The Fregeneda-Almendra pegmatite field of the Iberian Massif represents a typical expression of peraluminous rare-metal magmatism that occurred over western Europe at the end of the Variscan orogeny. It is the host for two main types of Li-mineralized intrusions, identified at the scale of the Variscan belt, including petalite- or spodumene-rich pegmatites, as well as Li-mica-rich pegmatites, for which the origin of mineralogical-chemical differences is not yet understood. Here, we provide cassiterite and columbite-group mineral (CGM) U-Pb ages along with oxide, mica and phosphate mineral compositions for Li-pegmatites from the Fregeneda-Almendra field in order to assess their petrogenesis and tectonic-magmatic context of emplacement. U-Pb geochronology indicates that petalite-rich and Li-mica-rich pegmatites were mostly emplaced sub-synchronously from 315 ± 6 to 308 ± 6 Ma, during strike-slip deformation and granitic magmatism within an anatectic dome bounding the pegmatite field. U-Pb data and pegmatite geographic zonation suggest that Li-pegmatites were sourced from buried equivalents of leucogranites and migmatites from the dome. Li-pegmatites experienced a complex crystallization including K-feldspar, petalite, topaz, Nb-Ta-Fe-Mn-rich cassiterite, amblygonite-group minerals (AGM) and CGM as early magmatic phases, followed by lepidolite for Li-mica-rich pegmatites. At the magmatic-hydrothermal transition, notably leading to the formation of Nb-Ta-Mn-Fe-poor cassiterite hosting CGM inclusions, earlier minerals were resorbed by muscovite and albite. A later F-rich hydrothermalism is locally reflected by zinnwaldite overgrowths on muscovite. Cassiterite, CGM and micas from petalite-rich pegmatites show lower Mn/Fe ratios and higher Ti contents, along with lower Zr-Ga contents for cassiterite, than that from Li-mica-rich pegmatites. Such behavior is consistent with a magmatic differentiation process whereby Ti content decreased and the degree of Mn-Fe geochemical fractionation and solubilities of Ga and Zr increased in the melts, possibly in relation with high fluorine activity. In Li-mica-rich pegmatites, AGM equilibrated with a melt with up to 2 wt% F, similar to that in equilibrium with lepidolite (1–3 wt%). In petalite-rich pegmatites, the relatively high F concentration of the melts equilibrated with AGM (≤ 1.5 wt% F) contrasts with the liquid equilibrated with muscovite (< 0.5 wt% F). This can be accounted for by muscovite crystallization after the exsolution of a F-rich aqueous phase at the magmatic-hydrothermal transition. Relatively similar F contents in the initial melts of petalite- and Li-mica-rich pegmatites support the hypothesis that the stability of lepidolite does not only involve high F but also a low H2O/F activity ratio. For the Fregeneda-Almendra Li-mica-rich pegmatites, this could be explained by a decrease of melt H2O solubility due to a relatively low pressure of emplacement.

Le champ de pegmatites de la Fregeneda-Almendra représente une expression typique du magmatisme peralumineux à métaux rares de la fin de l’orogenèse varisque. Il est l’hôte de deux principaux types d’intrusions minéralisées en Li incluant des pegmatites à pétalite ou spodumène, et à micas lithinifères (mica-Li), dont l’origine des différences cristallochimiques est mal comprise. Nous fournissons ici des âges U-Pb sur cassitérite et minéraux du groupe de la colombite (coltan) ainsi que la composition des oxydes, micas et phosphates de ces pegmatites afin d’évaluer leur pétrogenèse et leur contexte de mise en place. La géochronologie U-Pb indique que les pegmatites à pétalite et à mica-Li se sont mises en place de manière sub-synchrone de 315 ± 6 à 308 ± 6 Ma, lors d’une déformation en décrochement et du magmatisme granitique dans un dôme anatectique délimitant le champ de pegmatites. Les âges et la zonation géographique des pegmatites suggèrent que les leucogranites et migmatites du dôme représentent des équivalents de leur source. Les pegmatites à Li enregistrent une cristallisation complexe comprenant : feldspath-K, pétalite, topaze, cassitérite riche en Nb-Ta-Fe-Mn, minéraux du groupe de l’amblygonite (MGA) et coltans comme phases magmatiques précoces, suivies du lépidolite pour les pegmatites à mica-Li. À la transition magmatique-hydrothermale, conduisant notamment à la formation de cassitérite pauvre en Nb-Ta-Mn-Fe, les minéraux précoces ont été résorbés par la muscovite et l’albite. Un hydrothermalisme tardif riche en F est reflété par des surcroissances de zinnwaldite sur la muscovite. Les oxydes et les micas des pegmatites à pétalite présentent des rapports Mn/Fe et des teneurs en Zr et Ga plus faibles ainsi que des teneurs en Ti plus élevées que ceux des pegmatites à mica-Li. Ceci est cohérent avec un processus de différenciation magmatique durant lequel la teneur en Ti a diminué, et le fractionnement de Mn-Fe et les solubilités de Zr et Ga ont augmenté dans le liquide, possiblement en relation avec une augmentation de l’activité en F. Dans les pegmatites à mica-Li, les MGA étaient équilibrés avec un liquide silicaté contenant jusqu’à 2%pds de F, similaire à celui en équilibre avec le lépidolite (1–3%pds). Dans celles à pétalite, la concentration en F relativement élevée du liquide équilibré avec l’AGM (≤ 1,5%pds F) contraste avec celui équilibré avec la muscovite (< 0,5%pds F). Ceci peut s’expliquer par la cristallisation de la muscovite après l’exsolution d’une phase aqueuse riche en F. Des teneurs en F relativement similaires dans les liquides initiaux des pegmatites à pétalite et mica-Li suggèrent que la stabilité du lépidolite n’implique pas seulement une activité élevée en F mais aussi une faible activité en H2O relative au F. Pour les pegmatites à mica-Li de la Fregeneda-Almendra, ceci pourrait s’expliquer par une diminution de la solubilité de l’H2O dans le liquide parent due à une pression de mise en place relativement faible.

Lithium has become a strategic commodity as it represents a key component for green energy-storage devices, such as Li-ion batteries. In Europe, significant “hard-rock” Li resources are associated with peraluminous high phosphorous rare-metal granites (Linnen and Cuney, 2005) and LCT-type pegmatites (enriched in Li-Cs-Ta, Černý and Ercit, 2005) from the Variscan belt (Gourcerol et al., 2019) (Fig. 1). Mineralization of Li-bearing aluminosilicate phases such as spodumene, petalite or Li-micas (e.g., micas from the lepidolite series) along with Li-Al-phosphates (amblygonite-group minerals) typically occur disseminated within those intrusions. Moreover, rare-metal granites and pegmatites are commonly enriched in other strategic metals such as Sn, Ta and Nb generally hosted in oxide minerals including cassiterite, columbite-group and pyrochlore-group minerals (Linnen et al., 2012).

Peraluminous rare-metal pegmatites and granites (RMPGs) are related to crustal melting in late-orogenic setting and can reach metal (e.g., Li, Be, Ta, Nb, Sn, Cs) and halogen (F) concentrations that are up to three orders of magnitude higher than bulk upper crust and common peraluminous granites (Černý et al., 1985; Linnen and Cuney, 2005; Romer and Pichavant, 2021). Magmatic-hydrothermal processes including silicate melt-hydrosaline melt or melt-fluid immiscibility and alkali metasomatism can lead to significant element mobility and re-concentration at the end of magma crystallization (Kaeter et al., 2018; Müller et al., 2018; Ballouard et al., 2020a, 2020b). However, the occurrences of volcanic to sub-volcanic equivalents of RMPGs (Pichavant et al., 1987; Raimbault and Burnol, 1998) as well as experiments (Pichavant, 2022) suggest that their metal enrichment occurred dominantly at the magmatic stage. Rare-metal-enriched dykes or stocks, especially LCT pegmatites, commonly cluster in fields or belts with a lateral extent of ten’s kilometers or more and commonly define a geographic and geochemical zonation around peraluminous plutons (Černý, 1991). As such, pegmatites from the same field are commonly interpreted as coeval, cogenetic and issued from variable degrees of fractional crystallization of the same parent pluton (Černý and Ercit, 2005; Hulsbosch et al., 2014; London, 2018). However, spatially associated granites and pegmatites can feature distinct ages (Melleton et al., 2012, 2022; Müller et al., 2017) or geochemical and isotopic signatures (Antunes et al., 2013; Deveaud et al., 2015; Villaros and Pichavant, 2019) incompatible with parent-daughter relationships and suggesting variable partial melting conditions or source heterogeneities. On this basis, some studies have suggested that some LCT pegmatites are directly produced by small degree of partial melting of metasedimentary protoliths variably enriched in rare-metal and fluxing elements such as Li and F (Stewart, 1978; Shaw et al., 2016; Konzett et al., 2018b, 2018a; Gourcerol et al., 2019; Bonzi et al., 2021). Alternative models combining favorable partial melting conditions of rare-metal fertile protoliths followed by magmatic differentiation have also been proposed for the formation of LCT pegmatite deposits (Knoll et al., 2018, 2023; Ballouard et al., 2020a, Ballouard et al., 2023). Overall, no general consensus exists regarding the key processes controlling the genesis of RMPGs.

