New CA-ID-TIMS U–Pb zircon ages for the Altenberg–Teplice Volcanic Complex (ATVC) document discrete and coeval pulses of Variscan magmatic activity in the Eastern Erzgebirge (Eastern Variscan Belt)

The Altenberg–Teplice Volcanic Complex (ATVC) is a large ~ NNW–SSE trending volcano-plutonic system in the southern part of the Eastern Erzgebirge (northern Bohemian Massif, south-eastern Germany and northern Czech Republic). This study presents high precision U–Pb CA-ID-TIMS zircon ages for the pre-caldera volcano-sedimentary Schönfeld–Altenberg Complex and various rocks of the caldera stage: the Teplice rhyolite, the microgranite ring dyke, and the Sayda-Berggießhübel dyke swarm. These data revealed a prolonged time gap of ca. 7–8 Myr between the pre-caldera stage (Schönfeld–Altenberg Complex) and the climactic caldera stage. The volcanic rocks of the Schönfeld–Altenberg Complex represent the earliest volcanic activity in the Erzgebirge and central Europe at ca. 322 Ma. The subsequent Teplice rhyolite was formed during a relatively short time interval of only 1–2 Myr (314–313 Ma). During the same time interval (314–313 Ma), the microgranite ring dyke intruded at the rim of the caldera structure. In addition, one dyke of the Sayda-Berggiesshübel dyke swarm was dated at ca. 314 Ma, while another yielded a younger age (ca. 311 Ma). These data confirm the close genetic and temporal relationship of the Teplice rhyolite, the microgranite ring dyke, and (at least part of) the Sayda-Berggießhübel dyke swarm. Remarkably, the caldera formation in the south of the Eastern Erzgebirge (caldera stage of ATVC: 314–313 Ma) and that in the north (Tharandt Forest caldera: 314–312 Ma) occurred during the same time. These data document a large ~ 60 km NNW–SSE trending magmatic system in the whole Eastern Erzgebirge. For the first time, Hf-O-isotope zircon data was acquired on the ring dyke from the ATVC rocks to better characterize its possible sources. The homogeneous Hf-O-isotope zircon data from the microgranite ring dyke require preceding homogenization of basement rocks. Some small-scale melts that were produced during Variscan amphibolite-facies metamorphism show similar Hf-O-isotope characteristics and can therefore be considered as the most probable source for the microgranite ring dyke melt. In addition, a second source with low oxygen isotope ratios (e.g. basic rocks) probably contributed to the melt and possibly triggered the climactic eruption of the Teplice rhyolite as well as the crystal-rich intrusion of the ring dyke.


Introduction
At the end of Variscan orogeny, widespread magmatic activity occurred in many regions of the Bohemian Massif (Tischendorf 1989;Förster and Romer 2010;Breiter 2012;von Seckendorff 2012;Lützner et al. 2021). In the Erzgebirge/ Krušne Hory this magmatism is often related with large ore-bearing deposits (e.g. Breiter et al. 1999). There is still considerable debate about the magmatic ages and the tempo of intrusions although more and more age data became available over the last decade. CA-ID-TIMS zircon dating is especially suitable to establish the sequence and 1 3 periodic patterns of the Variscan magmatic activity because of its high precision (ca. ± 0.1% for individual zircon analyses and up to ± 0.03% for weighted mean sample ages; e.g. Schaltegger et al. 2015). For instance, CA-ID-TIMS zircon ages revealed multiple discrete phases of the Variscan magmatism in the Western Erzgebirge (Tichomirowa et al. 2019a). According to these data, the oldest granites (Aue-Schwarzenberg) intruded at ~ 323-322 Ma followed 2-4 Myr later by the granites of Bergen and Kirchberg. The highly evolved ore-bearing granites from the Eibenstock pluton intruded after a time lag of ~ 5 Myr at ~ 315-314 Ma. Breitkreuz et al. (2021) performed the first high precision CA-ID-TIMS data for granites and volcanics from the northern part of the Eastern Erzgebirge. Accordingly, the Niederbobritzsch granite (NBG) intruded approximately at the same time as the granites from Bergen and Kirchberg (320-318 Ma) while the volcanics from the Tharandt Forest caldera (TFC) were formed several Myr later (314-312 Ma), i.e. roughly coeval or slightly later than the Eibenstock granite in the Western Erzgebirge. However, the temporal relationship with other magmatic rocks in the Eastern Erzgebirge is still uncertain. The NBG-TFC magmatic system is located only about 10 km north of the ATVC and its temporal relationship to the ATVC still has to be determined.
Large occurrences of Variscan volcanics are absent in the Western Erzgebirge but widespread in the Eastern Erzgebirge (in addition to granites). The ATVC is interpreted to be the oldest volcanic occurrence of the late-to post-collisional Variscan magmatism in the Saxothuringian Zone of the Bohemian Massif (e.g. Hoffmann et al. 2013). It represents a giant collapse caldera. Recent publications suggest different lifetimes for this magmatic system, e.g. from ca. 324 to 300 Ma (Hoffmann et al. 2013;Casas-García et al. 2019;Tomek et al., 2022). According to Casas-García et al. (2019), the Teplice rhyolite was formed within a period between 325 and 317 Ma (the oldest dated Teichweg member yielded 323 ± 2 Ma while the youngest Cinovec member was dated 312 ± 4 Ma). Opluštil et al. (2016) dated the rhyolitic Třtěno ignimbrite by U-Pb CA-ID-TIMS on zircons at 313.41 ± 0.07 Ma. This pyroclastic density current is considered an extra-caldera equivalent of the Teplice rhyolite (Tomek et al. 2022). The Sayda-Berggießhübel dyke swarm (SBDS) was interpreted as a possible feeding system for caldera-related ignimbrites (Winter et al. 2008) and suspected to be older than the caldera collapse (Tomek et al. 2019). However, Schust (1980) and Wetzel (1984) proposed that the SBDS consists of three generations of pre-, syn-and post-caldera dykes. The microgranite ring dyke was considered as a cumulate-like magma from which the Teplice rhyolite was extracted. After the collapse of the caldera this magma intruded into feeder domains along the rims of the caldera (Müller and Seltmann 2002;Tomek et al. 2019Tomek et al. , 2022. Therefore, it is crucial to understand the temporal relationship between the SBDS, the Teplice rhyolite, and the microgranite ring dyke. In this study, we used high-precision zircon U-Pb CA-ID-TIMS dating to determine the ages of (i) the Teplice rhyolite (caldera stage) and the lifespan of the upper crustal Teplice magma body, (ii) the Sayda-Berggießhübel dyke swarm (SBDS), (iii) the microgranite ring dyke, and (iv) the precaldera stage (Schönfeld-Altenberg Complex). These data will allow us to better understand the sequence of magmatic activity in the Eastern Erzgebirge in time and space. It is also important for correlations with tuffs in Central Bohemian intramontane Permo-Carboniferous basins interpreted as outflows of the Teplice rhyolite, which are used as chronostratigraphic markers for the evolution of these sedimentary basins (Tomek et al. 2022). In addition, Hf-and O-isotope composition of zircons from the microgranite ring dyke is used to discuss possible source rocks for these magmas.

