The Golden Quadrilateral of the Apuseni Mountains (Romania) represents the richest Au(-Cu-Te) porphyry and epithermal district of Europe and the Western Tethyan metallogenic belt. The Au(-Cu-Te) mineralization is associated with Neogene calc-alkaline magmatism along graben structures growing during the late stages of the Alpine-Carpathian orogeny. We use zircon petrochronology to study the time-space distribution, sources, composition, and timescales of the Au(-Cu-Te)-mineralizing magmatism and explore its link to regional tectonics. Our own and published U-Pb zircon ages document ore-forming magmatic activity between ~13.61 and 7.24 Ma. In combination with available paleomagnetic data, the new zircon ages corroborate the hypothesis that the magmatism in the Golden Quadrilateral evolved in a tectonic environment dominated by major (up to 70°) crustal block rotation. Hafnium isotope composition of Neogene zircon (εHf between –2 and 10) supports the predominant origin of the magmas from a heterogeneous lithospheric mantle, which may have been fertilized during an earlier Cretaceous subduction event and possibly by concurrent Miocene subduction. Xenocrystic zircon shows involvement of crustal sources resembling European continental basement. Fertility indicators, including Eu/Eu* and oxygen fugacity based on zircon composition, show no systematic correlation with the mineralizing events and/or age. High-precision (isotope dilution-thermal ionization mass spectrometry) U-Pb zircon geochronology demonstrates that the magmatic systems exposed at district scale evolved over less than ~100 k.y. and that durations of hydrothermal mineralization pulses were even shorter.

Gold and copper in magmatic-hydrothermal deposits are hosted by shallow porphyry intrusions and epithermal vein systems but sourced from much larger cooling and fluid-exsolving magma reservoirs originating in the mantle and evolving from the lower to upper crust (Hedenquist and Lowenstern, 1994; Mungall, 2002; Sillitoe, 2010). Mantle melting and evolution of mineralizing magmas is commonly caused by concurrent subduction in active arc settings but may, in some cases, have been prepared by million-years-older subduction events (Richards, 2009; Pettke et al., 2010). Over the last decade, researchers have used zircon to gain a quantitative understanding of the factors controlling the ore-forming potential of magmas, including their source composition and timescales of geochemical and thermal evolution (e.g., von Quadt et al., 2011; Chelle-Michou et al., 2014, 2017; Loucks, 2014; Buret et al., 2016; Lu et al., 2016; Rezeau et al., 2016; Tapster et al., 2016; Li et al., 2017; Large et al., 2018, 2020, 2021). Unlike major rock-forming minerals, zircon is largely immune to hydrothermal alteration, making its chemistry a reliable proxy of magmatic conditions (e.g., Fu et al., 2009).

Zircon petrochronology combines petrogenetic information of trace element and isotopic composition with U-Pb geochronology based on variably precise dating methods such as in situ laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) and secondary ion mass spectrometry (SIMS) or bulk-grain chemical abrasion-isotope dilution-thermal ionization mass spectrometry (CA-ID-TIMS; e.g., Engi et al., 2017, and references therein; Schaltegger and Davis, 2017; Schoene and Baxter, 2017). When these single-crystal methods are applied to suites of magmatic rocks in clear field relationship to ore deposits, they can provide important constraints on the source(s), as well as cooling and differentiation histories of mineralizing magmas at depth prior to porphyry intrusion and hydrothermal ore deposition (Chelle-Michou et al., 2014; Buret et al., 2016, 2017; Large et al., 2021; Nathwani et al., 2021). Moreover, zircon trace element compositions (EuN/EuN*, Th/U, Yb/Dy, and calibrated proxies of oxygen fugacity) have even been applied to porphyry copper systems to discriminate inherently fertile from common barren magmas (e.g., Lu et al., 2016, 2017; Lee et al., 2017, 2021).

Here, we apply zircon petrochronology to study the Golden Quadrilateral of the Apuseni Mountains (Romania), where the calc-alkaline magmatism during the Neogene produced dozens of porphyry and epithermal deposits constituting the richest Au district of Europe and the Western Tethyan metallogenic belt (>100 Moz of Au; e.g., Udubașa et al., 2001; Vlad and Orlandea, 2004; Baker, 2019). The Au(-Cu-Te) ore district of the Golden Quadrilateral is thought to be caused by calc-alkaline magmatism sourced in a mantle that may have been metasomatized during both an earlier Mesozoic and a contemporaneous Neogene subduction event (e.g., Roșu et al., 1997, 2001, 2004a, b; Panaiotu, 1998; Seghedi, 2004; Neubauer et al., 2005; Harris et al., 2013; Holder, 2015; Nimis and Omenetto, 2015; Ene, 2020; Seghedi et al., 2022). We compiled an extensive zircon data set from Au-rich porphyry and epithermal deposits in the Golden Quadrilateral to explore the relationship between the Neogene magmatic activity, regional tectonic evolution, and mineralization.

The Golden Quadrilateral in Romania is a Neogene ore district located in the southern Apuseni Mountains, the island-like orogen embayed by the Carpathian arc (see Fig. 1; Schmid et al., 2008, 2020; Ustaszewski et al., 2008; Kounov and Schmid, 2013; van Hinsbergen et al., 2020). On a global scale, the Golden Quadrilateral is a small segment of the 3,500-km-long Western Tethyan metallogenic belt, which stretches from Slovakia to Turkey and hosts numerous volcanogenic massive sulfide, porphyry, and epithermal deposits of predominantly Cretaceous to Cenozoic age (e.g., Janković, 1997; Richards, 2015; Baker, 2019).

The porphyry and epithermal deposits in the Golden Quadrilateral are particularly rich in Au and Te and are confined to three sediment-filled Miocene grabens (i.e., basins): (1) the NW- to SE-trending Brad-Săcărâmb, (2) the WNW- to ESE-trending Zlatna, and (3) the small Roșia-Bucium basin in the northern part of the district (see Fig. 1; e.g., Ghitulescu and Socolescu, 1941; Udubașa et al., 2001; Cook and Ciobanu, 2004; Vlad and Orlandea, 2004; Baker, 2019). These basins host, among others, the largest epithermal Au deposit of the Western Tethyan belt (Roșia Montană), one of Europe’s richest Cu-Au porphyries (Roșia Poieni), historically important mining centers (Deva, Barza, Valea Morii, Săcărâmb, Zlatna, Baia de Arieș), and several recent exploration camps (Ivascanu et al., 2003; Kouzmanov et al., 2003, 2005, 2007; Milu et al., 2003, 2004; Vlad and Orlandea, 2004; Baker, 2019). More recently discovered Au-Cu mineralization, which is the focus of this study, includes the following: (1) the Au-Ag epithermal deposit Certej, Au-Cu porphyry deposits at Bolcana, Cireșata, Colnic, and Rovina, all hosted within the Brad-Săcărâmb basin, (2) the Au-Cu porphyry deposits at Stănija and Ciungi Stănija in the western part of the Zlatna basin, and (3) the porphyry-Au occurrences at Geamăna West, Bucium South, and Poenița, in the Roșia-Bucium basin (see Fig. 1). Gold and Cu grades are associated with disseminated and vein sulfides (chalcopyrite-bornite-pyrite ± molybdenite in porphyry, and additional sphalerite-galena-pyrrhotite-arsenopyrite Cu-Pb sulfosalts in epithermal veins) hosted in the Neogene intrusions and/or adjacent Cretaceous and Cenozoic sedimentary rocks (Table 1 with references; App. Table A1).