Moreover, the extraction and transport of rare-metal-enriched silicic magmas from their migmatitic or granitic sources to upper crustal levels are commonly seen as being the results of favorable tectonic conditions with their emplacement intimately controlled by deformation structures (Araújo et al., 2001; Demartis et al., 2011; Deveaud et al., 2013; Gourcerol et al., 2019; Ballouard et al., 2020a; Michaud et al., 2020). However, the few robust geochronological data available for RMPGs is a strong limitation for our understanding of their tectonic-magmatic contexts of formation. Traditional refractory phases with high-closure temperature U-Pb systems such as zircon are commonly rare, or when present, highly metamict in these geological environments.

The Fregeneda-Almendra (FA) pegmatite field, located in the Central Iberian Zone (CIZ) of the Variscan Iberian Massif at the boundary between Spain and Portugal, is a typical expression of peraluminous rare-metal magmatism that occurred at the end of the Variscan orogeny in western Europe (Roda-Robles et al., 1999, 2023; Vieira et al., 2011; Errandonea-Martin et al., 2022) (Fig. 1). This field hosts the two main types of LCT pegmatites identified in the Iberian Massif, including (i) spodumene- or petalite-rich, as well as (ii) Li-mica-rich pegmatites (Roda-Robles et al., 2016, 2018, 2023). If the stability of spodumene over petalite is known to be mostly pressure- and temperature-dependent (London 1984, 2018), few constraints exist on the conditions of crystallization of Li-micas such as lepidolite over other Li-aluminosilicates, except for the requirement of a Li-F-rich peraluminous melt (London and Burt, 1982) with possibly high alkali-fluorine salinity (Wise et al., 2012). Previous studies on Iberian pegmatites including the FA pegmatite field have suggested based on whole-rock (Roda-Robles et al., 2018, 2023), mica (Vieira et al., 2011) and phosphate mineral (Roda-Robles et al., 2010) compositions that Li-mica-rich pegmatites represent the most evolved residual melts of a fractional crystallization series including as more primitive members the petalite- and spodumene-rich pegmatites. However, the cogenetic character of such Li-pegmatites was questioned in a similar pegmatite field from Iberia (i.e., Barosso-Alvao) based on columbite-group mineral (CGM) geochemistry (Martins et al., 2011).

The pegmatites of the FA field were emplaced within a belt of metasedimentary rocks in contact to the south with a major anatectic dome and bounded by sinistral strike-slip shear zones; the dome mostly consists of syntectonic two-mica-bearing leucogranites and migmatites, referred to as the Figueira de Castelo Rodriguo-Lumbrales (FCL) complex (Díez Fernández and Pereira, 2016; Ferreira et al., 2019, 2020, 2022) (Fig. 2a). The pegmatites tend to define a zonal distribution around the FCL complex with a northward increase of Li-fertility, roughly perpendicular to metamorphic isograds, from barren pegmatites close to or within the anatectic complex to petalite-rich pegmatites, spodumene-rich and Li-mica-rich pegmatites in lower grade metasedimentary rocks (Roda-Robles et al., 1999; Vieira et al., 2011) (Fig. 2b). However, available muscovite and Li-mica 40Ar/39Ar geochronology of pegmatites from the FA field yielded ages from ca. 305 to 295 Ma (Roda-Robles et al., 2009, 2023). These Ar-Ar ages, as well as one U-Pb age of 302 ± 3 Ma obtained on a CGM crystal from a Li-mica-rich pegmatite (Roda-Robles et al., 2023), are significantly younger than the crystallization age of the main leucogranite intrusions from the FCL complex (ca. 319-311 Ma, zircon U-Pb; Ferreira et al., 2019), as well as U-Pb zircon ages of associated migmatites (ca. 344–315 Ma; Ferreira et al., 2020, 2022). Roda-Robles et al. (2009, 2023), thus, concluded for an absence of genetic connection between leucogranites from the FCL complex and Li-mineralized pegmatites from the FA field, and proposed a genetic link with non-outcropping late- to post-orogenic granites. However, such disconnection is strange considering the apparent spatial relationship between the FCL dome and the pegmatites (Fig. 2).

In this study, we use a new approach based on cassiterite U-Pb geochronology, coupled with CGM U-Pb dating along with oxide mineral, mica and phosphate geochemical compositions, on Li-mica- and petalite-rich pegmatites from the FA field. Our goal is to precisely determine the timing of emplacement of Li-pegmatites within the regional tectonic-magmatic framework, and to better decipher the petrogenetic processes involved during the crystallization of rare-metal-bearing mineral phases including Li-aluminosilicates.

The Central Iberian Zone

The Iberian Massif represents the south-western part of the European Variscan belt, a Paleozoic orogen resulting from the collision of supercontinents Laurussia and Gondwana, and involving several micro-continental blocks such as Iberia and Armorica (Matte, 1991; Kroner and Romer, 2013; Ballèvre et al., 2014; Martínez Catalán et al., 2021) (Fig. 1). The CIZ of the Iberian Massif largely consists of Precambrian to Cambrian metapelites and metagraywackes (the Schist-Graywacke Complex), overlain by Ordovician quartzites and intruded by voluminous Carboniferous to Permian granitoids, as well as Cambro-Ordovician igneous rocks (Dallmeyer and Garcia, 2012; García-Arias et al., 2018). Three main deformation events were recognized in the Iberian Zone including (i) collision-related thrusting and folding during Barrovian metamorphism between ca. 360 and 335 Ma (D1), (ii) crustal extension accommodated by low-dipping extensional shear zones associated with low-pressure high-temperature metamorphism from ca. 335 to 320 Ma (D2), and (iii) strike-slip tectonics along vertical shear zones associated with upright folding between ca. 320 and 295 Ma (D3) before and until the final closure of the Ibero-Armorican arc (Dallmeyer et al., 1997; Gutiérrez-Alonso et al., 2011; Martínez Catalán et al., 2014, 2021; Díez Fernández and Pereira, 2016; Dias da Silva et al., 2017, Pereira et al., 2017). Within the D3 event, Díez Fernández and Pereira (2016, 2017) suggested the existence of three distinct episodes of strike-slip deformations in Iberia occurring at ca. 317-311 Ma, 311-304 Ma and 304-295 Ma.

Voluminous granitoid magmatism in NW-Iberia occurred from ca. 345 to 285 Ma (Ábalos et al., 2002; Gutiérrez-Alonso et al., 2011, 2018; Martínez Catalán et al., 2014, 2021). Granitoids can be classified as three main groups, including (i) syn-D3 highly peraluminous two-mica- or muscovite-bearing granites and leucogranites, (ii) late-D3 peraluminous biotite-bearing granites locally hosting cordierite, as well as (iii) pre-D3 and late-D3 generations of metaluminous to slightly peraluminous biotite ± amphibole-bearing granodiorites and granites (Ribeiro et al., 2019). Those three granitoid types globally correspond to the muscovite-bearing peraluminous granites (MPG), cordierite-bearing peraluminous granites (CPG) and K-rich K-feldspar porphyritic calc-alkaline granitoids (KCG) of the general classification of Barbarin (1999) used in the map of Figure 1.

Despite their low volume compared to common granitoids, peraluminous Li-aplites or Li-pegmatites, hosting spodumene, petalite, Li-micas and/or Li-phosphates along with Sn-Nb-Ta oxide minerals, are widespread both in the Galicia-Trás-os-Montes and Central Iberian zones (Neiva and Ramos, 2010; Roda-Robles et al., 2016, 2018) (Fig. 1). They were mostly emplaced during D3-related tectonics from ca. 315 to 300 Ma (U-Pb on CGM; Melleton et al., 2022). Roda-Robles et al. (2018) proposed, on the basis of geochemical similarities, a genetic relationship between Li-pegmatites and highly peraluminous P-rich granite plutons. They distinguished two types of fertile granites including a S1-type series corresponding to two-mica-bearing leucogranites or MPG and a S2-series representing the most peraluminous terms of biotite ± cordierite-bearing granites or CPG.