Geological setting
The Erzgebirge/ Krušné hory is part of the Variscan orogenic belt and is located at the northern margin of the Bohemian Massif (Fig. 1). It forms a NE-SW orientated crystalline complex along the German-Czech border (Fig. 1b). Petrological studies distinguished several units with different P-T histories (e.g. Rötzler et al. 1998;Schmädicke et al. 1995;Willner et al. 1997), which together represent a tectonic stack of different crustal segments formed during the Variscan subduction and continent-continent collision (e.g. Schulmann et al. 2009Schulmann et al. , 2014Kroner and Görz 2010). Some of the eclogite-and granulite-facies rocks were buried to ultrahigh pressure (UHP) conditions at mantle depth (evidence from coesite and diamond: e.g. Schmädicke 1991;Massonne 2003). The ages suggested for peak (U) HP metamorphism cover a range from 360 to 330 Ma (e.g. Schmädicke et al. 1995Schmädicke et al. , 2018Kröner and Willner 1998;Tichomirowa et al. 2005;Tichomirowa and Köhler 2013). Gneisses form the core while mica schists and phyllites are located at the northern and western rim of the Erzgebirge.
In a late stage of the Variscan orogeny (Late Carboniferous-Early Permian) mainly acidic magmatism affected the entire Erzgebirge region. This led to the emplacement of voluminous granites , dyke swarms like the SBDS, and to extended volcanic systems like the ATVC and TFC in the Eastern Erzgebirge (Tischendorf 1989).
The ATVC is an elongated ~ NNW-SSE trending volcanoplutonic complex (Fig. 2). About two-thirds of the ~ 36 × 18 km large trap-door caldera is exposed at the present-day erosional level (e.g. Moesta 1928;Breiter et al. 2001;Tomek et al. 2022). The ATVC hosts pre-and post-caldera plutons and a thick sequence of volcanic rocks. The ATVC rocks were grouped into three evolutionary stages (Casas-García et al. 2019). (i) The S-type biotite monzogranites of the Fláje Massif were formed as pre-collapse intrusions (Štemprok et al. 2003;Breiter 2012). They were followed by sedimentary rocks, (trachy-)dacitic pyroclastics and related subvolcanic bodies, and rhyolitic volcaniclastic successions called the Schönfeld-Altenberg Complex (SAC, Walther et al. 2016). (ii) The caldera formation is strongly related to the SBDS, the Teplice rhyolite, and microgranite ring dykes. Consequently, these rocks can be roughly assigned to the caldera stage (in a broad sense). Around the ATVC and towards the north, more than 350 dykes constitute the ~ ENE-WSW to ~ NE-SW and ~ NNW-SSE to ~ N-S-striking rhyolite dyke swarm (SBDS; Fig. 2a). The dykes reach lengths from a few metres up to several tens of kilometres, and widths from decimetres to several hundred metres (Winter et al. 2008). They do not continue into the interior of the caldera and are thus often interpreted as preceding the climactic stage (except for the intra-caldera N-S striking dykes; Mlčoch and Skácelová 2010;Tomek et al. 2019). The crystalline basement of the caldera was reached in several boreholes and dips at ca. 20-40° towards east (Moesta 1928;Benek 1991;Mlčoch and Skácelová 2010). The intra-caldera Teplice rhyolite (TR) body in the eastern part of the ATVC represents the most extensive outcrop of volcanic rocks (Fig. 2). The largest subsidence of the TR occurred in the southeastern part (Mlčoch and Skácelová 2010). The maximal thickness of TR was determined in a borehole near the city Teplice with 1033 m (Mlčoch and Skácelová 2010). Based on detailed field mapping, lithofacies analyses, stratigraphy, and whole rock geochemistry, the TR can be divided into nine lithostratigraphic units that represent three main explosive phases (e.g. Breiter et al. 2001;Casas-García et al. 2019, 2021Tomek et al. 2022). Intercalated volcaniclasticsiliciclastic horizons document quiet periods between these eruptive phases (Fig. 2b). The TR occurs mainly as crystalpoor to crystal-rich rhyolitic ignimbrites but includes some rhyolitic and rhyodacitic lavas as well as rhyolite fallout tuffs (e.g. Lobin 1986;Breiter et al. 2001;Casas-García et al. 2019). The rapakivi-textured porphyritic microgranites are considered as contemporary intrusions forming the caldera ring dyke (Müller and Seltmann 2002;Tomek et al. 2018; "microgranite ring dyke" in Fig. 2a). (iii) The A-type granites represent the post-tectonic collapse magmatism (Breiter 2012). These are highly evolved, F-rich, albite, Li-mica, post-caldera granites of Schellerhau, Krupka, Sadisdorf, Altenberg, and Zinnwald/ Cínovec (Romer et al. 2007;Štemprok 2016) which often contain Sn-W greisen cupolas due to mineralization.
The ATVC erupted ca. 320 km 3 dense rock equivalent of intra-caldera volcanic facies (Casas-García et al. 2019) and at least 30 km 3 DRE of extra-caldera ignimbrites and fallout tuffs (Tomek et al. 2022). Therefore, it represents a moderately sized caldera with moderate volumes of volcanics.

