Introduction

Platinum Group Elements (PGE) mineral deposits have a direct relationship with the generation of mantle-derived mafic magmas (Naldrett 2004; Mungall 2007; Mungall and Naldrett 2008, and references therein), but their formation requires specific favourable conditions during mantle partial melting and subsequent magma evolution, in order to achieve substantial PGE concentrations (e.g., Mungall 2002; Maier et al. 2003; Naldrett 2004; Kinnaird et al. 2005; Sá et al. 2005; Mungall and Brenan 2014). On the other hand, the generation of gold mineral deposits can be associated to fairly different processes (magmatic, hydrothermal and sedimentary) and geological contexts, the most significant of which are the collisional, supra-subduction and Witwatersrand-type settings (e.g., Frimmel 2008). Although Au ores are mainly hosted in the crust and intermediate to felsic magmas frequently play a direct or indirect role in their formation (e.g., Hronsky et al. 2012; Groves et al. 2020), an increasing number of studies have highlighted the apparent connection between these deposits and the presence of an underlying noble metal-enriched subcontinental lithospheric mantle (e.g., Tassara et al. 2017; Holwell et al. 2019; Wang et al. 2020; Schettino et al. 2022). Similar conclusions have been drawn for PGE mineralizations (e.g., Alard et al. 2011; Akizawa et al. 2017).

Post-tectonic and/or post-orogenic mafic melts generated from a metasomatized subcontinental lithospheric mantle may show a spatial, time or genetic relationship with gold mineral deposits (e.g., orogenic and intrusion-related; Müller and Groves 2019). Such magmatic intrusive bodies are often volatile-rich silica undersaturated alkaline rocks of potassic or shoshonitic geochemical affinity (e.g., lamprophyres). When large enough, they may be relevant in ore genesis because they can promote hydrothermal circulation and crustal partial melting (e.g., Heidari et al. 2015). A similar connection between alkaline-shoshonitic mafic magmas and PGE deposits is rare, although a few examples have been documented (Barnes et al. 2008; Graham et al. 2017). In any case, the noble metal abundance in mantle-derived magmas is of key relevance when trying to establish the PGE and gold mineralizing potential of the lithospheric mantle under cratonic areas (e.g., Wang et al. 2020; Choi et al. 2020).

The Iberian Massif, which is part of the collisional European Variscan Belt, is characterized by the presence of abundant late-tectonic intrusion-related and orogenic gold deposits (e.g., Romer and Kroner 2018), and scarce PGE mineralizations (e.g., Piña et al. 2008) (Fig. 1a). Moreover, this terrane was crosscut by post-collisional mantle-derived magmas, which intruded mostly during the subsequent late-tectonic to post-tectonic extensional stages. These features make this terrane a suitable scenario for the study of the relationship between upper mantle geochemistry and noble metals mineral deposit generation in the upper crust.

Fig. 1
figure 1

a Zonation of the Iberian Massif and location of the main Au and PGE mineral deposits (modified after Romer and Kroner 2018). Gold deposits are represented in yellow: as stars, a wide band (IPB: Iberian Pyrite Belt) and a rectangle (the SE Volcanic Field). b: Schematic geological map of the Spanish Central System showing the Permian post-tectonic and post-orogenic dykes of shoshonitic microgabbros and alkaline lamprophyres and diabases. Sample location is shown as white circles together with the numbers of samples with PGE–Au whole-rock analyses

Full size image

The Spanish Central System (SCS) mountain range, located in central Iberia, offers one of the best examples of post-collisional magmatism of variable geochemical affinity in the Iberian Massif, including Permian mafic post-collisional shoshonitic and post-orogenic lamprophyric dykes derived from a heterogeneously metasomatized mantle (Orejana et al. 2008, 2020) (Fig. 1b). The nature of sulphides and the whole-rock PGE-Au contents in these magmas are still unknown. Our study is focused on the noble metal composition of such intrusions, which can help determine the mineralizing potential of the upper mantle under central Spain with respect to PGE and Au and allows proposing a tentative model for noble metal mantle enrichment in this region. The petrography and major element contents of magmatic sulphides in these alkaline rocks are also described, as these minerals most probably control the PGE distribution (e.g., Piña et al. 2013) and therefore their genesis is key to the discussion on regional primary sources of base and precious metal ores (e.g., Naldrett 2004 and references therein).

Geological setting

The Iberian Massif has been subdivided in several zones according to their stratigraphic, structural and petrologic characteristics (e.g., Martínez–Catalán 2012), with the Central Iberian Zone (CIZ) representing the innermost region (Fig. 1a). The CIZ is mainly composed of Neoproterozoic to Early Paleozoic metasedimentary rocks, Cambrian–Ordovician orthogneisses and Variscan felsic intrusions. These latter granitic bodies outcrop abundantly in the SCS mountain range, where a variety of mafic magmatic events have also been recorded, spanning from late Variscan to post orogenic rifting stages (e.g., Orejana et al. 2020; Villaseca et al. 2022). The Variscan mafic magmatism is restricted to small gabbroic–dioritic intrusions and dykes of calc-alkaline affinity (Orejana et al. 2009, 2020; Villaseca et al. 2022), coeval with the abundant felsic intrusions in the range ~ 316–294 Ma (e.g., Dias et al. 2002; Orejana et al. 2020; Bea et al. 2021 and references therein). This is followed by a short period of magmatic quiescence and the subsequent intrusion (~ 285 Ma; Orejana et al. 2020) of monzonitic microgabbroic dykes of shoshonitic affinity (Huertas and Villaseca 1994; Villaseca et al. 2004), which are restricted to a specific area within the SCS (Fig. 1b) and still can be related to the Variscan cycle. Both these and the calc-alkaline dykes share geochemical features characteristic of post-collisional magmatism, such as high LILE contents and Th/Yb ratios, low Ce/Pb values and negative Nb–Ta anomalies (e.g., Villaseca et al. 2022). Post-orogenic mafic–ultramafic alkaline dykes also crosscut the SCS mainly with a N–S strike (Fig. 1b). They have been subdivided into diabases, camptonitic lamprophyres and monzosyenitic porphyries based on petrographic and geochemical data (Orejana et al. 2008). Geochronological studies have yielded Permian intrusion ages in the range 274–264 Ma (Orejana et al. 2020 and references therein). The intrusion of the large gabbroic Messejana–Plasencia tholeiitic dyke (Fig. 1b), dated at 203 ± 2 Ma (Ar–Ar in biotite; Dunn et al. 1998), represents the last magmatic event recorded in the SCS and is linked to the opening of the Atlantic Ocean.

