Lu–Hf geochronology of ultra-high-pressure eclogites from the Tromsø-Nappe, Scandinavian Caledonides: evidence for rapid subduction and exhumation

Geochronology of ultra-high-pressure metamorphic rocks is able to constrain the timing and rates of subduction-zone processes. Lu–Hf garnet dating has the potential to yield information about the timing of the prograde evolution of subducting rocks under increasing pressure. In combination with other methods, it thus allows constraining the complete P–T–t path with high precision. Ultra-high-pressure eclogites from the Tromsø Nappe, the structurally highest tectonic unit of the Scandinavian Caledonides in northern Norway, were dated using Lu–Hf geochronology on garnet. A sample from Tromsdalstind yielded an age of 448.3 ± 3.6 Ma, interpreted as dating prograde garnet growth due to preserved zoning in the major-element and Lu contents of garnet grains. A sample from the diamond-bearing locality Tønsvika yielded an identical age of 449.4 ± 3.3 Ma. Garnet from this sample shows a weak zoning in Ca content and near-homogeneous Lu content. These ages are identical within error among each other and with published U–Pb ages of peak-eclogite-facies zircon and rutile/titanite from exhumation-related leucosome veins. Consequently, the entire subduction–exhumation cycle leading to the ultra-high-pressure eclogites lasted only very few millions of years during the Late Ordovician.


Introduction
Ultra-high-pressure (UHP) metamorphic rocks record pressures above the quartz-coesite equilibrium line, i.e., more than ~ 2.7 GPa, reflecting either subduction to mantle depths or, alternatively, strongly non-lithostatic pressure at shallower levels. Except for a few cases (e.g., Zermatt-Saas ophiolites in the Western Alps; Reinecke 1998), exposed UHP metamorphic rocks are mostly derived from continental crust of the lower plate in a collisional setting that is entering a subduction zone after the subduction of oceanic lithosphere. The mechanisms leading to the subduction and exhumation of continental crust are a matter of scientific debate (e.g., Kurz and Froitzheim 2002;Michard et al. 1993;Warren et al. 2008). Our understanding may be promoted by the study of the pressure, temperature, structural, and temporal evolution of HP and UHP rocks, in particular eclogites. Most geochronometers, like U-Pb dating on zircon and monazite, date either the pressure peak or stages along the retrograde P-T path. As these minerals are accessory and not unambiguously part of the peak-pressure assemblage, they may have grown at different stages of metamorphism, but their relation to the P-T evolution can be controlled via their trace-element contents (e.g., Rubatto and Hermann 2007). Information on the prograde path can be gained from Sm-Nd and Lu-Hf dating of garnet, a member of the eclogite peak-pressure assemblage which generally grows during pressure increase (e.g., Baxter and Scherer 2013). Lu-Hf geochronology is especially suitable, since Lu strongly fractionates into garnet, so that much of Lu is incorporated into the early grown garnet cores. This leads to strong variations in Lu/Hf ratios between different garnet fractions and between garnet, omphacite, and whole rock, resulting in a good chance for well-defined isochrons dating garnet growth. The method has been successfully applied in a variety of UHP terranes including the Scandinavian Caledonides (e.g., Cutts and Smit 2018;Kylander-Clark et al. 2007). Diffusion of Lu and Hf in garnet is slow, so that the isotopic system is generally not equilibrated after the pressure peak, except if extremely high temperatures prevail for a long time span or cooling is slow. In such cases, Lu-Hf geochronology may date a stage of cooling. This may be recognized from homogeneous, diffusionally equilibrated Lu contents throughout garnet grains. In this study, we applied Lu-Hf dating to UHP eclogites from the Tromsø Nappe, the structurally highest tectonic unit of the Caledonides in Northern Norway. For these rocks, a detailed petrological study (Janák et al. 2012(Janák et al. , 2013aRavna and Roux 2006) as well as U-Pb dating of zircon, rutile, and titanite (Corfu et al. 2003;Ravna et al. 2017) had been performed before, so that the P-T evolution and the timing of peak metamorphism and the retrograde path are well constrained, but chronological data for the prograde path are still scarce.

