The Spatial and Temporal Evolution of the Sadisdorf Li-Sn-(W-Cu) Magmatic-Hydrothermal Greisen and Vein System, Eastern Erzgebirge, Germany Open Access

Granite-related hydrothermal greisen deposits host a significant share of global Sn and W resources (Lehmann, 2021) and commonly occur within orogenic belts such as the Central Andean belt (Bolivia), the Yanshanian orogen (China), and the Variscan orogenic belt in Europe (Walshe et al., 1995; Mlynarczyk and Williams-Jones, 2005; Hou and Zhang, 2015). In addition to Sn, W, and minor base metals, greisen deposits host significant Li resources in Li-bearing micas (e.g., zinnwaldite and lepidolite series; Fedkin et al., 2001; Gourcerol et al., 2019).

The expected increase in demand and industrial relevance of Sn, W, and particularly Li (European Commission et al., 2020; USGS, 2022) has led to a recent resurgence in global exploration targeting Li-Sn-W greisen systems. Many greisen systems exhibit a distinct mineralogical and geochemical zoning—similar to magmatic-hydrothermal porphyry, skarn, and some epithermal systems (Meinert et al., 2005; Sillitoe, 2010; Burisch et al., 2019)—that usually includes Sn + W + Be + Bi ± Mo mineralization proximal to fertile granitic intrusions, whereas Cu + Pb + Zn ± Ag mineralization prevails in more distal environments. Examples with well-documented zonation include San Rafael, Peru (Mlynarczyk et al., 2003), Cligga Head, United Kingdom (Moore and Jackson, 1977), the Mole granite, Australia (Audétat et al., 2000), and the Sadisdorf deposit, Germany (Müller, 1887; Seltmann, 1984).

Cooling of the ore fluid along its fluid flow path has traditionally been assumed to cause the observed mineral zoning (Kelly and Turneaure, 1970; Liu et al., 2016). Admixture of meteoric fluids, fluid-rock interaction (Halter et al., 1998; Nambaje et al., 2020; Jiang et al., 2022) and phase separation (Sushchevskaya and Bychkov, 2010; Korges et al., 2018) may also play an important role. Yet direct evidence to elucidate the relative influence of these very different processes in greisen systems is scant.

The Sadisdorf prospect in the Erzgebirge/Krušné hory province (Erzgebirge hereafter) in Germany is an ideal case study to investigate how mineralogical zoning in greisen systems relates to the temporal and spatial evolution of the underlying magmatic-hydrothermal system because of its comparatively small size (~750 × 1,200 m), the preservation of both proximal pervasive greisen-type and distal vein-type mineralization, and the abundance of available drill core material. Apart from rather generic mineralogical and petrological descriptions and recent resource reports published by exploration companies, not much modern data is available for the Sadisdorf deposit. Thomas (1994) and Seltmann (1995) reported homogenization data for fluid inclusions in quartz and cassiterite for a wide range of temperatures (677° to ~100°C). However, these data sets must be regarded with caution because the concept of fluid inclusion assemblages (FIAs; Goldstein and Reynolds, 1994) was not used, and the petrographic context of reported fluid inclusions is poorly documented in both studies. Even on the province scale, the amount of modern fluid inclusion data for greisen deposits in the Erzgebirge is limited to the Hub stock (Dolníček et al., 2012), Geyer (Meyer et al., 2024), and Zinnwald (Korges et al., 2018). These studies provide important insights into the temporal evolution and ore forming processes in greisen systems of the Erzgebirge but offer limited clues with which to understand mineral zoning because they focus exclusively on the most proximal parts of the systems (endogreisen).

In this contribution, we integrate petrographic observations, fluid inclusion, and whole-rock geochemical analyses of spatially constrained samples from 12 drill cores from the Sadisdorf deposit as a case study to better understand (1) the magmatic evolution leading to magmatic-hydrothermal mineralization, (2) the alteration footprints and effects on element mobility, (3) the temporal and spatial fluid evolution, (4) the mineral/element zonation, and (5) the underlying mechanisms controlling ore formation in a shallow greisen system. It is expected that the insight obtained can be transferred to the exploration for buried greisen deposits well beyond the geographic limits of the Erzgebirge.