The geochemistry of the above SCS mantle-derived rocks reveals the involvement of enriched mantle sources of different characteristics during and after the Variscan continental collision (Orejana et al. 2008, 2009, 2020; Villaseca et al. 2022). Incompatible trace element contents and isotopic ratios (Sr–Nd–Hf) point to melting of a lithospheric mantle with a continental imprint in the case of the late-orogenic magmatism, which has also been identified in SCS pre-Variscan mafic rocks (Orejana et al. 2017). The introduction of such crustal component has been related either to oceanic subduction associated to the Cadomian continental margin or to crustal recycling during the Variscan collision (or both) (e.g., Orejana et al. 2020). On the contrary, the post-orogenic Permian alkaline dykes were extracted from deeper mantle sources metasomatized by melts/fluids ascending from the asthenosphere, which yield high LILE, REE and HFSE contents (and positive Nb–Ta anomalies). This mantle, which likely represents a transitional level between the deeper lithosphere and the asthenosphere, has proved to be somewhat heterogeneous, as reflected by the ocurrence of slightly different alkaline rock types (diabases and lamprophyres–monzosyenites) (Orejana et al. 2020).

The CIZ holds relevant mineral deposits which have been recognized and exploited since before Roman times, including widely distributed gold mineralizations (Fig. 1a). The CIZ gold deposits and those from other zones of the Iberian Massif have been classified in three main types (e.g., Romer and Kroner 2018, and references therein): 1) orogenic, 2) intrusion-related and 3) stratabound mineralizations. The orogenic gold deposits are hosted in Variscan shear-zones and were formed due to hydrothermal circulation during a subsequent reactivation of these structures. The mineralizations appear mainly in the form of quartz veins and are included in episyenites. The intrusion-related gold is spatially and genetically associated to small Variscan post-tectonic stocks of felsic to intermediate composition (granodiorites to monzodiorites), which promote intra- and extra-plutonic hydrothermal alteration and sulphide-Au precipitation (Romer and Kroner 2018). Stratabound gold appears either as paleo-placers or related to metal redistribution during the Variscan metamorphism of Cambro–Ordovician black shales and sandstones, giving rise to disseminated mineralizations and lithologically-controlled quartz veins. The proposed sources of these gold mineralizations are the massive sulphide deposits of SW Iberia (Iberian Pyrite Belt; South Portuguese Zone) and Cambrian to Ordovician metasedimentary layers deposited on the Cadomian basement (Romer and Kroner 2018).

PGE ore deposits are extremely rare in the Iberian Massif. Only one exceptional case has been documented: the Aguablanca Ni–Cu–PGE sulphide deposit (Ossa-Morena Zone; SW Iberia) (Fig. 1a). This mineralization is associated to an Early Carboniferous mafic stock which is part of the calc-alkaline Santa Olalla Igneous Complex (e.g., Piña et al. 2008). Apart from its uniqueness in SW Europe, other uncommon features of this deposit are its placement in a subvertical magmatic breccia and its association with a convergent plate margin setting. The mineralization is interpreted as derived from a sulphide melt segregated from an evolving underlying magma chamber.

Analytical methods

Ten samples representative of the different types of mafic dykes in the study area were selected for petrographic analysis and whole-rock PGE and Au determinations. These samples have sizes in the range 10–15 cm in diameter and include: 6 alkaline lamprophyres, 2 alkaline diabases and 2 shoshonitic microgabbros. Polished thin sections and rock powders were prepared at the Mineralogy and Petrology Department Sample Preparation Facility (Complutense University of Madrid, Spain). The major element composition of sulphides has been obtained with a JEOL Superprobe JXA 8900-M electron microprobe (Centro Nacional de Microscopía Electrónica “Luis Bru”, Complutense University of Madrid), equipped with four wavelength-dispersive mode spectrometers. The acceleration voltage was 20 kV and the beam current 50 nA. Counting times were 20 s on the peak and 10 s on each background position. The beam diameter ranged from 0 to 5 μm. Comparison with natural and synthetic reference materials (Galena, GaAs, CoNiCr alloy, Fe, Sb, Cd, Zn, Mo, Ni, Bi and Cu) allowed the determination of the absolute abundance of the following elements: Fe, S, Co, Ni, Mo, Cu, Zn, Cd, Pb, Bi, As and Sb. Detection limits for the analysed elements fall in the range 0.013–0.078 wt% and consider the average value plus 6 times the standard deviation. The ZAF program was used for data correction.

Whole-rock PGE and Au analyses were performed at the Geoscience Laboratories of the Ontario Geological Survey (Sudbury, Canada). The conventional nickel sulphide (NiS) fire-assay pre-concentration technique and the tellurium co-precipitation were employed before analyses by ICP-MS. Replicate analyses of the standards TDB-1, OREAS 681 y OREAS 13b, together with an in-house standard, were undertaken in order to assess analytical precision and accuracy. Detection limits for the elements analysed are: 0.025 ppb (Ir), 0.08 ppb (Ru), 0.05 ppb (Rh), 0.19 ppb (Pt), 0.2 ppb (Pd), 0.6 ppb (Au). Analytical error in these elements is in the range 1.5–14.3%.

