Introduction

The Lindås nappe, Bergen Arcs, Caledonides of W Norway, contains typical examples of localised transformation of a granulite-facies protolith into eclogite, controlled by the introduction of metamorphic fluid (Austrheim 1987; Boundy et al. 1992; Erambert and Austrheim 1993; Austrheim et al. 1997). Published 40Ar–39Ar, U–Pb, Rb–Sr, and Sm–Nd ages interpreted to date the eclogite-facies overprint span a large part of the Ordovician and Silurian periods from ca. 460–420 Ma (Cohen et al. 1988; Boundy et al. 1996, 1997a, b; Bingen et al. 2001b; Glodny et al. 2002). Available data do not constrain the timing of this overprint relative to the pivotal Wenlock epoch (428–423 Ma, Tucker and McKerrow 1995), which marks the onset of the Scandian (Silurian) collision between Laurentia and Baltica (Gee 1975). Controversy over the geochronology stems from incomplete achievement of equilibrium, on the microscopic and macroscopic scales, during the eclogite-facies overprint. Disequilibrium may result in preservation of pre-eclogite facies relic minerals (cores or inclusions, Gebauer et al. 1985), chemical inheritance from the pre-eclogite facies reagent minerals (Thöni and Jagoutz 1992), or occurrence of excess radiogenic Ar in K-rich minerals (Boundy et al. 1997a; Scaillet 1998).

Zircon is one of the most widely used minerals for geochronology because it is common, robust, rich in U, and poor in Pb. In metamorphic rocks, zircon commonly displays complex internal zoning (Vavra et al. 1999). In this situation, linking age data collected on different zones to petrologic or structural changes in the host rock may prove speculative. The trace element composition of zircon is additional information for establishing such links (Schaltegger et al. 1999; Belousova et al. 2002; Rubatto 2002; Whitehouse and Platt 2003). In this paper, samples of the Lindås nappe studied and dated by Bingen et al. (2001a, b) are revisited. New trace element and U–Pb geochronological data on zircon, collected by secondary-ion mass spectrometry (SIMS), are reported to establish better links between crystallization of zircon and eclogite-facies metamorphism and to resolve the timing of the eclogite-facies overprint. Implications for Caledonian orogenic evolution are discussed briefly.

Geological and geochronological setting

The Bergen Arcs is an arcuate structure. It is made of the Proterozoic Øygards gneiss complex covered by a set of Caledonian nappes, including the Lindås and Hardangerfjord nappes (Ragnhildstveit and Helliksen 1997; Boundy et al. 1996). The Hardangerfjord nappe is attributed to the Upper Allochthon. It includes Ordovician low-grade lithologies of oceanic affinity (Dunning and Pedersen 1988), unconformably overlain by Ashgill to Llandovery (449–428 Ma) shallow marine sediments (Os group). The Lindås nappe is attributed to the Middle Allochthon. It consists mainly of a meta-anorthosite complex and a banded gneiss complex. The meta-anorthosite complex comprises a range of mafic to felsic lithologies intruded between 1,237+43/-35 and 951±2 Ma (Bingen et al. 2001b).

The Lindås nappe was affected by penetrative Sveconorwegian granulite-facies metamorphism at <1 Gpa and 800–850°C (Austrheim 1987). This metamorphism took place around 930 Ma and was followed by comparatively slow regional cooling at an estimated rate of ca. 4°C Ma1 (Burton et al. 1995). The geochronological data base includes U–Pb ages of metamorphic zircon at 933±2 and 929±1 Ma (Bingen et al. 2001b), and Sm–Nd and Rb–Sr mineral isochron ages ranging from 929±9 to 835±7 Ma (Cohen et al. 1988; Burton et al. 1995; Kühn et al. 2000).

The granulite-facies complex was partially transformed to eclogite and amphibolite during the Caledonian orogeny. Eclogitization of the protolith was initiated along brittle fractures, interpreted as fluid pathways, and progressed as a reaction front migrating away from the fractures (Austrheim 1987; Boundy et al. 1992; Jamtveit et al. 1990). Eclogite-facies pseudotachylyte veins are locally reported supporting the brittle nature of the protolith at the onset of eclogitization, and the key role of earthquakes in initiating eclogitization (Austrheim et al. 1996; Bjørnerud et al. 2002). Eclogitization evolved into ductile eclogite-facies shear zones ranging from ca. 10 cm to 100 m in thickness (Boundy et al. 1992). The eclogite-facies assemblage records P–T conditions of 1.8–2.1 GPa—ca. 700°C (Jamtveit et al. 1990). The amphibolite-facies overprint occurred in a similar fashion. It started with static conversion of the mineral associations and developed into shear zones. P–T conditions are estimated at 0.8–1.2 GPa—690°C (Boundy et al. 1996). Some amphibolite-facies shear zones grade into late greenschist-facies shear zones.