The Erzgebirge/Krušné hory region (Fig. 1A) is a classic metallogenic province that hosts late-Paleozoic magmatic-hydrothermal systems such as greisen (Štemprok, 1967; Baumann and Tischendorf, 1976), skarn (Hösel, 2003; Burisch et al., 2019; Reinhardt et al., 2021), and epithermal vein-type mineralization (Swinkels et al., 2021) as well as Mesozoic hydrothermal five-element– and fluorite-barite–vein systems (Kuschka, 1972; Guilcher et al., 2021; Haschke et al., 2021; Burisch et al., 2022). The Erzgebirge comprises a series of metamorphic nappes that developed during the Variscan orogeny in Central Europe (i.e., the collision of Laurussia and Gondwana; Kroner et al., 2007; Linnemann and Romer, 2010). The metamorphic succession represents a series of sedimentary and igneous protoliths of Neoproterozoic to early Paleozoic age. Peak metamorphism for the different tectonometamorphic units occurred between 360 and 340 Ma (Kröner and Willner, 1998; Schmädicke et al., 2018; Collett et al., 2020) and was followed by rapid exhumation and synkinematic intrusion of granites to diorites aided by transtensional tectonics (Wenzel et al., 1997; Linnemann and Romer, 2010). The post-collisional stage (in older literature often referred to as late Variscan; e.g., Breiter et al., 1999) was marked by peraluminous magmatism (327–310 Ma; Romer and Meixner, 2014; Tichomirowa et al., 2019, 2022) and emplacement of voluminous, mostly S- and A-type syeno- and monzogranite batholiths, stocks, evolved porphyries, lamprophyres, and extensive rhyolitic and dacitic volcanic rocks (Tischendorf and Förster, 1990; Breiter et al., 1999; Förster et al., 1999). Later post-orogenic extensional tectonics were accompanied by (sub-) volcanic felsic magmatism between 305 and 286 Ma (Hoffmann et al., 2013; Luthardt et al., 2018; Zieger et al., 2019).

In the eastern Erzgebirge, the Altenberg-Teplice Caldera (ATC) offers the most extensive record of post-collisional magmatism. This magmatic complex is composed of caldera infill comprising rhyolitic and dacitic volcanic rocks, including ignimbrites, tuffs, and lavas that were deposited on top of metamorphosed Variscan basement rocks and the pre-caldera Fláje monzogranite intrusion (~328–317 Ma; Breiter, 2012; Tomek et al., 2019; Casas-García et al., 2019; Tomek et al., 2021). The ATC forms an elliptical structure (~35 × 18 km) delineated by a late-caldera porphyritic microgranite ring dike, which intruded in response to a trapdoor-style collapse of the caldera into the volcanic units (Tomek et al., 2019). Despite conflicting ages, the latest data suggest that the caldera complex formed between 325 and 310 Ma (Casas-García et al., 2019; Tomek et al., 2019, 2021; Tichomirowa et al., 2022). Biotite-topaz–bearing granitic rocks and alkali-feldspar Li-rich subvolcanic granite porphyries intruded after caldera collapse. The latter are closely associated with prominent Sn-W-(Li) mineralization, including in the Krupka, Cínovec-Zinnwald, Altenberg, and Sadisdorf localities (Fig. 1A). The ages of these deposits (327–310 Ma; Romer et al., 2007; Seifert and Pavlova, 2016; Ackerman et al., 2017; Zhang et al., 2017; Leopardi et al., 2024) suggest a close temporal relationship with the caldera development.

The Sadisdorf Li-Sn-(W-Cu) prospect is in the NE part of the ATC. Exploitation commenced in the 16th century with underground mining starting around 1638. After pausing for most of the 18th century, underground mining was reinitiated during the first half of the 19th century (Müller, 1887). The latest active mining period took place between 1937 and 1954 (Seltmann, 1984). Since then, the Sadisdorf prospect has been repeatedly explored for Sn and Li and for Mo and Cu as well. The most recent resource estimate is that the deposit contains 25 million tons (Mt) of ore at an average grade of 0.45% Li₂O (Lithium Australia, 2017) and 3.36 Mt of ore at an average grade of 0.44% Sn (Lithium Australia, 2017). The exploration license of Sadisdorf is currently held by Deutsche Lithium GmbH (as of June 2024).