Petrography

Alkaline lamprophyres and diabases

The lamprophyres and diabases are porphyritic (mostly 5–15 vol%), with the former displaying the higher phenocrysts abundance (occasionally up to ~ 40 vol%). Lamprophyres have been classified as camptonites (Orejana et al. 2008) and have only mafic phenocrysts such as clinopyroxene (diopside-augite), amphibole (kaersutite), phlogopite and oxides (ulvöspinel, Ti-magnetite, ilmenite), of variable size (from a few millimetres to several centimetres) (Fig. 2a, b). H2O-bearing minerals (amphibole and phlogopite–biotite) are restricted to the groundmass in the diabases, but these dykes include plagioclase phenocrysts which can be cm-sized (Fig. 2c). Olivine and chromite phenocrysts have also been recognised in accessory amounts, both in lamprophyres and diabases, although olivine is usually transformed to Mg-rich chlorite. Phenocrysts may be transformed to secondary phases (mainly chlorite) and partially to totally corroded, and they usually display complex zoning patterns (normal and reverse) (Orejana et al. 2007). The groundmass can be microcrystalline to cryptocrystalline and displays a similar mineral composition to that of the phenocrysts, as well as plagioclase in the case of the diabases. A variable typology of ocellar structures (feldspar-, chlorite- or carbonate-rich), is present both in the lamprophyres and diabases. Apart from chlorite, other low-T phases which can be found in the groundmass are sericite, calcite, zeolites (analcite) and epidote. These minerals are sometimes secondary in origin and associated to transformation of early crystallising minerals, but they can also represent late-stage crystallization from the lamprophyric-diabasic magma, once the exsolution of low-T volatile-rich fluids is reached (Rock 1991; and references therein).

Fig. 2
figure 2

Plane polarized (a, b, c, e)  and cross polarized (d, f) light images showing the main petrographic textures of the studied mafic dykes. a: alkaline lamprophyre (103,473); b: alkaline lamprophyre (103,818); c: diabase (81,843); d: shoshonitic microgabbro (112,907); e–f: altered ultramafic xenoliths in diabase 116,450. The white bar represents 500 \(\mu\)m. Amp: amphibole; Cpx: clinopyroxene; Pl: plagioclase

Full size image

Two of the analysed samples stands out for several specific petrographic features: (1) lamprophyre 103,818 for containing up to ~ 40 vol% phenocrysts (Fig. 2b) and a variable modal amount of opaque phenocrysts, such as ulvöspinel, Ti-magnetite and sulphides (pyrite, chalcopyrite and pyrrhotite) (up to 2 vol%), and (2) diabase 116,450 for the inclusion of small fragments (1–7 mm) of highly altered pyrite-bearing ultramafic xenoliths (Fig. 2e, f), which resemble other SCS enclaves (mainly clinopyroxenites and hornblendites) interpreted as deep-seated mafic cumulates (Orejana et al. 2006; Orejana and Villaseca 2008). The main alteration products of the mafic phases are Mg-rich chlorite and talc. These enclaves display a modal composition equivalent to the phenocryst assemblage of the host alkaline rocks, including scarce olivine pseudomorphs. Xenocrysts (clinopyroxene, quartz, garnet) and other types of xenoliths (mainly lower crust granulites) can also be found in these alkaline dykes.

Shoshonitic microgabbros

These dykes display porphyritic texture, with clinopyroxene (diopside-augite) and plagioclase (labradorite) phenocrysts up to ~ 0.5 cm in size, and a fine-grained holocrystalline groundmass including mainly these same phases, as well as biotite–phlogopite (Fig. 2d). The phenocrysts are euhedral to subhedral and variably altered to chlorite (cpx) and sericite (pl). The most common accessory minerals are ilmenite, apatite, magnetite, sulphides, zircon and titanite. The higher degree of alteration in some samples leads to formation of pseudomorphs with chlorite and calcite, which can also be present in the groundmass. Nonetheless, calcite can also be present as a primary magmatic mineral within ocelli or as an irregular interstitial phase. This suite of dykes also includes more evolved terms which may show quartz and K-feldspar, and amphibole (kaersutite) within the groundmass. Only the more mafic dykes were selected for the analytical work.

Sulphides petrography and chemistry

The sulphides recognised in the studied samples are pyrite, chalcopyrite, pyrrhotite and galena, pyrite being much more common than the rest. These sulphides are more abundant in the alkaline lamprophyres and diabases (which lack galena), whereas the shoshonitic microgabbros only display very small grains of chalcopyrite and galena, restricted to the groundmass. Pyrite crystals display a heterogeneous morphology (euhedral, subhedral and anhedral), sometimes forming clusters, and a size in the range 5–800 \(\mu\)m. According to petrography, pyrite can be classified in six groups: 1) within altered phenocrysts and cumulate xenoliths (Fig. 3a–c), 2) as inclusions within other phenocrysts (clinopyroxene, amphibole, apatite) (Fig. 3d–e), 3) as microphenocrysts (Fig. 3f), 4) in the groundmass (Fig. 3g), 5) surrounded by a Ti-magnetite rim (Fig. 3h), and 6) inside late-stage globular vesicles (Fig. 3i). Optical and back scattered electron (BSE) imaging do not reveal any zoning in pyrites. Chalcopyrite appears mainly associated to pyrite in phenocrysts or in altered enclaves, always representing low proportions (Figs. 3a, f), and pyrrhotite is rare and only found as small globules (5–50 \(\mu\)m) included in phenocrysts (Fig. 3d).