Published geochronological data interpreted to reflect eclogite-facies overprint in the Lindås nappe range from ca. 460–420 Ma. U–Pb data on zircon multigrain fractions from an eclogite and an enclosing granulite by isotope dilution thermal ionisation mass spectrometry (ID-TIMS) define a scattered discordia line (Bingen et al. 2001b). A lower intercept age of 456±7 Ma, obtained from a selection of fractions, was interpreted as the age of eclogite-facies overprint. Alternative lower intercept ages can be extracted from these data (see below). U–Pb data for titanite, epidote, and allanite in a garnet amphibolite and two samples of eclogite-facies marble yield 206Pb/238U ages ranging from 494 to 445 Ma, with an average value at 464 Ma (Boundy et al. 1997b). This estimate suffers from the low proportion of radiogenic Pb in the analysed minerals, and from possible inheritance. Rb–Sr and Sm–Nd data from eclogites and eclogite-facies veins yield mineral isochron ages of 425±4 and 422±10 Ma, respectively (Glodny et al. 2002). Boundy et al. (1996, 1997a) report hornblende 40Ar/39Ar ages of 455±2 and 448±4 Ma from an eclogite and a garnet amphibolite, and muscovite 40Ar/39Ar plateau ages of 433±1 to 429±1 Ma from an eclogite. These ages are interpreted as the timing of cooling and exhumation after eclogite-facies metamorphism.

Caledonian metamorphism in the Lindås nappe is also associated with intrusion of granite and pegmatite dykes. Two garnet-bearing pegmatite dykes yield Rb–Sr mineral isochron intrusion ages of 428±6 and 422±6 Ma and a trondhjemite dyke yields a zircon U–Pb lower intercept age of 418±9 Ma (Kühn et al. 2002). The dated dykes are affected by amphibolite-facies deformation, and consequently provide a maximum age for this overprint. The timing of amphibolite-facies overprint is also estimated by Rb–Sr mineral isochron ages of 409±8 Ma from an amphibolite shear zone (Bingen et al. 2001b), and 413±4 Ma from amphibolite-facies metamorphic veins (Glodny et al. 2002). The data support the theory that amphibolite-facies overprint post-dates eclogite-facies overprint, in accordance with petrographic and field observations.

Analytical methods

Zircon crystals were mounted in epoxy resin, polished to approximately half thickness, and imaged with a cathodoluminescence (CL) detector in a scanning electron microscope (Fig. 1). Thirty-eight U–Th–Pb analyses in 24 zircon grains were collected in one eclogite sample (BH2) by SIMS using the Cameca IMS 1270 instrument at the Nordsim laboratory, Swedish Museum of Natural History, Stockholm (Table 1). The analytical method, data reduction, error propagation, and assessment of results are outlined in Whitehouse et al. (1997, 1999). The analyses were performed with a ca. 2 nA O2-beam and a spot size of ca. 20 μm. As the main objective of this study is to improve the age estimate for Caledonian eclogite-facies overprint, a long total counting time of 25 min optimised for Palaeozoic zircon was chosen. The calibration was performed using Kipawa reference zircon with an age of 993 Ma (Stern 1997). The error in the U/Pb ratio in Table 1 includes propagation of the error on the calibration curve obtained by regular analysis of the reference zircon (a total of 27 analyses divided in two sessions). Two correction methods for common Pb were applied—a correction based on the 204Pb concentration and present-day common Pb isotopic composition (204Pb correction scheme) and a correction assuming concordance between 206Pb/238U and 207Pb/206Pb ages (207Pb correction scheme). The Isoplot program (Ludwig 2001) was used to regress the isotopic data. All regression ages are presented at the 95% confidence level.

Fig. 1
figure 1

Selected CL images of zircon from eclogite BH2 with location of SIMS analyses. Plain ellipses U–Pb analyses (±2σ), dashed ellipse trace element analyses. a, b Zircon crystals made of a large Proterozoic core and minor Caledonian rims. The U–Pb analysis is from Bingen et al. (2001a). b The convex boundary of the rim is interpreted as evidence of a replacement–recrystallization process. c–e Zircon crystals dominated by Caledonian rims. Relationships between Proterozoic micro-cores, weakly-zoned rims, and oscillatory-zoned rims are illustrated. The euhedral oscillatory-zoned rim is interpreted as an overgrowth. The labels of analyses provide the link to Tables 1 and 2. One analysis of an oscillatory-zoned rim is not reported in Table 1, because levels of U, Th, and Pb were below detection limits (e)

Full size image
Table 1 SIMS Th–U–Pb data on zircon from eclogite sample BH2
Full size table

Analyses of rare earth elements (REE), Y and Hf were performed on zircon using the same SIMS instrument and same epoxy-resin mount (Table 2). Eighteen analyses of zircon from the eclogite sample (BH2) were performed (crystals dated in this study or in Bingen et al. 2001a) as were 19 analyses of zircon from three samples of the enclosing meta-anorthosite complex. Data were acquired by the method outlined in Whitehouse and Platt (2003) operating under energy filtering, at low mass resolution, and with a spot size of ca. 30 μm. The NIST SRM 610 reference glass was used for calibration, with working values recommended by Pearce et al. (1996). The Geostandards zircon 91500 was used as control material (Table 2). Uncertainties are evaluated and discussed in Whitehouse and Platt (2003). The average analytical error ranges from ca. ±10% (1σ) for light rare earth elements (LREE) to ca. ±5% for the other REE. The concentration of La is low in some of the zircon overgrowths (ca. 0.1 ppm La). The error in these analyses is larger and might amount to ca. 50%. Errors in Y and Hf analyses are typically ca. ±2 and ±4%, respectively, at the abundance levels encountered.