The mineralized intrusions of the Golden Quadrilateral pierce the southern part of the Cretaceous-age Apuseni orogen, which was assembled mainly from pre-Mesozoic basement units (Dacia and Tisza megablocks) and their Mesozoic cover, ophiolite nappes derived from Jurassic ocean basins and oceanic arcs, and intruding calc-alkaline rocks of the Apuseni-Banat-Timok-Srednogorie belt (e.g., Săndulescu, 1984; Dallmeyer et al., 1999; Bortolotti et al., 2002, 2004; Pană et al., 2002; Nicolae and Saccani, 2003; Seghedi, 2004; Márton et al., 2007; Zimmerman et al., 2008; Balintoni et al., 2009; Dunkl et al., 2009; Gallhofer et al., 2017; Reiser et al., 2017). The latter is a magmatic arc related to the then N-dipping subduction of the Neotethys (Eastern Vardar) ocean that during the Mesozoic separated the Africa-derived microplate Adria from Europe-derived Tisza and Dacia megaunits in front of the European craton (Moesia block sensu lato; e.g., Schmid et al., 2008, 2020; van Hinsbergen et al., 2020). The subduction of Neotethys during Late Cretaceous to Paleogene times gave rise to major porphyry-related ore deposits including Majdanpek, Bor, and the Srednegorie district farther toward the south and east (e.g., Janković, 1997; Berza et al., 1998; Ciobanu et al., 2002; von Quadt et al., 2005; Gallhofer et al., 2015). During the Neogene (Late Miocene to Pliocene), the Apuseni orogen was overprinted by block rotation and extension, resulting in NW-SE–oriented graben structures filled by siliciclastic sedimentary rocks, marls, and freshwater limestones and a variety of volcanic rocks and subvolcanic intrusions associated with the hydrothermal deposits of the Golden Quadrilateral (see Ghitulescu and Socolescu, 1941; Ianovici et al., 1969; Merten et al., 2011, and references therein).

The porphyry-type and epithermal Au(-Cu-Te) mineralization in the Golden Quadrilateral is related to an ~7-m.y.-long phase of postorogenic calc-alkaline magmatism (Lemne, 1983; Pécskay et al., 1995; Roșu et al., 1997, 2001, 2004a, b; Kouzmanov et al., 2005, 2007; Holder, 2015; Ene, 2020). Besides shallow (<1.5 km) mineralized porphyries, this magmatic event produced barren intrusions, extensive pyroclastic deposits, and lava flows (e.g., Ianovici et al., 1969; Vlad and Orlandea, 2004). The magmatic rocks are predominantly andesitic in composition, containing abundant hornblende and biotite, but also including subordinate basaltic andesites, dacites and rhyolites, and rare pyroxene- and garnet-bearing trachybasalts and trachyandesites (e.g., Seghedi, 2004; Seghedi et al., 2022). Apatite, zircon, Fe-Ti oxides, and Cr spinel make up the most common accessory phases (Seghedi, 2004). The majority of the Neogene intrusions in the Golden Quadrilateral exhibit uniform calc-alkaline arc-like signatures—i.e., enrichment in large ion lithophile elements (LILEs), depletion in high field strength elements (HFSEs) and Ti and Nb, and steep light rare earth element (LREE) to heavy rare earth element (HREE) profiles (e.g., Roșu et al., 2004a; Harris et al., 2013; Seghedi et al., 2022). However, a few smaller magmatic centers, located mainly in the Roșia-Bucium basin, show more alkaline compositions, characteristic of extension-related mantle melting and in line with geophysical evidence for thinned crust below the northern, most extended part of the graben structures (Takács et al., 1996; Seghedi, 2004; Bala et al., 2017). Sporadic adakite-like intrusions, indicative of lower-crustal fractionation, are confined to the southeastern segment of the Brad-Săcărâmb basin and Deva (Rosu et al., 2004a; Harris et al., 2013).

The calc-alkaline compositions in the Golden Quadrilateral may be related to one or two subduction events in the Carpathian region. The Golden Quadrilateral magmas may have formed by postsubduction extensional melting of a lithospheric mantle modified by earlier subduction processes, which may have started in the Jurassic (ophiolite nappes) and extended to the Late Cretaceous (Neotethys subduction and Apuseni-Banat-Timok-Srednogorie arc; Roșu et al., 2001, 2004a; Schmid et al., 2008, 2020; Harris et al., 2013; van Hinsbergen et al., 2020). Alternatively, they may represent true arc magmas associated with westward-directed subduction of the oceanic lithosphere of Alpine Tethys, preserved as relicts in the Ceahlău-Severin ophiolites and geophysically indicated by a subvertical slab below the Eastern Carpathians (Vrancea seismic zone; e.g., Linzer, 1996; Linzer et al., 1998; Mason et al., 1998; Nemcok et al., 1998; Wortel and Spakman, 2000; Sperner et al., 2002; Carminati et al., 2012; Nimis and Omenetto, 2015).

Fifty-four samples of Neogene intrusions and some basement rocks were collected from drill core and outcrops (App. Table A2). Intrusions with well-established spatial and temporal relationships to mineralization were sampled, whenever possible, from minimally altered, homogeneous rock intervals (i.e., away from fractures and intrusive contacts).

The studied Neogene intrusions are markedly uniform in their mineralogy. The majority correspond to andesites containing varying proportions of plagioclase, hornblende, and biotite phenocrysts embedded in a plagioclase-rich groundmass. Common accessory phases observed both in thin section and mineral separates include apatite, zircon, and magnetite. Groundmass quartz occurs as a minor phase (<5 vol %) in few andesite samples, while quartz phenocrysts (<2 mm) are only present in the host-rock dacites at Bolcana (sample SMB-12) and Roșia Montană (RM19) among the Neogene samples, in Jurassic basement rhyolites from Bolcana and Certej (SMB-22, A17-19, and A17-23), and in regional samples of Mesozoic granitoids (DG072, DG115, DG109, DG101, DG091, DG112, DG100, and DG110; Gallhofer et al., 2015, 2017). More mafic rocks, including syenites, basalts, and gabbros corresponding to the Jurassic ophiolites, were sampled from drill core at Certej (A17-13 and A17-17) and outcrops in the broader area of the southern Apuseni Mountains (Gallhofer et al., 2017). Because of the petrographic uniformity of the Neogene intrusions, the nomenclature used in exploration was favored over the petrographic names in the text; both are provided in Appendix Table A2.

