Introduction

The Cap de Creus peninsula (NE Spain) in the easternmost end of the Pyrenean Axial Zone is a remarkable center of interest for geologists because of its well-exposed outcrops that reveal a complex tectonic, metamorphic and magmatic history. Several episodes of deformation and regional metamorphism are accompanied by calc-alkaline and peraluminous magmatism and affected a late Proterozoic series of metasediments and metavolcanics during the Variscan orogeny. An important group of mineralized granitic pegmatites is associated with these tectonic events.

The study of granitic pegmatites in Europe is gaining interest because these highly fractionated rocks are important sources of industrial minerals and strategic metals such as Li and the high field strength elements Nb-Ta and Sn (Linnen et al. 2012). In Europe rare-element granitic pegmatites are abundant in the Variscan terrains (French Massif Central, Iberian Massif, Moldanubian domains of Czech Republic, Slovakia and Germany). The Cap de Creus area is one of these fields.

Studies on the geology of Cap de Creus essentially dealt with structural aspects and tectonic interpretations of the Variscan orogeny (Carreras 1975, 2001; Carreras and Druguet 1994, 2013; Druguet and Hutton 1998; Druguet 2001; Fusseis et al. 2006) and the petrology of the peraluminous pegmatites (Corbella and Melgarejo 1993; Alfonso et al. 1995, 2003; Alfonso and Melgarejo 2008). Apart from the recent study by Druguet et al. (2014) that dated a granodiorite and a quartz diorite intrusion, geochronology of the Variscan in Cap de Creus was only indirectly inferred from other studies in the Eastern Pyrenees (e.g., Aguilar et al. 2014).

The peraluminous pegmatite swarm that crops out in the Cap de Creus area consists of different types of pegmatites, ranging from type I (K-feldspar pegmatites, least evolved) to type IV (albite pegmatites, most evolved), distributed along zones of increasing intensity of the deformation and metamorphic grade. Consequently, dating the Cap de Creus pegmatites could better frame the geological history of the Variscan orogeny in the Pyrenees and particularly could help to place a timeline on the succession of tectonic events. The pegmatites contain U-bearing accessory minerals such as zircon, xenotime and columbite-group minerals (CGM). The aim of our study is to date these minerals in order to constrain the pegmatite emplacement. Age correlations with published geochronological data on peraluminous granite and migmatitic rocks from the area may also help to establish a model for the anatectic versus granitic origin of the mineralized pegmatites during the Variscan orogeny.

Columbite-tantalite is well suited for U-Pb age determination of pegmatite emplacement. It generally contains around 500 ppm U, but values up to 10,000 ppm are not uncommon, and it accommodates low common Pb. Solid state U-Pb diffusion is also thought to be minor in CGM (Romer and Wright 1992). Columbite-tantalite is a common primary magmatic mineral phase in pegmatites, and its refractory nature makes it resistant to hydrothermal alteration and weathering. Large crystals several hundred microns in size allow detailed characterization of their internal textural features, thus permitting precise location of the ablation laser spots with respect to possible late precipitation phases, metamict zones and U-bearing inclusions. Columbite-tantalite is a good alternative to date pegmatites where zircon is too U-rich and highly metamict. The present study gives ages from both zircon and CGM associated together in the same samples, using a thorough sorting out of metamict grains.

Geological setting

General features

The study area is located in the Northern Cap de Creus peninsula (Fig. 1). It consists of metasedimentary rocks (metagreywackes, metapelites, rare quartzites) with minor metavolcanic intercalations. The protolith of this sequence is referred to as the Cadaqués series (Navidad and Carreras 1995) and is considered Neoproterozoic in age. During the Variscan, the rocks of the Cadaqués series were affected by polyphase deformation with three main deformation episodes (D1, D2, D3; Druguet 2001), the two first ones occurred during the prograde LP-HT regional metamorphism and the last one during late shearing events under retrograde conditions. Metasediments show a gradient from the chlorite-muscovite zone in the south (out of the map in Fig. 1) to the sillimanite-K feldspar zone in the north. Locally, migmatites were formed in the sillimanite-K feldspar zone (Druguet et al. 1997). High to medium grade schists in the northern part of the area are extensively intruded by pegmatite (Fig. 1; Carreras and Druguet 1994; Bons et al. 2004).

Fig. 1
figure 1

a Geological map of the central and eastern Pyrenean Axial Zone (modified from Druguet et al. 2014). b Geological map of the north Cap de Creus area, showing the distribution of the four pegmatite types (modified from Corbella and Melgarejo 1993) within the three metamorphic zones (Sil-Mu: sillimanite-muscovite; Sil-Kfs: sillimanite-K-feldspar; Crd-And: cordierite-andalusite; Bt: biotite; the chlorite-muscovite zone is further to the south, out of the map). The main shear zones of the mylonite belt and the sampled pegmatite localities are also shown. Labels 2a, 2b and 2c correspond to the locations of photographs in Fig. 2. Modified from Druguet and Carreras (2006)