Table 2 SIMS trace element data in zircon
Full size table

Samples and results

Granulites and amphibolite

Two samples of granulite from the Proterozoic meta-anorthosite complex of the Lindås nappe were re-investigated in this study. These include a garnet granulite (BH5), and a two-pyroxene granulite (BH10). A garnet amphibolite (BH8) resulting from static transformation of a granulite during the Caledonian orogeny was also re-investigated. U–Pb systematics of zircon of these samples is reported in Bingen et al. (2001a, b). Zircon occurs as large (40–500 μm) prismatic, rounded, and anhedral grains randomly distributed in the rock and as platy micro-grains (ca. 10 μm) forming corona textures at the surface of ilmenite. Zircon crystals display a magmatic core and a metamorphic rim. The magmatic core is characterized by oscillatory to sector zoning, and the metamorphic rim by bright CL contrast. A CL-dark U-rich inner core is observed in some crystals. ID–TIMS analyses of magmatic zircon populations in samples BH10 and BH8 yield ages of 951±2 and 957±11 Ma, interpreted to reflect intrusion of the plutons. Analyses of small rounded metamorphic zircon grains in samples BH8 and BH5 yield ages of 933±2 and 929±1 Ma, interpreted as the age of granulite facies metamorphism. SIMS U–Pb analyses in the three samples (Bingen et al. 2001a) yield equivalent age estimates for both events and provide ranges of the Th/U ratio of 1.03–1.48 for the magmatic core and 0.40–0.74 for the metamorphic rim (Fig. 2).

Fig. 2
figure 2

Th/U ratio of zircon versus age (±1σ) in granulites, amphibolite, and eclogite samples of the Lindås nappe. Data from Bingen et al. (2001a) and this work (Table 1)

Full size image

Rare earth element analyses were performed on magmatic zircon cores and metamorphic rims for the three samples. Hf content ranged from 7,510 to 8,430 ppm in magmatic cores and from 8,250 to 13,200 ppm in metamorphic rims. The Y and REE content of cores and rims overlaps, although lower values are systematically found for the metamorphic rims in each sample. The lowest REE and Y contents are recorded in metamorphic rims of garnet-bearing samples. In a chondrite-normalized Matsuda diagram, all analyses of zircon define very similar and parallel REE patterns, characterized by heavy rare earth element (HREE) enrichment (223≤LuN≤ 2,480; 6.5×10−4≤LaN/YbN≤3.6×10−3), a Ce positive anomaly and a negative Eu anomaly (0.38≤Eu/Eu*≤0.64) (Fig. 3a).

Fig. 3
figure 3

Chondrite-normalized REE patterns of zircon determined by SIMS. Data from Table 2. a granulites BH10, BH5 and amphibolite BH8. b Proterozoic cores in eclogite BH2. c Caledonian rims in eclogite BH2. d Selected analyses of Caledonian rims in eclogite BH2 to show intragrain variation

Full size image

Eclogite

Eclogite sample BH2 (Bingen et al. 2001a, b) has no visible fabric and probably results from static transformation of the mafic garnet granulite exposed in the same outcrop (sample BH5). The eclogite contains garnet, omphacite, clinozoisite, phengite, blue amphibole, rutile, apatite, and zircon, and lacks any significant post-eclogite-facies overprint. Garnet generally has an inherited granulite-facies core, enriched in pyrope component (similar textures are described in Erambert and Austrheim 1993). Zircon recovered from heavy mineral separates consists of single crystals or polycrystalline aggregates made of euhedral well-terminated prisms, platy prisms, multifaceted grains, and anhedral grains. Zircon grains generally consist of an anhedral core surrounded by an anhedral to euhedral rim. The core–rim structure is not visible in all grains. In thin section zircon is observed as ca. 60–100 μm grains randomly distributed in the rock and as platy micro-grains (ca. 10 μm in length and a few micrometers thick) forming corona structures at some distance around rutile included in garnet. The zircon grains forming coronas were probably situated at the surface of ilmenite, as commonly observed in the mafic granulites, and were left in situ during the static eclogite-facies breakdown of ilmenite to rutile (see photo and discussion in Bingen et al. 2001a).

Proterozoic zircon cores

Two types of Proterozoic zircon core are observed in eclogite BH2—large cores and platy micro-cores.