Fig. 3
figure 3

Back scattered electron images of sulphides from the SCS mafic dykes. a–c: altered phenocrysts and mafic enclaves (samples 104,395, 116,450 and 103,673); d, e: inclusions in phenocrysts (samples 103,818 and 104,529); f: phenocrysts (sample 103657A); g: groundmass (sample 103,811); h: pyrite within Ti-magnetite (sample 103,817); i: low-T vesicle (sample 103657A). The number indicated in the scale bar at the bottom-left of each image represent size in microns. Py: pyrite, Ccp: chalcopyrite, Po: pyrrhotite, Ap: apatite, Cpx: clinopyroxene, Ilm: ilmenite, Mag: magnetite, Chl: chlorite, Tlc: talc

Full size image

Electron microprobe data of pyrite, chalcopyrite and pyrrhotite have been included in Table 1. Pyrite is characterised by a moderate heterogeneity in Fe (44.1–46.9 wt%) and S (51.9–54.7 wt%), with S/Fe atomic ratios in the range of 2.003–2.054. These values are always higher than the ideal stoichiometric ratio of 2, which implies that Fe is more readily substituted by base metals than S by As, Sb or other elements. Deviations from the mentioned stoichiometry are restricted to 0.14–2.69%, with the higher values being related to higher Ni contents.

Table 1 Representative major and minor element composition of the main sulphides from the SCS mafic dykes
Full size table

S and Fe concentrations are relatively homogeneous in the six pyrite groups (Fig. 4a, b). Pyrite clearly related to late-stage evolution (groundmass, pyrites in Ti-magnetite and vesicles) shows lower Ni values (sometimes below detection limit) when compared to pyrite inclusions and pyrite associated with altered xenoliths, which can reach values up to 1.16 wt% (Fig. 4c). Pyrite phenocrysts display Ni values similar to those of groundmass pyrites. Most base metal concentrations are close to or below detection limit (Table 1), and their homogeneous contents do not allow establishing significant differences between pyrite groups. It is remarkable that Cu and As show low values (Table 1), almost always below or very near the analytical detection limit.

Fig. 4
figure 4

Box plot displaying the Fe, S and Ni composition (wt%) of pyrites from the SCS alkaline dykes. The central box is the middle 50% of the data. The line and circle in each box represent the median and mean value, respectively. The key for each pyrite group is indicated in diagram a; inc: pyrite inclusions within fresh primary minerals (amphibole, clinopyroxene, apatite), phen: phenocrysts, grm: groundmass pyrite, Ti-mg: pyrite with rim of Ti-magnetite, alt: pyrite in altered enclaves or phenocrysts, ves: pyrite within vesicles

Full size image

Chalcopyrite, which is always associated to pyrite from altered enclaves and phenocrysts, shows a homogeneous composition, with S = 34.3–34.6 wt%, Fe = 29.7–30.4 wt% and Cu = 33.6–33.8 wt% (Table 1). The pyrrhotite inclusions do not show a significant compositional variation either (S = 39.28–39.89 wt%; Fe = 58.57–58.77 wt%), and their Ni concentrations only reach minor amounts (Ni = 0.27–0.31 wt%) (Table 1).

Noble and base metal geochemistry

PGE and Au contents determined in the present study are included in Table 2, together with Cu, Co and Ni concentrations previously obtained in the same samples (Orejana et al. 2008, 2020). Overall, the Ir, Rh and Ru contents are very low. Ir and Rh values are below detection limit in 8 out of 10 samples, ranging from 0.11 to 0.56 ppb and from 0.1 to 0.16 ppb, respectively, in lamprophyre 103,818 and diabase 116,450. Ru contents are also low, although above detection limit in all samples (0.08–0.93 ppb). The Pt, Pd and Au contents are significantly higher and more variable than the rest of PGE: Pt = 0.25–4.42 ppb; Pd =  < 0.27–5.62 ppb and Au = 2–29 ppb. Ni, Co and Cu contents only reach moderate concentrations (Ni = 27–320 ppm; Co = 29–58 ppm; Cu = 21.6–70 ppm; Table 2).

Table 2 Noble and base metal composition of the SCS mafic dykes
Full size table

When plotted against Mg# [Mg/(Mg + Fe2+)] (values taken from Orejana et al. 2008, 2020), an apparent positive correlation can be identified for Pd and Pt, as well as most base metals (Cr, Cu and Ni), whereas Au and IPGE do not show a clear tendency (Fig. 5; Ir and Rh have not been plotted due to the scarcity of data). It is noteworthy that the high Pd–Pt concentrations of sample 103,818 represent an anomaly with respect to the rest of lamprophyres, and the high base and noble metal values shown by sample 116,450, also fall out of any evolutionary trend. Considering the special petrographic features described in these two samples, this anomalous chemical composition is likely associated to higher relative modal proportions of sulphide minerals. Due to the distinct composition of these samples, not representative of primary melts, they have not been considered to further discuss the nature of the mantle sources and the metasomatic events.

Fig. 5
figure 5

Noble and base metal composition of the SCS mafic dykes plotted against Mg#. Samples 116,450 and 103,818 have been highlighted where they display a distinct composition with respect to other samples (see text for further explanation)

Full size image

The relative abundance of PGE and Au in the three dyke groups with respect to the Primitive Mantle values (McDonough and Sun 1995) is broadly similar and characterized by a marked fractionation between the Ir-group PGE (IPGE; Ir and Ru) and the Pd-group PGE (PPGE; Rh, Pt and Pd), the latter displaying much higher values (Fig. 6). The analytical detection limit values of Ir (0.025 ppb) and Rh (0.05 ppb) have been used to draw the thick dashed grey line in Fig. 6, which can be considered as an approximation to the likely maximum abundances of Ir and Rh in samples lacking Ir and Rh concentrations. The exceptions to the above general pattern are lamprophyre 103,818, which stands out due to its higher degree of enrichment in Pt and Pd, and diabase 116,450, which shows a flat Primitive Mantle-normalized pattern for most PGE (Fig. 6). Au shows a variable degree of enrichment in the normalized spectrum, giving rise to marked positive anomalies and abundances of 2–29 × Primitive Mantle. Co also shows a moderate positive anomaly, although its relative abundances are below 1 (Fig. 6). Cu abundance is lower than that of Au and its concentration is similar to that of the Primitive Mantle, except for sample 116,450 (diabase) which displays slightly higher values.