Designations pre-, syn-, and postmineralization denote the relative timing of the emplacement of the Neogene intrusions with respect to the bulk of the mineralization. These are adopted from exploration and updated with our observations and rely on deposit-scale core logging and assay results. Intrusions denoted as premineralization host the earliest, barren A-type quartz veins but never truncate sulfide-bearing veins or higher-grade disseminated mineralization. Synmineralization intrusions both host and crosscut sulfide veins and elevated metal grades. Postmineralization intrusions crosscut higher-grade zones and carry significantly lower grades compared to the pre- and synmineralization intrusions or are fully barren. Host-rock intrusions predate the premineralization porphyries based on crosscutting relationships. The term “unclear” is used for those intrusions whose spatial and temporal relationship to mineralization could not be established with confidence.

Sample aliquots for whole-rock geochemistry were mixed with an Li-based fluxer (66% Li tetraborate and 34% Li metaborate) in 1:5 proportion to produce glass beads. Major element oxide concentrations were analyzed on the glass beads using a wavelength-dispersive spectroscopy (WDS) Panalytical Axios X-ray fluorescence (XRF) instrument. Trace element concentrations were analyzed with a 115-μm spot size on an Excimer laser system coupled to a NexION 2000 quadrupole inductively coupled plasma-mass spectrometer. Data reduction was done using the MATLAB-based software SILLS (Guillong et al., 2008), with CaO concentrations measured by XRF serving as internal standard.

Zircon crystals were concentrated in a procedure including rock disintegration using the high-voltage pulse system SelFrag, sieving, processing on a Franz magnetic separator, density separation with methylene iodide, and picking under a binocular. The zircon crystals were heated in a muffle furnace at 900°C for 48 h to anneal radiation-damaged domains prior to their removal by leaching (chemical abrasion; Mattinson, 2005; von Quadt et al., 2014), embedded in epoxy mounts, ground, and polished. Cathodoluminescence (CL) imaging of the internal texture of zircon crystals was carried out on a JEOL JSM-6390 scanning electron microscope (SEM) equipped with a Deben Centaurus panchromatic CL detector.

Uranium-lead isotope ratios and trace element concentrations were analyzed simultaneously in situ on the same zircon volumes using a S155-LR ASI Resolution Excimer laser ablation system coupled to a Thermo Scientific Element-XR sector field inductively coupled plasma-mass spectrometer. The analyses were carried out with a laser spot size of 29 µm at a repetition rate of 5 Hz and a fluence of 2 J cm–2. Raw output data were processed with the IgorPro-based Iolite v3.6 software (Paton et al., 2010, 2011), and signal treatment was done utilizing the VizualAge software package (Petrus and Kamber, 2012). During signal treatment, analyses with unstable age signals longer than one-third of the integration time were discarded, whereas shorter unstable age signals and those with anomalous common Pb and Al, P, Ti, and La intensities (indicative of mineral/melt inclusions) were filtered out. Individual 206Pb/238U dates were corrected for initial 230Th/238U disequilibrium in the 238U-206Pb decay scheme by employing a Th-U mineral-melt partition coefficient of 0.2 (Wotzlaw et al., 2014). Weighted mean 206Pb-238U dates were calculated using IsoplotR (Vermeesch, 2018) and are reported with the associated 2σ statistical uncertainty (95% confidence level) and external uncertainty that includes a 1.5% excess intersession variance of the measured U/Pb and Pb/Pb in the secondary zircon standard AUSZ7-5 (Horstwood et al., 2016; von Quadt et al., 2016).

Trace element concentrations in zircon were quantified using the stoichiometric concentration of Si in zircon of 15.2 wt % as the internal standard. Titanium concentrations were corrected relative to the 91500 zircon reference material (Szymanowski et al., 2018). Zircon Eun/Eun* for the ore-related Neogene intrusions and Jurassic basement rocks were calculated relative to the concentrations of Eu, Sm, and Gd in C1 chondrite by McDonough and Sun (1995). Ti-in-zircon crystallization temperatures and melt oxygen fugacity relative to the fayalite-magnetite-quartz buffer (ΔFMQ) were calculated from trace element concentrations (Ti, Ce, U) after the calibration of Loucks et al. (2020), assuming a fixed lithostatic pressure of 200 MPa and aTiO2=0.75 for all studied samples, and aSiO2=0.7 or aSiO2=1 for quartz-free or rocks containing magmatic quartz, respectively. Variation in intensive parameters aTiO2 and aSiO2 and water content during differentiation are not accounted for in our calculations, and our discussion in the following is solely based on empirical observations.

Hafnium isotope composition of zircon was analyzed on the same spots ablated for U-Pb geochronology or adjacent on domains with similar CL appearance. The analyses were carried out on an ASI Resolution ArF Excimer laser ablation system coupled to a Nu Plasma 2 multicollector-inductively coupled plasma-mass spectrometer, employing a 50-μm laser spot size, repetition rate of 5 Hz, and a fluence of 4 J cm–2. The data were processed with the IgorPro-based Iolite v3.5 software using an in-house data reduction scheme. The initial zircon 176Hf/177Hf ratios and εHf(t) values were calculated with the in situ U-Pb dates, 176Lu decay constant of 1.867 × 10 –11 year –1 (Scherer et al., 2001), and 176Hf/177Hf and 176Lu/177Hf values of 0.282785 and 0.0336, respectively, for the chondritic uniform reservoir (CHUR; Bouvier et al., 2008).

Zircon crystals displaying homogeneous CL appearance and no inheritance detectable in reconnaissance LA-ICP-MS dating were selected for high-precision U-Pb geochronology using ID-TIMS. Following the standardized procedure, the zircon crystals were plucked out of epoxy mounts, chemically abraded in HF at 180°C for 12 h (Mattinson, 2005), cleaned in multiple cycles with HNO3 and HCl, spiked with the 202Pb-205Pb-233U-235U EARTHTIME (ET2535) tracer solution (Condon et al., 2015; McLean et al., 2015), and dissolved in HF at 210°C within 60 h (for a detailed description of the laboratory procedure, see Wotzlaw et al., 2014). Uranium and lead were separated from the matrix elements by ion-exchange chromatography using an HCl-based single-column chemistry modified after Krogh (1973). High-precision U-Pb isotope analyses were carried out on a Thermo TRITON Plus TIMS instrument in static measurement mode, with Faraday cups connected to 1013 Ω amplifiers (von Quadt et al., 2016; Wotzlaw et al., 2017). Data reduction was carried out using the Tripoli and ET_Redux softwares (Bowring et al., 2011) with algorithms of McLean et al. (2011). Individual U-Pb dates were calculated relative to the published calibration of the ET2535 tracer solution (Condon et al., 2015) using the decay constants of Jaffey et al. (1971) and are reported with their 2σ uncertainties (95% confidence level).