Large cores

Large (60–100 μm) cores have variable internal zoning pattern. Some cores have sharp oscillatory to sector zoning, commonly truncated by bright CL replacement rims (Fig. 1a). These are similar to the main population of large zircons reported in the enclosing meta-anorthosite complex, i.e. zoned magmatic core with granulite-facies metamorphic rims. Other cores display a comparatively dark CL contrast and poorly visible zoning (Fig. 1b), and are similar to some CL-dark cores in the enclosing meta-anorthosite complex. U–Th–Pb SIMS analyses of six cores reported in Bingen et al. (2001a) yield 238U–206Pb ages ranging from 934±36 to 777±32 Ma (204Pb-corrected ages). One additional analysis reported in this study yields an age of 847±26 Ma (Table 1). The age range overlaps with the timing of intrusion of the anorthosite suite (957–951 Ma) and the Sveconorwegian granulite-facies metamorphism (933–929 Ma). Some of the cores showing a dark CL contrast and a younger apparent age were probably affected by post-Sveconorwegian Pb loss or underwent partial recrystallization during post-Sveconorwegian cooling. The Th/U ratios of the cores range from 0.31 to 0.89 and the U content from 82 to 592 ppm (this work and that of Bingen et al. 2001a). Their Th/U ratio is in the range recorded for magmatic and metamorphic zircon in the meta-anorthosite complex (0.36–1.48) (Fig. 2).

Platy micro-cores

Platy micro-cores are a few micrometers thick and ca. 10 μm long (Figs. 1c–e). These have a dark CL contrast and correspond morphologically to platy micro-zircon grains observed in corona structures at the surface of ilmenite in granulite-facies rocks, or around rutile in eclogite-facies rocks. Two analyses were attempted (Table 1), but due to the small size of the micro-cores the analytical spots overlap Proterozoic and Caledonian material. They yield mixed apparent 238U–206Pb ages of 572±18 and 469±14 Ma (Fig. 1c) and Th/U ratios of 0.24 and 0.30 (Fig. 2), intermediate between the values for large cores (934–777 Ma, 0.31≤Th/U≤0.89) and Caledonian rims (452–383 Ma, Th/U≤0.13, see below). The data thus support the interpretation that platy micro-cores have a Sveconorwegian age and were presumably formed at the surface of granulite-facies ilmenite. These cores are commonly present in the centre of zircons grains dominated by Caledonian rim material (Figs. 1c–e).

Trace element analyses were realized in two large-zoned cores and three large CL-dark cores. Hf content ranges from 4,660 to 8,790 ppm, and Y content from 175 to 679 ppm. The REE patterns of the cores are characterized by HREE enrichment (6.8×10−4≤LaN/YbN≤6.3×10−3), a positive Ce anomaly, and a negative Eu anomaly (0.54≤Eu/Eu*≤0.74). The REE patterns of zircon cores in the eclogite BH2 overlap with, and are similar to, Sveconorwegian zircon in the enclosing meta-anorthosite complex (Fig. 3b).

Caledonian zircon rims

Two types of Caledonian zircon rim are observed in eclogite BH2: weakly-zoned rims with comparatively bright CL signal, and oscillatory-zoned rims with variable CL signal (Fig. 1e). The oscillatory-zoned rims correspond to overgrowth textures, and systematically give an euhedral habit to the zircon grain. If both types of rim are observed in one zircon grain, the oscillatory-zoned rim always surrounds the weakly-zoned rim. Locally, domains with very bright CL signal truncate both types of Caledonian rim. This type of domain is late and seems to replace the previously formed rims.

Fifteen SIMS U–Th–Pb analyses of weakly-zoned rims and twenty analyses of oscillatory-zoned rims are reported in Table 1. These new data replace the 12 less precise SIMS analyses on the same type of material reported in Bingen et al. (2001a). The U content ranges from 29 to 192 ppm and all analyses are characterized by low Th content (≤25 ppm) and low Th/U ratio (Th/U≤0.13, Fig. 2). 207Pb corrected 206Pb/238U ages range from 452±38 (±2σ) to 383±32 Ma (Table 1). Although it is clear that the oscillatory-zoned rim is more peripheral (Figs. 1c, e), analyses on weakly-zoned rims and oscillatory-zoned rims yield totally overlapping age results, so there is no statistical justification to treat them separately. No correlation can be established between the apparent age of data points and their U content or Th/U ratio (Fig. 2). Consequently, both types of rim were regarded as a single age population, within analytical resolution, and an average age estimate was calculated from the entire data set. Regression of the 35 data points in a Tera-Wasserburg concordia diagram, uncorrected for common Pb, but anchored at a common Pb value with present-day composition (207Pb/206Pb=0.84±0.08, Stacey and Kramers 1975), yields an intercept age of 423±5 Ma with an MSWD of 2.8 (Fig. 4). Three data points are responsible for a significant part of the scatter in this calculation. Point 16b giving an age of 450±14 Ma corresponds to a faint core structure and points 12b and 05a giving comparatively young ages of 393±12 and 383±32 Ma possibly correspond to zircon affected by post-Caledonian Pb-loss. Removing these three points from the age calculation yields a statistically valid intercept age of 423±4 Ma with an MSWD value of 1.7 (Fig. 4). This age is regarded as the best estimate for the crystallization of the rims.