Previous geochemistry and discussion on source rocks
A lot of bulk rock geochemical data exists for ATVC rocks (e.g. Breiter et al. 2001Breiter et al. , 2012Müller and Seltmann 2002;Štemprok et al. 2003;Breiter 2012;Walther et al. 2016;Casas-García et al. 2019, 2021. Breiter (2012) distinguished two types of magmas: (i) strongly peraluminous P-rich (S-type) magmas that mainly form the early precaldera stage plutons (Fláje and Niederbobritzsch granites) but also the SAC, and (ii) slightly peraluminous P-poor magmas (A-type) that form the Teplice rhyolite and the porphyritic dykes. Consequently, the later second magma type is strongly correlated with volcanic activity. Breiter (2012) interpreted that these two magma types resulted from mixing of different source rocks (quartzo-feldspathic rocks and metapelites), variations of their P-T conditions, and their water contents. According to melt inclusion studies the ATVC magma accumulated in the middle crust (Thomas 1992;Müller et al. 2006). Various authors suggested mixing of different magmas based on variations of bulk rock chemistry and their mineral chemistry as well as core-rim structures of feldspar phenocrysts and quartz (e.g. Müller and Seltmann 2002;Müller et al. 2005;Breiter 2012). Although some authors proposed mixing of crustal magmas with a more basic melt (e.g. Müller et al. 2008;Seifert 2008), most authors exclude mixing with basic mantle-derived melts because of the exceptional scarcity of basic rocks in the Erzgebirge, the generally low contents of Mg, V, Ni, Co Cr in these igneous rocks, and the absence of mafic microgranular enclaves in the ATVC rocks (e.g. Breiter et al. 2012;Casas-García et al. 2021). Consequently, partial melting of quartzo-feldspathic metagreywacke, gneiss, metapelite, and intermediate igneous rocks have been proposed as the main source for the ATVC magmas (e.g. Breiter 2012;Breiter et al. 2012;Casas-García et al. 2021). The Cadomian basement of the Erzgebirge consists mainly of Late Neoproterozoic to Early Cambrian igneous and sedimentary rocks and Lower Paleozoic volcano-sedimentary successions. Most authors agree with the suggestion of Förster and Romer (2010) that an intensely layered, heterogeneous crustal pile with abundant fertile (qtz-fsp) lithologies (gneisses, greywackes), metapelites and subordinate mantle-derived intermediate to basic rocks that form the Cadomian basement can be considered as the most probable source rocks of Variscan igneous melts. This suggestion is based mainly on the Sr-, Nd-, and Pb-isotope compositions of these Variscan magmatic and pre-Variscan basement rocks (Förster and Romer 2010).

Samples
We present the results from 8 samples from the ATVC. Table 1 lists the sample names and localities. Figure 2 shows their locations and lithostratigraphic positions. From the pre-caldera stage, we analysed one sample (19Schön; from the Schönfeld ignimbrite). Three further samples were analysed from the Teplice rhyolite (Sö30,19Hart,19Lug). Sample Sö30 is part of the Teichweg member, and samples 19Hart and 19Lug represent the Lugstein-Pramenáč member of the TR (Fig. 2). From the microgranite ring dyke two samples were analysed (19Elend,Frau). Finally, for the SBDS, we analysed two samples (19Biene,19Turm).

Zircon separation
Zircons have been extracted from all eight samples by the usual procedure (crushing, Wilfley table, Frantz magnetic separator, heavy liquids, final handpicking). Selected zircons were characterized using secondary electron (SE) and cathodoluminescence (CL) images.
In general, we separated 150-200 grains from each sample, except from sample 19Turm, where we could find only 60 small grains (even after a second attempt of zircon separation). Then, for each sample ca. 30-50 zircon grains with the most idiomorphic forms (with sharp edges and even surfaces) were chosen from these overall zircon populations for further zircon dating. This was done to exclude xenocrystic zircons from further dating since the CA-ID-TIMS method is very time consuming and therefore we tried to avoid dating inherited zircons. The selected idiomorphic zircon grains usually show magmatic oscillatory zoning as shown in Fig. 3 and in the inset of Fig. 4 for sample Frau.

U-Pb Zircon dating by SHRIMP
Zircons from one sample (Frau) were analysed by the SHRIMP II technique (Sensitive High mass Resolution Ion MicroProbe) at the Centre of Isotopic Research (VSEGEI, St. Petersburg, Russia). Each analysis consisted of five scans through the mass range. The spot diameter was about 18 μm, and the primary beam intensity was about 4 nA. The data have been reduced like those used by Williams (1998,  and references therein), using the SQUID Excel Macro of Ludwig (2000). The zircon standard Temora 2 was used for reference of the U/Pb ratio and concentrations (Black et al. 2003). Corrections for common lead (Pb c ) were made using measured 204 Pb and by applying the Pb evolution model of Stacey and Kramers (1975). Uncertainties given for individual analyses (ratios and ages) are at the 1σ level, for calculated Concordia ages at the 2σ level.
To trace the accuracy of any dating method, secondary reference materials should be dated together with the unknowns (Horstwood et al. 2016). Our SHRIMP data were acquired in 2011. This is the reason why a secondary standard was not analysed by the SHRIMP method.