Whole-rock geochemistry

Whole-rock major and trace element compositions of the Neogene intrusions and basement rocks from the Golden Quadrilateral are reported in Appendix Table A3. Although least altered rocks were sampled at each site, nearly all of the Neogene samples still exhibit a significant alteration overprint, which is detectable in their high loss on ignition (LOI) values (mean = 4.24 wt %). Because the original concentrations of alkali metals and silica are more likely to be disturbed by alteration, we used the classification diagrams based on immobile trace elements for our samples (Winchester and Floyd, 1977, modified by Pearce, 1996; Ross and Bédard, 2009).

The Neogene intrusions exhibit a uniform composition and geochemical affinity (Fig. 2). The majority are characterized by consistent Zr/Ti and Nb/Y ratios (Zr/Ti = 0.02–0.04 and Nb/Y = 0.32–0.57) with very limited scatter and classify as andesites and basaltic andesites (Fig. 2A). These results are consistent with their mineralogical composition and earlier data (e.g., Roșu et al., 2004a; Seghedi, 2004; Harris et al., 2013; Ene, 2020). Intrusions from Poenița and Geamăna West are more alkaline and plot into the trachyandesite field (Nb/Y values of 0.86 and 1.07, respectively). The host-rock intrusion from Bolcana also falls into the trachyandesite field (Nb/Y = 0.73), although it was petrographically classified as dacite, as it contains conspicuous macroscopic quartz phenocrysts. Except at Bolcana, no pronounced shift in composition was observed between successive Neogene intrusions within individual deposits. The most evolved samples in the region are the basement magmatic rocks at Bolcana and Certej, which correspond to rhyolites (Zr/Ti ~ 0.085). The least evolved rocks are basalts sampled from drill core within the ophiolite basement at Certej (Zr/Ti ~ 0.008). All our samples apart from the basalts, which are tholeiitic, display a calc-alkaline affinity (Th/Yb mean value of 2.62), with no adakite-like features as reported for a few intrusions in the Bolcana-Certej-Săcărâmb and Deva areas (Fig. 2B; e.g., Roșu et al., 2004a; Seghedi, 2004; Harris et al., 2013).

CL imaging

The imaged zircon crystals vary in size mostly between 30 and 200 µm along the c axis and occasionally reach up to ~350 µm. They are mostly bipyramidal and exhibit oscillatory zoning from core to rim (App. Fig. A1). Neogene zircons from Certej, Bolcana, Rovina, Colnic, Stănija, and Ciungi Stănija exhibit similar characteristics, and only those from Certej are depicted in Appendix Figure A1. Conversely, zircons from Geamăna West and Poenița display a more varied CL appearance. For instance, one subpopulation from Geamăna West has bright Ordovician and Neogene cores and darker overgrowths of Triassic and Neogene age, respectively. Another subpopulation, which corresponds to the oldest, Archean-Proterozoic dates, consistently displays dark-gray to black tones from core to rim. Zircons from Poenița locally host euhedral bright and dark cores with inclusions and fractures, whereas those from the basement rhyolites exhibit oscillatory, sector, and patchy zoning and besides bipyramids, also appear in needle-like shapes.

In situ U-Pb zircon geochronology

Results of the in situ zircon U-Pb geochronology are reported in Appendix Table A4. The 230Th-corrected 206Pb/238U weighted mean dates of 36 samples, encompassing the Neogene intrusions and basement rocks in the Golden Quadrilateral, are reported in Table 2 and shown in Figure 3.

The weighted mean dates of the Neogene samples fall into three time intervals. Most of the samples yield dates between ~13.2 and 12.1 Ma, which we call phase I. These include the intrusions at Bucium South (Roșia-Bucium basin), Stănija, and Ciungi Stănija (Zlatna basin), the host-rock intrusion at Bolcana, and the entire magmatic suites at Rovina, Colnic, and Certej (Brad-Săcărâmb basin). All intrusions at Cireșata and the premineralization intrusions at Bolcana yield resolvable younger dates, between ~11.5 and 10.7 Ma (phase II). The weighted mean dates of the intrusions at Geamăna West and Poenița (Roșia-Bucium basin) fall between ~8.2 and 7.3 Ma (phase III).

The weighted mean dates of the host-rock, pre-, syn-, and postmineralization intrusions from the individual deposits in the Golden Quadrilateral mostly overlap within the uncertainty of LA-ICP-MS. For example, in the Au-Cu porphyry deposits Colnic and Cireșata, the weighted mean dates of the relatively oldest host-rock porphyry are equivalent to those of the late post-ore dikes. One exception is the Au-Cu deposit Bolcana, in which the oldest host-rock intrusion yields a date at 13.0 ± 0.06|0.20 Ma (where the slash separates internal|external uncertainties), whereas the ore-related but premineralization porphyries yield younger dates between ~11.1 and 10.8 Ma. Two samples of the basement rhyolites from Bolcana and Certej yield weighted mean dates at 158.63 ± 0.66|2.47 and 156.74 ± 0.60|2.43 Ma, respectively.

Several samples from the deposits in the Zlatna and Brad-Săcărâmb basin contain a zircon population between ~14 and 13 Ma, which overlaps with individual zircon dates from the Bucium South porphyry and host-rock intrusion at Bolcana (Fig. 3). Also, older concordant zircon analyses from the Geamăna West intrusion at ~8.25 Ma overlap with the main population of zircon dates from the Poenița intrusion. Older concordant zircon analyses (n = 56) identified in the Neogene intrusions cover a range of ages from the Mesoarchean to Cenozoic (App. Fig. A2A), with most analyses spanning the Neoproterozoic, Permian-Triassic, and Late Cretaceous-Oligocene. Notably, no concordant zircon dates were recorded within the age range of 2.1 to 1 Ga.

Zircon trace element geochemistry

In situ zircon trace element compositions are reported in Appendix Table A4. Zircon EuN/EuN*, ΔFMQzircon, and Ti-in-zircon crystallization temperatures for various deposits in the Golden Quadrilateral are compared in Figure 4.

The EuN/EuN* ratios of Neogene zircon vary significantly across the individual deposits (Fig. 4A). The lowest values, mostly in the range of 0.2 to 0.4 (median values of 0.29, 0.31, and 0.34, respectively) are observed in the Au-Cu porphyry deposits Colnic, Stănija, and Ciungi Stănija. Conversely, the youngest dated zircons in the Golden Quadrilateral, from the trachyandesite intrusions at Poenița and Geamăna West, as well as intrusions at Certej, display the highest EuN/EuN*, with median values of 0.59, 0.79, and 0.62, respectively. Zircons from the basement rhyolites exhibit EuN/EuN* between 0.32 and 0.47 (median = 0.41) in Bolcana and somewhat higher EuN/EuN*, between 0.45 and 0.90 (median = 0.56) in Certej.