The Figueira de Castelo Rodriguo-Lumbrales anatectic complex and associated Fregeneda-Almendra pegmatite field

The Figueira de Castelo Rodriguo-Lumbrales (FCL) anatectic complex is a more than 50 km long E-W to NE-SW oriented dome-like structure separated from lower grade metasedimentary rocks by two D3 sinistral shear zones including the Huebra shear zone to the north and Juzbado–Penalva do Castelo shear zone to the south (Fig. 2a). The two-mica-bearing peraluminous leucogranites are locally deformed and bear a roughly E-W oriented S3 sub-vertical foliation along the northern and southern edges of the FCL complex (Díez Fernández and Pereira, 2016). Most leucogranites were emplaced between 317 ± 2 and 313 ± 2 Ma (U-Pb zircon), and a late-D3 magmatic pulse marked by the emplacement of a muscovite-rich leucogranite intrusion occurred between 308 ± 3 Ma (Díez Fernández and Pereira, 2017) and 300 ± 2 Ma (Ferreira et al., 2019). The beginning of emplacement of leucogranites at ca. 319-317 Ma was interpreted to mark the evolution from D2-related extension to D3-transpression, and thus a transition from vertical to horizontal shortening (Díez Fernández and Pereira, 2016). Ferreira et al. (2020) describe gradual transition from metatexites, diatexites, to syn-D3-leucogranites within the FCL complex, with peak low pressure-high temperature metamorphic conditions for calc-silicate rocks interlayered with metapelites of 760 ± 50 °C and 5.0 ± 1.0 kbar (Pereira et al., 2017). U-Pb zircon ages of diatexites and metatexites from the FCL complex overlap with that of leucogranites and range from 344 ± 3 to 315 ± 1 Ma with peak metamorphism estimated at 317 ± 2 Ma (Ferreira et al., 2022). Carboniferous migmatites and leucogranites from the FCL complex are separated from the overlying lower grade Neoproterozoic to Cambrian metasedimentary rocks by the Pinhel shear zone, a D2 extensional structure that was folded during D3 tectonics and eventually localized strike-slip shear zones such as the Huebra structure (Díez Fernández and Pereira, 2016) (Fig. 2a). The overlying metasedimentary envelope located to the north and south of the FCL complex was uprightly folded in relationship with both D1 and D3 deformations, and the core of major E-W oriented synclines is marked by the occurrence of Ordovician quartzites (Díez Fernández and Pereira, 2016).

The ∼20 km long and up to 6 km wide Fregeneneda-Almendra pegmatite field occupies a roughly E-W oriented belt of Neoproterozoic to Cambrian metasedimentary rocks (Schist-Graywacke Complex) located to the north of the FCL complex and to the south-west of the two-mica leucogranite of Saucelle (López Plaza et al., 1982; Villar et al., 1990; Ferreira da Silva and Luisa Ribeiro, 1994; Roda-Robles et al., 1999, 2023; Vieira et al., 2011) (Fig. 2b). The supracrustal rocks mostly consist of metapelites, metagraywackes and quartzites with minor calc-silicate rocks. The roughly E-W-oriented sedimentary layering (S0) of metasedimentary rocks was affected by D1 and D3 folds with globally WNW-ESE oriented axes (Figs. 2b and 3), and S1 and S3 fabrics typically host a M1-M3-related greenschist-facies mineral assemblage (Díez Fernández and Pereira, 2016). The core of the main syncline structure occurring to the north of the FA field is filled by Paleozoic metasedimentary rocks. This suggests that the sedimentary series was not overturned and that the thickness of the metasedimentary envelope increases toward the north (i.e., a northward deepening of the D2 Pinhel extensional shear zone, bounding migmatites and leucogranites) (Díez Fernández and Pereira, 2016) (Figs. 2b and 3). Metamorphic isograds in metasedimentary rocks were interpreted as striking perpendicular to the contact with the FCL by Dias da Silva (2013). However, several studies have shown that the metamorphic grade of metasedimentary rocks increases toward the south and that the variably folded metamorphic isograds are oriented roughly parallel to the contact with the FCL complex (Carnicero, 1982; Ferreira da Silva et al., 1990b, 1990a; Vieira, 2010) (Figs. 2b and 3). Index minerals defining those isograds evolve from (i) chlorite, (ii) biotite, (iii) andalusite ± cordierite to (iv) sillimanite toward the south and define a typical sequence of low pressure − high temperature metamorphism (Carnicero, 1982). Those M2-related index minerals are characteristic of S2 fabrics that developed in relation with deep crust exhumation, migmatization and leucogranitic magmatism during regional extension (Díez Fernández and Pereira, 2016).

Minor pegmatites were emplaced within leucogranites of the FCL and the majority was emplaced within Neoproterozoic to Cambrian metasedimentary rocks. Three main groups of pegmatites, along with ten subtypes, were defined in the FA pegmatite field (see Vieira et al., 2011 and Roda-Robles et al., 2023 for more detailed descriptions) (Fig. 2b). Group 1, including four subtypes (i.e., T1-4), consists of intra- to peri-granitic barren pegmatite dykes bearing quartz, feldspar, muscovite, tourmaline ± andalusite ± garnet. They are commonly deformed and generally strike parallel to the roughly E-W oriented S3 foliation within the metasedimentary rocks and the leucogranites. Group 2 and Group 3 pegmatite dykes are generally sub-vertical and mainly crosscut the foliations of metasedimentary rocks with an NNE-SSW to NE-SW orientation, although their orientation and dip angles are locally variable, in particular close by the FCL complex (Figs. 2b and 4a) (Villar et al., 1990). Group 2 weakly-mineralized pegmatites (T5-6) are characterized by the presence of Fe-Mn phosphates along with amblygonite-group minerals (AGM; [Li,Na)Al(PO4)(F,OH)]), and variable proportion of columbite-group minerals (CGM) and cassiterite. Group 3 pegmatites also host AGM and are characterized by the presence Li-bearing aluminosilicates and a higher amount of cassiterite and CGM (Fig. 4). Vieira et al. (2011) identified four subtypes of Li-pegmatites (i.e., Group 3) corresponding to petalite-rich (T7, Fig. 4a, b, c), spodumene-rich (± petalite-bearing) (T8), Li-mica-rich and spodumene-bearing (T9), and Li-mica-rich and spodumene-free (T10) (Fig. 4d, e, f) pegmatites. Those Li-pegmatite subtypes (T7 to T10) were interpreted as defining a progressive fractional crystallization series, with Group 2 weakly mineralized pegmatites as the least evolved members (Roda-Robles et al., 2010, 2023; Vieira et al., 2011). The degree of host-rock M2 metamorphism, defined by the presence of index minerals, seems to exert a primary control over the Li-pegmatite subtypes; petalite-rich pegmatites (T7) were mostly emplaced within the andalusite to biotite zones, whereas spodumene-rich pegmatites (T8) and Li-mica-rich ± spodumene-bearing pegmatites (T9-10) were dominantly emplaced in lower grade metasedimentary rocks within the biotite to chlorite zones (Vieira et al., 2011) (Figs. 2b and 3). Moreover, a particular type of cassiterite-rich quartz veins, hosting muscovite and minor K-feldspar, albite and CGM, occurs in the Feli open pit to the north of the pegmatite field (Fig. 2b). Those veins are folded, discordant to the foliation of metasedimentary rocks and crosscut by undeformed Li-mica-rich spodumene-bearing pegmatite dykes (T9) (Roda-Robles et al., 2010, 2023;,Vieira et al., 2011). Drilling in this location has revealed the presence of a hidden muscovite-bearing granitic body that was suggested as potential source for Feli cassiterite-rich veins (Roda-Robles et al., 2009, 2023) (Fig. 3).

Detailed analytical protocols along with mineral compositions and U-Pb isotope data are provided in Supplementary files. Thick section chemical mapping was done by µ-XRF. Oxide mineral backscattered electron (BSE, for CGM and cassiterite) and cathodoluminescence (CL, for cassiterite) imaging were performed using a scanning electron microscope. Higher resolution chemical mapping of oxide minerals was performed by energy-dispersive X-ray spectroscopy (EDS) or wavelength-dispersive X-ray spectroscopy (WDS) using a scanning electron microscope and electron microprobe, respectively. Major and minor elemental abundances were determined by electron microprobe for CGM, cassiterite, micas and phosphates, and trace element abundances were measured by laser ablation-inductively coupled plasma mass spectrometry (LA-ICP-MS) for cassiterite. Cassiterite U-Pb geochronology was done by LA-ICP-MS using Yankee cassiterite as primary reference material (Carr et al., 2020), and Jian-1 cassiterite as control material (Tapster and Bright, 2020), following the method of Carr et al. (2023). Columbite-group mineral U-Pb dating was performed by LA-ICP-MS using zircon crystals as reference (GJ1, Jackson et al., 2004) and control (Plešovice, Sláma et al., 2008; M127, Nasdala et al., 2016) materials following the method described in Melleton et al. (2022). At the time of the analysis, no internationally recognized reference material was available for CGM. However, Melcher et al. (2008) did not observe significant matrix-dependent U-Pb fractionation when employing zircon as a primary reference material during CGM dating by LA-ICP-MS. This method generally provides age results consistent with those obtained through conventional matrix-matched U-Pb dating of other minerals such as monazite or apatite (Fosso Tchunte et al., 2018; Marcoux et al., 2021). Moreover, further investigations have revealed that matrix effects are primarily associated with tantalite compositions (Legros et al., 2019), and CGM analyzed during this study mostly range in compositions from Fe- to Mn-columbite. U-Pb diagrams and ages were produced with IsoplotR (Vermeesch, 2018) with errors quoted at the 2-sigma level, ages including the estimated long-term reproducibility of 1.9% for cassiterite (Carr et al., 2023).

Samples were collected from Nb-Ta-Sn-oxide-mineral-bearing pegmatites over the FA area. Large quantities of cassiterite, along with CGM or rare pyrochlore-groupminerals, were found in eight pegmatites, including four petalite-rich (T7 of Vieira et al., 2011), two Li-mica-rich-spodumene-bearing (T9: Feli, P010) and two Li-mica-rich spodumene-free (T10: Ramala-West and −East) pegmatites (Figs. 2b and 4). For the remainder of this paper, we refer to the T9 and T10 pegmatites as Li-mica-rich pegmatites following the Iberian scale classification of Roda-Robles et al. (2018). Pegmatite sample mineralogy along with GPS coordinates are provided in Table 1 and a paragenetic sequence summarizing our textural observations and interpretations is provided in Figure 5.