Full size image

The oldest deformation in the area (D1) led to the development of a first continuous and penetrative N-S trending bedding-parallel schistosity (Sl) in the metasediments prior to the metamorphic climax. Later intense and inhomogeneous D2 deformation led to folding and shearing of S1 with upright or steeply inclined axial surfaces which trend approximately NE-SW in less deformed areas and E-W in more deformed areas. Parallel with the increase of metamorphic grade, the intensity of the D2 event increases from south to north, where a 200 m thick E–W trending zone of high strain is observed and S1 is transposed into a steeply dipping composite S1/S2 foliation with a few relics of tight to isoclinal D2 folds (Druguet and Carreras 2006). L2 lineations are generally steeply plunging towards the NW. D2 structures formed around peak metamorphic conditions, as shown by the presence of synkinematic sillimanite and by partial melting of metasediments. A third episode of deformation of unknown age occurred under retrograde (greenschist facies) metamorphic conditions and was characterized by strain localization that gave rise to a network of D3 shear bands with predominantly reverse-dextral movement. These form the classical Cap de Creus shear zones and mylonites which overprint and therefore postdate all the preexisting structures (Carreras 2001). Unambiguous field relationships show that pegmatites intruded after D1 and before D3, that is more or less contemporaneously with D2 (Fig. 2).

Fig. 2
figure 2

Field photographs showing the relationships between pegmatite dykes and deformation phases (see location on Fig. 1). a Pegmatitic vein crosscutting bedding/S1 and being slightly folded by D2. Zone of low D2 strain south of Puig de Culip. b Syntectonic pegmatite dykes are folded by D2 in a zone of strong D2 strain, as shown by a penetrative composite S1/S2 fabric. North of Puig de Culip. c Mylonitic foliation (S3) affecting schists and a pegmatite body at the margin of a late dextral zone. NW Cala Culip

Full size image

Published geochronological data of Cap de Creus

The Cadaqués metasedimentary series, although not directly dated, is inferred to be older than the El Port de la Selva gneiss (located about 5 km west of the Punta dels Farallons, Fig. 1b), whose igneous precursor intruded the metasedimentary series and was dated at 553.0 ± 4.4 Ma (Castiñeiras et al. 2008). Zircons from the Tudela migmatite (northern part of the Cap de Creus, Fig. 1b) yield inherited ages from the Precambrian protolith, with two main age clusters at c. 2.9–2.2 Ga and c. 730–542 Ma (Druguet et al. 2014). However, based on field structural relationships, Druguet et al. (2014) interpret that the migmatization event was synchronous with the emplacement of a syntectonic quartz diorite from the Tudela migmatitic complex dated at 298.8 ± 3.8 Ma by these authors. The western and southern granitoid stocks, known as Rodes and Roses stocks respectively (Fig. 1a), consist of granodiorite and tonalite and were emplaced within lower-grade rocks at 290.8 ± 2.9 Ma (Druguet et al. 2014). For further geochronological data in the Pyrenees, the reader can refer to Laumonier et al. (2004); Cocherie et al. (2005); Casas et al. (2010); Liesa et al. (2011), Aguilar et al. (2014), Denèle et al. (2014) and Casas et al. (2015).

Pegmatites

Four types of peraluminous, lithium-cesium-tantalum-family pegmatites (after the classification of Černỳ and Ercit 2005) were distinguished among the ~400 bodies that crop out in Cap de Creus. This distinction was made on mineralogical and textural criteria (Corbella and Melgarejo 1993). Type I pegmatites are barren with graphic textures and a relatively simple concentric structure roughly consisting of border, first intermediate and second intermediate zones; in addition to biotite and muscovite, peraluminous minerals as cordierite, sillimanite, andalusite, almandine and schorl are very common in all these zones. Xenotime is associated with zircon in this pegmatite type. Type II pegmatites are transitional with the most evolved pegmatites; the main differences with type I are the occurrence of a well developed quartz core and the existence of late albite units. In addition to the above mentioned peraluminous minerals, these pegmatites may contain chrysoberyl, gahnite, green beryl, Ca-Fe-Mn-Mg-phosphates and some Be- and Al-phosphates. Nb-rich minerals of the columbite group are scarce in all units as well as wolframite, Sc-rich rutile and uraninite. The internal structure of type III pegmatites is more complex with large quartz cores and well developed albite and quartz-muscovite replacement units. Biotite is absent and schorl is scarce; garnet is enriched in the spessartine component. White beryl, montebrasite and Li-Fe-phosphates are common in the second intermediate unit and in the albite or quartz-muscovite replacement veins, as well as Ta-rich minerals of the columbite group, cassiterite and uraninite. Type IV pegmatites are the most evolved in the field. In addition to the above mentioned units, they may also contain late Al-phosphate veins. Beryl or chrysoberyl, montebrasite and Li-Mn phosphates are common. Ore minerals consist of Ta-Mn rich members of the columbite group, as well as cassiterite, tapiolite and aeschynite. Following the classical pegmatite classification (Černỳ and Ercit 2005), type II pegmatites belong to the beryl-columbite subtype, type III belong to the beryl-columbite-phosphate subtype, and type IV belong to the albite subtype. Type I pegmatites are nearly sterile and may be considered as pegmatitic granite rather than pegmatite sensu stricto.