Zircon ΔFMQ values also scatter considerably (Fig. 4B). Zircon from Certej exhibits the most positive values (median = 0.8) among the Neogene samples, followed by Bolcana (median = 0.4), whereas the most negative values are observed in zircon from Cireșata (median = –0.8, individual analyses at –1.4) and Ciungi Stănija (median = –0.7). The ΔFMQ values of zircon from the basement rhyolites are on average higher than those from the Neogene (trachy-)andesite and dacite intrusions, with median values of 0.9 at Bolcana and 0.7 at Certej.

Ti-in-zircon temperatures of most of the Neogene intrusions range between 690° and 780°C. Somewhat higher temperatures exceeding 800°C are recorded in zircon from Rovina, Colnic, Certej, and Bolcana. Rhyolite-hosted Jurassic zircons collectively exhibit higher temperatures than the Neogene zircons, varying between 780° and 890°C and 815° and 890°C at Certej and Bolcana, respectively.

High-precision (ID-TIMS) zircon geochronology

High-precision U-Pb dates of single zircon crystals are reported with their 2σ uncertainties in Appendix Table A5 and depicted in Figure 5.

The 230Th-corrected 206Pb/238U dates from the porphyry deposit Rovina range between 12.630 ± 0.016 and 12.526 ± 0.021 Ma (Fig. 5A). The dates obtained for the premineralization so-called Glam breccia (polymict but sediment clast-dominated contact breccia associated with shallow porphyry emplacement) and C1 porphyry overlap entirely within uncertainty, with equivalent youngest analyses at 12.583 ± 0.030 and 12.574 ± 0.015 Ma, respectively. Younger dates, spanning from 12.539 ± 0.015 to 12.526 ± 0.021 Ma, are resolved within the synmineralization intrusive magmatic breccia. The distribution of zircon dates from Colnic extends to that of Rovina and is bracketed by the oldest date from the Colnic porphyry (premineralization) at 12.527 ± 0.011 Ma and the youngest date from the late-mineral dike (synmineralization) at 12.448 ± 0.041 Ma. Zircon dates from intrusions at Cireșata range between 11.475 ± 0.010 Ma in the early mineral porphyry and 11.432 ± 0.009 Ma in the late postmineral dike, all overlapping within uncertainty. The same is true for the intrusions in Certej, where the oldest zircon date is recorded in the premineralization Hondol-deep porphyry at 12.782 ± 0.014 Ma and the youngest in the Hondol-shallow porphyry at 12.705 ± 0.013 Ma.

Zircon Hf isotope composition

In situ zircon Hf isotope compositions also include data of zircon from the Jurassic mid-ocean ridge basalts (MORB), gabbros and island-arc granitoids, and Cretaceous arc magmatic rocks from the broader area of the Apuseni Mts. (App. Table A6; Fig. 6)

The εHf(t) of Neogene zircons from different deposits in the Golden Quadrilateral varies mostly from 2 to 10. The most positive zircon εHf(t) values are recorded in the Roșia-Bucium basin: for the Cetate dacite (8) and quartz andesite (4.8) at Roșia Montană, for the biotite-bearing diorite (7.7) and host diorite (5.5) at Roșia Poieni, and for the intrusions at Geamăna West (4.4) and Poenița (3.9). The most negative values are documented for the intrusions at Ciungi Stănija (–1.5) and Stănija (–0.8) in the Zlatna basin. Distinctly negative εHf(t) values, between –10 and –7.4, are identified in zircon xenocrysts entrained in the trachyandesite intrusion in Geamăna West, which yielded concordant dates at ~28 Ma. The εHf(t) values of cores and rims of Neogene zircon display no apparent trend.

The εHf(t) values of zircon from the Mesozoic samples generally display greater scatter. The εHf(t) values of zircon from the Cretaceous (~80 Ma) magmatic rocks from the southern and northern Apuseni Mts. vary between –6 and 7.8, around a median value of 1.7. Jurassic (~158 Ma) MORB gabbros and island-arc granitoids systematically exhibit more positive zircon εHf(t), with median values of 9.6 and 7.5, respectively.

Tectonic framework for the Neogene magmatic-hydrothermal mineralization in the Golden Quadrilateral

The U-Pb zircon ages from this study, together with those from the literature, constrain the Neogene magmatic activity in the Golden Quadrilateral to an ~6.5 m.y. period between 13.61 ± 0.07 and 7.24 ± 0.04 Ma (Kouzmanov et al., 2005, 2007; Rocha, 2013; Gallhofer et al., 2015; Holder, 2015; Brunner, 2018; Müller, 2018; Ene, 2020). Each basin contains clusters of intrusions within small areas that have overall similar ages (small, <<1-m.y. variation in single-crystal ID-TIMS ages and overlapping less precise LA-ICP-MS ages; Figs. 1, 7; App. Table A7). On the other hand, among different small areas, the intrusions may differ significantly in age, indicating that brief events of magma emplacement jumped, apparently at random, between the basins and between small areas within the same basin (e.g., Roșia Montană, and Roșia Poieni in the Roșia-Bucium basin; Rovina, Colnic, Brad, Cireșata, and Valea Morii in the Brad-Barza area of the Brad-Săcărâmb basin; Certej, Bolcana, and Săcărâmb further to the southeast in the same basin; Figs. 1, 7). The small age range of intrusions in each deposit and of individual zircon in any one intrusion sample (Fig. 5) is interpreted to indicate small, short pulses of magma generation in the Golden Quadrilateral, consistent with the heterogeneous zircon and whole-rock isotopic composition and only sporadic occurrence of alkaline and adakite-like magmas. This contrasts the long-lived, presumably lower-crustal fractionation of large magma volumes characterizing giant porphyry Cu ± Au ore camps such as Tampakan (Rohrlach et al., 2005; Parra-Avila et al., 2022), El Salvador (Lee et al., 2017), Kadjaran (Rezeau et al., 2016, 2019), or Rio Blanco-Los Bronces (Large et al., 2024) and Quellaveco (Nathwani et al., 2021).

In Figure 8, we plot paleomagnetic declination against new U-Pb zircon ages for a subset of Neogene intrusions in the wider area of the Golden Quadrilateral along with previous K-Ar results, which despite analytical uncertainties show a striking correlation between age and rotation (Roșu et al, 2004a). The intrusions emplaced prior to ~12 Ma experienced larger-scale clockwise rotation (up to 70°), whereas the younger intrusions generally record only minor rotation (±~30°). Precise and accurate zircon ages largely support the older Ar dating and confirm that the Neogene intrusions in the broader Apuseni Mts. and north in the Maramureș Mts. were emplaced during large-scale clockwise rotations of the Tisza-Dacia megablock into the Carpathian embayment (Pătrașcu et al., 1994; Panaiotu 1998, 1999; Roșu et al, 2004a). Higher positive values in older intrusions may reflect early block rotation facilitated by the retreat of the Carpathian slab, which regionally resulted in opening of detachments and steep normal faults in the neighboring Pannonian basin and shearing along major strike-slip faults throughout the Carpathian region (e.g., Tischler et al., 2007; Ustaszewski et al., 2008; Matenco and Radivojević, 2012). The formation of ore deposits in the Golden Quadrilateral does not seem preferentially related to any particular period of block rotation.

