, the komatiite samples with the highest Ti concentrations (and also highest Ti/Ti*) are typically the most LREE depleted (Fig. 9b). This combination of the most elevated Ti concentrations and lowest LREE/MREE in some Jeesiörova samples is not found among the Barberton- or Munro-type rocks. Consequently, it is difficult to envision known melting scenarios in which the high concentrations of Ti and other moderately incompatible elements in Jeesiörova rocks could have been produced from a mantle source, similar in degree of LREE depletion to that of the Munro komatiites, solely as a result of lower degrees of melting. It is more likely that, regardless of the absolute degree of melting of the Jeesiörova mantle source, the higher Ti concentrations (Fig. 2b) are due to process(es) other than differential degrees of melting relative to the Munro komatiites. For example, it is possible that either the mantle source for the Jeesiörova komatiites had higher relative concentrations of Ti and other moderately incompatible elements than the Abitibi mantle source and/or the primary Jeesiörova magmas underwent subsequent enrichments in these elements en route to the surface. These two possibilities are discussed below.
Consideration of alternative petrogenetic processes
Assimilation of phases rich in moderately incompatible elements
First, we consider the possibility that the overall convex-up REE patterns of the Jeesiörova komatiites (Fig. 3) were the result of selective enrichment of the primary magmas in MREE and other moderately incompatible elements (e.g. Ti, Zr) relative to both LREE and HREE. For example, the high concentrations of Ti, Zr and MREE in the Jeesiörova komatiites could, potentially, have resulted from selective assimilation of phases rich in moderately incompatible elements during ascent of the magma through the lithosphere. The high Ti and Zr abundances in the Jeesiörova komatiites are coupled with slightly suprachondritic Zr/Hf ratios (typically 42–46 relative to the chondritic value of 36: McDonough & Sun, 1995). Moreover, the rocks show a broad positive correlation between Ti/Ti* and Zr/Zr*, and samples that have the highest values of Ti/Ti* also typically have the largest positive Zr anomalies (Zr/Zr* up to 1·7: Fig. 7a). Some Ti-rich minerals (e.g. rutile, armalcolite, loveringite) produced by metasomatism within the sub-continental lithospheric mantle (SCLM), as sampled by peridotite xenoliths, commonly show high abundances of HFSE and also suprachondritic Zr/Hf ratios (up to 58: Kalfoun et al., 2002). Despite very low modal abundances, these minerals can account for >50% of HFSE budgets in some whole-rock peridotite xenoliths (Kalfoun et al., 2002). Ti-rich metasomatic mineral assemblages (rutile, ilmenite, armalcolite) have also been documented in harzburgitic oceanic mantle xenoliths from the Kerguelen Islands that are likely to have crystallized at temperatures of 1200°C, consistent with the relatively low melting temperatures for these minerals (Grégoire et al., 2000).
Selective assimilation of phases rich in moderately incompatible elements by low-Ti primary melts in the generation of the Jeesiörova parental magmas is consistent with some experimental studies. For example, the dissolution rate experiments of Wagner & Grove (1997) suggest that Ti-rich oxide minerals, such as ilmenite, melt out 3 times faster than clinopyroxene in a typical mineral assemblage in the lunar mantle. Wagner & Grove inferred that some high-Ti lunar magmas were produced from low-Ti primitive magmas via assimilation of ilmenite-rich assemblages at shallow levels. Similarly, high-pressure and high-temperature melting experiments on terrestrial picritic basalt compositions suggest that near-solidus melts produced from armalcolite (± ilmenite)-bearing peridotites have very high TiO2 contents (
20 wt %: Xirouchakis et al., 2001) consistent with preferential non-modal melting of Ti-rich phases.