Scandinavian Caledonides
The Scandinavian Caledonides represent a stack of nappes emplaced southeast to eastward onto the margin of Baltica, due to its collision with Laurentia during the Scandian phase in Silurian-to-Early Devonian time (430-400 Ma;Fossen 2010;Gee et al. 2008). The nappes are grouped into four allochthons based on their structural level in the nappe stack which, in most cases, corresponds to their palaeogeographic affiliation (Lower, Middle, Upper, and Uppermost;Gee et al. 1985;Strand and Kulling 1972). The Lower and Middle Allochthons are interpreted to be derived from proximal and distal parts of the Baltican margin, respectively. The Upper Allochthon is derived from oceanic crust and volcanic arcs of the Iapetus Ocean, and the Uppermost Allochthon from the continental margin of Laurentia. The basement of Baltica is not only exposed in the foreland; it also re-appears as large tectonic windows through the allochthons, such as the Western Gneiss Region (Fig. 1). There, the basement has been overprinted to various degrees by Caledonian deformation and metamorphism. Basement and allochthonous inliers of the Western Gneiss Region reached UHP conditions, leading to the formation of coesite (e.g., Smith 1984), diamond (Dobrzhinetskaya et al. 1995), and majoritic garnet (Scambelluri et al. 2008;Van Roermund and Drury 1998). UHP metamorphism in the Western Gneiss Region occurred at ~ 430-400 Ma (e.g., DesOrmeau et al. 2015;Griffin and Brueckner 1980;Krogh et al. 2011;Kylander-Clark et al. 2007, 2009Terry et al. 2000), i.e., during the Scandian phase. In recent years, two more UHP metamorphic units have been identified in the allochthons. The first one is the Tromsø Nappe (Uppermost Allochthon), where thermobarometry and thermodynamic modeling (Janák et al. 2012;Ravna and Roux 2006) as well as the finding of microdiamond in gneisses (Janák et al. 2013a) indicate UHP metamorphism at ~ 452 Ma, i.e., pre-Scandian (Corfu et al. 2003). The second one is the Seve Nappe Complex in Jämtland, Västerbotten, and Norrbotten representing the upper part of the Middle Allochthon (Gee et al. 2008), where the pre-Scandian UHP metamorphism was documented by thermobarometry and thermodynamic modeling of various rock types, and the occurrence of microdiamonds (Bukała et al. 2018;Fassmer et al. 2017;Gilio et al. 2015;Janák et al. 2013b;Klonowska et al. 2016Klonowska et al. , 2017Majka et al. 2014;Petrík et al. 2019). The tectonic framework of the pre-Scandian HP/UHP metamorphism in the allochthons is still unclear. Suggestions for the Seve Nappe Complex include subduction under an island arc (Dallmeyer and Gee 1986), a microcontinent (Roberts 2003), a composite of arc and microcontinent (Brueckner and Van Roermund 2004), or Laurentia (Gilio et al. 2015). UHP metamorphism of the Tromsø Nappe is thought to be related to the Taconian/Grampian Orogeny at the Laurentian margin (Brueckner and Van Roermund 2007;Corfu et al. 2003;Roberts and Gee 1985). Alternatively, it has been suggested that the Tromsø Nappe belonged to a part of the continental margin of Baltica, which had already been subducted before the Scandian collision, and was emplaced by an out-of-sequence thrust during the Scandian phase (Janák et al. 2012).