Fig. 6
figure 6

Primitive Mantle-normalized abundances of noble and base metals in the SCS mafic dykes. Primitive Mantle values after McDonough and Sun (1995). Au–PGE composition of kimberlites from the Bastar craton (Chalapathi Rao et al. 2014) and calk-alkaline lamprophyres from the Yilgarn craton (Choi et al. 2020) have been plotted for comparison. Question marks in Ir and Rh indicate that the composition of these elements is below detection limit: the dashed line has been drawn considering these values and represents an estimated maximum composition

Full size image

The above described low PGE contents contrast with the abundances shown by mafic rocks (kimberlites) from the Bastar craton (India), extracted from a PGE-enriched subcontinental mantle source (Chalapathi Rao et al. 2014) (Fig. 6). Other alkaline and calk-alkaline magmas derived from a similarly metasomatized mantle source also display slightly higher PGE values (Choi et al. 2020). These latter authors additionally document the occurrence of a variable, positive Au anomaly, which nevertheless does not reach the high abundances of our samples (Fig. 6).

Discussion

Nature of sulphides in the SCS mafic dykes

Sulphides in the SCS alkaline lamprophyres and diabases, although generally accessory, may reach significant modal amounts. Several features point to a magmatic origin: presence of pyrrhotite and pyrite inclusions within fresh primary phenocrysts (clinopyroxene, amphibole and apatite) (Fig. 3d, e), occurrence of disseminated pyrite (either as phenocrysts or in the groundmass) (Fig. 3f–g), coexistence with magmatic Ti-magnetite (Fig. 3h) and low values for most trace metals (e.g., Co, Cu, Zn, As, Sb, Pb and Bi; Table 1). Moreover, they do not show any textural evidence of alteration by hydrothermal fluids.

Models on the evolution of magmatic sulphides are based on experimental studies of ternary or quaternary systems involving Ni, Fe, Cu and S (Naldrett 2004, and references therein). The absence of Ni-rich phases, such as pentlandite, in the studied SCS mafic dykes likely suggests that Ni could have been exhausted almost completely before sulphide exsolution, and the system Cu–Fe–S would be the most appropriate reference model. The experiments and observations of natural assemblages predict the formation of a Cu-rich (Ni-poor) intermediate solid solution (iss) from, or in conjunction with, a Fe-rich monosulphide solid solution (mss) (Jensen 1942; Kullerud et al. 1969; Naldrett 2004). The presence of pyrrhotite and pyrite inclusions in mafic phenocrysts implies that exsolution of such sulphide melt is not a late-stage process but happened during fractionation at depth. The scarce pyrrhotite globules found within mafic phenocrysts likely formed by exsolution from mss. On the other hand, pyrite inclusions and phenocrysts could derive from mss or iss (Naldrett et al. 1967; Naldrett 2004; Dare et al. 2011). The predominance of pyrite over pyrrhotite in the studied rocks is not a usual feature of sulphide melts segregated from most mafic magmas, but it is not rare in ultramafic–mafic volatile-rich alkaline rocks (Graham et al. 2017) akin to the SCS diabases and lamprophyres. Thus, pyrite inclusions and phenocrysts could be considered primary phases, and their preferred equilibration over pyrrhotite is likely related to relatively high ƒS2 in the liquid. The broadly low Cu and Ni concentrations in pyrite and pyrrhotite (Table 1; Fig. 4c) (and scarcity of chalcopyrite) also suggest that this sulphide saturation was a relatively early process, as Ni is compatible during silicate liquid fractionation (Koshlyakova et al. 2022 and references therein).

The presence of pyrite within mafic–ultramafic cumulate xenoliths could in principle favour a primary origin during crystal settling. However, pyrite in these enclaves is only found in altered zones, mostly in their boundaries or where magma infiltration is apparent (Figs. 2e, f and 3b). Such feature, as well as the absence of reaction textures in the pyrite itself, calls for a second process of sulphur exsolution occurring during xenolith alteration. Mineral assemblages including chlorite, talc and pyrite, resulting from reaction of mafic–ultramafic rocks and hydrous fluids, have been described elsewhere (e.g., Smith 1995). The alkaline magma ascent, and its associated pressure decrease, could have provoked volatile exsolution (H2O and CO2), which promoted vesiculation and alteration of suspended solids (xenoliths and phenocrysts) (e.g., Rock 1991). The consumption of H2O and CO2, resulting in chlorite and talc formation during the xenoliths transformation, would reduce the oxygen fugacity, increase ƒS2 and facilitate sulphide crystallization. Such reactions, leading to pyrite stabilization, have already been described in mafic rocks (e.g., Kanitpanyacharoen and Boudrau 2013; Piña et al. 2013). The late-stage volatile-rich (hydrothermal) fluids responsible for this autometasomatic transformation would also explain the formation of pyrite in the groundmass, included in vesicles and associated to altered mafic phenocrysts. Likewise, a late-stage process is consistent with pyrite crystals included in Ti-magnetite from sample 103,817 (Fig. 3h). Ti-magnetite is a late crystallizing mineral and its formation may cause redox changes in the melt (a decrease in oxygen fugacity), which in turn may trigger S saturation (e.g., Jenner et al. 2010). Chalcopyrite, which is scarce and associated to pyrite phenocrysts and pyrite formed during late-stage alteration, likely resulted from iss breaking down during cooling, as both chalcopyrite and pyrite can coexist in equilibrium at temperatures above 150 ºC (Naldrett 2004).