Fig. 4
figure 4

Tera-Wasserburg concordia diagram showing SIMS data on eclogite BH2 (data uncorrected for common Pb, with 1σ error cross). Regression of 32 of the 35 SIMS analyses of zircon rims, anchored at a common Pb value with present-day isotopic composition, yield an intercept age of 423±4 Ma (MSWD=1.7)

Full size image

Thirteen trace element analyses help to characterize the Caledonian rims. Weakly-zoned rims (six analyses) and oscillatory-zoned rims (six analyses) overlap in trace element composition. Hf content ranges from 6,820 to 13,900 ppm, and Y content between 8 and 1,120 ppm. In a chondrite-normalized Matsuda diagram, rims define HREE-enriched patterns with a positive Ce anomaly. Content in LREE is low, with a limited spread (0.17<PrN<0.80). In contrast, the HREE abundance displays a large spread (10≤LuN≤2,769; 4.8×10−4≤LaN/YbN≤6.2×10−2; Fig. 3c). The REE patterns are concave between La and Sm, with a minimum abundance for Pr or Nd, and convex between Sm and Er. In grain 17 one analysis in a weakly-zoned inner rim and one analysis in an oscillatory-zoned outer rim demonstrate a general decrease of REE content toward the rim of the crystal, especially marked in the HREE between Dy and Lu (Fig. 3d). In grain 19 it was possible to place one analytical spot within the bright CL domain truncating an oscillatory-zoned rim. This domain could not be dated because of the very low U and Pb content. The trace element content is lower than for all the other overgrowths analysed, except for the HREE. This domain displays a specific REE pattern characterized by very low Nd content and a relative enrichment in HREE.

Discussion

Trace element signature of eclogite-facies zircon

In the samples collected from the Lindås nappe (Bingen et al. 2001a, b) a Caledonian age signature is recorded only in zircon from the eclogite sample (BH2) and not from granulites and amphibolites (four samples). This suggests that the Caledonian zircon rims are related to the eclogite-facies overprint. The absence of significant post-eclogite facies mineralogy in eclogite BH2 makes it unlikely that zircon rims are related to exhumation of the complex through the amphibolite-facies domain. Although the eclogite-facies overprint in the Lindås nappe is related to introduction of fluid, eclogitization of the granulite-facies protolith has been shown to be isochemical, within analytical resolution, except for loss on ignition (Rockow et al. 1997). As a consequence, the formation and composition of metamorphic zircon can be interpreted in a chemically closed system on whole-rock scale.

Two alternative modes of formation of the Caledonian zircon rims are conceivable; either they were formed by replacement–recrystallization of the Proterozoic zircon (Pidgeon et al. 1998; Schaltegger et al. 1999; Hoskin and Black 2000), or they represent overgrowth precipitated from ZrO2 and SiO2 liberated from other minerals (Fraser et al. 1997; Degeling et al. 2001). Textural evidence shows that both processes were probably operative in eclogite BH2. Weakly-zoned rims locally display a convex boundary relative to large Proterozoic cores (Fig. 1b), indicative of a replacement–recrystallization process. In contrast, the widespread occurrence of euhedral, oscillatory-zoned rims suggests that most rim textures are overgrowths. The eclogitization reaction involved sub-solidus breakdown of the two-pyroxene+garnet+plagioclase+ilmenite assemblage of the granulite-facies protolith to form the garnet+omphacite+rutile eclogite-facies assemblage. This reaction liberates SiO2 and probably ZrO2, leading to precipitation of zircon. Breakdown of ilmenite is especially important, because ilmenite might be a significant reservoir of Zr in mafic lithologies (references and arguments in Bingen et al. 2001a). As we demonstrate below, the trace element composition of zircon rims in eclogite BH2 supports the theory that they crystallized during the eclogite-facies overprint. Three geochemical parameters display significant variation in the zircon populations analysed in this study: (1) the HREE content, (2) the LREE and Th content, and (3) the Eu anomaly.

Heavy rare earth elements

Zircon generally displays HREE-enriched REE patterns (Hinton and Upton 1991; Hoskin and Ireland 2000; Hanchar et al. 2001; Belousova et al. 2002). The higher compatibility of HREE relative to LREE is a consequence of their smaller ionic radius, closer to that of Zr (0.84 Å). Crystallization of garnet in a metamorphic environment is known to influence the HREE content of zircon (Schaltegger et al. 1999; Rubatto 2002; Rubatto and Hermann 2003; Whitehouse and Platt 2003). Garnet is a mineral having a comparatively large partition coefficient for HREE (Bea et al. 1997; Hermann 2002; Otamendi et al. 2002). Consequently, crystallization of major amount of garnet during a metamorphic reaction depletes the reacting volume in HREE, and results in a HREE-depleted signature for co-precipitated minerals.