At Tønsvika, eclogites occur as lenses within marbles, clinopyroxenites, and diamond-bearing garnet-biotite gneisses, intruded by carbonatites. Eclogite is essentially Fig. 2 Simplified geological map of the Tromsø area, modified after Zwaan et al. (1998). Black squares mark the sample areas bi-mineralic (garnet + omphacite), with minor amphibole, white mica, rutile, carbonate, and quartz. In contact with carbonatite, the eclogite is strongly enriched in phlogopite and carbonate minerals (Ravna et al. 2017). The peak-pressure metamorphism was dated using U-Pb geochronology on zircon at 452.1 ± 1.7 Ma (Corfu et al. 2003). Peak P-T conditions of this eclogite were constrained to 3.4 GPa at 740 °C (Ravna and Roux 2006), and a diamond-bearing gneiss, sampled only a few meters away from our eclogite sample, reached 3.5 ± 0.5 GPa and 770 ± 50 °C as determined by thermodynamic modeling (Janák et al. 2013a). U-Pb-zircon dating of the carbonatite yielded ages of 454.5 ± 1.1 Ma (upper intercept age of prismatic zoned zircon) and 451.6 ± 1.7 Ma (concordia age of equant, subrounded zircons; Ravna et al. 2017). These authors proposed that the carbonatite was generated in situ by partial melting of carbonated eclogite under UHP conditions. The zircon ages of Ravna et al. (2017) thus give an additional constraint for the pressure peak of metamorphism.
At Holmevatn, chromium-rich kyanite eclogites (Janák et al. 2015) occur as layers and lenses of centimeter-to-meter size in serpentinized garnet peridotite ). Metamorphic P-T conditions of these rocks are ca. 2.6-3.0 GPa and 750-800 °C Janák et al. 2015). The garnet peridotite experienced two temperature maxima after peak-pressure metamorphism ): 750-800 °C at 1.8 GPa and > 750 °C at 1.0 GPa. The latter is considered to be related to the juxtaposition of the Tromsø Nappe onto the hot Skattøra Migmatite Complex (Krogh et al. 1990;Ravna and Roux 2006;Skjerlie 2002).
At Tromsdalstind, the largest eclogite body within the Tromsø Nappe builds up the uppermost part of the mountain (Fig. 2). The mafic rocks of Tromsdalstind are interpreted to be metabasalts or gabbros, indicating a magmatically active setting (Corfu et al. 2003). Besides mafic rocks this body also contains tonalitic/trondhjemitic orthogneiss. One of those gneisses gave a protolith zircon age of 493 + 5/− 2 Ma (Corfu et al. 2003). This suggests that the intrusive activity was contemporaneous with the early generation of ophiolites and arc-related magmatism on both sides of the Iapetus Ocean (Dunning and Pedersen 1988). Corfu et al. (2003) also dated titanite and rutile in eclogites, leucosomes in eclogites, and calc-silicate rocks, giving ages between 448.8 ± 1.4 Ma and 451.0 ± 1.4 Ma that according to these authors postdate the eclogite-forming event. P-T conditions of kyanite-phengite eclogites from Tromsdalstind calculated from thermobarometry and thermodynamic modeling are 3.2-3.5 GPa at 730-780 °C (Janák et al. 2012). These UHP conditions are supported by microstructural features like polycrystalline quartz inclusions in omphacite indicating breakdown of coesite and needle-like inclusions of SiO 2 in omphacite, which may have been exsolved from Ca-Eskola clinopyroxene (Janák et al. 2012). These authors also constrained a clockwise P-T path with a maximum temperature of ca. 800 °C at 2.2-2.7 GPa (Janák et al. 2012). Heating during decompression led to partial melting which produced amphibole-bearing pegmatite veins (Corfu et al. 2003). Stevenson (2005) described two stages of migmatization of eclogite post-dating peak-pressure metamorphism. Krogh et al. (1990) proposed a second episode of metamorphism at low pressure following post-UHP decompression, involving renewed pressure increase from 0.8 GPa up to 1.0-1.1 GPa at maximum temperatures of 665 °C, which they related to an episode of thrusting and emplacement of the Nakkedal Nappe Complex onto the Lyngen Nappe Complex (Upper Allochthon).
The tight grouping of ages demonstrates that the UHP event with subsequent uplift and partial melting of eclogite happened within only a few million years (Corfu et al. 2003). To add constraints on the timing of subduction of the Tromsø Nappe, we dated UHP eclogites from Tromsdalstind and Tønsvika using Lu-Hf geochronology on garnets.