The two sulphur saturation processes mentioned above have not led to noticeable differences in pyrite composition, except for Ni. Pyrites from the groundmass, vesicles and inclusions in Ti-magnetite display undetectable or low Ni concentrations, when compared to pyrite inclusions in mafic minerals (Fig. 4c). The relatively high Ni contents of pyrite in altered mafic minerals (Fig. 4c) could be explained by a variable incorporation of these elements from the transformed mafic phases.

Evaluation of post-magmatic processes on PGE–Au geochemistry

There is evidence of post-magmatic transformation of the SCS shoshonitic and alkaline dykes, recorded in the form of low-T phases (chlorite, sericite, calcite, zeolites, epidote) which can be found mainly in the groundmass, in ocellar structures (vesicles) and within altered (pseudomorphosed) phenocrysts. Although circulation of hydrothermal fluids is commonly associated with shoshonitic and lamprophyric mafic magmatism (e.g., Müller and Groves 2019), much of the alteration found in these rocks is caused by metasomatism induced by the volatile-rich melt fraction remaining in the last stages of crystallization, typical of lamprophyric and related rocks (Rock 1991; Mitchell 1995). Whatever the origin, the question is to what extent such alteration processes may have affected the PGE and Au concentrations.

Wang et al. (2008) have suggested that hydrothermal fluids circulation may be responsible for PGE mobilization (and reprecipitation) at low temperatures. Taking into account that secondary minerals abundance in the studied rocks is broadly lower than 5 vol% (only some shoshonitic dykes show alteration up to 10–15%), any modification of their PGE composition due to such hydrothermal process seems to be negligible. In any case, the role played by alteration in the noble metals content can be tested using the Alteration Index (AI) of Ishikawa et al. (1976) [[(K2O + MgO)/(Na2O + K2O + CaO + MgO)] × 100], and its potential correlation with PGE–Au. As shown in Fig. 7, there is no clear correlation between AI and Ru, Pt, Pd or Au, suggesting that noble metals were mostly immobile during alteration. This conclusion is reinforced by the positive slope of the Primitive Mantle-normalised pattern from Pt to Pd (Fig. 6), which is consistent with a magmatic context (Gan and Huang 2017). Flat patterns or those displaying negative slope could account for a relevant role of alteration, as Pd is the most mobile PGE (Barnes et al. 1985). Accordingly, the PGE–Au contents of the studied shoshonitic and alkaline dykes can be considered reliable indicators of the noble metal geochemical signature of the mantle sources underlying the Spanish Central System at the time of emplacement.

Fig. 7
figure 7

Nobel metal (Au, Pt, Pd and Ru) concentrations vs. the Alteration Index (AI) of Ishikawa et al. (1976) [[(K2O + MgO)/(Na2O + K2O + CaO + MgO)] × 100] in the studied SCS mafic dykes

Full size image

PGE abundance and behaviour

Several processes have been suggested to explain an IPGE–PPGE fractionation in mafic magmas such as that observed in our samples: fractionation of Ir–Ru alloys (Maier and Barnes 1999) or other early crystallizing minerals (e.g., Barnes and Picard 1993), low degrees of mantle partial melting (e.g., Lorand et al. 2008) and sulphide segregation (e.g., Amosse et al. 1990). The compatible behaviour of Os, Ir and Ru during mafic magma fractionation (Barnes et al. 1985) can give rise to small Os–Ir–Ru sulphide and alloy inclusions in high-temperature minerals, such as chromite and olivine (e.g., Zhou 1994). Ru exhibits a more compatible behaviour with Mg-rich phases than the rest of PGE and, therefore, it shows higher relative abundances in olivine (Dale et al. 2009). However, the lack of a negative Ru anomaly in the Primitive Mantle-normalized spectrum and of a positive correlation of Ru and Mg# of the studied rocks (Figs. 5d and 6), suggest that IPGE fractionation during deep-seated crystal fractionation did not played a key role in the PGE behaviour.

Low degree partial melting of sulphide-bearing mantle sources could also account for PGE fractionation, as these elements are mainly concentrated in Fe–Ni monosulphide phases and intergranular sulphides (e.g., pentlandite, chalcopyrite, isocubanite) (Lorand et al. 2008). This second group of sulphides is relatively enriched in PPGE with respect to IPGE and melts preferentially for low degrees of partial melting with respect to monosulphides, which are retained in the residual mantle (Ballhaus et al. 2006). Therefore, the low degrees of partial melting proposed for the studied rocks (1–2%; Villaseca et al. 2022) are in accordance with this possibility.

Alternatively, sulphur saturation in magmas and subsequent sulphide segregation could also account for the IPGE–PPGE fractionation due to the different partition coefficients of these elements (e.g., Barnes et al. 1985). Considering that S and Cu show a similar behaviour during magmatic differentiation, Cu can be considered a proxy for sulphur. Several geochemical fingerprints, such as a negative correlation of Cu/Pd ratio and Pd concentration and a positive correlation of the Pd/Cr and Pt/Y ratios (Gan and Huang 2017), could be related to an early sulphide saturation and segregation, which would lead to a progressive decrease of Pt–Pd with respect to Cr–Y–Cu. This correspondence can be observed in the SCS mafic dykes (Fig. 8a, b) and points to an early crystallization of sulphides, which could also explain the positive correlation of Mg# with respect to Cu, Pt and Pd (Fig. 5). The fact that Cr and Ni also decrease towards lower Mg#, suggests that both mafic silicates fractionation and sulphide segregation justify the geochemical variability observed in the SCS mafic dykes. The presence of sulphides as inclusions within mafic phenocrysts in the alkaline lamprophyres and diabases supports this possibility.