The Neogene intrusions within the Brad-Săcărâmb basin (~12.9–9.7 Ma) tend to become progressively younger in a northwest to southeast direction (Figs. 1, 7). This younging trend may reflect gradual, “zipper-like” basin opening propagating from northwest to southeast, in continuation of the regional Békés-Zărand graben. Small-scale extensional melting of the mantle may have been focused at the root of the growing graben structure, with magmas subsequently channeled toward the upper crust and emplaced at the hinge of the propagating structure. The observed age progression toward the hinge in the largest basin, combined with the paleomagnetic data, likely reflects local and short-lived melt formation driven by rotation-induced extension, although concurrent subduction hydration cannot be excluded as an additional driver for melting. Isotopic heterogeneity of magmas in the Golden Quadrilateral (and occurrences of alkaline and adakite-like intrusions) further supports the local melting hypothesis over district-wide crustal fractionation.

Tracing the sources of Neogene magmas in the Golden Quadrilateral

The Neogene magmas across the Golden Quadrilateral were derived from a heterogeneously metasomatized lithospheric mantle followed by differential assimilation of preexisting crust in various basin segments (εHf(t)zircon between –2 and 10; Fig. 6). The most primitive Hf isotope compositions are identified in zircon from the adakite-like intrusion at Deva (median εHf = 10.6; southernmost part of the Golden Quadrilateral), from calk-alkaline magmatic rocks at Roșia Montană (8 and 4.9) and Roșia Poieni (7.7 and 5.5), and from alkaline intrusions at Poenița (3.9) and Geamăna West (4.4 and 4.7; Kouzmanov et al., 2007; Ene, 2020). Except for Deva, all are located in the Roșia-Bucium basin but include the oldest as well as the youngest Neogene intrusions in the Golden Quadrilateral. Furthermore, the Cu-Au porphyry deposits Roșia Poieni and Deva and the Au-Ag epithermal deposit Roșia Montană represent the largest deposits in the district. These magmas are only marginally less primitive than the depleted mantle melts parental to Jurassic MORB gabbros (9.6) and melts producing island-arc granitoids derived from a metasomatized mantle (7.5; Bortolotti et al., 2002, 2004; Gallhofer et al., 2017). We interpret the alkaline magmas at Poenița and Geamăna West as rare asthenospheric melts in the Golden Quadrilateral and local isotopic heterogeneities among calc-alkaline magmas to reflect variable metasomatism of their mantle source and/or minor crustal assimilation. Most of the intrusions in the Zlatna and Brad-Săcărâmb basins (with the exception of Coasta Mare, Haitău, and Ledișoiu) exhibit more evolved, crustal-like εHf(t)zircon (mostly in the range between –1 and 2; Fig. 6). They partly overlap with xenocrystic zircons sampling the composition of the local basement (Ene, 2020) and with Cretaceous arc intrusions in the Apuseni Mts. (Apuseni-Banat-Timok-Srednogorie belt) that had evolved over longer residence time in the crust (Gallhofer et al., 2015).

Zircon εHf values of Neogene intrusions are consistent with the published mantle-like whole-rock Sr-Nd-Pb and clinopyroxene δ18O (5.5–5.6‰) compositions of magmatic rocks from the broader Apuseni Mts. and regionally with mantle-derived spinel-lherzolite xenoliths and asthenosphere-derived alkali basalts from the Pannonian (e.g., Downes et al., 1992; Embey-Isztin et al., 1993; Mattey et al., 1994; Dobosi et al., 1998; Seghedi et al., 2007; Harris et al., 2013). While two studies reported a progression from more evolved toward more depleted younger magmas throughout the south Apuseni Mts. (based on whole-rock K-Ar ages and Sr-Nd-Pb isotopes; Roșu et al., 2004a; Harris et al., 2013), we observe no such trend in the zircon data.

Ample interaction of the Neogene magmas in the Golden Quadrilateral with the basement in the broadest sense is attested to by zircon xenocrysts, which are most abundant in intrusions at Colnic and in the Roșia-Bucium basin (see Fig.6; App. Fig. A2; Ene, 2020). The age spectrum of xenocrystic zircon fingerprints Neoproterozoic, Permian through to Triassic, and Cretaceous-Cenozoic crustal sources (see App. Fig. A2A, B). The Permo-Triassic population matches the age of the Apuseni basement (e.g., Dallmeyer et al., 1999; Pană et al., 2002), and the Cretaceous-Paleogene population likely reflects the presence of Apuseni-Banat-Timok-Srednogorie–related magmatic rocks and/or lower-crustal cumulates. The Triassic and Eocene-Oligocene populations fingerprint previously unrecognized crust. Cross-basin differences partly reflect locally predominant rocks, e.g., Cretaceous flysch in the Roșia-Bucium basin and Neoproterozoic-Cambrian sources in the basement of the Brad-Săcărâmb and Zlatna basins (see App. Fig. A2B). The near absence of Jurassic-age xenocrysts indicates that the exposed Jurassic ophiolites (Brad-Săcărâmb and Zlatna basins) are restricted to the upper crust above the level of significant assimilation by the Neogene magmas.

The age spectrum of zircon xenocrysts from the Golden Quadrilateral does not resemble the detrital record of the Dacia megablock in the Inner Carpathians, on which the Golden Quadrilateral rests, but rather matches the more distant Dobrogea and Moesia blocks and partly the Dacian basement of the South Carpathians (External Carpathians; see App. Fig. A2C, D, and references therein). Diagnostic features include pronounced Neoproterozoic and Permian-Triassic peaks but rare mid-Ordovician zircon xenocrysts (e.g., Balintoni et al., 2010b, 2011a, b; Balintoni and Balica, 2016; Ducea et al., 2018; Roban et al., 2023).

Several studies proposed melting of the lower crust rather than the mantle as the dominant source for the Neogene magmas in the Golden Quadrilateral. This hypothesis was based on findings of rare andesite-hosted xenocrystic garnets in the Golden Quadrilateral that resemble garnets in granulite xenoliths in Pannonian alkali basalts (Seghedi et al., 2007). More recently, Ene (2020) argued that the enrichment in the fluid-immobile element Th in some Neogene intrusions is inconsistent with a metasomatized mantle source and that the composition of the Neogene magmas may be produced by up to 90% extension-driven melting of lower-crustal mafic cumulates. Nonetheless, this interpretation has not yet been tested against thermal constraints (cf. Heinonen et al., 2022; Storck et al., 2021; Chang and Audetat, 2023).