Generally, Li-pegmatites do not display well-defined large-scale zoning, for example a wall zone or a quartz core that are common to other pegmatites deposits (London, 2014). However, petalite-rich pegmatites commonly have a heterogeneous texture with large (> 2 cm) subhedral to anhedral crystals of K-feldspar and petalite, occurring in a finer-grained (< 1 cm) matrix mostly consisting of albite, quartz, muscovite, phosphate and oxide minerals (Fig. 4b, c). K-feldspar phenocrysts show resorption textures and are partially replaced by albite (Fig. 6a, b), and AGM are corroded by muscovite (Fig. 7a). Moreover, F-apatite, along with rare Al-Sr-Ca-F phosphates (probably a solid solution between crandallite and goyazite ; Roda-Robles et al., 2010), commonly occur as alteration products around AGM and Fe-Mn-phosphates (Figs. 6a, b and 7b). Moreover, as previously noted by Vieira et al. (2011), zinnwaldite (a Fe-F-Li-rich trioctahedral mica, Fig. 8) was found as small secondary overgrowths on muscovite in the Bajoca petalite-rich pegmatite (Fig. 7a, b).

Li-mica-rich pegmatites (or aplo-pegmatites) commonly show unidirectional solidification textures forming a layering marked by the alternation of Li-mica-rich and quartz-albite-rich millimeter to centimeter thick zones parallel to the contact with the country-rock (Figs. 4e and 6c). K-feldspar was found as anhedral crystal relics, partly replaced by albite, Li-mica and quartz, within the Ramalha-W pegmatite, and as inclusions within topaz in the Ramalha-E pegmatite (Fig. 7c). Most topaz and AGM appear to have crystallized relatively early as they are partially resorbed by Li-micas (Fig. 7c, d).

Amblygonite-group minerals in both pegmatite types do not contain detectable Na and overlap in F concentrations between 0.25 and 0.45 atoms per formula unit (apfu) (Supplementary File 2). However, AGM from P010 and Ramalha-E Li-mica-rich pegmatites reach slightly higher F concentrations than that from the petalite-rich Bajoca pegmatite (< 0.4 apfu). Some AGM crystals in the P010 pegmatite have low F contents < 0.3 apfu and possibly experienced subsolidus alteration (e.g., London and Burt, 1982).

Dioctahedral mica flakes in petalite-rich pegmatites have a typical muscovite composition, Ti contents up to 0.02 apfu (based on 22 oxygen atoms) and relatively low F concentrations below 0.3 apfu (Fig. 8). Conversely, micas in Li-mica-rich pegmatites are mostly trioctahedral Li-muscovite to lepidolite (classification of Tischendorf et al., 1997, not shown) with 1 to 3 apfu F, and Ti contents < 0.01 apfu. Those micas are locally patchy zoned with dioctahedral micas (muscovite) having relatively low F of ∼0.1–1 apfu. Micas in Li-mica-rich pegmatites are enriched in Mn relative to Fe (Mn minus Fe ≥ 0 apfu) compared to that from petalite-rich pegmatites (Mn − Fe < 0 apfu) (Fig. 8).

Cassiterite and CGM were observed together in most samples, except in Plombal petalite-rich and Ramalha-W Li-mica-rich pegmatites where CGM are rare or absent (Tab. 1; Figs. 6, 9 and 10). Pyrochlore-group minerals with a fluorcalciomicrolite composition were found in the Ramalha-W pegmatite (Fig. 9e). In the studied samples, oxide minerals are generally coarser in petalite-rich pegmatites than in Li-mica-rich pegmatites. Cassiterite commonly forms large, 0.5 to 5 mm in size, subhedral to anhedral short prismatic crystals whereas CGM generally occur as numerous small, tens of micrometers to 1 mm in size, elongated prismatic crystals. Cassiterite crystals are commonly partially resorbed by albite and micas (Figs. 6a). Columbite-group minerals tend to cluster with AGM, especially in petalite-rich pegmatites, suggesting that most CGM crystallized along with or slightly after AGM (Figs. 6 and 9a). However, some CGM probably crystallized later as reflected by the local occurrence of albite-quartz-CGM-bearing veins crosscutting K-feldspar (Fig. 6b).

Columbite-group minerals are characterized by variably complex internal textures and are commonly overgrown with rim structures or show intra-crystal mantles with bright BSE signals that are enriched in Ta relative to Nb (Fig. 9). In CL images, cassiterite crystals from petalite- and Li-mica-rich pegmatites commonly show a core characterized by dark color and growth zonation (Fig. 10). This low-CL cassiterite core is locally brecciated and overgrown by cassiterite with yellow CL color that may have patchy or oscillatory zoning. The high-CL cassiterite hosts inclusions of concentrically zoned CGM, albite, mica, quartz and rare AGM, whereas the low-Cl cassiterite is inclusion-free. The high-CL cassiterite is depleted in Nb, Ta, Mn and Fe compared to the low-CL cassiterite (Fig. 10).

Columbite-group minerals from petalite-rich pegmatites are mostly columbite-Fe, with Mn/(Mn+Fe)apfu ratios, or Mn*, between ∼0.05 and 0.60, whereas the large majority of CGM from Li-mica-rich pegmatites are columbite-Mn with Mn* values close to 1 (Fig. 11a). The range of Ta/(Nb+Ta)apfu ratios, or Ta*, of CGM within each pegmatite sample is comparable and vary from ∼0.05 to 0.6, with the highest Ta* values generally occurring in CGM rims or mantles. Columbite-group mineral inclusions within high-CL cassiterite from the Bajoca petalite-rich pegmatite have slightly lower Mn* values (0.03–0.13) than discrete CGM crystals from the same pegmatite (∼0.1–0.50). However, such compositional difference is not observed between CGM inclusions in cassiterite and discrete CGM from Li-mica-rich pegmatites. Columbite-group minerals from the different petalite-rich pegmatites show distinct geochemical signatures. Their mean Mn* and Ta* values increase from the Bajoca pegmatite emplaced in the andalusite zone (Mn mean* = 0.21, Ta mean* = 0.14) to the Hinojosa (0.33, 0.20) and Ramalha (0.45, 0.33) pegmatites emplaced within the biotite zone (Fig. 2b). In Figure 11b, the Sn concentrations of CGM broadly correlate with their Ti contents for petalite-rich pegmatites, and CGM from the Bajoca pegmatite reach higher Ti concentrations than that from other pegmatites of the same type. Columbite-group minerals from Li-mica-rich pegmatites are characterized by lower Ti contents than CGM from petalite-rich pegmatites, but have a similar range of Sn concentrations that do not correlate with their Ti contents. Few CGM were analyzed for their Zr concentrations; CGM from the Feli Li-mica-rich pegmatite contain minor amount of Zr up to 0.03 apfu (0.7 wt%) whereas Zr is below 0.01 apfu (0.07 wt%) in Bajoca petalite-rich pegmatite CGM (Supplementary File 2).

Cassiterite from both pegmatite types reaches relatively high Nb + Ta concentrations up to 0.1 apfu (∼7 wt%) that correlate with Mn + Fe contents (Fig. 12a). However, cassiterite from both pegmatite types defines distinct geochemical trends, whereby compositions from Li-mica-rich pegmatites tend to plot above the CGM substitution line and are relatively enriched in Nb + Ta at given Mn + Fe contents compared to that from petalite-rich pegmatites (Fig. 12a). Cassiterite from Li-mica-rich pegmatites are characterized by significantly higher Mn/Fe (0.1–90) ratios, Zr (300–2500 ppm) and Ga concentrations (> ∼10 ppm) along with lower Ti/Zr ratios (< ∼0.6) and Ti concentrations (30–640 ppm) than cassiterite from petalite-rich pegmatites (Mn/Fe < ∼0.1; Ti/Zr > ∼0.6; Zr ∼130–1000 ppm; Ga < ∼10 ppm; Ti ∼200–3400 ppm; Figs. 12c and 13). However, cassiterite from both pegmatite types shows similar Nb/Ta ratios (0.3–9.0) as well as Nb and Ta concentrations (Fig. 13). At the pegmatite field scale, cassiterite roughly evolves toward lower Ti concentrations and Ti/Zr ratios, and higher Mn/Fe ratios from petalite-rich pegmatites emplaced in the andalusite zone (Bajoca-Plombal) to that emplaced in the biotite zone (Hinojosa-Ramalha) (Figs. 12c and 13). Regarding Li-mica-rich pegmatites, cassiterite from the Feli pegmatite, emplaced in the chlorite zone, is characterized by higher Mn/Fe ratios than that from P010 (biotite zone) and Ramalha-W (biotite-chlorite zone transition) pegmatites. In the Bajoca pegmatite, early, low-CL, cassiterite shows significantly higher Nb/Ta ratios as well as Ga, Nb, Ta, Mn and Fe concentrations than late, high-CL, cassiterite, whereas differences in terms of Mn/Fe ratios and Ti concentrations are only minor between both cassiterite generations (Figs. 12b,c and 13).