It might also be expected that significant interaction of the parental komatiite melts with the SCLM would lead to a reduction in the initial 187Os/188Os of the magmas. This is because it is commonly observed that the SCLM has depleted 187Os/188Os relative to the contemporaneous convecting upper mantle as a result of prior Re depletion (Walker et al., 1989). For example, combined Os–Nd isotope characteristics have been used to support up to 50% contribution of lithospheric components in the generation of 190 Ma Karoo picrites (Ellam et al., 1992). This interpretation was based on the correspondence of the Karoo picrites in an Os–Nd isotope diagram to a calculated mixing array plotting between assumed plume (both high Os and Nd) and SCLM (both low Os and Nd) end-members. Another possible manifestation of such a process may be found in the 2·7 Ga Al-depleted ultramafic rocks from Boston Creek (Abitibi greenstone belt), which have strongly depleted initial Os isotopic compositions [Os(I) = –3·8 ± 0·5: Walker & Stone, 2001]. The chondritic Os isotopes, combined with the depleted mantle-like Nd isotopic composition of the Jeesiörova rocks, on the other hand, argues against bulk mixing of large proportions of these components, as a mixing array rather than uniform Nd–Os isotopic compositions for these rocks would be expected.
The uniform chondritic initial Os isotopic composition of the Jeesiörova rocks may suggest that either (1) assimilation of metasomatized lithospheric minerals was not involved, (2) insufficient time between the metasomatic event(s) that produced the lithospheric minerals and their melting precluded significant enrichment or depletion in 187Os relative to that of the contemporaneous convecting upper mantle, or (3) Os concentrations in the putative metasomatized mineral assemblages were significantly lower than in the primitive Jeesiörova magmas. Consistent with the last possibility, the phases rich in Ti (picroilmenite) in alkaline ultramafic magmas from Norseman–Wiluna belt, Western Australia, have very low concentrations of Os (0·02–0·03 ppb: Graham et al., 2002) that are two orders of magnitude lower than those in whole-rock Jeesiörova komatiites (Table 2). In the absence of Os concentrations and isotopic compositions for metasomatized mineral assemblages in xenoliths from the SCLM that the Jeesiörova magmas must have traversed, the likelihood of the last two possibilities [(2) and (3)], cannot yet be evaluated.
Enrichment of the Jeesiörova mantle source by basaltic melts
The high concentrations of moderately incompatible elements in the Jeesiörova rocks, combined with their radiogenic initial Nd isotopic composition (Nd +4: Hanski et al., 2001) and LREE-depleted nature, may, alternatively, have resulted from enrichment of the mantle source by basaltic melts or incorporation of young recycled mafic oceanic crust, similar to depleted normal mid-oceanic ridge basalts (N-MORB), into the mantle source. This possibility is consistent with the higher average abundance of most incompatible elements in the Jeesiörova komatiites relative to typical Al-undepleted and Al-depleted komatiites. The Os concentrations of MORB (e.g. 0·23–31·8 ppt: Schiano et al., 1997) are 2–3 orders of magnitude lower than in likely depleted peridotitic mantle sources for komatiites (e.g. empirical estimates of 3·9 ppb Os for the source of Abitibi komatiites: Puchtel et al., 2004), so the chondritic initial Os isotopic composition of the mixed source for the Jeesiörova komatiites could essentially represent that of the peridotite component. The long-term 143Nd-enriched initial isotopic composition of the Jeesiörova rocks, on the other hand, may reflect contributions from both LREE-depleted N-MORB and peridotite source components. Furthermore, enrichment of the peridotite source by mafic melts may lead to an increase in Re concentrations, so this process could account for higher relative Re concentrations in these komatiites. The high Re/Os ratios of the mantle source, over time, would potentially lead to growth of suprachondritic 187Os/188Os. Therefore, the chondritic initial 187Os/188Os of the Jeesiörova rocks suggests that mafic enrichment of the source could not have occurred more than 200 Myr prior to komatiite production (see Walker et al., 1991).