XRF and electron microprobe analysis
Mechanical sample preparation was done at the Institute of Geosciences, University of Bonn (Germany). One aliquot of each eclogite sample was crushed and powdered to analyze the bulk rock composition of both eclogite samples with a PANanalytical Axios X-ray fluorescence spectrometer. The remaining aliquot was crushed for digestion and subsequent Lu and Hf separation via column chemistry to future isotope measurements. Electron microprobe analyses were carried out using a JEOL 8200 Superprobe at the Institute of Geosciences in Bonn. A beam current of 15 nA and an accelerating voltage of 15 kV were set for single spot measurements. Well-defined natural minerals (Fe-magnetite, Mg-olivine, Ca-anorthite, Na-scapolite, Si/Al-garnet, and K-sanidine) and pure metals were used as standards (Jarosewich et al. 1980). Major-element distribution maps of Ca, Mg, Fe, and Mn in garnets were obtained with a beam current of 50 nA, an accelerating voltage of 15 kV, and a measurement time of 100 ms per point.

Laser ablation analysis
The trace-element contents in selected garnet grains were measured via LA-ICP-MS using a Thermo Scientific X-Series 2 Q-ICP-MS, coupled to a Resonetics RESOlution M50-E 193 nm Excimer Laser Ablation System at the Institute of Geosciences, University of Bonn (Germany), following the procedure described in Kirchenbaur et al. (2012) and Sandmann et al. (2014). Laser ablation spot sizes were 44 and 58 µm, depending on the garnet grain size, and 12-13 spots per garnet profile were chosen manually to avoid inclusions. Measurements were performed with a laser repetition rate of 15 Hz and a laser fluency at the sample surface of 7 J/ cm 2 . Nuclides of 29 Si, 43 Ca, 47 Ti, 55 Mn, 91 Zr, 177 Hf, 178 Hf, 180 Hf, and 175 Lu were analyzed, measuring for 15 s on the gas background and 30 s on each measured point, followed by 15 s of gas background to allow the complete washout of the sample's signal. Count rates were normalized to 29 Si, as the internal standard, and to NIST-612 as an external glass reference material (Jochum et al. 2011). Normalized count rates were then converted into concentrations using the procedure described by Longerich et al. (1996).

Lu-Hf isotope measurements
The crushed parts of the eclogites were sieved to retrieve garnet-and omphacite-rich mineral fractions for Lu and Hf isotope measurements. Only the fractions with the highest amount of optically inclusion-poor minerals were separated with a Frantz L-1 magnet separator. Fractions of 250-355 µm and 355-500 µm were used for handpicking of nearly inclusion-free grains of garnet and omphacite. Threeto-four whole-rock powders, three garnet, and one omphacite separate were measured for each sample. A mixed 176  Hf tracer was added to the samples prior to digestion. Two different procedures for sample digestion were applied: (1) a tabletop digestion procedure after Lagos et al. (2007) was applied to the mineral separates and 2-3 of the whole-rock powders, using 6 ml of a mixture of conc. HF and conc. HNO 3 (2:1), which was left for at least 48 h at 120 °C. Afterwards, 1 ml of HClO 4 was added to the samples and dried down at 180 °C. The samples were then dissolved in 6 N HCl at 120 °C and dried down again. The whole digestion procedure was repeated until complete digestion was accomplished.
(2) One whole-rock powder of each sample was digested in a Parr® bomb for 4 days at 180 °C using a 1:1 mixture of conc. HF and conc. HNO 3 to achieve complete dissolution, including Hf-rich phases like zircon or rutile. All separates and whole-rock splits were dissolved in 5 ml 3 N HCl to apply the one-column chemistry after Münker et al. (2001), using Eichchrom® Ln-spec. Eluted Lu and Hf cuts were dried down at 120 °C on a hotplate and organics were removed with 1 ml of a mixture of 0.14 N HNO 3 (Lu) or 0.56 N HNO 3 /0.24 N HF (Hf) and 30% H 2 O 2 (9:1). Hf cuts were further purified by applying a clean-up chemistry, after Lagos et al. (2007).
All Lu and Hf isotope measurements were carried out using a Thermo Scientific Neptune Multicollector ICP-MS at the joint Cologne-Bonn isotope facility. The measured 176 Hf/ 177 Hf was given relative to the Münster AMES standard ( 176 Hf/ 177 Hf = 0.282160), which is isotopically indistinguishable from the JMC-475. The instrumental bias on Hf was corrected using a 179 Hf/ 177 Hf ratio of 0.7325 and the exponential law. Isobaric interferences on 176 Hf and 180 Hf were accounted for by measuring 173 Yb, 175 Lu, 181 Ta, and 183 W and using their natural isotopic compositions . Likewise, the interferences on 176 Lu were corrected by monitoring 173 Yb and 177 Hf. To account for the mass bias correction of Lu ratios, naturally occurring Yb in the Lu cuts was used Vervoort et al. 2004). The typical external reproducibility for Hf measurements is ± 0.4 epsilon units. Isochrons were calculated using Isoplot 4.1 (updated version of Isoplot 2.49, Ludwig 2001). A decay constant of λ 176 Lu = 1.867 × 10 -11 a −1 Söderlund et al. 2004) and 2 σ uncertainties for 176 Lu/ 177 Hf and 176 Hf/ 177 Hf ratios were used for isochron calculations.