The four types of pegmatites occupy different zones parallel to the tectono-metamorphic zoning (Fig. 1): types I and II occur in high-grade and high-D2 strain rocks of the migmatite and sillimanite-muscovite zones that lie along the northern coast, whereas types III and IV occur in medium-grade metamorphic rocks of the cordierite-andalusite zone to the south. The size and frequency of the pegmatite bodies decrease from type I to type IV (Corbella and Melgarejo 1993): along the northern coast, large stocks of pegmatitic granite may reach a length of several hundreds of meters and a width of more than 50 m, whereas 2 km further to the southwest, only a few bodies outcrop with a maximum length of 30 m.

In general, pegmatite dykes follow the main S2 foliation, and are locally affected by late shearing. Some dykes that were emplaced oblique to S2 foliation are affected by ductile folding.

Sampling and analytical procedure

Rock samples were collected on six dykes representative of type I, III and IV pegmatites (Fig. 1), but only two locations (L3 and L7) were relevant for dating. Polished sections and thin sections were prepared for mineralogical description and investigation of the mineral textures to determine the primary magmatic versus secondary nature of the dated minerals. In samples where zircon suitable for dating was observed, a larger sample volume was crushed and zircon was separated using a standard separation procedure: 1) gravity separation using either a shaking table or a gold pan; 2) heavy liquid separation using tetrabromo-ethane; 3) magnetic separation to eliminate the metamict zircon grains; 4) heavy liquid separation using metyleniodide. Thirty to 40 separated grains per sample were mounted in lines in epoxy resin blocks that were subsequently polished. In L7 sample, xenotime occurs in direct contact with zircon and was therefore dated together with it. Because of their coarse-grained habit, CGM crystals were directly dated on the polished sections.

Back-scattered electrons (BSE) images were taken for each zircon and CGM grain, and crack- and inclusion-free domains were selected for the laser spots. In sample L7 where zircon shows complex textures, additional cathodoluminescence (CL) images were taken in order to highlight the metamict parts of the altered domains.

Quantitative chemical analyses were carried out with a Cameca SX50 electron probe micro-analyzer (EPMA) using a 15 kV accelerating voltage, 20 nA beam current, 1 μm beam diameter, 10 s and 5 s acquisition times on peak and background respectively, natural and synthetic calibrant materials (Ta, Nb, and W metals, cassiterite, zircon, hematite, wollastonite, MnTiO3, ScPO4, UO2, ThO2, Pb glass), and ZAF correction procedures.

Uranium-lead dating of CGM, xenotime and zircon was carried out in-situ at the Goethe University of Frankfurt (GUF) using a slightly modified method as the one previously described in Gerdes and Zeh (2006, 2009) and Zeh and Gerdes (2012). Thermo-Scientific Element II sector field ICP-MS was coupled to a Resolution M-50 (Resonetics) 193 nm ArF Excimer laser (CompexPro 102, Coherent) equipped with two-volume ablation cell (Laurin Technic, Australia). Samples were ablated in a helium atmosphere (0.6 l/min) and mixed in the ablation funnel with 0.7 l/min argon and 0.02 l/min nitrogen. Signal strength at the ICP-MS was tuned for maximum sensitivity while keeping oxide formation below 1 %. The laser was fired with 5.5 Hz at a fluence of about 2–3 J cm−2. This yielded with the above configuration at a spot size of 30 μm and depth penetration of 0.6 μm s−1 a sensitivity of 11,000–13,000 cps/μg g−1 for 238U. Raw data were corrected offline for background signal, common Pb, laser induced elemental fractionation, instrumental mass discrimination, and time-dependent elemental fractionation of Pb/U using an in-house MS Excel© spreadsheet program (Gerdes and Zeh 2006, 2009).

Laser-induced elemental fractionation and instrumental mass discrimination were corrected by normalization to the reference zircon GJ-1 (0.0982 ± 0.0003; ID-TIMS GUF value). Repeated analyses of the reference zircon Plesovice and 91,500 (Slama et al. 2008; Wiedenbeck et al. 2005) during the same analytical session yielded an accuracy of better 1 % and a reproducibility of <2 % (2 SD). The same applies to monazite run as secondary standards normalized to GJ-1 using the same analytical setting and tune parameter except of the spot size: 15 μm relative to 33 μm for GJ-1. Repeated analyses (n = 9) of the reference monazite Manangotry and Moacir (Horstwood et al. 2003; Gonçalves et al. 2016) yielded an accuracy of around ~1 % and reproducibility of 2–3 % (2 SD). This is in line with previous studies at GUF that have shown that LA-SF-ICP-MS with non-matrix matched standardization can yield precise and accurate U–Pb ages for different phosphate minerals (e.g., Meyer et al. 2006; Millonig et al. 2013 and references therein). Thus no correction for phosphate matrix have been applied for xenotime analysis. However, in case of CGM the Coltan 139 (Gäbler et al. 2011) was used as matrix matched standard. More details on the operating conditions and instrument settings are given in Gerdes and Zeh (2006, 2009) and in data Tables 1 and 2. All uncertainties are reported at the 2sigma level.