Fig. 8
figure 8

Cu/Pd ratio vs. Pd (a), Pt/Y ratio vs. Pd/Cr ratio (b) and Cu vs. Ni and Pt (c, d) in the studied SCS mafic dykes

Full size image

The anomalous composition of diabase 116,450, which shows the highest IPGE values (Figs. 5 and 6), is likely related to the inclusion of highly altered pyrite-bearing ultramafic xenoliths. These enclaves are deep-seated cumulate xenoliths genetically related with the SCS lamprophyric and diabasic melts (Orejana et al. 2006; Orejana and Villaseca 2008). The reaction of these cumulates with the volatile-rich magma leads to formation of relatively abundant pyrite (as explained above), which can incorporate important IPGE and Rh amounts substituting Fe in its structure (Duran et al. 2015; Piña et al. 2013). The higher Ni, Cu, Co and Au contents of this sample compared to the rest of analysed rocks (Fig. 5a, f–h, Fig. 8c–d and Table 2) reinforces this idea and highlights the control exerted by sulphides on all chalcophile elements during magma fractionation. Similarly, the excessive Pt–Pd enrichment in lamprophyre 103,818 (Figs. 5b, c and 6) is probably associated with its petrographic characteristics. The abundance of phenocrysts in this sample (~ 35–40 vol.%), including pyrite, implies certain degree of crystal accumulation. The relative PPGE enrichment of the primary lamprophyric melt, together with the strongly compatible behaviour of Pt–Pd with late-stage crystallizing sulphides, can explain the anomalously high concentrations of these elements in this sample.

In summary, both early sulphur segregation and low degrees of mantle partial melting can justify the general pattern of PGE fractionation and the low PGE contents observed in the studied samples.

Gold abundance and behaviour

Au, Pt and Pd are chalcophile during magmatic processes. However, the gold enrichment shown by the SCS alkaline and shoshonitic dykes (concentrations 2–29 times higher than that of the Primitive Mantle) is much higher with respect to that of Pt and Pd (Fig. 6). These data suggest that one or more enrichment processes have occurred in the mantle source and that these processes have been more efficient for Au than for PGE.

The metasomatized lithospheric mantle is generally enriched in gold with respect to the primordial mantle, and the source of this gold is usually associated with fluids or melts derived from the descending plate in a subduction context (e.g., Groves et al. 2020), or to alkaline melts ascending from a mantle plume (e.g., Webber et al. 2012; Tassara et al. 2017). The SCS post-tectonic and post-orogenic shoshonitic and alkaline mafic dykes display highly incompatible trace element concentrations indicative of mantle metasomatism by infiltration of agents of contrasting nature: deep asthenospheric magmas for the alkaline lamprophyre sources and a subduction-derived metasomatic agent for the shoshonitic microgabbros (Orejana et al. 2020; Villaseca et al. 2022). Each of these two possibilities would produce different geochemical fingerprints in the partial melts. High La/Sm ratios imply the existence of a broad trace element enrichment, whereas high Th/Yb ratios support subduction and recycling of continental components in the mantle; on the other hand, high Nb/Yb ratios are consistent with metasomatism caused by magmas coming from the deep asthenospheric mantle, without evident crustal influence (e.g., Wilson 1989). Moreover, a Co positive anomaly (with respect to Ni and Ir) could be associated to subduction-related metasomatism (Holwell et al. 2019). An overall positive correlation exists in the analysed samples between Au and La/Sm (Fig. 9a). However, the alkaline lamprophyres and the shoshonitic microgabbros exhibit contrasting features with respect to Th/Yb and Nb/Yb ratios, with the lamprophyres displaying lower Th/Yb and higher Nb/Yb ratios with respect to the shoshonitic dykes (Fig. 9b, c). Moreover, the Th/Yb ratios display a broad positive correlation with respect to Au contents, whereas no clear connection seems to exist between Au and the Nb/Yb values.

Fig. 9
figure 9

La/Sm, Th/Yb and Nb/Yb ratios vs. Au contents in the studied SCS mafic dykes. The dashed line represents the broad correlation between Au concentrations and La/Sm and Th/Yb ratios. La/Sm, Th/Yb and Nb/Yb ratios have been calculated considering the data of Orejana et al. (2008, 2020)

Full size image

The above data suggest that Au enrichment in the subcontinental lithospheric mantle under central Spain was mainly controlled by subduction processes, which are more efficient regarding Au enrichment than infiltration and reaction with alkaline melts coming from the convecting mantle. The fact that all samples yield Au and Co positive anomalies, including the alkaline lamprophyres and diabases (Fig. 6), might indicate that the metasomatism related to entrainment of deep magmas overprinted an older subduction-related metasomatic event (Fig. 10). The high (Au/Cu)N ratios (1.4–17.3; Table 2), reflected in the Au peak shown in Fig. 6, mean a higher addition of Au relative to Cu, a feature which could not be ascribed only to entrainment of sulphide-saturated melts. Although very high (Au/Cu)N values (~ 80–375) have been related to a secondary hydrothermal alteration of the mafic rock in the vicinity of Au mineralizations (Müller and Groves 2019, and references therein), this possibility is unlikely in the case of the SCS shoshonitic and alkaline mafic dykes, as they are fresh, yield much lower (Au/Cu)N ratios and are far from Au mineralizations. Alternatively, stronger Au enrichment in the mantle, relative to Cu, may be expected due to infiltration of S-rich fluids, as gold is more compatible with this hydrous phase than Cu (Pokrovski et al. 2014). This is in accordance with recent studies that emphasize the presence of restricted mantle domains with high heterogeneity, sometimes associated with variable reactive melt/rock ratios (Wang et al. 2022, and references therein), and suggests that the Au–Cu relative abundances shown by the SCS dykes are a characteristic inherited from their mantle source.