Proterozoic zircons of this study are characterized by HREE-enriched REE patterns with comparatively high HREE content (220≤LuN≤2480; 4.9×10−4≤LaN/YbN≤6.3× 10−3, Figs. 3a and b). Among the four samples analysed, three contain granulite-facies garnet—garnet granulite BH5, amphibolite BH8 and eclogite BH2 (inherited pyrope-enriched garnet cores). Some analyses of zircon rims in sample BH8 and zircon cores in samples BH2 display a flattening of the REE pattern between Dy and Lu (0.65≤DyN/YbN≤0.83, Figs. 3a, b and 5a), suggesting coeval precipitation of zircon and garnet from the same reservoir. The Caledonian zircon rims in eclogite BH2 display a much larger range of HREE content than the Proterozoic zircons (10≤LuN≤2,769, Fig. 3c). Some of the analyses, especially those with the highest REE content, display a smooth increase in normalized values between Tb and Lu (0.23≤DyN/YbN≤0.31) very similar to that of most Proterozoic zircons, whereas other analyses display a flattening or even a decrease of abundance between Tb and Lu (0.71≤DyN/YbN≤2.6, Figs. 3c and 5a). A decrease of abundance of HREE between the inner and outer rim of a crystal can be documented (LuN from 81 to 10 in grain 17, Figs. 1c, 3d and 5a). The large variation of HREE and the flattening of the HREE pattern for the least enriched domains are characteristic features of coeval precipitation of zircon with garnet from a common reservoir of limited volume (i.e. closed system sub-solidus environment) (Schaltegger et al. 1999; Rubatto 2002; Rubatto and Hermann 2003; Whitehouse and Platt 2003). Although oscillatory-zoned rims are more peripheral than the weakly-zoned rims, the two types cannot be distinguished on the basis of HREE content. This overlap suggests that both types of rim formed from local reservoirs from which HREE were progressively fixed in garnet as eclogite-facies garnet-forming reactions proceeded. Coeval precipitation of zircon and garnet during eclogite-facies overprint is consistent with petrographic observation. In thin sections of eclogite BH2, zircon grains with euhedral rims are commonly included in eclogite-facies garnet (see photograph in Bingen et al. 2001a).

Fig. 5
figure 5

Trace element composition of zircon. a DyN/YbN vs. Eu/Eu* diagram. b PrN vs. Eu/Eu* diagram. Eclogite-facies zircon rims in eclogite BH2 have a signature characterized by a lack of negative Eu anomaly, by low Pr content, and variable DyN/YbN ratio. The variations of the Eu/Eu*, PrN and DyN/YbN parameters attributable to co-precipitation of feldspar, clinozoisite, or garnet, and to increase of oxygen fugacity during metamorphism are qualitatively indicated by arrows

Full size image

Th and LREE

Thorium and LREE are larger ions (1.05–1.16 Å) than U and HREE (0.94–1.00 Å). They are thus less compatible in zircon (Hanchar et al. 2001). Zircon displays a lower LREE/HREE ratio and a lower Th/U ratio than the whole rock (Schärer 1984). Metamorphic zircon is generally, although not systematically, characterized by a low Th/U ratio (Rubatto 2002). No consensus interpretation exists for this feature. The Th/U ratio is commonly lower for amphibolite- or eclogite-facies zircon (Gebauer et al. 1997; Rubatto et al. 1999; Vavra et al. 1999) than for granulite-facies zircon (Vavra et al. 1999; Möller et al. 2003; Whitehouse and Platt 2003). Co-precipitation of zircon with an epidote group mineral during metamorphism is probably one of the most effective processes depleting zircon in Th and LREE. Epidote group minerals, including allanite, are stable in amphibolite and eclogite-facies conditions, but tend to break down during granulite-facies metamorphism (Bingen et al. 1996; Hermann 2002). They are an important carrier of LREE and Th in eclogite-facies rocks (Hermann 2002; Zack et al. 2002). LREE and Th substitute for Ca in epidote group minerals via a coupled substitution. Different epidotes have slightly different crystal-chemical properties; in the epidote-allanite solid solution, the most compatible REE is La, whereas in zoisite and clinozoisite it is Nd (Frei et al. 2003). Apatite might be a significant repository for Th and LREE under granulite-facies conditions (Bingen et al. 1996), but not under amphibolite- or eclogite-facies conditions (Bingen et al. 1996; Hermann 2002; Zack et al. 2002).

Among the samples from this study, Proterozoic magmatic cores in granulites BH10 and BH5 and amphibolite BH8 have a Th/U ratio ranging from 1.03 to 1.48, and Proterozoic metamorphic rims have a significantly lower Th/U ranging from 0.40 to 0.74 (Fig. 2, Bingen et al. 2001a). Pr, representing LREE, ranges from 1.3 to 10.2 times the chondritic abundance in magmatic cores and from 0.74 to 5.3 in metamorphic rims (Fig. 5b). The Proterozoic zircon cores in the eclogite BH2 define overlapping ranges for the Th/U ratio (0.31–0.84, Fig. 2) and for the PrN value (0.71–1.85, Fig. 5b), supporting their origin as both magmatic and metamorphic zircon. Caledonian rims in eclogite BH2 are characterized by low Th content and a Th/U ratio lower than 0.13 (Fig. 2). They are also characterized by comparatively low LREE (0.17<PrN<0.80; Figs. 3c and 5b) and a specific concave REE pattern with minimum enrichment for Pr or Nd (Fig. 3c). A decrease of abundance of LREE, especially of Pr and Nd, between the inner and outer rim of a crystal is documented (PrN decreases from 0.63 to 0.17 in grain 17 and NdN from 0.56 to 0.10, Figs. 1c, 3d, 5b). In eclogite BH2, clinozoisite is part of the eclogite-facies assemblage, and thus represents a sink for Th and LREE, especially Pr and Nd (Frei et al. 2003), during crystallization of the eclogite-facies assemblage. The low Th content of zircon rims associated with a concave REE pattern with minimum enrichment for Pr or Nd, and the decrease of Pr and Nd content toward the outer rim, are arguments for co-precipitation of zircon and clinozoisite during eclogite-facies overprint. Apatite is present in all samples analysed in this study, and might be another exchange reservoir for Th and LREE during zircon crystallization, especially during granulite-facies metamorphism.