Table 1 Age data obtained by LA-ICP-MS for pegmatite L3
Full size table
Table 2 Age data obtained by LA-ICP-MS for pegmatite L7
Full size table

One zircon age was duplicated at the Laboratoire Magmas et Volcans of Clermont-Ferrand, equipped with an Excimer 193 nm laser coupled to a quadrupole Agilent 7500 ICP-MS, using zircon GJ-1 as reference material (analytical techniques described in Paquette et al. 2014). However, the Thermo-Scientific Element II sector field ICP-MS in Frankfurt is more adapted to Hercynian ages since it has a better precision on the U/Pb ratios and its higher sensitivity allows better correction for common Pb. For CGM ages, an external manganotantalite crystal (Coltan 139; Gäbler et al. 2011) from Madagascar was used to correct for matrix-dependent U/Pb elementary fractionation. This reference was used to date CGM from African pegmatites with the goal to fingerprint illegally mined coltan (Melcher et al. 2008, 2015). The Coltan 139 reference is a large manganotantalite crystal that is isotopically and chemically homogeneous at the micrometer scale, has a U concentration of about 1600 ppm and low common Pb (Gäbler et al. 2011). It displays an intercept age of 505.6 ± 3.4 Ma obtained by LA-ICP-MS and verified by ID-TIMS. Where necessary, the various textural domains of zircon and CGM were dated, and most crystals were measured in both core and rim for comparison. Concordia diagrams were plotted using Isoplot 3.7 (Ludwig 2008).

Results

Dated pegmatites and minerals

Six pegmatite dykes (Fig. 1) were studied but only two of them displayed CGM, xenotime and/or zircon crystals that were suitable for dating. Textural and chemical features of CGM and zircon were studied in all pegmatites where they were observed. In type I pegmatites L1 and L5, no CGM was found and most zircon crystals were too small (c. 10 μm) to be dated. In type II pegmatite L2, zircon crystals were too small and altered to be dated, but CGM displayed a few prismatic crystals that could be dated in Clermont-Ferrand. However, standardization on zircon lead us to exclude those CGM ages. In type IV pegmatite L4, the zircon crystals were larger (c. 100 μm), but they were highly metamict and rich in uraninite inclusions, and displayed uninterpretable ages due to loss of radiogenic Pb on one hand and entry of common Pb on the other. Although CGM are also common in type IV pegmatites, they are generally associated with the aplitic albite units and their crystals were too fine-grained to be dated. The two dated pegmatites belong to type I (L7) and type III (L3). Their geographic coordinates and description are given in Table 3. Pegmatite L3 (type III) is a well-zoned pegmatite located in the cordierite-andalusite metamorphic zone. The dyke is 100 m long and 10 m wide. Pegmatite L7 (type I) is a large (200 × 20 m) homogeneous dyke emplaced in the sillimanite-K-feldspar zone.

Table 3 Description of the two dated pegmatite localities
Full size table

Textural and chemical features of CGM and zircon

In type II to type IV pegmatites, columbite-group minerals occur as millimeter to centimeter-sized tabular crystals included in major mineral constituents like mica and albite (Fig. 3a) and sometimes arranged in “star shape” (Fig. 3d-e). Backscattered electron images reveal complex chemical zoning including simple progressive zoning, oscillatory zoning and patchy zoning. Bizonal crystals with broad bands showing sharp chemical contrasts between a dark Nb-rich core and a bright thin Ta-rich rim are common (Fig. 3d-e). Other Nb-Ta-minerals associated with CGM include wodginite, cassiterite and microlite. Zircon and CGM may be found intimately associated, either as intergrowth (Fig. 3b) or as inclusions (Fig. 3c). Zircon is mostly found as fine-grained (<1 mm) euhedral crystals disseminated in major silicate minerals. It can be slightly zoned with concentric bands (Fig. 3h), but radiation damage generally masks this zoning (Fig. 3i). Metamict and inclusion-rich crystals such as the ones shown in Fig. 3i-j were discarded for age dating. Zircons from the type I pegmatite L7 are coarser-grained and present complex oscillatory zoning with zones of porous inclusion-rich zircon (Fig. 3f-g); these two types of zircon zones are later distinguished as primary versus secondary based on their geochemistry. Xenotime occurs in direct contact with pegmatite L7 zircons. It shows resorbed textures and systematically occurs near zircon cores, which evokes exsolution during dissolution-reprecipitation of primary zircon.

Fig. 3
figure 3

BSE pictures showing CGM inclusions within mica in pegmatite L4 (a), the intimate and cogenetic association between CGM and zircon in pegmatite L2 (b) and L3 (c), the star-shape habit of CGM at different scales, and its bizonal chemical zoning in pegmatite L2 (d) and L3 (e), complex zoning in type-I zircon and its association with xenotime in pegmatite L7 (f-g), various zircon habits, from slightly zoned (h, pegmatite L3) to highly metamict (i, pegmatite L4) and inclusion-rich (j, pegmatite L5). In Fig. 3e, laser-spot locations for U-Pb dating are marked with circles. ms: muscovite, zrn: zircon, xtm: xenotime

Full size image

CGM chemistry includes minor concentrations of TiO2 (<1.7 wt%), WO3 (<1.4 wt%), SnO2 (<1 wt%), ZrO2 (<0.5 wt%), UO2 (<0.4 wt%) and Sc2O3 (<0.2 wt%). EPMA analyses show a large range of compositions (Table 4) that plot in the ferrocolumbite to ferrotantalite parts of the CGM quadrilateral (Fig. 4). Core to rim variations illustrate the common Ta over Nb enrichment that is generally observed during CGM fractionation. Nb-Ta fractionation is also visible from type II to type IV CGM, and can be illustrated in a Rayleigh-type Nb/Ta vs. Ta2O5 fractionation diagram (Fig. 5). Fe-Mn fractionation leads to a general Fe enrichment over Mn.