Fig. 10
figure 10

Cartoon illustrating the geodynamic setting of gold enrichment under the Central Iberian Zone and generation of post-tectonic and post-orogenic mafic dykes, either during the Cadomian subduction (A) or late- to post-Variscan stages (B). The possible origin of orogenic and intrusion-related gold deposits in the CIZ is also represented in diagram B

Full size image

Noble metal fertility of the subcontinental lithospheric mantle under central Spain

The results presented above point to the presence of an Au-enriched mantle under central Spain, but the same is doubtful for the PGE. Even though alkaline-carbonatitic melts from the asthenosphere can be responsible for PGE enrichment in the lithospheric mantle (e.g., Alard et al. 2011; Akizawa et al. 2017), these elements always display subchondritic concentrations, most of them 10 to 100 times lower than chondrite (Fig. 6). The study of González-Jiménez et al. (2014) on mantle xenoliths from central Spain describes a Fe–Ni monosulphide solid solution with low PGE concentrations, interpreted as a residual phase after partial melting. No sign of PGE refertilization has been found in these mantle xenoliths. It is thus not surprising that PGE deposits are a rarity in the Iberian Massif, with only one exceptional example found in the Ossa-Morena Zone (the Aguablanca Ni–Cu–(PGE) sulphide deposit), where assimilation of sulphide-rich metasediments seems to have played a relevant role (Piña et al. 2008).

Most recent models highlight that Au endowment in certain metallogenic provinces is related to enrichment by fluids derived from a subducting oceanic plate (e.g., Holwell et al. 2019; Groves et al. 2020), whereas few examples have been explained as linked with continental subduction (Deng et al. 2020). The mantle under central Spain has been likely subjected to several enrichment events, including the Neoproterozoic oceanic subduction under the northern Gondwana margin (the Cadomian continental arc) and recycling of continental components during the Variscan continental collision (Orejana et al. 2020; Villaseca et al. 2022) (Fig. 10). Although both geodynamic contexts could be advocated for Au enrichment under the Central Iberian Zone, it is likely that oceanic subduction might have left a more significant imprint due to its longer lifespan and widespread extension.

Contrary to the PGE, gold deposits are relatively abundant in the Iberian Massif (e.g., Romer and Kroner 2018) (Fig. 1a). These authors summarize the main features of Paleozoic gold in the Variscides and establish three main types of gold deposits in Iberia (orogenic, intrusion-related and stratabound) associated to two auriferous sources: massive sulphide deposits of the Iberian Pyrite Belt and Cambrian–Ordovician siliciclastic sediments deposited on the Cadomian basement. Mobilization of gold contained in the volcanic massive sulphide deposits and in sediments with anomalous Au quantities during the low or medium grade Variscan metamorphism would be the main form of gold transfer and concentration to eventually yield gold deposits in the crust (Romer and Kroner 2018) (Fig. 10b).

However, the Au abundances shown by the mafic dykes analysed in this study highlight that the lithospheric mantle under central Spain should be considered another potential source of gold. It has been noted that potassic and shoshonitic igneous rocks (including alkaline lamprophyres) may coexist spatially and temporally with important orogenic and intrusion-related gold deposits (Rock and Groves 1988), although this association does not always imply the existence of a genetic link (e.g., Müller and Groves 2019). Accordingly, our data can be meaningful only regarding the study of the lithospheric mantle as a potential source of noble metal ores. In this sense, it is remarkable that the studied mafic dykes contain Au concentrations (2–29 ppb) higher than those of equivalent calc-alkaline and ultramafic lamprophyres (Au = 0.4–13.81 ppb) outcropping in areas with relevant gold deposits, such as the Yilgarn craton in Australia (Choi et al. 2020) (Fig. 6). These data indicate the relevance of the subcontinental lithospheric mantle under central Spain as a potential source of gold deposits. This kind of connection has already been proposed in SE Spain, where metasomatic Au-rich base-metal sulphides included in peridotite xenoliths have been considered the main metal source of the Miocene middle- to high-K calc-alkaline volcanism and, eventually, of the associated crustal gold mineralizations (Schettino et al. 2022). Such contribution of the metasomatized lithospheric mantle as a source of gold endowment in the upper crust has also been highlighted in previous studies (e.g., Tassara et al. 2017; Holwell et al. 2019; Wang et al. 2020).

Most gold deposits in the Central Iberian Zone seem to involve Au-bearing Cambrian–Ordovician metasediments. Such sources imply gold provenance from emerged neighbouring regions of the West-Gondwana Orogen (Romer and Kroner 2018) (Fig. 10a). However, a direct relation of the Iberian gold deposits with the underlying mantle is still possible in the case of intrusion-related deposits, as some Variscan stocks are associated to calc-alkaline mantle-derived magmas, including gabbroic–dioritic–monzonitic terms (e.g., Cuesta et al. 1998; Crespo et al. 2000; Martín-Izard et al. 2000) (Fig. 10b).

Conclusions

Pyrite and chalcopyrite are the main sulphides in the SCS post-collisional and post-orogenic mafic dykes and formed due to sulphur saturation in two distinct stages: an early stage characterised by crystal fractionation at depth, and a second and late stage related to magma ascent, CO2 and H2O exsolution, vesiculation, Ti-magnetite crystallization and cumulates alteration.

The PGE contents are generally low and fractionated from very low IPGE to moderate PPGE. These features seem to be associated to very low degrees of partial melting in the mantle sources and sulphide segregation at depth.

All analysed samples display moderate to high gold enrichment, comparable with equivalent mafic rocks extracted from other lithospheric mantle sections under gold-endowed provinces. The geochemistry of the SCS samples favour a subduction-related metasomatic event as the most likely geodynamic context for Au refertilization in the mantle.

The gold abundances shown by the mafic dykes analysed in this study highlight that the lithospheric mantle under central Spain should be considered a potential source of gold, as already pointed out for the mantle under SE Spain.