Cassiterite is variably enriched in common Pb and plots in concordant to highly discordant position in Tera-Wasserburg diagrams (Fig. 14). In all samples, cassiterite analyses define discordia lines allowing to calculate lower intercept ages mostly ranging from 315 ± 6 Ma to 308 ± 6 Ma (Tab. 2). In Bajoca and Hinojosa petalite-rich pegmatites as well as in Li-mica-rich pegmatites, some discordant analyses plot outside of the discordia lines defined by most analyses. Those data, interpreted to reflect secondary disturbances of the U-Pb isotopic composition such as the loss of U or Pb (Fig. 14a), were excluded from lower intercept age calculations. Two samples, including Ramalha-W and Ramalha-E Li-mica-rich pegmatites, provide intercept ages of 312 ± 7 Ma and 310 ± 25 Ma with high MSWD values of 6.8 and 4.6, respectively, reflecting individual geological age dispersion (Fig. 14). However, those intercept ages are comparable to intercept ages yielded by other samples, supporting their geological significance.

The discordia lines defined by cassiterite analyses mostly return initial 207Pb/206Pb(0) compositions between 0.83 ± 0.02 and 0.87 ± 0.01 globally in agreement with the model terrestrial Pb isotope composition at 310 Ma (i.e., 207Pb/206Pb(310) ∼0.86; Stacey and Kramers, 1975) (Fig. 14 and Tab. 2). However, cassiterite from the Plombal pegmatite yields a weakly constrained and abnormally high 207Pb/206Pb(0) ratio of 1.09 ± 0.7, due to the sub-concordant to concordant character of most of the analyses (Fig. 14c). Therefore, although identical within error, we recommend the use of the concordia age of 315 ± 6 Ma over the lower intercept age of 314 ± 6 Ma for the Plombal pegmatite. In the Bajoca pegmatite, cassiterite yields also similar concordia and lower intercept ages of 317 ± 6 and 315 ± 6 Ma, respectively (Fig. 14a). We nevertheless recommend the use of the lower intercept age over concordia age for data interpretation in this case as it takes into account a higher number of analyses and yield a geologically reasonable 207Pb/206Pb(0) of 0.87 ± 0.01. Analyses performed on high-CL cassiterite from the Bajoca pegmatite are strongly discordant partly due to low U concentrations generally < 1 ppm, and provide a poorly constrained lower intercept age of 300 ± 27 Ma (MSWD = 22) (Fig. 14a, Supplementary material 4). In contrast, low-CL cassiterite (U ∼ 5–25 ppm) yields a lower intercept age of 315 ± 6 Ma (MSWD = 0.3) identical to that calculated by combining both cassiterite generations from this pegmatite.

Discrete CGM crystals of ∼100–500 μm in size were handpicked from the petalite-rich Bajoca pegmatite (MB1001b) and Li-mica-rich Feli pegmatite (MF1001C). No differences in terms of U-Pb isotope composition can be attributed to CGM zoning. Columbite-group mineral analyses from the Bajoca pegmatite plot in sub-concordant to slightly discordant position in the Tera-Wasserburg diagram and yield a mean 206Pb/238U age of 309 ± 3 Ma (n=8, MSWD = 0.5; Fig. 15a). Analyses of CGM from the Feli pegmatite are concordant to sub-concordant. The seven most concordant analyses (discordance ≤ 3%) yield a concordia age of 308 ± 5 Ma (MSWD = 1.0) identical to the lower intercept age of 308 ± 4 Ma (MSWD = 1.2) and mean 206Pb/238U age of 308 ± 3 Ma (MSWD = 0.4) calculated using all analyses (n = 15, Fig. 15b). Columbite-group mineral U-Pb ages are identical within error to the cassiterite lower intercept U-Pb ages obtained on the Bajoca (315 ± 6 Ma) and Feli (310 ± 6 Ma) pegmatites (Fig. 14).

Oxide mineral crystallization

Cassiterite crystals from Li-pegmatites of the FA field are locally partially resorbed by albite (Fig. 6a), and commonly show evidence for metasomatic overprint with low-CL cores brecciated by high-CL cassiterite relatively depleted in Nb, Ta, Fe and Mn (Figs. 10 and 12). The low-CL cassiterite cores, that are inclusion-free, are interpreted to have crystallized at the magmatic stage, along with or slightly after K-feldspar phenocrysts (Fig. 5). Cassiterite with likely magmatic origin has been documented in several peraluminous rare-metal granites and pegmatites (Cuney et al., 1992; Linnen and Cuney, 2005). Its crystallization, reflecting a decrease of Sn solubility, has been interpreted to reflect the low solidus temperature of the flux-rich parental melts of those intrusions along with an increase of oxygen fugacity (Linnen et al., 1996; Bhalla et al., 2005; Pichavant, 2022). In contrast, the late, high-CL, cassiterite of the FA pegmatites notably hosts inclusions of albite, muscovite, CGM and quartz (Fig. 10). This suggests that the high-CL cassiterite formed in relationship with the albitization process that affected K-feldspar, and this metasomatism was also accompanied by the crystallization of micas and quartz (Figs. 6a-b). The Nb-Ta-Fe-Mn component of CGM inclusions/exsolutions in high-CL cassiterite likely originates, at least partly, from the destabilization of early, low-CL, cassiterite cores that are relatively enriched in Nb-Ta and Fe or Mn (Fig. 10). However, a significant part of those elements might have been brought by the albitizing fluid (or hydrosaline melt), as suggested by the local occurrences of albite-quartz-CGM-bearing veins crosscutting K-feldspar in petalite-rich pegmatites (Fig. 6b).

In contrast to low-CL cassiterite, a well constrained U-Pb age cannot be obtained from high-CL cassiterite alone, partly due to its relatively low U, and thus, low radiogenic Pb concentrations (Fig. 14a). Despite the absence of absolute age constrains, albitization is a well-documented magmatic-hydrothermal phenomena of rare-metal pegmatite systems worldwide including the FA pegmatites (Errandonea-Martin et al., 2022). Such auto-metasomatism is commonly thought to occur during or at the end of pegmatite-forming magma crystallization due to the exsolution of a Na-rich hydrosaline melt or fluid (Dewaele et al., 2011; Kaeter et al., 2018; Müller et al., 2018; Ballouard et al., 2020a; Errandonea-Martin et al., 2022). Therefore, the age difference between magmatic, low-CL, and magmatic-hydrothermal, high-CL, cassiterite might be indistinguishable considering the precision of LA-ICP-MS dating technique and the duration for the crystallization of pegmatite bodies generally expected to be lower than 1 Ma (e.g., London, 2018).

Like cassiterite, discrete CGM crystals from FA Li-pegmatites commonly show complex zoning with small overgrowths or intra-crystal mantles with relatively high Ta* values, representing mostly dissolution-reprecipitation textures (Figs. 9 and 11). Such features may be the effect of disequilibrium crystallization (Van Lichtervelde et al., 2018) or the interaction with an aqueous media, including a hydrosaline melt or late-magmatic fluid phase (Kaeter et al., 2018; Ballouard et al., 2020a). Therefore, CGM likely crystallized both at the magmatic stage (i.e., discrete CGM cores) and magmatic-hydrothermal stage (i.e., discrete CGM rims and CGM in inclusion in cassiterite or in veins) (Fig. 5). U-Pb dating of zoned and unzoned discrete CGM crystals yield consistent U-Pb ages with low MSWD values (≤ 1.0, Fig. 15) suggesting that the time differences between these two stages cannot be resolved with the current method precision.