Eu anomaly

Zircon in magmatic and metamorphic environments commonly shows a negative Eu anomaly (Hoskin and Ireland 2000; Belousova et al. 2002). The comparatively large Eu2+ ion (R I =1.17 Å) is not accommodated easily in the zircon lattice and therefore the magnitude of the Eu anomaly depends on the Eu2+/Eu3+ ratio or oxidation state during crystallization. It also depends on crystallization of feldspar, as Eu2+ is selectively substituted for Ca in plagioclase. Crystallization of feldspar either before or during a metamorphic reaction results in a depletion of Eu in the reacting volume, which results in a negative Eu anomaly in co-precipitated minerals. Breakdown of plagioclase during a reaction might result in the opposite effect.

Proterozoic zircons of this study display a negative Eu anomaly. The Eu/Eu* ratio ranges from 0.38 to 0.64 in zircon from granulites BH10 and BH5 and amphibolite BH8, and has a largely overlapping range from 0.54 to 0.74 in zircon cores of eclogite BH2 (Figs. 3a, b and 5a). In contrast, the Caledonian zircon rims in eclogite BH2 do not display a significant negative Eu anomaly. The Eu/Eu* ratio ranges from 0.83 to 1.79 with an average value of 1.10 close to unity (Figs. 3c and 5a). The lack of a significant Eu anomaly in the Caledonian rims is consistent with precipitation of zircon under comparatively oxidizing conditions, not in equilibrium with feldspar, not from a rock-volume previously depleted in Eu by crystallization of feldspar. It argues for crystallization of zircon during eclogite-facies overprint, during which breakdown of feldspar occurred (Rubatto 2002; Sun et al. 2002; Rubatto and Hermann 2003).

Timing of eclogite-facies overprint in the Lindås nappe

Published ID-TIMS data on zircon multigrain fractions from eclogite BH2 (ten fractions) and the enclosing granulite BH5 (six fractions, Bingen et al. 2001b) define a scattered discordia line with a MSWD of 12. A statistically valid discordia line derived from a selection of the coarsest fractions gives intercept ages at 931±2 and 456±7 Ma (MSWD=1.3) and was regarded as providing the best estimate for eclogite-facies overprint by Bingen et al. (2001b). The new SIMS data giving evidence for crystallization of zircon rims at 423±4 Ma (Fig. 4) suggest that the assumption used to derive the age of 456±7 Ma, i.e. existence of a post-Caledonian surface Pb loss event affecting fractions with the youngest apparent age, is incorrect. Revised regression analysis of the ID-TIMS data focussing on those fractions with the greatest proportion of Caledonian zircon is thus proposed here. The four fractions of the eclogite with the largest euhedral rims (youngest apparent ages), together with six fractions of near concordant Proterozoic zircon from the enclosing granulite define a discordia line with intercept ages at 931±2 and 422±3 Ma (MSWD=1.16). This line is interpreted as a mixing line, with the lower intercept reflecting formation of the euhedral rims. This revised regression analysis discards the six fractions from eclogite BH2 with the largest proportion of Proterozoic core (oldest apparent ages). These fractions plot in a position compatible with post-Sveconorwegian Pb loss or partial recrystallization during post-Sveconorwegian cooling (above the discordia line in a conventional concordia diagram). As the Proterozoic core is physically surrounded by rims, the air abrasion method, used to prepare fractions for ID-TIMS analysis (Krogh 1982), cannot eliminate the effect of surface disturbance in the core. As a conclusion, published ID-TIMS data and the new SIMS data provide consistent evidence for crystallization of zircon rims at 422±3 and 423±4 Ma respectively, assuming post-Sveconorwegian disturbance of the U–Pb system in the core.

The zircon crystallization age of 423±4 Ma (Fig. 4) is in agreement with recent Rb–Sr and Sm–Nd mineral isochron ages of 425±4 and 422±10 Ma (Glodny et al. 2002). Glodny et al. (2002) argue that the mineral isochrons record crystallization of the eclogite-facies assemblage and are not affected by post-crystallization isotope exchange. The zircon age is significantly younger than muscovite 40Ar/39Ar plateau ages of 433±1 to 429±1 Ma from an eclogite sample (Boundy et al. 1996). The interpretation of eclogite overprint at 423±4 Ma proposed in this work is not compatible with the interpretation proposed by Boundy et al. (1996) of cooling of the Lindås nappe after eclogite-facies metamorphism at 433–429 Ma. To clear up this incompatibility, paired 40Ar/39Ar and U–Pb data sets from key localities should be collected.