Zircon dating by CA-ID-TIMS
Selected zircon grains (ca. 30-50 per sample) were annealed for 96 h at 900 °C, and subsequently chemically abraded for 12 h at 210 °C with concentrated HF and HNO 3 in a pressurized Parr dissolution vessel. This procedure dissolves crystal domains with strong radiation damage which are suspected to have experienced post-crystallization lead loss (Mattinson 2005). Afterwards, the acid together with dissolved zircon material was completely pipetted out and 3.5 N HNO 3 was added to the remaining zircons fragments and left for 30 min at 50 °C to remove surface lead. Several cleaning cycles with water combined with repeated ultrasonic treatment were conducted before single zircon fragments were selected for further processing. Single zircon fragments were washed with 3.5 N HNO 3 and transferred into cleaned microcapsules with a small drop of this fluid and four drops of concentrated HF. Samples were spiked with a 205 Pb-233 U-235 U-tracer solution (ET535 at TU Bergakademie Freiberg, Condon et al. 2015). For final dissolution, the microcapsules were placed in pressurized Parr dissolution vessels and heated Zircon SEM-(upper row) and CL-images (lower row) for selected zircon grains from sample Frau. Note that the same grains are shown in the upper and lower row. For dating with CA-ID-TIMS the most idiomorphic grains were chosen (outlined grains 1 and 2) to avoid dating of xenocrystic zircons to 200 °C for 48 h, followed by drying at 130 °C and then re-dissolution in 6 N HCl for 24 h at 200 °C to transfer them into chlorides. After repeated drying, the samples were dissolved in ten drops of 3.1 N HNO 3 and transferred into micro-columns for column chemistry. U and Pb were separated from the rest of the sample by anion exchange chromatography using HCl and H 2 O. The U and Pb containing fraction was loaded on pre-degassed rhenium filaments with a drop of silica gel (Gerstenberger and Haase 1997) and measured with a Finnigan Mass Spectrometer MAT 262 using a secondary electron multiplier (SEM). Alternatively, the samples were measured on an IsotopX Phoenix Mass Spectrometer using a SEM Daly ion counter. The comparability of the results of both machines´ was proven by repeated measurement of zircon standards 91500 (Wiedenbeck et al. 1995) and Temora 2 (Black et al. 2004). The published ages of Temora 2 were determined as 416.8 ± 0.3 Ma (Black et al. 2003). Our date of 417.3 ± 0.6 Ma (data in Table 3 and Figures in Käßner et al. 2021) perfectly matches these values. Additionally, the accuracy of zircon U-Pb CA-ID-TIMS ages was monitored by dating the standard 91500. This standard was determined to be 1062.4 ± 0.4 Ma (Wiedenbeck et al. 1995) or 1063.6 ± 0.3 Ma (Schoene et al. 2006). Our weighted mean 206 Pb/ 238 U-age of 1064.6 ± 1.3 Ma (data in Table 3 and Figures in Käßner et al. 2021) is within 0.1% of the accepted values. Based on the results of standard dating, we presume the here presented CA-ID-TIMS ages to be accurate on the 0.1% level.

Zircon geochemistry: Hf-and O-isotopes
For sample Frau, in addition to in situ U-Pb zircon ages analysed by SHRIMP, the Hf-and O-isotope composition was determined from the same spots of these zircon grains. Hafnium isotopes were measured on a Thermo-Finnigan Neptune multi-collector ICP-MS coupled to a Resonetics 193 nm ArF excimer laser (CompexPro 102, Coherent) system at Goethe-University Frankfurt (GUF) (Gerdes and Zeh 2006). Spots of 40-60 μm in diameter were ablated with a repetition rate of 5.5 Hz and an energy density of 5 J/cm 2 during 55 s of data acquisition. All data were adjusted relative to the JMC475 standard ( 176 Hf/ 177 Hf = 0.282160; Vervoort and Blichert-Toft, 1999) and quoted uncertainties are quadratic additions of the within-run precision of each analysis and the reproducibility of JMC475 (2 SD = 0.0028%, n = 6). We verified the accuracy and external reproducibility by repeated analysis of the reference zircons, Temora and GJ-1. They yielded 176 Hf/ 177 Hf ratios of 0.282689 ± 0.000023 (2 SD, n = 11 for Temora), 0.282012 ± 0.000014 (2 SD, n = 8 for GJ-1). This is in perfect agreement with previously published results (e.g., Gerdes and Zeh 2006;Sláma et al. 2008) and with the LA-MC-ICP-MS long-term average (2006-2012) of GJ-1 (0.282010 ± 0.000025; n > 800) reference zircon at GUF.
Zircon oxygen isotopes were measured with the Cameca IMS 1280 multicollector ion microprobe at the Swedish Museum of Natural History (Heinonen et al. 2015), utilizing a ~ 2 nA Cs + primary ion beam together with a normal incidence low-energy electron gun for charge compensation, medium field magnification (~ 80x), and two Faraday detectors (channels L2 and H2) at a typical mass resolution of ~ 2500. Measurements were performed in pre-programmed chain analysis mode with automatic field aperture and entrance slit, centered on the 16 O signal. The magnetic field was locked using NMR regulation for the entire analytical session. Each data acquisition run comprised a 20 × 20 μm pre-sputter to remove the Au layer, followed by the centering steps, and 64 s of data integration performed using a non-raster, ~ 10 × 10 μm spot. In the measurement chain, every set of four unknowns was followed by two bracketing analyses on the 91500 standard zircon. A δ 18 O value of + 9.86 ‰ (SMOW, Wiedenbeck et al. 2004) was assumed for the 91500 zircon in data normalization, and small linear-drift corrections were applied to each session. The external reproducibility of ± 0.3 ‰ (1 SD) based on measurements on the standards was propagated into the overall uncertainty for each analysis. SHRIMP analyses for sample Frau in a Concordia (Tera-Wasserburg) diagram. Each analysis is represented by an ellipse with its 2σ analytical uncertainty. The dotted ellipse is considered as an outlier and is not included in the Concordia age calculation. The Concordia age is shown as a red ellipse with 2σ uncertainty (including decay constant uncertainty); MSWD is given for equivalence and concordance. One selected zircon grain from this sample is shown to illustrate the idiomorphic form and internal zonation: OM optical microscopy, CL cathodoluminescence

Zircon U-Pb age determinations by SHRIMP
The results of sample Frau are given in Table 2 and shown in a Concordia diagram (Fig. 4). The single measurements range from 301 ± 3 to 327 ± 4 Ma but except for the youngest age (301 Ma), all other dates gave a narrow range between 317 and 327 Ma (Table 2). A Concordia age of 322.5 ± 2.5 Ma (MSWD of concordance = 0.03) was calculated from nine spot analyses excluding the measurement with the youngest age (301 Ma, Fig. 4).