Timing and tectonic-magmatic context of pegmatite-forming magmatism

Cassiterite and CGM U-Pb geochronology of both petalite-rich and Li-mica-rich pegmatites from the FA field yields ages from 315 ± 6 Ma (Bajoca and Plombal petalite-rich pegmatite cassiterite) to 308 ± 6 Ma (Hinojosa petalite-rich pegmatite cassiterite) that are identical within uncertainty (Fig. 16 and Tab. 2). Those ages are slightly older but overlap within uncertainty with the CGM U-Pb age of 302 ± 3 Ma obtained by Roda-Robles et al. (2023) on one crystal from the Carmen Li-mica-rich pegmatite from the FA pegmatite field (Fig. 16). The cassiterite and CGM U-Pb systems yield an older age range than the previously applied muscovite or Li-mica 40Ar/39Ar geochronometer with ages from 305 ± 4 to 295 ± 4 Ma for the pegmatite field, and ages between 297 ± 4 and 295 ± 4 Ma for petalite- and Li-mica-rich pegmatites (Roda-Robles et al., 2009, 2023; Vieira, 2010) (Fig. 16). Pegmatite cassiterite and CGM U-Pb ages overlap with zircon U-Pb ages of the main two-mica-bearing leucogranite facies from the FCL anatectic complex emplaced from 317 ± 2 to 313 ± 2 Ma (Ferreira et al., 2019), and with the youngest ages of zircon crystallization in FCL migmatites (344 ± 3 to 315 ± 1 Ma; Ferreira et al., 2022) (Fig. 16). The mica 40Ar/39Ar age range of FA pegmatites is comparable to: (i) the muscovite 40Ar/39Ar age of 311 ± 4 Ma of a FCL leucogranite (Roda-Robles et al., 2009; Vieira, 2010), (ii) apatite U-Pb ages from 314 ± 7 to 288 ± 14 Ma obtained on FCL migmatites and leucogranites (Ferreira et al., 2019), as well as (iii) the emplacement ages for late-D3 muscovite-rich leucogranite intrusions to the southern border of the FCL complex from 308 ± 3 to 300 ± 2 Ma (zircon U-Pb; Díez Fernández and Pereira 2017; Ferreira et al., 2019). Apatite U-Pb ages were interpreted by Ferreira et al. (2019; 2022) as reflecting the exhumation and cooling of the FCL dome below the closure temperature of the apatite U-Pb system between ∼485 and 465 °C (Cherniak et al., 1991) in relation with D3 transpressive tectonics. Like apatite, the mica 40Ar/39Ar ages may have recorded the cooling of pegmatites below temperatures of ∼550-350 °C, corresponding to the closure of muscovite 40Ar/39Ar system depending on grain size and cooling rate (Harrison et al., 2009). However, the temperatures of emplacement reported for pegmatite-forming magmas are commonly low, between 550°C and 450°C (London, 2014), and within the range of closure temperatures defined for the muscovite 40Ar/39Ar geochronometer. Therefore, a more likely hypothesis to explain the relatively young ages yielded by the mica 40Ar/39Ar system within the FA pegmatite field is a partial or complete open-system behavior of this geochronometer during a late-D3 thermal event, in relationship with magmatic pulses marked by the emplacement of late leucogranite intrusions within the FCL complex until 300 ± 2 Ma (Ferreira et al., 2019). Although Pb diffusion data in cassiterite remains scarce, Zhang et al. (2011, 2014) have reported relatively high closure temperatures > 500 °C for cassiterite grains. To our knowledge, no closure temperature data are available for the CGM U-Pb system. However, pegmatite geochronology studies based on different isotopic systems generally report mica 40Ar/39Ar (and apatite U-Pb) ages significantly younger than CGM and cassiterite U-Pb ages, whereas CGM and cassiterite U-Pb ages are commonly similar to zircon U-Pb ages (e.g., Van Lichtervelde et al., 2017;,Li et al., 2019; Hao-Yu et al., 2023). Therefore, the cassiterite and CGM U-Pb systems are much more likely to have recorded the age of emplacement of Li-pegmatites from the FA field compared to mica Ar-Ar ages.

The CGM U-Pb age of 302 ± 3 Ma for a FA Li-mica-rich pegmatite (Roda-Robles et al., 2023) is surprisingly young (despite some overlap with our age data). However, excess variance or long–term reproducibility for CGM U-Pb ages has not been defined yet, meaning that associated uncertainties are likely underestimated. Considering a more realistic uncertainty of ∼2% for LA-ICP-MS U-Pb age data (Horstwood et al., 2016; Carr et al., 2023), the error of CGM U-Pb ages obtained for FA pegmatites would be of ± 6 Ma. Therefore, although it is possible that Li-pegmatite-forming magmatism in the FA field was pulsatile over a period from ca. 315 to 302 Ma, all cassiterite and CGM U-Pb ages (i.e., from 315 ± 6 to 302 ± 6 Ma) are almost identical within uncertainty around 310-308 Ma.

Overall, our U-Pb CGM and cassiterite dating suggests that both petalite-rich and Li-mica-rich pegmatite dykes from the FA field were mostly emplaced sub-contemporaneously at the end of the main pulse of leucogranite intrusions and migmatization in the FCL anatectic complex (i.e., within the precision of method). Pegmatite-forming magmatism from 315 ± 6 to 308 ± 6 Ma occurred just after D2 extension and contemporaneously with D3 strike-slip tectonics that affected the leucogranites during and after their crystallization (Fig. 16) (Díez Fernández and Pereira, 2016). Within or close to the FCL anatectic complex, in the sillimanite and andalusite M2 metamorphic zones, pegmatite dykes are commonly barren, variably deformed and strike roughly W-E to WNW-ESE parallel to the S3 foliation of their host-rock. In contrast, pegmatites emplaced at higher structural level, from the andalusite to chlorite M2 zones, have a higher rare-metal potential and commonly strike N10 to N30 perpendicular to the S3 axial plane foliation of metasedimentary rocks (Villar et al., 1990; Vieira et al., 2011) (Fig. 2b). Due to their different orientations, Group 1 barren conformable pegmatites were interpreted as significantly older than discordant Group 2 weakly mineralized and Group 3 Li-Sn-Nb-Ta-mineralized pegmatites (e.g., López Plaza et al., 1982;,Ferreira da Silva and Luisa Ribeiro, 1994; Vieira et al., 2011). Such structural relationship is not in conflict with our U-Pb dataset as well as zircon U-Pb ages for leucogranites from the FCL complex (Ferreira et al., 2019), considering the precision of the LA-ICP-MS method (< 2% or ± 6 Ma). However, available U-Pb ages, indeed, suggest that most leucogranites in the FCL complex as well as most pegmatites from the FA field, including barren pegmatites locally intruding FCL leucogranites, were emplaced over a period of about 10 Ma (ca. 317-308 Ma), with possibly a last pulse ca. 302 Ma ago related to late- to post-D3 magmatism (Roda-Robles et al., 2023).

In our view, the various pegmatite orientations over the FA field are, thus, most likely the expression of an evolution of the same D3 strike-slip deformation regime at different structural levels (Figs. 2b and 3). In such sinistral transpressive framework, barren pegmatites in deeper ductile crustal levels would have been mostly emplaced as ∼E-W shear veins within the S3 foliations of leucogranites or metasedimentary rocks, whereas in shallower crustal levels, the Li-rich pegmatite-forming magmas would have been dominantly channeled into semi-brittle to brittle fracture networks including mostly ∼NNE-SSW tension gashes or locally ENE-WSW Riedel shears oblique to the foliations of metasedimentary rocks (Fig. 2b).

The apparent structural level of emplacement of D3-related pegmatites, marked by M2 metamorphic isograds, evolves toward the north along with their Li fertility (Figs. 2b and 3); this could be related to the capacity of vertical migration of their parent melts, largely controlled by their viscosity and relative enrichment in fluxing components such as Li, F, P and H2O (Bartels et al., 2011; Thomas and Davidson, 2012; Fiege et al., 2018). Indeed, the switch from D2 extension to D3 strike-slip tectonics around 317 Ma was likely a progressive process, between a stress field dominated by gravity forces toward horizontal shortening dominated by tectonic forces, and the orientation of deformation axes have probably not fundamentally changed during this transition (Díez Fernández and Pereira, 2016). That is why, although variably folded during D3 compression, M2 isograds and S0 bedding strike roughly parallel to the contact with the FCL complex and S3 fabrics (Fig. 2). Moreover, as suggested by U-Pb zircon dating of migmatites (Ferreira et al., 2022), peak high temperature-low pressure M2 metamorphism in the FCL complex at 317 ± 2 Ma likely preceded but was not completely disconnected with the beginning of pegmatite-forming magmatism at 315 ± 6 Ma. Thus, we suggest that M2 metamorphic zones, recorded by metasedimentary host-rocks of pegmatites, represent relevant proxies of the vertical distance of migration of the D3-related pegmatite-forming magmas from their sources located in the deeper crust (Fig. 3).

The cassiterite and CGM U-Pb ages of pegmatites from the FA field between 315 ± 6 and 302 ± 6 Ma overlap with CGM U-Pb ages obtained for other LCT pegmatites from the Iberian Massif, ranging from 312 ± 4 to 300 ± 4 Ma, as well as cassiterite or CGM U-Pb ages for LCT pegmatites and peraluminous rare-metal granites from the Armorican Massif and the French Massif Central (Fig. 1) (Gourcerol et al., 2019; Marcoux et al., 2021; Melleton et al., 2022). This confirms that the FA pegmatite field represents the local expression of a major and widespread late-orogenic tectonic-magmatic-metallogenic event occurring at the scale of the European Variscan belt.

Pegmatite-forming melt evolution as seen through CGM and cassiterite compositions

U-Pb dating of cassiterite and CGM supports a possible link between FA pegmatites and FCL leucogranites and late migmatites. However, the petrogenetic relationship between those rocks, including different pegmatite types, deserve to be assessed using suitable geochemical tracers. The composition of CGM is widely used as a marker of differentiation in rare-metal pegmatites and granites. Indeed, Mn/Fe and Ta/Nb ratios are known to generally increase in the residual melt during fractional crystallization processes involving micas and/or oxide minerals, as Nb and Fe are more compatible (or less incompatible) than Ta and Mn, respectively (Černý and Ercit, 1985; Linnen and Keppler, 1997; Stepanov et al., 2014; Ballouard et al., 2016, 2020a, 2020b; Garate-Olave et al., 2020). The cassiterite composition as a marker of the magmatic evolution remains unclear, probably because magmatic cassiterite is not as common as CGM in rare-metal granites and pegmatites. Moreover, cassiterite can only accommodate a limited number of elements in its lattice, requiring the use of analytical techniques with low detection limits such as LA-ICP-MS over common electron microprobe analyses, to assess significant compositional variabilities.