Table 4 Chemical compositions of columbite-group minerals as determined by EPMA and structural formulae calculated for 6 oxygens
Full size table
Fig. 4
figure 4

Chemical variations of CGM in the columbite quadrilateral. Arrows indicate core to rim variations in single samples

Full size image
Fig. 5
figure 5

Rayleigh fractionation trends for zircon and CGM of the different pegmatite types

Full size image

Zircon chemistry reveals high concentrations of UO2 (up to 1.6 wt%), and HfO2 concentrations ranging from 2.1 to 6.1 wt%, which slightly increase from type I to type III and IV pegmatites (Table 5). Figure 5 illustrates this Zr/Hf fractionation trend. In type I zircon from the L7 pegmatite (Fig. 3f-g), three types of zircon zones were distinguished based on backscattered images and show distinct chemistry (Table 5; see Fig. 6 for outline of zircon zones). The highly porous and inclusion-rich cores have negligible UO2 and Y2O3 concentrations, whereas the oscillatory zones (zr1) have low UO2 and Y2O3 concentrations (0.6 and 0.1 wt% in average). On the Zr/Hf fractionation trend (Fig. 5), these two zircon zones plot on a continuous trend which can be interpreted as magmatic fractionation. These zircon zones are therefore interpreted as primary. Alternatively, the patchy zones (zr2), which crosscut the oscillatory zones, have high UO2 and Y2O3 concentrations (up to 3 and 1.9 wt% respectively) and low totals due to metamictization. They also display high levels of impurities (P, Ca and Fe). The P + Y component is negatively correlated with Zr + Hf + Si (apfu); its incorporation is explained by the vector P5+ + Y3+ = Si4+ + Zr4+, which reflects the solid solution between zircon and xenotime (Fig. 6). This third zircon type plots outside the Zr/Hf fractionation trend and is therefore interpreted as secondary. Few EPMA analyses of xenotime revealed UO2 and ThO2 concentrations of 4 and 0.1 wt% in average, respectively.

Table 5 Chemical compositions of zircon as determined by EPMA
Full size table
Fig. 6
figure 6

BSE picture of a zircon crystal (pegmatite L7) showing an inclusion-rich core, oscillatory-zoned primary zircon (zr1) and patchy-zoned secondary zircon (zr2), as well as xenotime (xtm). The graph shows Si + Zr + Hf vs. P + Y apfu contents for the different zircon zones

Full size image

U-Pb dating

Full age data is available in Table 1 (pegmatite L3) and Table 2 (pegmatite L7). Table 6 summarizes the number of grains that were analyzed, the total number of analyses and the number of analyses that were used to calculate Concordia ages. Concordia ages are given except when they are too few, in this case intercept ages are given.

Table 6 Synthesized age data
Full size table

For pegmatite L3 (type III), U-Pb analyses of zircon and CGM reported in Concordia diagrams (Fig. 7) spread over a large range of isotopic ratios along discordant lines, indicating extensive lead loss. For CGM, 6 points plot on the Concordia line and give an age of 301.9 ± 3.8 Ma (MSWDC+E = 1.3 with C + E = concordance + equivalence). No distinction can be made between the two main BSE zones (see laser spot locations in Fig. 3e). For zircon, only one age plots on the Concordia line but the 26 discordant ages define an upper intercept at 298.7 ± 5.7 Ma (MSWD =1.5). The BSE images of the analyzed zircon grains (Figs. 3j and 8) reveal highly porous and inclusion-rich crystals, supporting lead loss as the cause of the dispersion on the Discordia line. The duplicate analyses performed in Clermont-Ferrand (inset of Fig. 7b), indicate a combination of discordance and common Pb contribution. The eight remaining concordant analyses display a Concordia Age of 297.3 ± 2.1 Ma (MSWDC+E = 1.6, n = 8). A second, smaller group of concordant analyses displays a younger age (ca. 275 Ma).

Fig. 7
figure 7

U-Pb data in Concordia diagrams for L3 CGM and zircon. The inset of b) shows duplicate ages from LMV Clermont-Ferrand. Data-point error ellipses are 2σ

Full size image
Fig. 8
figure 8

Laser spot locations for U-Pb dating in pegmatite L3 zircon. Backscattered electron images. The 100 μm scale bar is valid for all pictures

Full size image

In pegmatite L7 (type I), oscillatory-zoned (primary, zr1 in Fig. 6) and porous (secondary, zr2 in Fig. 6) zones of zircon were thoroughly distinguished during laser spot location (Fig. 9), and the age results display two age groups (Fig. 10). In primary zircon (zr1), 19 of the 28 U-Pb ages plot on the Concordia and display an age of 296.2 ± 2.5 Ma (MSWDC+E = 1.7, n = 19). For secondary zircon (zr2), the 36 analyses are spread over a large range of isotopic ratios and 12 of them plot on the Concordia line and give an age of 290.5 ± 2.5 Ma (MSWDC+E = 0.7). Eight of the 12 xenotime U-Pb analyses plot on the Concordia and display an age of 292.9 ± 2.9 Ma (MSWDC+E = 0.76).