Zircon U-Pb age determinations by CA-ID-TIMS
Zircon U-Pb CA-ID-TIMS isotopic results for all eight samples are presented in Table 3 and shown as 206 Pb/ 238 U ranked age plots in Fig. 5. For each sample, 8 to 21 grains were analysed by this method. Mean sample ages representing the crystallization were calculated from established age clusters with the software ET Redux (Bowring et al. 2011). The error includes the internal 2σ measurement error, the tracer calibration uncertainty, and the uncertainty of the decay constant (z error according to Schoene et al. 2006).
From sample 19Schön (Schönfeld ignimbrite) 15 zircon fragments (left after the pre-treatment with chemical abrasion) were dated (Fig. 5a). Four single grain ages are distinctly older and interpreted as xenocrystic zircons (ca. 519-518 Ma, ca. 412 Ma, ca. 338 Ma). The remaining 11 single ages vary from 312.6 Ma to 325.0 Ma. Four measurements (analyses 1, 5, 9, mp3) form a cluster with identical ages between 321.6 and 322.6 Ma. These four measurements were used to calculate a weighted mean age of 322.1 ± 0.4 Ma (MSWD = 2.6). Although one measurement (analysis 12) overlaps with this age cluster we did not include it because of its large measurement error. Five further measurements gave younger ages (between 312.6 and 320.0 Ma) that we consider as Pb loss that was not completely erased by the CA-pretreatment. Widmann et al. (2019) and Tichomirowa et al. (2019a) have shown that the "standard" leaching time of 12 h is for many samples too short to completely remove Pb loss. The remaining slight Pb loss often results in a tail towards younger ages and can explain these five younger measurements. Analysis mp2 (ca. 325 Ma) is distinctly older than the calculated age cluster and dated an antecrystic zircon (zircons incorporated during intrusion from slightly older magma batches according to Miller et al. 2007). We consider the    Altogether, 21 single grain ages were determined from sample Sö30 (Fig. 5b). In the first attempt (analyses numbers 1-8 in Table 3) we applied the standard leaching time (12 h) during chemical abrasion. The single ages vary between 299.7 ± 0.4 and 318.5 ± 1.5 Ma (except for one distinctly older age at 340.7 ± 6.5 Ma). We suggested that the large age scatter towards younger ages resulted from remaining Pb loss. For this reason, we applied an additional leaching procedure (24 instead of 12 h, analysis numbers b-1 to b-9 in Table 3). The single ages from zircons leached 24 h show less scatter (from 306.4 ± 0.6 to 315.8 ± 0.3 Ma) and support our suggestion of Pb loss for the youngest single dates. However, the variation of these single ages is still too large to exclude an important impact of Pb loss and to ensure a correct mean age calculation. Therefore, we applied an additional leaching step (36 h instead of 12 h) to the remaining zircon fragments. From the 36 h leached zircons only four measurements were possible (analyses numbers c-1 to c-4) resulting in a very narrow range between 313.0 ± 0.4 to 314.5 ± 0.5 Ma again supporting the interpretation that the younger single ages are too young because of Pb loss in this sample. Six single measurements gave identical (within error) dates between 313.9 ± 0.4 and 314.5 ± 0.5 Ma (all from fractions b and c) and define a mean age of 314.2 ± 0.4 Ma.
From sample 19Hart, 13 single grain analyses were performed (Fig. 5b). From the first attempt (12 h standard leaching, numbers 1-4 in Table 3) we got variable single dates between 307.0 ± 0.5 Ma to 312.9 ± 0.3 Ma. We applied additional leaching (24 h instead of 12 h) to verify that Pb loss did not play a role for these unexpected young ages. The 24 h leached zircons (analyses numbers b-1 to b-9 in Table 3) recorded single ages between 313.2 ± 0.4 and 315.1 ± 0.5 Ma (except one xenocrystic age at ca. 607 Ma) and are thus older than all measurements from the first leaching step (12 h). Identical ages were obtained from five measurements (all from fractions b, i.e. 24 h leaching) resulting in a weighted mean age of 314.1 ± 0.4 Ma (MSWD = 0.7). One slightly older age (315.1 ± 0.5 Ma) is considered as an antecrystic zircon grain.
For sample 19Lug, 17 measurements were performed (Fig. 5b). After 12 h leaching, we got variable single ages from 307.6 ± 0.7 to 314.2 ± 0.4 Ma (analyses numbers 1-9 in Table 3). After applying additional 12 h leaching (24 h instead of 12 h) most dates vary between 311.5 ± 0.5 and 316.2 ± 2.4 Ma (except one older age at 329.1 ± 0.6 Ma considered as an antecrystic zircon grain). Six measurements yielded identical ages but we decided to exclude the measurement with the highest error (316.2 ± 2.4 Ma). The weighted mean age from the remaining five measurements results in 314.2 ± 0.5 Ma (MSWD = 0.3) considered as the Two samples were dated from the adjacent microgranite ring dyke. All age data of sample 19Elend cluster between 310.5 ± 0.4 and 313.6 ± 0.6 Ma ( Table 3: 11 analyses). The oldest five single dates yielded identical ages resulting in a mean age of 313.1 ± 0.5 Ma (MSWD = 1.0). A tail towards younger ages is considered as slight Pb loss (n = 6; Fig. 5c). The second sample Frau also shows a generally small variation of single ages (between 312.2 ± 0.7 and 315.2 ± 0.4 Ma; Table 3; Fig. 3c). One single slightly older age (315.2 ± 0.5 Ma) is considered as antecrystic zircon. From the remaining eight measurements the oldest four dates form a cluster with a mean age of 314.3 ± 0.4 Ma (MSWD = 0.03). The other four measurements form a tail towards younger ages implying slight Pb loss.
Two samples were dated from the SBDS. Twelve measurements were performed for sample 19Biene. After 12 h leaching, the single ages vary from 306.9 ± 0.2 Ma to 309.1 ± 0.4 Ma. These ages are significantly younger than for all other samples. A second leaching step (24 h instead of 12 h) was applied to reduce possible Pb loss. The three analyses from this data set yielded even slightly younger ages (304.7 ± 0.3 to 306.3 ± 0.3 Ma, Table 3: analyses numbers b-1 to b-3). Then, two groups of zircons were separated: small and large grains. For the small grains, 24 h of leaching was applied to ensure that the grains were not completely dissolved during this step. From the remaining parts of the zircons, only three measurements were possible yielding ages between 310.7 ± 1.1 and 311.0 ± 0.4 Ma (Table 3, analyses numbers bk-1 to bk-3). For the large grains, 36 h leaching was applied. The two measurements yielded ages of 308.5 ± 0.4 and 310.6 ± 0.6 Ma (Table 3: analyses numbers cg-1 and cg-2). The four oldest dates form a cluster with a mean age of 310.9 ± 0.5 Ma (MSWD = 0.4). All other measurements form a tail to younger ages considered as Pb loss (Fig. 5d).
In general, the zircon yield for the sample 19Turm was rather low and the zircons were small. Seven measurements yielded ages between 312.4 ± 0.4 and 314.3 ± 0.8 Ma with one significantly older age (323.5 ± 0.5 Ma) considered as Fig. 5 Single grain zircon analyses and 206 Pb/ 238 U weighted mean dates for analysed samples. Each horizontal bar represents a single zircon grain analysis including its 2σ analytical (internal) uncertainty. Grey bars are not included in the weighted mean calculation. Green vertical bars represent the weighted mean age, with the associated 2σ uncertainty. Uncertainty of the weighted mean dates is reported as ± x/y/z, with x-as 2σ internal, y-2σ external uncertainty including tracer calibration and z-2σ external uncertainty including 238 U decay constant uncertainty; MSWD mean square of weighted deviates ▸ a xenocrystic zircon. From newly separated zircons from the heavy mineral fraction, only one measurement was possible (analysis number N in Table 3) yielding an age at 313.9 ± 0.4 Ma (Fig. 5d). The four oldest dates form a cluster with a mean age of 313.8 ± 0.5 Ma (MSWD = 2.1).