Utility of zircon trace element signatures as tools for exploration in the Golden Quadrilateral

Low EuN/EuN* anomalies in zircon as indicator of high magmatic H2O and high Ce anomalies as proxies for high magmatic oxygen fugacity have been proposed to discriminate fertile (i.e., ore-forming) from barren intrusions and magmatic districts (e.g., Ballard et al., 2002; Burnham and Berry, 2012; Trail et al., 2012; Dilles et al., 2015; Lu et al., 2016, 2017; Lee et al., 2017, 2021; Loucks et al., 2020). Figure 4 and Appendix Figure A3 test the applicability of these indicators on our example of a porphyry-epithermal district with clear Au affinity, located in an unusual tectonic setting. Pre- and synmineralization intrusions are taken to represent fertile magma batches that exsolved ore-forming hydrothermal fluids. Host-rock and postmineralization intrusions are regarded as barren magma batches emplaced prior to and/or after ore formation.

Results show neither cross- nor intradeposit patterns that could be correlated with the Au-mineralizing events at this small scale (see Fig. 4; App. Fig. A3). Between individual deposits, zircon EuN/EuN* values range widely, from a median value of ~0.2 in the Au-rich porphyry deposits Stănija and Colnic, to ~0.6 in the Au-Ag epithermal deposit Certej, and to even higher values in the Au porphyry prospect Geamăna West (~0.8). EuN/EuN*zircon values of the Au-mineralizing intrusions in the Golden Quadrilateral significantly overlap with the values for the Jurassic island-arc rhyolites, which are unmineralized throughout the wider region. EuN/EuN*zircon values of some Neogene deposits also overlap with the Pannonian Miocene Bogács ignimbrite of the same age, which might approximate an erupted compositional equivalent of the ore-forming magmas in the Golden Quadrilateral (i.e., calc-alkaline andesitic and dacitic compositions; Lukács et al., 2021).

The substantial spread in zircon ΔFMQzircon across the deposits in the Golden Quadrilateral (median values between 0.8 in Certej and –0.8 in Cireșata) suggests that the Au-rich porphyry and epithermal mineralization may be associated both with oxidized and more reduced magmas (Fig. 4B; Loucks et al., 2020). Such large cross-deposit variation is surprising, given the uniform andesitic composition of nearly all ore-related intrusions (see Fig. 2; App. Table A3), and is consistent with the interpretation of small magma batches evolving in isolation. The barren Jurassic rhyolites exhibit, on average, the highest values in the region, further illustrating the disassociation between elevated zircon ΔFMQ and mineralization potential. The lowest ΔFMQzircon (i.e., most reduced) compositions are recorded for the Bogács ignimbrite, but they plot only marginally below the values for the Au-rich porphyry deposit Cireșata. Contrary to Holder (2015), we observe no correlation of zircon EuN/EuN* or ΔFMQ and magmatic age across the district.

Within the individual deposits, zircon EuN/EuN* and ΔFMQ do not generally correlate with ore-productive magma batches but locally reflect evolving melt composition (see App. Fig. A3). In the Au-rich porphyry deposit of Bolcana, zircons from the barren rhyolite and host-rock dacite, among others, can be discriminated based on their EuN/EuN* and ΔFMQ values from zircon of the fertile premineralization (shell and crowded amphibole-biotite) andesites. This comparison may suggest that, on the scale of a single deposit, barren magma batches can be distinguished from fertile ones provided there is an accompanying shift in melt composition. Yet, melt composition alone cannot explain the strongly contrasting EuN/EuN*zircon and ΔFMQ of otherwise compositionally similar andesitic intrusions in Cireșata and Certej. This suggests that additional factors, such as timing of zircon saturation, as well as abundance of other cocrystallizing REE-bearing phases (e.g., titanite, apatite, monazite, hornblende) also play an important role (e.g., Buret et al., 2016; Loader et al., 2017, 2022). In summary, zircon EuN/EuN* and ΔFMQ do not seem diagnostic for Au-mineralizing magma pulses.

Ti-in-zircon crystallization temperatures

Zircons from the ore-related Neogene intrusions in the Golden Quadrilateral record relatively low temperatures between 690° and 780°C, with no pronounced cross-deposit differences (Fig. 4C; App. Fig. A4). Similar and even lower temperatures were reported for the giant ore Cu porphyry systems in the Andes, such as El Salvador (Chile) and Quellaveco (Peru), and at Yerington (Nevada; Dilles et al., 2015; Nathwani et al., 2021), whereas higher temperatures are recorded in barren volcanic rocks of the Apuseni area (Jurassic rhyolites at Certej, 780°–890°C, and Bolcana, 815°–890°C, and the regional Bogács ignimbrite, 800°–896°C; Lukács et al., 2021). These differences may reflect zircon saturation in relatively cool but H2O-rich magmas giving rise to ore formation. Decreasing Ti-in-zircon temperatures with time in Roșia Poieni (from median 767° to 739°C), Roșia Montană (732°–720°C) and Stănija (761°–748°C) magmatic centers are consistent with increasing H2O content from premineral to syn- and postmineral magmatic phases.

Lifetime of magmatic centers and duration of magmatic-hydrothermal ore formation

The total spread of zircon high-precision ages in a mineralized magmatic center indicates the minimum lifetime of magmas exposed in a composite intrusion (Schaltegger et al., 2009), whereas the youngest zircons in single intrusions that pre- and postdate hydrothermal veining and ore formation bracket the maximum duration of ore-forming hydrothermal activity (von Quadt et al., 2011). These two timescales relating to an individual magmatic-hydrothermal ore deposit are both shorter than the duration of magmatism on the scale of an entire mineral province. The latter relates to tectonic processes generating magmas in the mantle and lower crust and is also thought to precondition the thermal state of the upper crust in the lead-up to generating an ore-forming magmatic-hydrothermal system (e.g., Lee et al., 2017; Large et al., 2021).