Fig. 9
figure 9

Laser spot locations for U-Pb dating of pegmatite L7 zircon. Backscattered electron images. The 100 μm scale bar is valid for all pictures

Full size image

Discussion

Columbite versus zircon dating

In-situ U-Pb geochronology by LA-ICPMS on columbite-group minerals has been developed in the 2000s (Smith et al. 2004) and is now widely applied for pegmatite age determination (Melcher et al. 2008, 2015; Dewaele et al. 2011; Melleton et al. 2012; Deng et al. 2013). However, most geochronological studies have used zircon standardization, arguing that matrix-dependent effects are low (Melcher et al. 2008). Some of these geochronological results are Precambrian in age, and are therefore less sensitive to U/Pb fractionation. Che et al. (2015a, b) recently evaluated the effect of matrix-dependent fractionation by comparing U-Pb ages obtained on CGM using two different references (Zircon 91,500 and Coltan 139), and noticed a significant matrix effect leading to approximately 7–15 % younger ages where zircon references were used compared to the CGM reference. In our study, CGM ages were obtained using zircon GJ-1 primary reference and Coltan 139 was only used as a secondary control reference, therefore the Concordia age of 301.9 ± 3.8 Ma may be slightly shifted on the Concordia, which may explain the slight difference between that age and the zircon Concordia age of 297.3 ± 2.1 Ma obtained in the same sample. However, the two ages are coeval within error, and the reliability of the CGM age is evidenced by the fact that Coltan 139 yields a correct age when normalized to GJ-1 zircon in the same analytical series (Table 1).

Dating pegmatite emplacement

The primary magmatic origin of CGM and zircon dated at c. 299 Ma is evidenced by several indicators. Textural relationships between zircon, CGM and major silicate minerals (feldspars, muscovite or garnet) indicate that zircon and CGM are cogenetic and crystallized at the pegmatitic stage. The patchy zones in L7 zircon (Figs. 3f-g and 6) are an exception; they are interpreted as secondary post-solidus phases. In CGM, the sharp chemical contrast between Nb-rich cores and Ta-rich rims is not reflected by a detectable change in their ages. In pegmatite L7 (type I) zircon, the porous and inclusion-rich cores have similar chemistry compared to primary magmatic oscillatory-zoned zircon (Fig. 6), which reflects their pegmatitic origin instead of being inherited cores. Generally, the high Zr (and HFSE in general) solubilities in F-Li-P-rich pegmatitic melts (Linnen 1998) lower the chances to find inherited zircon cores in pegmatites. Consequently, it can be stated with high confidence that the obtained zircon and CGM ages dated at c. 299 Ma are representative of the pegmatite emplacement. The parallel trends followed by the Zr/Hf and Nb/Ta Rayleigh fractionation curves (Fig. 5) are also evidences that CGM and zircon both followed magmatic fractionation trends and therefore crystallized together at the magmatic stage (Hulsbosch et al. 2014).

The different zones of pegmatite L7 (type I) zircon reflect several crystallization stages. Oscillatory-zoned zircon may have crystallized at magmatic stages, whereas the patchy zones, which crosscut the oscillatory zones (see Fig. 3f), could represent a secondary either magmatic or hydrothermal stage of crystallization. The highly porous nature of patchy zircon zones suggests a replacement texture, whereas its elevated U content indicates that it is strongly metamict. Such patchy textures could also be the result of the metamictization and alteration of the most U-rich zircon bands in the oscillatory zoning, as previously shown by Paquette et al. (2003), which would explain that the patchy zones follow the growth banding of magmatic zircon. In the Zr/Hf fractionation trend (Fig. 5), patchy zircon plots outside the Rayleigh trend at high Zr/Hf ratios, suggesting a non-magmatic origin, whereas all other zircon zones plot on a continuous Rayleigh curve typical of magmatic fractionation. The cores are also highly porous and inclusion-rich, and may represent primary zircon which has undergone dissolution-reprecipitation processes with exsolution of its U and Y contents during the secondary event. The high Y + P concentrations of patchy zircon, negatively correlated with Zr + Hf + Si, suggest an important xenotime component, therefore implying that patchy zircon and xenotime are co-genetic. Whereas limited amounts of Y were integrated in primary magmatic zircon, Y was probably added by fluids during post-solidus alteration, and precipitated as Y-rich secondary zircon and xenotime replacing and overgrowing primary zircon. An alternative explanation is that xenotime was a primary magmatic phase like oscillatory-zoned zircon and it was dissolved and recrystallized during the hydrothermal event, with some Y being remobilized and integrated into secondary zircon. The U/Pb ratio of xenotime therefore dates the hydrothermal event. The slight age difference between primary (296.2 ± 2.5 Ma) and secondary (290.5 ± 2.5 Ma) zircon and its associated xenotime (292.9 ± 2.9 Ma) suggests that the secondary (hydrothermal?) event took place after pegmatite emplacement.