Sample description
For Lu-Hf-dating, we selected two eclogite samples from Tønsvika and Tromsdalstind which are described below. Detailed petrology and P-T path reconstructions of eclogites from both locations can be found in Ravna and Roux (2006) and Janák et al. (2012). Major-element compositions of both samples are typical for metabasalts and shown in Table 1.

Tromsdalstind
The peak-pressure assemblage of the Tromsdalstind eclogite comprises garnet, omphacite, rutile, quartz, and probably epidote (Fig. 3a, c). Representative microprobe analyses of the major minerals in this sample can be found in Table 2. Omphacite is partly converted to symplectites of plagioclase and Na-poor clinopyroxene (diopside), with or without amphibole. Elongated quartz rods in omphacite (Fig. 3d), which have been also observed in kyanite eclogite from the same locality (Janák et al. 2012), suggest exsolution from the more silicic Ca-Eskola clinopyroxene stable at UHP conditions (e.g., Katayama et al. 2000;Smith 1984).
Most garnets contain a few inclusions, mostly concentrated in the garnet cores. Inclusions comprise clinopyroxene, amphibole, epidote, quartz, rutile, and ilmenite. Zoning of major elements in garnet differs between garnet grains in the sample. Fe is only weakly zoned (decreasing from rim to core) to unzoned in all garnets, while Mg is more strongly zoned and always increases towards the rim. Therefore, FeO/(FeO + MgO) has always a bellshape pattern in profiles through garnet grains (Fig. 4). Ca content in all analyzed garnets decreases from core to rim (Figs. 4, 5). Mn shows an enrichment in the core in comparison with the rim in large garnet grains (Fig. 4), while small garnet grains are nearly unzoned with respect to Mn (Fig. 5). The general composition of garnet is Alm 43-52 Gr 18-32 Py 19-37 Sps 0.5-1.9 . Amphibole occurs at the rims of some garnets. In some garnet grains, there are veins filled with matrix minerals: omphacite with a matrix-like composition, amphibole with a composition as those at garnet rims, and albite. Clinopyroxene in the matrix shows homogenous compositions with Jd 41-43 and can be classified as omphacite. Omphacite grains of similar chemical composition can be found also as inclusions in garnet. In symplectitic intergrowth with plagioclase, omphacite is relatively Na-poor (Jd 22-28 ). Amphibole occurring as inclusions in garnet shows highly variable compositions and can be classified as taramite, barroisite, or sadanagaite. Some of these inclusions have a remarkable high Ti content of up to 4 wt%. At some of the garnet rims, amphibole with sadanagaite composition occurs in a flame-like structure, where the two minerals seem to be intergrown with each other (Fig. 3d). Amphibole grains at garnet rims and in flame-like structures around garnets correspond to sadanagaite. Amphibole can also be found in symplectites with plagioclase after omphacite, where it can be classified as taramite. Plagioclase is found only in symplectitic intergrowth with omphacite or amphibole and classifies as oligoclase with a varying chemical composition (Ab 79-87 ). Epidote is found only as inclusions in garnet, so it is difficult to determine if it is part of the peakpressure paragenesis or a prograde mineral which was not stable at peak-pressure conditions.