The oldest rocks in the Sadisdorf prospect (Fig. 1B) are Neoproterozoic para- and orthogneisses (Seltmann, 1984). The basal (tectonostratigraphically lowest) part of the sequence consists of paragneiss (metagraywackes, locally referred to as two-mica and biotite gneisses) with intercalations of amphibolite lenses and minor quartzite layers. This succession outcrops in the NW and SE of the study area (Fig. 1B) and has also been encountered at depth in several drill cores. The central portion of the study area is mainly composed of a succession of migmatitic metavolcanic rocks (muscovite gneiss and schist, locally referred to as muscovite gneiss) and metagraywackes (muscovite-biotite-plagioclase gneiss, locally referred to as “dense gneiss”). These rock types outcrop along a NE-SW-trending belt (Fig. 1B; Seltmann, 1984).

Toward the southwest of the study area there are post-Variscan sandstones and conglomerates related to the early stage of the ATC. These sedimentary rocks unconformably overlie the metamorphic basement rocks along a NE-SW-striking structure (Fig. 1B). They are covered by the ~323 to 300 Ma old (Hoffmann et al., 2013; Casas-García et al., 2019; Tomek et al., 2021; Tichomirowa et al., 2022) syn-caldera Teplice rhyolite (Seltmann, 1984).

The Sadisdorf composite granite (G1–G4; Figs. 1B, 2) intruded both the metamorphic basement and the Teplice rhyolite. Early intrusive stages comprise biotite syenogranite (G1; 280 × 210 m in size) followed by a ring-like biotite monzogranite (G2) emplaced along the northern contact of G1 (Fig. 2B, C). Finger-like apophyses of G2-like composition emanate radially from this intrusion (Seltmann, 1984; Breiter et al., 1999). Following the G2 intrusion, a small (~70 × 60 m) porphyritic monzogranite (G3) intrusion was emplaced along the G1 and G2 contact (Fig. 2B, C; Seltmann, 1984; Breiter et al., 1999). These early granitic intrusions (G1–G3) are texturally similar to the early intrusive phases (SG1–SG2) of the post-caldera Schellerhau granite, which is 3.5 km southeast of the Sadisdorf granite (Müller et al., 2000). At depth, the syeno- and monzogranites (G1–G3) are intruded by a larger (~430 × 340 m) alkali-feldspar microgranite (G4; Fig. 2B, C; Seltmann, 1984). This alkali-feldspar microgranite is assumed to be the fertile intrusion that caused magmatic-hydrothermal mineralization at Sadisdorf. However, because absolute ages are currently not available for any of the four intrusions that define the Sadisdorf magmatic system, this genetic relation remains speculative.

Based on melt inclusion analyses, Thomas (1994) proposed that G4 was emplaced at paleodepths between 2.6 and 1.5 km, forming a cupola (Fig. 2B, C) with steep flanks. It is separated from the surrounding country rocks by a quartz-dominated pegmatitic carapace (i.e., stockscheider; Müller, 1887; Seltmann, 1984) of variable thickness (0.05–5 m; Fig. 2B, C). The pegmatitic carapace consists mostly of massive quartz with abundant unidirectional solidification textures of quartz crystals in a fine-grained granitic matrix, and locally coarse-grained Li-Fe mica and K-feldspar nests in the apical part of the cupola. The quartz-dominated pegmatite transitions into more feldspar-rich zones toward the flanks of the cupola. Below the contact with the pegmatitic carapace, the G4 granite is variably greisenized. Zones of intense greisen alteration overlap with and eventually transition into feldspathized zones dominated by albite ± K-feldspar at depth (Seltmann, 1984). The extent of the feldspathized zone at depth is unknown owing to the lack of deep drilling. A complex pipe-like magmatic-(hydrothermal) breccia body extends above the apex of the G4 granite toward the surface; it is exposed in the old mine workings (Seltmann, 1984; Märtens et al., 1992). Toward the NNW-SSE this breccia body grades into cataclastic gneiss (Märtens et al., 1992). The upper portions of the breccia mostly contain clasts of gneiss and quartz, whereas granite clasts (G1–G3) dominate in the deeper portions. Based on the nature of the clasts, some authors (Seltmann, 1984; Märtens et al., 1992) have speculated that the breccia pipe formed in discrete pulses following each intrusive event.