Zircon geochemistry: Hf-and O-isotope ratios
The zircons from sample Frau have uniform Hf and very homogeneous O isotopic compositions (Table 4). The hafnium isotope ratio 176 Hf/ 177 Hf varies from 0.282570 to 0.282610 (corresponding to εHf(t) = − 0.6 to + 0.9 (t = 320 Ma)), and the oxygen isotope value (δ 18 O) varies from 6.0 to 6.6 ‰. Hf model ages range from 1.1 to 1.2 Ga.

Comparison of zircon ages from the same samples/locations obtained by different methods (LA-ICP-MS, SHRIMP, CA-ID-TIMS)
Dating of the same samples (or samples from the same location) with different zircon dating methods should result in identical ages. In this section, we discuss zircon ages obtained from the same sample locations while in the next section we compare our new ages with published ages from other localities of the ATVC.
Two  Fig. 7a). Therefore, only few selected measurements could be used for the Concordia age calculation. We suggest that this selection can be the reason for the age bias between the two different dating methods. The single zircon CA-ID-TIMS measurements of this sample (Sö30) showed the largest within sample variations due to the highest degree of Pb loss even after 12 h and 24 h leaching (Fig. 5b). Therefore, it is not easy to get an accurate zircon age for this sample without applying CA (as is the usual procedure for LA-ICP-MS dating).
Sample Frau was dated by SHRIMP (322.5 ± 2.5 Ma) and by CA-ID-TIMS (314.3 ± 0.4 Ma). These ages do not overlap and the difference is ca. 2.6% (8.2 Ma). The accuracy of our SHRIMP age cannot be assessed, because it was determined without a parallel measurement of a secondary standard in 2011. Thus, the uncertainty is only a statistical error, excluding external components. The absolute age resolution of SHRIMP has been estimated as 1-2% (Schaltegger et al. 2015). Tichomirowa et al. (2019a) compared CA-ID-TIMS with SHRIMP and LA-ICP-MS ages from the same samples of Variscan granites from the Western Erzgebirge. While single SHRIMP ages based on ca. 10 measurements can show deviations from CA-ID-TIMS ages (from 0 to ± 3%), the mean age of several samples from the same pluton (corresponding to > 30 SHRIMP data points) resulted in a SHRIMP age identical with CA-ID-TIMS data. Therefore, we suggest that the small number of single SHRIMP data points for sample Frau (n = 9) may explain the deviation to the CA-ID-TIMS age. . In a few cases distinct ages were published as the possible too young K-Ar ages by Müller et al. (2005), one too young Ar-Ar age (Seltmann and Schilka 1995), and one too old Pb-Pb zircon age for the microgranite ring dyke (Kempe et al. 1999).

Timescale of Variscan magmatic activity in the ATVC (pre-caldera and caldera stage)
The new zircon data show that there is a significant time gap of ca. 8 Myr between the pre-caldera phase and the caldera phase. In contrast to the supposed 11 Myr duration for the Teplice rhyolite (Casas-García et al. 2019), the CA-ID-TIMS data (our data: 314.2 ± 0.4 Ma, 314.1 ± 0.4 Ma, 314.2 ± 0.5 Ma, and that of Opluštil et al. 2016: 313.4 ± 0.4 Ma) suggest a much shorter lifetime of only 1-2 Myr for various members of the Teplice rhyolite. However, it should be noted that our three samples all represent the last TR eruption stage 3 (Fig. 2) while the lithostratigraphic position of the extra-caldera equivalent of TR is unclear (Tomek et al. 2022). Similar durations (1-2 Myr) have recently been obtained by CA-ID-TIMS ages for other Variscan plutonic and volcanic suites (e.g. Tichomirowa et al. 2019a;Lützner et al. 2021;Breitkreuz et al. 2021;Käßner et al. 2021).
According to the new age data, the microgranite ring dyke (314.3 ± 0.4 Ma, 313.1 ± 0.5 Ma) was formed approximately during the same time interval as the Teplice rhyolite. These data support the idea of the coeval formation of the microgranite ring dyke with the climactic stage of the Teplice ignimbrite as was supposed earlier (Müller and Seltmann 2002;Müller et al. 2005;Tomek et al. 2019Tomek et al. , 2022. Table 4 LA-MC-ICPMS Lu-Hf isotope data of zircon from sample Frau Quoted uncertainties (absolute) relate to the last quoted figure. The effect of the inter-element fractionation on the Lu/Hf was estimated to be about 6% or less based on analyses of the GJ-1 and Plesoviče zircon. Accuracy and reproducibilty was checked by repeated analyses (n = 30 and 20, respectively) of reference zircon GJ-1 and Plesoviče (data given as mean with 2 standard deviation uncertainties)