Despite a portion of CGM may have crystallized in relation with albitization processes, their Mn-Fe signature appears to be globally inherited from the initial pegmatite-forming melt. Indeed, the Mn* values of discrete CGM from petalite-rich pegmatites evolve consistently with the M2 metamorphic grade of their country-rocks that we suggest as an approximate indicator of the structural levels of emplacement of pegmatites (Fig. 3). Explicitly, CGM from the Bajoca pegmatite emplaced within the andalusite zone have a less evolved geochemical signature, i.e., lower Mn*, than CGM from the Hinojosa and Ramalha pegmatite emplaced at higher structural level within the biotite zone (Fig. 11). In contrast, Ta* variability in the CGM seems to mostly reflect internal evolution processes occurring during pegmatite-forming magma crystallization at the site of emplacement. Tin and Ti concentrations also decrease in CGM from the Bajoca to the Ramalha and Hinojosa petalite-rich pegmatites (Fig. 11b). The decrease of Ti concentrations, an element with a typically compatible behavior in peraluminous granitic melts, likely reflects magma differentiation. The rough correlation between Sn and Ti contents is more difficult to interpret but might reflect the coupled substitution of Ti4+ and Sn4+ in the structure of CGM.

Similarly, Ti concentrations, along with Ti/Zr and Mn/Fe ratios, appear to change in cassiterite along with the structural level of emplacement of pegmatites, whereas those geochemical parameters have remained globally unchanged during metasomatism, as reflected by low-CL and high-CL cassiterite from the Bajoca pegmatite (Fig. 13). Cassiterite from the Bajoca pegmatite emplaced in the andalousite zone generally has higher Ti contents and Ti/Zr ratios along with lower Mn/Fe ratios than other pegmatites emplaced at higher crustal level in the biotite zone such as the Hinojosa and Ramalha pegmatites. Additionally, cassiterite from the Feli Li-mica-rich pegmatite located in the chlorite zone, at the northern extremity of the FA field, is globally characterized by higher Mn/Fe ratios than that from other pegmatites emplaced in the biotite zone (Figs. 12c and 13).

The evolved signature of cassiterite from the Li-mica-rich and spodumene-bearing Feli pegmatite contrasts with interpretations of Viera et al. (2011) based on mica compositions suggesting that this pegmatite subtype is less evolved than Li-mica-rich and spodumene-free subtype (i.e., Ramalha locality). Indeed, muscovite to lepidolite from a Li-mica-rich and spodumene-bearing pegmatite (Viera et al., 2011) have intermediate compositions in terms of Mn-Fe and F systematics between micas from petalite-rich and Li-mica-rich spodumene-free pegmatites (Fig. 8). More samples, coupled with precise estimations of the pressure of pegmatite emplacement, would be needed to reconcile the discrepancies between cassiterite and mica compositions.

In the meantime, available data on CGM and cassiterite are globally consistent with an evolution of their Mn/Fe ratios and Ti compositions in extrinsic relationship with the vertical distance of migration of the pegmatite-forming magmas away from their source(s) that may correspond to buried equivalents of rocks, including leucogranites, from the FCL anatectic complex. Such compositional variation in oxide minerals might have occurred in intrinsic relationship with the evolution of their parent melts. In such scenario, the most evolved pegmatite-forming magmas, with high Mn/Fe ratios and low Ti concentrations (i.e., crystallizing high Mn/Fe and low Ti oxide minerals), could have emplaced at the highest structural levels because they were also the most enriched in fluxing components that influenced their respective capacities of vertical migration within the crust.

The petrogenesis of distinct Li-pegmatite types from the Fregeneda-Almendra field as seen through Mn-Fe and F geochemical fractionation

U-Pb geochronology supports the coeval emplacement and possible cogenetic relationship between the distinct types of Li-pegmatites from the FA field including petalite-rich and Li-mica-rich dykes, as suggested by previous studies based notably on micas, phosphate, feldspar and whole-rock compositions (Roda-Robles et al., 1999, 2010, 2023; Vieira et al., 2011). Moreover, cassiterite and CGM from those two pegmatite types show significantly distinct compositions: oxide minerals from Li-mica-rich pegmatites are notably characterized by important enrichment in Mn over Fe, low Ti concentrations and, at least for cassiterite, high Zr and Ga contents, compared to that from petalite-rich pegmatites (Figs. 11, 12 and 13). Such a distinct composition is particularly marked by the Mn* signature of CGM (Fig 11a). Comparisons with CGM and cassiterite data from the literature for the FA pegmatite field (Roda-Robles et al., 1999; Vieira, 2010) (Figs. 11 and 12) confirm this tendency and also indicate that CGM and cassiterite from spodumene-rich pegmatites, weakly mineralized pegmatites and quartz-cassiterite-rich veins have compositions comparable to that from petalite-rich pegmatites.

A similar composition gap in terms of Mn* compositions was observed for CGM between Li-mica-rich pegmatites and petalite/spodumene-rich or weakly mineralized pegmatites from the Barroso-Alvao pegmatite field (Galicia Trás Os Montes Zone of the Iberian Massif), that were emplaced sub-synchronously with FA pegmatites (Martins et al., 2011) (Fig. 1). These authors suggested that this compositional gap might reflect the existence of distinct melt batches, and thus that Li-mica rich pegmatites may not represent a more evolved equivalent of petalite/spodumene-rich pegmatites. However, in the case of the FA field, such compositional gap is partially filled by CGM from a spodumene-bearing Li-mica-rich pegmatite (Vieira, 2010) (Fig. 11a) suggesting that the distinct oxide mineral compositions shown by the different Li-pegmatite types could reflect a continuous magmatic evolution. Indeed, strong Mn over Fe enrichment is a common feature of CGM from lepidolite-bearing pegmatites worldwide, and such geochemical fractionation have been interpreted to reflect the high F activity, and more specifically the high alkali-fluorine over HF chemical potential (μalkF/μHF), of their parental melts (Černý, 1989, 1992; Wise et al., 2012). Moreover, F notably increases the solubility of zircon in silicate melts (Keppler, 1993; Aseri et al., 2015), and this may explain the higher concentrations of Zr, and probably Ga, of cassiterite from FA Li-mica-rich pegmatites compared to that from petalite-rich pegmatites (Fig. 13). In the FA field, the high F activity of Li-mica-rich pegmatite-forming magmas, relative to that of petalite-rich muscovite-bearing pegmatites, is at first order suggested by the presence of lepidolite with up to 8 wt% F, also characterized by enrichment of Mn over Fe (Fig. 8), as well as topaz (Fig. 7c) or fluorcalciomicrolite (Fig. 9e) for the Ramalha Li-mica-rich pegmatites. In the Ramalha-E pegmatite, topaz is partially resorbed by lepidolite (Fig. 7c) suggesting a decrease of HF chemical potential, stabilizing lepidolite over topaz (Wise et al., 2012), during magma crystallization.

In the F vs Mn minus Fe diagram (Fig. 8), micas from pegmatites of the FA field (data from this study and Vieira et al., 2010), as well as from potential source leucogranites from the FCL complex and buried Feli intrusion (data from Vieira, 2010) do not define chemical trends compatible with a fractional crystallization series. Consistently with the evolution of Mn/Fe ratios of oxide minerals, the Mn − Fe values of micas increase from (i) granites and barren pegmatites, (ii) petalite/spodumene-rich and weakly mineralized pegmatites to (iii) Li-mica-rich pegmatites. Moreover, mica Mn − Fe values anticorrelate with mica Ti contents (not shown). However, the muscovite from weakly mineralized and petalite/spodumene-rich pegmatites, as well as few dioctahedral micas from Li-mica-rich pegmatites, is generally depleted in F (≤ 0.15 apfu) compared to that from granites and barren pegmatites (0.15–0.50 apfu). Such anti-correlation between Mn − Fe and F contents is not consistent with a magmatic evolution process during which F is expected to behave incompatibly and be enriched in the residual liquid.

In order to resolve the discrepancy shown by the F composition of micas, we have calculated in Figure 17a the F concentration of the granitic melts in equilibrium with dioctahedral (muscovite) and trioctahedral (Li-muscovite to lepidolite) micas from Li-pegmatites of the FA field based on available partition coefficients KdFmica/granitic melt (Icenhower and London, 1995; Pichavant et al., 2016; Villaros and Pichavant, 2019). The calculated melts in equilibrium with Li-muscovite to lepidolite from Li-mica-rich pegmatites have relatively high F concentrations between ∼0.7 and 3.0 wt%, and such composition matches that of the silicate liquid in equilibrium with fresh AGM from those pegmatites (∼1–2 wt% F), calculated using the regression of London et al. (2001) (Fig. 17b). The F composition of those melts were mostly above that of the silicate liquid equilibrated with muscovite from the Feli leucogranite (0.5–1.3 wt%) considering the uncertainty on partitioning coefficients. However, the liquid equilibrated with muscovite from the petalite-rich pegmatites was strongly depleted in F (<0.1 to ∼0.5 wt%) similar to or below that of the melt in equilibrium with the muscovite from the FCL leucogranite (∼0.2–0.4 wt%). Those melt compositions contrast with that calculated based on the F concentrations of AGM from the Bajoca-petalite-rich pegmatite that have equilibrated with a significantly enriched melt with 0.8 to 1.5 wt% F.