Tønsvika
The peak-pressure mineral assemblage of the eclogite from Tønsvika consists only of garnet and omphacite with minor rutile (Fig. 3b, e, Table 2). This eclogite shows almost no retrogression apart from some scarce symplectites of Napoor clinopyroxene and plagioclase after omphacite. Garnet grains are anhedral without retrogression at their rims. They show rather homogenous compositions with respect to major elements, except for a slight increase of Fe and a decrease of Mg at the outer rims, and a faint zoning of Ca, showing a slightly Ca-richer core and Ca-poorer rim (Fig. 6). This Ca zoning is similar but weaker than in the sample from Tromsdalstind. Overall garnet composition is Alm 43-53 Gr 22-26 Py 21-33 Sps 0.2-0.8 . Inclusions in garnet are rare, comprising clinopyroxene, amphibole, paragonite, biotite, plagioclase, pyrite, titanite, rutile, zircon, and carbonate. The composition of clinopyroxene inclusions in garnet corresponds to omphacite with Jd 26-27 or augite and is similar to those in the matrix, whereas clinopyroxene in the symplectites is diopside (Jd 13 ). Amphibole occurs only as inclusions in garnet and therefore is not regarded as a peakpressure phase. Both clino-and ortho-amphibole have been identified as sadanagaite and gedrite. Inclusions of plagioclase in garnet are nearly pure albite (Ab 98 ), whereas plagioclase in symplectites with clinopyroxene has a composition of Ab 93 . Pyrite is abundant in this sample and occurs both in the matrix and as inclusions in garnet. Rutile and titanite are often associated with pyrite, but occur also as tiny inclusions in garnet. Carbonates, mostly dolomite, are quite abundant in the matrix and as inclusions in garnet (Fig. 3f).

Lu distribution in garnet
In Figs. 4, 5, and 6, Lu profiles are plotted through the same garnet grains that were measured with electron microprobe. The large garnet grain No. 1 from Tromsdalstind (Fig. 4), which exhibits prograde growth zoning in Mn, shows an asymmetric Lu profile with a maximum close to one rim, a shoulder in the core, a minimum between the core and the other rim, and a small secondary maximum close to the  (Fig. 5) with a homogenous Mn content shows a Lu peak in the core and two equally high peaks immediately at the rim of garnet. In the garnet grain from the Tønsvika eclogite (Fig. 6), the Lu content is almost homogeneous, varying between 0.4 and 0.55 ppm.

Lu-Hf geochronology
Lu and Hf isotopic compositions of both studied eclogite samples are shown in Table 3. For the sample from Tromsdalstind, we plotted one bomb-digested and two table-top-digested whole-rock splits, one clinopyroxene separate, and three garnet separates on an isochron which yields an age of 448.3 ± 3.6 Ma (MSWD = 1.19, n = 7). For the Tønsvika eclogite, we also obtained a statistically good isochron with all measured data points which gives an age of 449.4 ± 3.3 Ma (MSWD = 1.14, n = 8). One of the garnet separates has a distinctly larger error, because uncertainties also include propagated errors from interference corrections and interferences of Lu and Yb on Hf were higher than in the other separates. As the larger error does not influence the age, we see no reason for excluding this point from the isochron. Both isochrons are shown in Fig. 7. The ages of both eclogites are identical within their uncertainty.