Fig. 10
figure 10

U-Pb concordia diagrams for pegmatite L7 primary zircon, secondary zircon and xenotime. Data-point error ellipses are 2σ

Full size image

Implications for the geochronology of late Variscan tectonics

The five obtained ages define two groups of statistically distinct ages that lie between 296 and 302 Ma for the first, and 290 and 293 Ma for the second (Fig. 11). Despite their small overlap when 2-sigma error bars are considered, the two age groups remain distinct. These results have important implications for the geochronology of late Variscan tectonic events in the Cap de Creus. In a recent paper, Druguet et al. (2014) obtained similar results on syntectonic quartz diorite from the Tudela migmatitic complex, dated at 298.8 ± 3.8 Ma, and granodiorite from the Roses pluton, dated at 290.8 ± 2.9 Ma (U-Pb zircon geochronology using SHRIMP). They concluded that the D3 ductile deformation extended into the Lower Permian as a transitional stage between the Variscan and Cimmerian cycles. Taking their age results and error bars into account, mean values of 298.9 ± 6 Ma and 292.4 ± 4 Ma are calculated for the two age groups (Fig. 11). The 298.9 Ma age corresponds to the emplacement and primary crystallization of the pegmatitic melts, regardless of type I or III, and is coeval with migmatization. The 292.4 Ma age correlates with zircon replacement and xenotime crystallization as a consequence of late hydrothermal reactions that affected the pegmatites after their crystallization. However, no other evidence of this late hydrothermal event was observed in our study, and its correlation with the granodiorite emplacement remains very hypothetic.

Fig. 11
figure 11

Statistical distribution of the 5 ages obtained in this study compared with the two ages obtained by Druguet et al. (2014)

Full size image

The pegmatites were emplaced near the peak of metamorphism, therefore the pegmatite age is contemporaneous or slightly younger than this metamorphic peak. Type I to type III pegmatites, were emplaced in the same time span, around 299 Ma. The high standard deviations on in-situ U-Pb geochronological methods do not permit the different pegmatitic pulses to be distinguished, although field evidences (early, syn and late D2 emplacement) support a multiple-emplacement history for the pegmatites. Alfonso et al. (1995) report columbite-tantalite crystals showing primary, pre-deformation oscillatory zoning broken during the deformation and subsequently overgrown by a “post-tectonic” Ta-rich rim. Field structural data indicate that the pegmatites are syn-tectonic with D2 and are affected by (and therefore predate) D3. The idea that this late deformation event could have occurred after the Carboniferous-Permian limit, concomitant with the hydrothermal event at the origin of zircon and xenotime recrystallization dated at c. 292 Ma, has to be further investigated.

Although the relationship between D2 and D3 deformations has been extensively investigated (Druguet 2001; Carreras et al. 2004), the lower geochronological limit for the Variscan tectonic event remains unclear. After Laumonier et al. (2015), this orogeny extended from Namurian to Stephanian times (c. 325–300 Ma) in the Pyrenees. However, the latest published data (Druguet et al. 2014) and our geochronological results indicate that the upper limit of the Variscan orogeny in Cap de Creus extended into the Early Permian. In this area, the granitoids and magmatic rocks are clearly syntectonic (syn- to late-D2 and pre-D3), and dated at c. 299 and 291 Ma (Late Carboniferous and Early Permian) by Druguet et al. (2014). Druguet et al. (2014) dated the migmatization event at c. 299 Ma based on field structural relationships that indicate that the dated quartz dioritic magmas are synchronous with migmatites, in agreement with field relationships. Although the migmatites themselves only present inherited zircons with Precambrian ages (542 Ma at the earliest), strong field evidences in Cap de Creus and elsewhere in the Pyrenees indicate that they are Variscan. In the Pyrenees, a few granites were dated between 292 and 300 Ma (e.g., 298.5 ± 1.8 Ma for a syn-D2 leucogranite from the Albera massif, using Th-U-Pb geochronology on monazite by electron probe microanalysis, Laumonier et al. 2015). Our geochronological results, yielding two distinct age groups at c. 299 and 292 Ma, suggest that the last stages of the magmatic events in the Variscan Pyrenees could have extended into the Early Permian.

The anatectic vs. granitic origin of pegmatites

The anatectic (melting of country rock) versus granitic (extreme fractionation of granitic melt) origin of pegmatites is still strongly debated (see London 2008 for a synthetic view of this topic), especially in cases where no potential parental granite is observed and the pegmatites are associated with migmatites, which is the case in Cap de Creus. The distribution of the pegmatites and their fractionation trends indicate an origin by differentiation of a granitic melt originating from the north of the peninsula, whereas their spatial association with migmatites has been used to argue for an anatectic origin. The common absence of visible granite in the vicinity of granitic pegmatites is generally explained by the extreme mobility and the low solidus temperatures of the highly-fluxed melts that can travel through considerable distances before the onset of dyke crystallization (Baker 1998). In general, pegmatites showing a zoned distribution with increasing fractionation degrees and mineralogical complexity are classically interpreted as being genetically related to a single melt source which evolved with fractional crystallization (London 2008). Arguments for a granitic origin of the Cap de Creus pegmatites are 1) their zonal distribution with sterile bodies located near the zone of maximum deformation in the high metamorphic zones, and fractionation degree increasing when moving toward the south down the metamorphic gradients, and 2) progressive geochemical trends in feldspar, micas (Alfonso et al. 2003), and in Nb-Ta oxides (Alfonso et al. 1995) from type I to type IV pegmatites. The source granite could have been emplaced during the main deformation event in the migmatized area, and would now be hidden further to the north of the peninsula, or translated to the southeast by late dextral shear zones.