Timescale of Variscan magmatic activity in the Eastern Erzgebirge
According to the new data, the SAC (ca. 322 Ma) is roughly contemporaneous with the Flaje granite (Tomek et al., 2022;Fig. 6) and slightly older than Niederbobritzsch granite (ca. 320-318 Ma; Breitkreuz et al. 2021). The older age of the SAC implies a higher erosion level (at or close to surface) already at 322 Ma in the southern part of the Eastern Erzgebirge (Fig. 7). The SAC age dates the first appearance of (sub)volcanic (mainly dacitic) activity in the Eastern Erzgebirge at ca. 322 Ma representing the earliest preserved post-Variscan volcano-sedimentary rocks in central Europe. However, the main volcanic activity with the largest volumes of erupted volcanic rocks took place ca. 7-8 Myr later. Two calderas formed during this stage: the ATVC-caldera in the south (mainly at 314-313 Ma) and the Tharandt Forest caldera (ca. 315-312 Ma), which is located ca. 10 km north of ATVC (Fig. 7 (Breitkreuz et al., 2021, this study) for various Variscan igneous rocks. The geological map is based on Hoth et al. (1995) and Sebastian (2013) formed ca. 5-8 Ma earlier), could have caused a continuous increase in upper crustal temperatures and led to an elevated geotherm and produced a ductile halo in the host rocks in the pre-eruption level for the later voluminous eruptions. According to de Silva and Gregg (2014), this preheating of the upper crust is considered a major prerequisite for longliving caldera-forming magma reservoirs. The coeval age of the ring dyke system (ca. 314-313 Ma; two samples from this study and one sample from Breitkreuz et al. 2021) supports its genetic relationship to the climactic stage. These mainly subvertical ring dykes mark the rim of the ATVC caldera structure and mainly follow two directions: (i) especially in the south they form the NNW-SSE trending rims of the ATVC, but (ii) they also occur in NE-SW directions, south from the TFC (Fig. 7). The dykes from the SBDS (ca. 314 Ma and 311 Ma) mainly follow the same two directions (Fig. 7). Consequently, these dykes probably played an important role during and slightly after the climactic stage as supposed before (e.g. Tomek et al. 2019). Therefore, they can be interpreted (i) as the former magma chamber roof (as suggested by Christiansen 2001 for the Yellowstone Volcanic supereruption), (ii) as faults propagating downwards into the magma chamber from an uplifted and extended roof (de Silva and Gregg 2014), or (iii) as postclimactic and post-collapse dykes that erupted along faults bounding the flanks of a resurgent dome (Folkes et al. 2011). Mlčoch and Skácelová (2010) interpreted the block west of the Teplice rhyolite body up to the western ring dyke (they call it "Altenberg Block") as a roof pendant. Here, mainly gneisses from the basement crop out as well as remnants of low-grade Early Paleozoic rocks (phyllites/mica schists and metabasites) that occur also in the basement of the SAC. In this "Altenberg Block" the Teplice rhyolite has been eroded due to the eastward trap-door tilt and the downsagging of the caldera floor (Benek 1991;Tomek et al. 2019). Accordingly, the ring dykes most likely represent syn-climactic dykes propagating downwards into the magma chamber in the sense of de Silva and Gregg (2014).
Remarkably, during or slightly after the climactic stage of the ATVC and TFC, intense plutonic magmatic activity took also place further to the east along the NNW-SSE trending Intra-Sudetic line as evidenced by the large Karkonosze Pluton (Kryza et al. 2014: ca. 313-312 Ma) and by smaller granitic plutons in Lusatia (Käßner et al. 2021: ca. 313-312 Ma). Together with the dominant NNW-SSE trend of the ATVC and their porphyritic ring dykes this underlines the important role of the WNW-ESE extensional stress field during this period (Edel et al. 2018).
Until now precise CA-ID-TIMS dates for the post-caldera granite magmatism of the ATVC (evolutionary stage 3 of the ATVC rocks according to Casas-García et al. 2019) are missing. The magmatic activity may have lasted until about 305 Ma (according to the age compilation done by Tomek et al. 2022;their Fig. 12), although contradictory ages still do not allow a robust time to estimate for these post-collapse igneous rocks.