Neither of the two petrogenetic processes discussed above, however, is consistent with all aspects of the Jeesiörova komatiites. For example, the presence of uniform negative Nb anomalies in the Jeesiörova rocks (Figs 3 and 7b) cannot be readily reconciled with models of trace element enrichment via addition of basaltic melt or recycled oceanic crust to their mantle source. Present-day N-MORB does not have negative Nb anomalies, and, therefore, recycling of oceanic crust would require that the LREE-depleted primary Jeesiörova asthenospheric melts possessed negative Nb anomalies to begin with: a possibility for which there are no known analogs (Hofmann, 1988; Salters & Stracke, 2004, and references therein). Similarly, neither assimilation of hypothetical lithospheric mantle phases nor prior basaltic melt enrichment of the source can cause the apparent HREE depletion of the Jeesiörova komatiites, which is commonly explained in terms of fractionation of majorite garnet during the generation of the mantle-derived magmas (Hofmann, 1988; Gurenko & Chaussidon, 1995; Arndt et al. 1997). Furthermore, if either of the petrogenetic processes suggested above occurred in combination with majorite fractionation at some stage in the generation of the Jeesiörova rocks, it remains unclear as to why their HREE depletion is neither accompanied by correspondingly low Al2O3 contents (Fig. 2a) nor correlated with other geochemical parameters suggestive of majorite fractionation (Fig. 7a–d), similar to those commonly observed in Barberton-type komatiites.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND ROCK...SAMPLESANALYTICAL TECHNIQUESRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The Jeesiörova komatiites have subchondritic Al2O3/TiO2 ratios and HREE depletion (Gd/YbCN >1), similar to those typically observed in Al-depleted komatiites. We have demonstrated that the subchondritic Al2O3/TiO2 in the Jeesiörova rocks is not a result of depletion in Al, but rather is due to high concentrations of Ti at a given MgO content relative to both Al-undepleted and Al-depleted komatiites. The Jeesiörova rocks also have overall high concentrations of moderately incompatible elements relative to these two types of komatiite and, thus, represent a chemically distinct type of komatiite. The whole-rock major and trace element characteristics (e.g. absence of depletion in Al corresponding to HREE depletion), and redox state (
FMQ) estimated from the chemical composition of chromite, are not consistent with either of the petrogenetic models suggested for the Al-depleted rocks, majorite fractionation in the deep mantle or hydrous melting in a supra-subduction zone setting. The high relative concentrations of Ti and similarly incompatible elements in these rocks may suggest either (1) assimilation of metasomatized mineral assemblages rich in moderately incompatible elements present in the lithospheric mantle or (2) contamination of the source of the primary asthenospheric melts by young LREE-depleted oceanic crust. The characteristic negative Nb anomalies in the Jeesiörova rocks, however, are not consistent with any of these models. Nevertheless, the combined Os–Nd isotopic characteristics suggest that their mantle source, regardless of the mechanism responsible for the distinctively high concentrations of moderately incompatible elements, is indistinguishable from the contemporaneous convecting upper mantle. |
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING AND ROCK...SAMPLESANALYTICAL TECHNIQUESRESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data on this paper are available at Journal of Petrology online.
|
ACKNOWLEDGEMENTS |
|---|
We are grateful to Pentti Kouri for performing mineral separation, and Lassi Pakkanen and Bo Johanson for providing microprobe analyses. A.G. dedicates this work to an unidentifiable academician, from whom even a humble attempt to make a little contribution is always duly recognized. A.G. thanks Mike Lesher and Al Brandon for encouragement, Dante Canil for providing his spreadsheet that helped easy comparison of D values for V and their corresponding fO2 values for different buffers, and Harry Becker and Bill McDonough for their comments on an earlier version of the manuscript. We thank Igor Puchtel, Steve Barnes and J. Mungall for their critical but very constructive reviews. We also thank Nick Arndt for his editorial handling, incisive comments and helpful suggestions, and Marjorie Wilson for her constructive suggestions and meticulous editing of our manuscript. This work was supported by NSF Grant 9909197, and this support is gratefully acknowledged.
*