Dating fluid activity in subducted continental crust

The dry granulite-facies nature of the protolith in the Lindås nappe has implications for the interpretation of geochronological data on the eclogite-facies overprint. In such a protolith the eclogite-facies reactions are kinetically inhibited unless a water-bearing fluid has access to the reaction site (Austrheim 1987; Jamtveit et al. 1990; Austrheim et al. 1997). Initiation of eclogitization along fractures is evidence of such control. Development of eclogite-facies hydrous minerals, namely phengite, clinozoisite, and amphibole, together with garnet and omphacite, within and along the fractures, attests to the introduction and consumption of fluid. Euhedral and oscillatory-zoned omphacite crystals are reported along fractures in eclogites (Austrheim et al. 1997). These indicate the presence of a free fluid phase, in accordance with fluid inclusion studies (Andersen et al. 1991). The oscillatory-zoned nature of the outer zircon rims is consistent with crystallization of zircon in the presence of a free fluid phase. Oscillatory- to sector-zoned zircon is reported in eclogite-facies rocks in the Alps (Rubatto and Hermann 2003).

Once initiated, eclogitization might have propagated rapidly in the dry protolith. Formation of fractures was maintained and enhanced by the volume reduction associated with eclogite-facies reactions (Austrheim et al. 1996; Jamtveit et al. 2000; Bjørnerud et al. 2002). Ultimately, reduction of the strength of the rock was the limiting factor in the production of new fractures and in the conversion of the protolith into eclogite (Bjørnerud et al. 2002), and not pressure and temperature conditions. Based on a reservoir-flux system model, Bjørnerud et al. (2002) estimate that the eclogitization process took place in <1 Myr. These various observations imply that the protolith might have been in the eclogite-facies P–T domain for a significant period before it started to react. Zircon is a product of eclogitization. Consequently, the zircon U–Pb age of 423±4 Ma (Fig. 4) records fracturing, fluid infiltration, and propagation of eclogitization in the metastable crust. It does not necessarily record entrance in the eclogite-facies P–T domain, or the maximum P–T conditions attained by the sample. It is also unlikely to record exit from the eclogite-facies P–T domain, because eclogite-forming reactions were inhibited before this point (Bjørnerud et al. 2002).

Implications for Caledonian orogenic evolution

In Scandinavia, the architecture of the Caledonian orogen mainly results from the Siluro-Devonian Scandian oblique collision between Laurentia and Baltica (Gee 1975; Fossen and Dunlap 1998; Cocks and Torsvik 2002; Roberts 2003), and subsequent Devonian extension (Andersen 1998; Fossen and Dunlap 1998). The Scandian collision post-dates formation of the youngest dated ophiolite complex (Solund–Stavfjorden ophiolite at 443±3 Ma; Dunning and Pedersen 1988). The onset of the Scandian collision is well constrained in the Wenlock epoch (428–423 Ma) by a number of independent constrains (Gee 1975; Andersen et al. 1998; Fossen and Dunlap 1998; Roberts 2003; Cocks and Torsvik 2002). The zircon data reported in this study yield a Wenlock–Ludlow age of 423±4 Ma (Fig. 4) for eclogite-facies overprint in the Lindås nappe, linking this overprint to the onset of the Scandian event. The Lindås nappe has a probable Baltica ancestry (Bingen et al. 2001b; Corfu and Andersen 2002). Consequently, the eclogite-facies overprint in this nappe relates to subduction to a depth of ca. 70 km (1.8–2.1 Gpa) or more of the crystalline crust of Baltica under the Laurentia plate. Rb–Sr mineral isochron ages for amphibolite-facies overprint in the Lindås nappe constrain the timing of exhumation of the nappe to mid-crustal level at 413±4 or 409±8 Ma (Bingen et al. 2001b; Glodny et al. 2002). Transport of the nappe pile on to the Baltica continent in S Norway is estimated by a group of muscovite and biotite 40Ar/39Ar plateau ages between 415±2 and 408±6 Ma (Fossen and Dunlap 1998). Available geochronological data thus suggest that exhumation of the Lindås nappe to mid-crustal level, its juxtaposition to the Upper Allochthon Hardangerfjord nappe containing low-grade lithologies, and its final thrusting on to the Baltica continent as part of the nappe pile took place between 415 and 408 Ma. Exhumation of the Lindås nappe and its transfer to the orogenic wedge thus took place during convergence of the orogen.

Conclusions

The trace-element content of metamorphic zircon can be qualitatively related to crystal-chemical properties of zircon and to partition of trace elements between minerals during crystallization of a metamorphic assemblage. Zircon rims in eclogite from the Lindås nappe have a trace element signature characterized by low Th/U ratio, low LREE content (minimum enrichment for Pr and Nd), variable HREE enrichment, and no Eu anomaly. These characters support coeval formation of zircon rims with garnet and clinozoisite in a sub-solidus system during eclogite-facies metamorphism. SIMS analyses of the rims give an age of 423±4 Ma, interpreted as the age of eclogite-forming reactions in the presence of a fluid phase. The age records fluid–rock interaction leading to eclogitization and not necessarily maximum pressure–temperature conditions reached by the sample. Eclogite-facies overprint in the Lindås nappe took place at the onset of the Scandian (Silurian) collision between Laurentia and Baltica, and corresponds to subduction of crystalline basement of Baltica ancestry. Exhumation of the Lindås nappe to amphibolite-facies conditions and its transfer to the orogenic wedge took place by 409±8 Ma during convergence of the orogen.