Possible source rocks of Variscan magmatic rocks: what information can Hf-and O-isotopes from zircons add?
For sample Frau, Hf-and O-Isotope values were determined on the same spot locations where the SHRIMP ages were produced. This ensures that all analyses are determined from zircons formed in the Variscan melt (no xenocrystic ages). Zircon is an early crystallizing mineral in granites and very robust against late-stage (e.g. hydrothermal) overprints. In contrast to bulk rock chemistry, 176 Hf/ 177 Hf isotope ratios in zircons are not affected by fractionation processes and hence represent the melt composition (Chen and Zheng 2017). Melt production always results in increased Hf-isotope compositions in newly formed zircons compared to the source (Chen and Zheng 2017). For our samples, all Hf isotope analyses show a remarkable homogeneity in their epsilon values but also in their δ 18 O values (Fig. 8). Bindemann and Valley (2002) have shown that homogenous δ 18 O values are common for long-lived voluminous pyroclastic eruptions. They interpreted this large-scale δ 18 O melt homogeneity as evidence for magma chambers longevity and convection. In turn, the convective stirring of large magma bodies requires underplating and/or recharge of magmas over long periods. From geophysical measurements (seismic measurements, Bouger gravity low, e.g. Conrad et al. 1994) as well as from many drill cores (Mlčoch and Skácelová 2010;Berger et al., 2011) it is suggested that Variscan magmatic rocks continue for several kilometers downwards in the Erzgebirge. In this context, it should be noted that granulitic and small scale amphibolitic melts were produced in the gneisses already during the Variscan metamorphism (Tichomirowa et al., 2018). Thus, the preceding Variscan metamorphism could contribute to an increase in the upper crustal geotherm in the Erzgebirge.
Although Hf-O studies on zircons from the Bohemian Massif are still limited, zircons from the microgranite ring dyke plot in a specific narrow field (Fig. 8) that has slightly higher epsilon Hf values compared to Variscan granites from the Western Erzgebirge, the Koenigshain granite from Lusatia, and to zircons with the highest values from rhyolites from the NE German basin ( Fig. 8; Pietranik et al. 2013;Słodczyk et al. 2018;Tichomirowa et al. 2018Tichomirowa et al. , 2019bKäßner et al. 2021 (Fig. 8, "af melting" arrow). However, these metamorphic quartzo-feldspathic rocks still have higher δ 18 O values (ca. 7‰) compared to those from the ring dyke (ca. 6‰). Therefore, in addition to these fertile metamorphic rocks another source with low δ 18 O values is necessary to achieve the oxygen isotope composition of the ring dykes. Basic melts are the most probable second source for the Frauenstein microgranite. A mafic melt input supports the concept of de Silva and Gregg (2014) that requires recharge/s of the magma chamber to keep magma temperatures high enough and to prevent viscous magma death. Eventually, this magma recharge can promote roof uplift and final eruption of such crystal-rich magmas like the microgranite ring dyke. In detailed studies that compare zoning of feldspar phenocrysts in the Teplice rhyolite and the microgranite ring dyke, Müller and Seltmann (2002) and Müller et al. (2005) have shown that clear indications for an input of basic magma are found only in the latest stages of the Teplice rhyolite (stage "TR3c") as well as in the microgranite ring dyke. In the ring dyke, the second stage of microgranite intrusion is chemically less evolved compared to the first intrusion stage, again confirming an input of a basic melt (Müller and Seltmann 2002). In contrast, Casas-García et al. (2021) disprove a basic melt input for the Teplice rhyolite and the microgranite ring dyke mainly on the base of homogeneous Nd and Pb isotope fingerprints that "represent intermediate ranges between crustal and mantle components of the Cadomain and Lower Paleozoic basement". The zircon Hf model ages of the microgranite ring dyke (Table 4: 1.2-1.1 Ga) agree with Nd model ages (T DM2 : 1.3-1.2 Ga) from many other rocks of the ATVC (Walther et al. 2016;Casas-García et al. 2021) and their homogeneous ε Nd -values (-1.9 to -3.7; Casas-García et al. 2021). These Nd model ages are younger than those of Cadomian basement rocks from the Erzgebirge (Kröner et al. 1995;Tichomirowa et al. 2001;Mingram and Rötzler 1999: grey gneisses = 1.8-1.6 Ga, Ordovician red gneisses: 1.6-1.5 Ga). No Nd isotope data were published for Paleozoic sedimentary rocks (shale, arkose) from Saxothuringia of the Bohemian Massif. Cambro-Ordovician shales and pelites from other regions of the Variscan belt (Iberia, French Massif Central) have typical Nd model ages between 1.8 and 1.4 Ga (Nägler et al. 1995;Simien et al. 1999;Vila and Pin 2016) that are distinctly older compared to the analysed Variscan granites and rhyolites. Unfortunately, no Nd model ages were published for metamorphic melts in the Erzgebirge, but their zircon Hf model ages are 1.3-1.2 Ga (Tichomirowa et al. 2018). If metamorphic quartzo-feldspathic rocks were the major source of caldera-stage ATVC melts, then the slightly younger model ages require a minor input from a juvenile source (e.g. mantle melts). Indeed, old xenocrystic zircons (> 350 Ma) are scarce for these rocks and the very few xenocrystic ages between 323 and 341 Ma (Table 3) support the assumption that metamorphic quartzo-feldspathic  Tichomirowa et al. (2018). Abbreviations: af-melting-amphibolite facies melting, gf-melting-granulite-facies melting according to Tichomirowa et al. (2018) rocks could be the dominant source. In contrast, the earlier pre-caldera Flaje granite and SAC volcanics have higher abundances of xenocrystic pre-Variscan zircons (Hoffmann et al. 2013;Tomek et al. 2022; Table 3) pointing to another dominant source with a higher contribution of pre-Variscan basement rocks for their melts.

Conclusions
The obtained high-precision U-Pb zircon ages document a substantial time gap (7)(8) between the pre-caldera Schönfeld-Altenberg Complex (SAC) and the main volcanic caldera stage of the ATVC. The volcanic rocks of the SAC represent the earliest volcanic activity in the Erzgebirge and central Europe at ca. 322 Ma, indicating a (sub)surface erosion level in the Southern Erzgebirge already at this time. According to the new high-precision zircon data, the Teplice rhyolite body formed during a relatively short time interval between 314 and 313 Ma. At the same time, the microgranite ring dyke as well as (at least) part of the rhyolitic dykes of the SBDS were formed. Accordingly, the new age data support the previously suggested close genetic and temporal relationship between the TR, the microgranite ring dyke, and the SBDS. The age of the ring dyke (314-313 Ma) can now be considered as the best estimate for the climactic stage of the ATVC, as the ring dyke is generally accepted as a syncollapse intrusion.
Compared to Variscan igneous rocks of the northern part of the Eastern Erzgebirge, the SAC formed slightly earlier (ca. 322 Ma) than the deeply intruded Niederbobritzsch granite (320-318 Ma). Remarkably, the main magmatic activity leading to the formation of calderas in the ATVC and the Tharandt Forest was mainly contemporaneously (at ca. 314-312 Ma). These data document a large and long-standing upper crustal silicic magmatic system in the whole Eastern Erzgebirge stretching about 60 km along a NNW-SSE trending zone that is parallel to the Elbe Fault Zone. Possibly, the preceding Variscan metamorphism (ca. 340 Ma) as well as the pre-climactic plutons (Niederbobritzsch, Flaje: ca. 322-318 Ma) and volcanics (SAC: ca. 322 Ma) played an important role in a continuous increase of the upper crustal temperature and thereby affected the thermal and mechanical behaviours of the upper crustal host rocks as supposed as a necessary prerequisite for later voluminous and long-living magma systems (e.g. Folkes et al. 2011;de Silva and Gregg 2014).
We present the first zircon Hf-O-isotope data from the microgranite ring dyke of the ATVC. These data confirm the generally accepted crustal origin of the Variscan melts. However, the zircon Hf-O isotope ratios from the widespread occurring quartzo-feldspathic Cadomian basement rocks (gneisses) are too variable and too different to assume that these rocks could be the dominant source rocks. Consequently, preceding homogenization by melting is a prerequisite. The exceptional homogeneity of O isotopes can be interpreted as evidence for long-lasting convective stirring in the magma body. Smalls scale in situ melts during Variscan amphibolite-facies metamorphism produced fertile quartzofeldspathic rocks with already pre-homogenized Hf-O isotope ratios and with model ages of about 1.2-1.1 Ga and are probably the dominant source of microgranite ring dyke melts. In addition, a second and more juvenile source (basic melts) is necessary to explain the low oxygen isotope ratios of the crystal-rich microgranite syn-climactic ring dyke.