Subduction and exhumation of the Tromsø Nappe
Garnets from the Tromsdalstind eclogite retained their Lu growth zonation even where major elements were partly thermally re-equilibrated (Fig. 5). In grain No. 1, we interpret the strongly variable Lu contents as a proof that it has not been homogenized after garnet growth, although the shape of the profile is difficult to interpret. Such irregular Lu distribution patterns may potentially be interpreted as directly inherited from Lu-rich and Lu-poor minerals that were present before garnet grew, in situations of low intergranular REE mobility (Moore et al. 2013). In grain No. 2, we interpret the central peak as resulting from the strong fractionation of Lu into garnet as it began to grow. The maxima at the rim may be explained by a progressive temperature increase during the final stage of garnet growth, leading to an increase of the diffusion rate and thereby allowing the diffusion of Lu from the more distant parts of the rock into the garnet grain (diffusion-limited REE uptake; Skora et al. 2006). Alternatively, resorption of garnet and back-diffusion of Lu into the remaining garnet could also lead to a Lu increase at the rim (Kelly et al. 2011), but the well-preserved shape and major-element zonations of this garnet grain speak against a significant resorption. As Lu is concentrated in the core on one hand and near the rim on the other hand, the resulting age is likely related to prograde garnet growth. In contrast to Tromsdalstind, the garnet in the eclogite from Tønsvika shows nearly homogeneous Lu content. This could be explained by post-growth diffusional equilibration. In this case, the age would likely represent a cooling age. However, two observations speak against this: first, the age for Tønsvika is identical within error, but the mean age is slightly older than for Tromsdalstind, which would be difficult to reconcile with a post-peak age for Tønsvika. Furthermore, a weak but clear prograde zoning is seen in the Ca content of garnet (Fig. 6). If a part of the original Ca zoning survived diffusional equilibration, the even more immobile Lu is unlikely to have become completely equilibrated.
Under these circumstances, we find it more logical to assume that the uniform Lu content is an original feature of the garnet grain. This particular garnet grain could have started growing relatively late, when most of the available Lu had already been used up in the cores of earlier grown garnets. Alternatively, if a narrow Lu peak existed in the core of the grain, the exact center with the Lu peak may have been missed by the cutting of the thin section. Cutts and Smit (2018) described similarly homogenous profiles that they interpreted as being recrystallization during prograde highgrade metamorphism. Therefore, we also interpret the age from Tønsvika to represent prograde garnet growth (Fig. 8).
The Lu-Hf ages for both eclogite samples from the Tromsø Nappe overlap with the ages obtained by U-Pb

Conclusion
We obtained the first Lu-Hf isochron ages for eclogites from the Tromsdalstind and Tønsvika localities in the Tromsø Nappe. The samples yielded ages of 448.3 ± 3.6 Ma and 449.4 ± 3.3 Ma which we interpret as dating prograde garnet growth. All ages for prograde, peak, and retrograde parts of the UHP metamorphic cycle in the Tromsø Nappe overlap around 454-448 Ma. The maximum duration of the subduction-exhumation cycle is thus c. 6 Ma, and probably less. The palaeogeographic framework of subduction and  Ravna and Roux (2006). Different dating methods and their attribution to different parts the PT path are marked in different colors. Metamorphic facies grid is from Okamoto and Maruyama (1999). BS blueschist facies; EA epidote amphibolite facies; AM amphibolite facies; HGR high-pressure granulite facies; Lw-EC lawsonite eclogite facies; Ep-EC epidote eclogite facies; Amp-EC amphibole eclogite facies; Dry-EC dry eclogite facies. The quartz-coesite line is calculated from thermodynamic data of Holland and Powell (1998). b Graphic presentation of isotopic ages with the same color coding as in part (a), which shows a fast subduction-exhumation cycle. Lu-Hf ages are from this study and U-Pb ages are from Corfu et al. (2003) UHP metamorphism in the Tromsø Nappe, i.e., whether the processes occurred along the Baltican or the Laurentian side of Iapetus, is still unclear. More work on the structural and pressure-temperature evolution of the Tromsø Nappe and adjacent tectonic units is needed to sort this out.
Acknowledgements Open Access funding provided by Projekt DEAL. This work was supported by the German Science Foundation (DFG) Grant no. FR700/18-1 to N. F. and by the Slovak Research and Development Agency project APVV-18-0107 to M.J. We would like to thank two anonymous reviewers for their constructive criticism which strongly improved the paper and gave it a clearer focus.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.