The compositional differences between the melt in equilibrium with muscovite and AGM for the Bajoca pegmatite likely reflects the depletion in F of the liquid during pegmatite crystallization as AGM is partially resorbed by muscovite in petalite-rich intrusions (Fig. 7a, b). Therefore, AGM are probably early magmatic near-liquidus phases that crystallized from a F-rich melt whereas muscovite might have crystallized near solidus from a F-poor melt that experienced the exsolution of a F-rich fluid (or hydrosaline melt) (Fig. 5). Evidence for such F-rich fluids include the presence of zinnwaldite overgrowths with up to 8 wt% F on muscovite in the Bajoca-pegmatite (Figs. 7a, b and 8), and strong F enrichment from few hundred ppm to up to 0.3 wt% of its metasomatized metasedimentary country-rock (Errandonea-Martin et al., 2022). The modeling presented in Figure 17 suggests that the melt F concentration decreased from ∼1.0 to 0.2 wt% between AGM and muscovite crystallization. Therefore, up to 80 wt% of the F budget of the initial melt might have been lost in an immiscible aqueous phase upon crystallization. Amblygonite-group minerals are also partially replaced by Li-muscovite and lepidolite in Li-mica-rich pegmatites (Fig. 7d) but both crystallized from a F-rich melt with ∼1–3 wt% F (Fig. 17). This indicates that F was mostly trapped in trioctahedral micas before melt-fluid (or silicate melt-hydrosaline melt) immiscibility and exsolution, although metasomatism involving aqueous media is notably recorded by strong F enrichment in Li-mica-rich pegmatite country-rocks (up to 0.7 wt%, Errandonea-Martin et al., 2022).

The fact that the initial melt composition of petalite-rich pegmatites exhibits only slightly lower F concentrations (≤ ∼1.5 wt%), although not significantly different from those of Li-mica-rich pegmatite-forming melts (≤ ∼2 wt%), suggests that lepidolite stability is influenced not solely by the magma F content, but also by other physicochemical parameters. Relatively high oxygen fugacity, between ∼NNO+3 and NNO+1, is a prerequisite to crystallize Li-micas, or muscovite, relative to biotite in granitic magmas as suggested by crystallization experiments (Pichavant et al., 2016; Pichavant, 2022). However, neither petalite-rich nor Li-mica-rich pegmatites host magmatic biotite, but both contain early cassiterite with likely magmatic origin, suggesting that the parental melts of both pegmatite types were characterized by similarly high fO2. Alternatively, the pressure of magma emplacement might have played a significant role for lepidolite crystallization as Li-mica-rich pegmatites were globally emplaced at higher structural levels than petalite-rich pegmatites (Figs. 2 and 3). The pressure could have directly affected the stability of lepidolite or more likely affected the water concentration of the melt, as H2O solubility is known to increase with pressure. The decrease of H2O/F melt activity ratio, or more precisely μHF/μalkF as suggested by Wise et al. (2012), from petalite- to Li-mica-rich pegmatites was possibly the trigger for lepidolite (and Li-muscovite) crystallization. Experimental work would be needed to confirm this hypothesis.

Overall, oxide mineral, phosphate and mica compositions along with CGM and cassiterite U-Pb geochronology support a cogenetic relationship between petalite-rich and Li-mica-rich pegmatites. In such a framework, the Mn/Fe and Ti compositions of minerals suggest that Li-mica-rich pegmatites represent a more differentiated equivalent of petalite-rich pegmatites, as suggested by previous studies (e.g., Vieira et al., 2011; Roda-Robles et al., 2023). The F composition of the initial melts of both pegmatite types was relatively similar although probably slightly higher in Li-mica-rich pegmatites. U-Pb geochronology and mineralogical-chemical zonation of the pegmatite field suggests that the source of pegmatites could correspond to buried equivalents of rocks from the FCL complex. Our data do not allow to favor either a residual or anatectic origin for Li-pegmatites as the timing of their emplacement overlaps with the zircon crystallization ages of both migmatites and granites in the complex (Fig. 16). Equivalents of the leucogranite intrusion located below the Feli mine could have represented fertile parent magmas, notably considering the high F concentration of Feli granite muscovite (data from Vieira, 2010) (Fig. 17). However, without zircon U-Pb dating, the potential relationship between such evolved leucogranite intrusions and Li-pegmatites remains uncertain.

U-Pb dating of cassiterite and CGM from pegmatites from the FA field demonstrates the sub-coeval emplacement of two main types of Li-Sn-Nb-Ta-rich pegmatite intrusions, including Li-mica-rich and petalite-rich pegmatites, in relation with D3-strike-slip tectonics from ca. 315 to 308 Ma (Fig. 16). A late pulse of Li-pegmatite-forming magmatism possibly occurred ca. 302 Ma ago (Roda-Robles et al., 2023). Several other Li-pegmatite fields from Iberia were also emplaced during this period (Melleton et al., 2022) (Fig. 1), supporting late-orogenic transcurrent tectonic framework as a particularly favorable context for the emplacement of rare-metal pegmatite intrusions. Indeed, such context likely favors the generation of the vertical dilation structures (e.g., tension gashes) necessary for flux-rich magma ascent. Intimate relationship between strike-slip tectonics and peraluminous rare-metal magmatism is also documented, for example, in the Argemela area, Iberian belt (Michaud et al., 2020) (Fig. 1), the Variscan Echassières district, French Massif Central (Gagny and Jacquot, 1987) (Fig. 1), the Neoproterozoic Seridó belt, Brazil (Araújo et al., 2001) and the Mesoproterozoic Orange River pegmatite belt, South Africa (Ballouard et al., 2020a).

The geographic zonation of pegmatite around the FCL anatectic dome and U-Pb geochronology suggest a genetic link between buried equivalents of ca. 317-313 Ma old leucogranites (or migmatites) from the dome and FA rare-metal magmatism. Cassiterite, CGM and micas show decreasing Ti contents as well as increasing Mn/Fe ratios and Zr-Ga contents (for oxide minerals) from petalite- to Li-mica-rich pegmatites. This is consistent with a magmatic differentiation process, notably leading to an increase of the F activity of parental melts. Geochemical analyses of coexisting phosphates and micas suggest a decrease of F melt concentrations from ≤ 1.5 to < 0.5 wt% between near-liquidus AGM and near-solidus muscovite crystallizations in petalite-rich pegmatites. Most of the initial F melt budget may have been lost to an aqueous phase during the magmatic-hydrothermal transition. Late circulation of F-rich fluids is reflected by the local occurrence of zinnwaldite overgrowths on muscovite, and strong enrichment in F of pegmatite country-rocks (Errandonea-Martin et al., 2022). The melts equilibrated with AGM and Li-muscovite to lepidolite in Li-mica-rich pegmatites had similarly high F concentration ≤ ∼2–3 wt% indicating that F was largely trapped in Li-micas before the magmatic-hydrothermal transition. Relatively similar F concentrations in the initial melt of both petalite- and Li-mica-rich pegmatites support the hypothesis that lepidolite stability depends not on only on F activity but also on other parameters such as a relatively low H2O activity (μHF) relative to alkaline-fluorine salinity (μalkF), as previously suggested by Wise et al. (2012). We tentatively propose that the lower H2O activity in the melt of Li-mica-rich pegmatites from the FA field, compared to that of petalite-rich pegmatites, was due to a lower pressure of emplacement that could have decreased water solubility.

The post-doctoral fellowship of C. Ballouard was supported by the French National Research Agency through the national program ‘Investissements d’avenir’ with the reference ANR-10-LABX-21-RESSOURCES21, the Grand Est region via the program ‘Jeunes chercheurs’ 2020 (19_GE9_066) and the ANR TRANSFAIR (ANR-21-CE01-0022-01). The Master’s thesis of F. Parisot was funded by the ANR-10-LABX-21-RESSOURCES21 program. Additional funding was provided by the ERA‐MIN/0001/2017–LIGHTS project. The LA-ICP-MS laboratory of GeoRessources in Nancy was funded by the Labex Resources 21, the Région Lorraine and the European Community through the FEDER program. R. Schmitt participated to the sampling and provided some field photographs. We acknowledge E. Roda-Robles and A. Lima for discussions and support during sampling. We are grateful to E. Moreira for her help during LA-ICP-MS analyses as well as to A. Petrelli for LIGHTS project management. A. Flammang and J. Moine are thanked for thick section preparation. Earlier versions of this manuscript benefited from the scientific comments of M. Sirbescu, J. Cempírek, A. Acosta-Vigil and one anonymous reviewer. We thank M. Van Lichtervelde and one anonymous reviewer for their comments on the present version, and we are grateful to M. Poujol for editorial handling.

Cite this article as: Ballouard C, Carr P, Parisot F, Gloaguen É, Melleton J, Cauzid J, Lecomte A, Rouer O, Salsi L, Mercadier J. 2024. Petrogenesis and tectonic-magmatic context of emplacement of lepidolite and petalite pegmatites from the Fregeneda-Almendra field (Variscan Central Iberian Zone): clues from Nb-Ta-Sn oxide U-Pb geochronology and mineral geochemistry, BSGF - Earth Sciences Bulletin195: 3.

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.