The youngest ID-TIMS dates constrain the emplacement age of individual intrusions (e.g., Schaltegger et al., 2009; von Quadt et al., 2011; Large et al., 2020). The Au-rich magmatic-hydrothermal systems in Rovina, Colnic, Cireșata, Bolcana, and Certej evolved over brief timescales of tens of thousands of years to a few hundred thousand years (Fig. 5A; App. Table A5). The magmatic activity associated with the Au-Cu porphyry mineralization at Rovina lasted for a minimum of ~104 ± 26 k.y., as bracketed by the oldest zircon (12.630 ± 0.016 Ma) in the premineralization Glam breccia and the youngest zircon (12.526 ± 0.021 Ma) in the premineralization intrusive breccia (intrusive magmatic breccia). At Colnic, zircon crystallization is recorded over 79 ± 42 k.y., from the age of the oldest zircon from the Colnic porphyry (12.527 ± 0.011 Ma; premineralization) to the youngest zircon from the late-mineral dike (12.448 ± 0.041 Ma; synmineralization). The youngest zircons in the Colnic porphyry (12.502 ± 0.015 Ma) and in the late-mineral dike at Colnic (12.448 ± 0.041 Ma) constrain the maximum duration of hydrothermal mineralization to 54 ± 44 k.y. (Fig. 5). High-precision ages from the Au-Cu porphyry deposit Colnic lie in perfect continuity with the ones from Rovina, suggesting that zircons from both deposits might have crystallized in sequence from a common magmatic system. However, this assumption is not supported by the distinctly lower Eu/Eu* in zircons from Colnic (0.2–0.35 compared to 0.45–0.6) and their, on average, more mantle-like Hf isotope signatures (see Figs. 4A, 6). The ore-forming magmatic activity in Cireșata lasted as short as 43 ± 13 k.y., within which all porphyry phases—from the barren wall-rock porphyry (host rock) to the late postmineral dike—were emplaced almost synchronously, as shown by overlapping ranges of high-precision zircon ages. Hydrothermal ore formation in Cireșata is bracketed by the youngest zircons of the early-mineral intrusion (11.431 ± 0.012 Ma) and postmineral dike (11.432 ± 0.009 Ma) to timescales shorter than 15 k.y. At Certej, the oldest (12.782 ± 0.014 Ma) and youngest (12.705 ± 0.013 Ma) zircon ages from the premineralization Hondol-deep and Hondol-shallow porphyry, respectively, record 77 ± 19 k.y. of magmatic zircon crystallization, but the age and duration of subsequent ore deposition are not further constrained.

The Au-dominated mineral province Golden Quadrilateral records a 6.5-m.y.-long duration of calc-alkaline magmatism but much shorter duration of magmatic-hydrothermal activity in each deposit, constrained to less than ~100 k.y. These two timescales are similar to other Cu ± Au mineral provinces containing small to midsized porphyry deposits worldwide (e.g., Halter et al., 2004; Chelle-Michou et al., 2014, 2015; Buret et al., 2016; Tapster et al., 2016; Large et al., 2018). What may be peculiar in the Golden Quadrilateral is that the minimum duration of magmatic zircon crystallization in each mineralized center, including distinctly pre-ore host intrusions and volcanic rocks, is unusually brief compared to these published examples. This is consistent with the interpretation that the magma reservoirs eventually saturating ore-forming hydrothermal fluids are small and cool rapidly in the upper crust. Geochronological constraints do not exclude that the magmas were prepared over longer periods at greater depth, as suggested by variably adakitic magma compositions, but the observed differences in geochemical characteristics of closely neighboring magmatic centers of similar age make a long-lived, large common reservoir in the lower crust a rather unlikely case for the Golden Quadrilateral.

Zircon petrochronology and rock geochemistry that were used to investigate the time-space patterns, sources, composition, and duration of the Neogene calc-alkaline magmatism associated with the Au-rich porphyry and epithermal mineralization in the Golden Quadrilateral (Apuseni Mts., Romania) led to the following findings:

  1. New and published U-Pb zircon ages show that the ore-related magmatism in the Golden Quadrilateral took place almost continuously between 13.61 ± 0.07 (Roșia Montană) and 7.24 ± 0.04 Ma (Geamăna West) but in brief events jumping in time between different segments of three extensional basins. The U-Pb ages confirm a correlation of intrusion age with paleomagnetic declination at the scale of the entire ore district, indicating magma emplacement synchronous with rotation-induced extension in a southwest-northeast direction. The magmatism in the Golden Quadrilateral evolved during the rotation of the Tisza-Dacia megablock, which triggered the opening of the basins in the Golden Quadrilateral and partial melting of the subjacent mantle and provided pathways for the ore-forming magmas to the upper crust. Within the largest basin (Brad-Săcărămb), a SE-directed younging trend in magmatic ages tracks the propagation of an underlying graben structure from the neighboring Pannonian basin into the Golden Quadrilateral.

  2. The Neogene magmas in the Golden Quadrilateral were predominantly derived from a locally heterogeneous lithospheric mantle and incorporated various proportions of the preexisting crust (εHfzircon between –2 and 10). The most mantle-like magmas in the Golden Quadrilateral are associated with the oldest and the youngest magmatic activity in the Golden Quadrilateral, and they gave rise to the three largest deposits in the region: Roșia Poieni, Roșia Montană, and Deva. Age spectra of xenocrystic zircon more closely correspond to European continental basement rather than the Dacian megablock in the Inner Carpathians, on which the Golden Quadrilateral rests today.

  3. Europium anomaly (EuN/EuN*) and ΔFMQ of zircon, commonly used as fertility indicators for porphyry Cu systems, reveal no systematic patterns with respect to the ore-forming events in the Golden Quadrilateral. High and low EuN/EuN* and ΔFMQzircon values derived from zircon trace element concentrations imply that the Au-rich porphyry and epithermal mineralization in the Golden Quadrilateral is not systematically associated with unusually H2O-rich or oxidized magmas.

  4. The maximum duration of magmatic-hydrothermal activity in individual deposits in the Golden Quadrilateral is bracketed by high-precision (ID-TIMS) U-Pb zircon geochronology to below 100 k.y. This finding, together with small-scale trace element variations, indicates separate magmatic fluid sources of small to moderate dimensions.

The authors wish to thank Paul Ivășcanu, Tim Baker, Călin Tămaș, Zsolt Kulcsár, Réka Dénes, Randy Ruff, Sorin Halga, and Tim Fletcher for their guidance during field work and sampling in the Golden Quadrilateral and for sharing with us the unpublished mine geologic data and their knowledge of local geology. We further thank the teams of Eldorado Gold, Barrick Gold, Carpathian Gold (Euro Sun Mining), and Deva Gold for granting access to the exploration sites and for their logistical support. We extend our gratitude to Vivian Santana Rocha for acquiring the preliminary in situ zircon U-Pb and trace element data, Stefan M. Schmid for reading an early version of the manuscript, Remy Lüchinger and Lydia Zehnder for their help with sample preparation, and David Holder for sharing his data set with us. Field work in the Golden Quadrilateral (Romania) conducted by S.M., M.B., and L.M. was partly funded by the Grubenmann-Burri travel grant of ETH Zurich. Finally, S.M. acknowledges the generous financial support of ETH Zurich (ERDW scholarship) received for the duration of his M.Sc. studies.

Sava Markovic is currently pursuing a Ph.D. degree in the Mineral Resource Systems group at ETH Zurich (Switzerland). His current research combines field work with high-precision zircon geochronology in exploring the dynamics of Sn mineralization in peraluminous granitic systems of the Peruvian Eastern Cordillera. He previously worked on several Cu-Au porphyry deposits of the Late Cretaceous-Neogene Apuseni-Banat-Timok-Srednogorie belt in the Balkans. Sava obtained his B.Sc. degree in geology from his hometown University of Belgrade (Serbia) and holds an M.Sc. degree from ETH Zurich.

Gold Open Access: This paper is published under the terms of the CC-BY-NC license.

Supplementary data