Arguments for the anatectic origin are provided by stable isotope constraints (Damm et al. 1992). The authors conclude that the pegmatites are derived from anatexis of the metapelitic rocks at shallow crustal levels, but their study only takes into consideration the type I pegmatites near the lighthouse of Cap de Creus. In the Albera massif about 50 km northwest of the Cap de Creus peninsula, the peraluminous granites were interpreted as anatectic by Autran et al. (1970). Several hundred pegmatite dykes occur concentrically and zonally around muscovite-biotite granite stocks, close to their boundaries with the Precambrian orthogneisses and the Paleozoic series, therefore Autran et al. (1970) suggested an anatectic origin for the pegmatites as well. However, Malló et al. (1995) argue for an origin by magmatic fractionation of the Albera pegmatites based on the geochemical trends of the accessory minerals (phosphates and Nb-Ta-oxides). Malló et al. (1995) specify that the pegmatite source would be the anatectic muscovite-biotite leucogranites. The pegmatites in the Albera massif and in the Cap de Creus peninsula are comparable in their structure, mineralogy, geochemistry and regional distribution, which suggests a common origin.

The high fractionation degrees and highly mineralized nature of type III and IV pegmatites suggest an origin by extreme magmatic fractionation rather than in situ partial melting. Arguments are provided by the experimental work of London and Evensen (2002) and Evensen and London (2002) that shows taking the example of Be, that beryl saturation in pegmatites only occurs after extended crystal fractionation of large magma batches (>95 % crystallization), themselves originating from low partial melting of a fertile sedimentary source. Moreover, the continuous fractionation trends displayed by zircon and CGM indicate a genetic affiliation between all Cap de Creus pegmatites. However, this does not preclude that type I pegmatites, which are unmineralized and would be better named as pegmatitic granite, could be anatectic in origin. The large pockets of that pegmatitic granite observed in the north of the peninsula could have resulted from the partial melting of high-grade schists concomitant to the development of the migmatitic complexes. Their differentiation could have produced evolved pegmatitic melts that migrated down the metamorphic gradient and crystallized up to 3 km away from their source (Fig. 1). However, the presence of peraluminous granites associated with pegmatites in the Albera massif, also dated around 299 Ma (298.5 ± 1.8 Ma for a leucogranite, Laumonier et al. 2015), suggests that a peraluminous granite could also be the source of the Cap de Creus pegmatites. Therefore, to confirm one or the other model, geochemical and isotopic signatures of the migmatites, unmelted sedimentary units, granitic intrusives and pegmatites are necessary. 3D modeling of the pegmatite batch distribution in relation to the structural context of emplacement may also help quantifying the magma volumes implied in anatexy versus granitic fractionation (Demartis et al. 2011; Deveaud et al. 2013).

Conclusion

The U-Pb dating of magmatic zircon and columbite-group minerals in the Cap de Creus reveals that at least pegmatite types I and III were emplaced at c. 299 Ma. Although field relations clearly show that the various pegmatite types are not all simultaneously emplaced, our data demonstrate that they were formed and emplaced during the latest stages of the Variscan orogeny, more or less synchronously with the D2 deformation event and the associated thermal peak. Secondary zircon and xenotime that probably formed during a late post-solidus hydrothermal event, were dated at c. 292 Ma. This age correlates with the intrusion of late post-D2 calc-alkaline granites. This late hydrothermal event could be related to the D3 localized deformation event that is clearly post-magmatic, which would imply that the Variscan deformation was still active during the Early Permian. However, the age of the D3 event is to date unconstrained, and correlating the hydrothermal event with the D3 deformation event on one side, and the granodiorite emplacement on the other side, remains very hypothetic because of the important overlap (considering uncertainties) between the different ages.

Our results are in agreement with recent geochronological results from granitic rocks in the Variscan Eastern Pyrenees, which show that the peak of magmatic activity is well dated around 306 Ma. The pegmatitic melts were also generated simultaneously with the partial melting of the metasediments in high-grade metamorphic zones, but the extremely fractionated character of the most evolved pegmatite types III and IV suggests an origin by extreme magmatic fractionation rather than in situ partial melting. An alternative model would be that the most evolved pegmatitic melts could have originated from the extreme fractionation of low volumes of anatectic melts, but the presence of peraluminous granites associated with pegmatites in the Albera region, also dated at 298.5 Ma, is an evidence that peraluminous magmatism was active at that time in the area, and therefore suggests that a peraluminous granite could be the source of the most evolved Cap de Creus pegmatites.