A series of NE-SW–trending rhyolitic porphyry dikes (G5) represent the youngest magmatic event in the area (Fig. 2). These rhyolitic dikes not only crosscut the metamorphic host rock and the G1 to G4 phases, but also the magmatic-hydrothermal mineralization. It is assumed that these dikes belong to the youngest stage of the extensive Sayda-Berggießhübel dike swarm, which strikes northeast-southwest across the eastern Erzgebirge and postdates the mineralization (Müller, 1887; Seltmann, 1984; Wetzel, 1984).

The bulk of the magmatic-hydrothermal mineralization of the Sadisdorf deposit occurs in two distinct sites identified as Kupfergrube and Zinnklüfte (Fig. 2A). These two sites are separated by a small valley striking north-northeast to south-southwest. Mineralization around Kupfergrube is centered on the surface exposure of the composite Sadisdorf intrusion described above. The Zinnklüfte area, on the other hand, is marked by the occurrence of discrete, SW-NE–trending vein/stockwork zones that can extend up to 800 m horizontally along strike (Fig. 1B).

The mineralization at the Kupfergrube site occurs as: (1) Li rich mica, cassiterite, and wolframite disseminated along the greisen-altered portions of the microgranite (G4; endocontact mineralization); (2) disseminated to semi-massive wolframite within greisen-altered portions of the pegmatitic carapace; and (3) cassiterite-wolframite to Cu-Zn-Pb-As sulfide veins and stockworks that occur around the G4 granite intrusion in earlier intrusive phases (G1–3), metamorphic country rocks, and the breccia pipe above the G4 granite intrusion (exocontact mineralization; Seltmann, 1984). The veinlets define a stockwork zone that extends from the apex of the G4 intrusion to their present-day surface within 40 to 60 m of the G4 contact.

In the Zinnklüfte area, the mineralization lacks a clear spatial relationship to a host intrusion, although granitic rocks have been intersected in two drill holes at ~380 m below the exposed mineralization. The mineralization occurs as millimeter-thin cassiterite- and/or sulfide-bearing quartz veinlets forming stockworks and isolated breccia bodies associated with three NE-SW–trending structures (Fig. 1B). Table 1 presents a historical classification of the mineralization by previous authors (Müller, 1887; Seltmann, 1984).

There are minor cassiterite-bearing veins at Hoher Hau (Fig. 1B) and a series of Cu-Pb-Zn veins (locally associated with Sn mineralization) and younger barite-fluorite veins across the area (e.g., at Eulenwald, Perlschacht, Mittlere Löwe, and Erzengel; Fig. 1B; Müller, 1887; Seltmann, 1984). These minor occurrences were not considered in this study.

Previous geochronological age data were limited to the Kupfergrube 219 site with a wide range of ages from 326 ± 8 Ma (wolframite-fluorite U-Pb thermal ionization mass spectrometry [TIMS]; Kempe and Belyatsky, 1997), 326.1 ± 3.4 Ma (cassiterite U-Pb laser ablation-inductively coupled plasma-mass spectrometry [LA-ICP-MS]; Zhang et al., 2017), and 321.4 ± 3.8 Ma (Re-Os on molybdenite; Ackerman et al., 2017) to 310.5 ± 3.5 Ma (Ar-Ar on Li-mica; Seifert and Pavlova, 2016). However, Leopardi et al. (2024) recently provided robust U-Pb LA-ICP-MS ages of cassiterite ranging from 315 to 311 Ma, which indicate that the mineralization at the Zinnklüfte and the Kupfergrube sites occurred at the same time. These new ages relate ore formation at Sadisdorf to magmatism in conjunction with the Teplice-Altenberg Caldera collapse at around 314 to 313 Ma (Tichomirowa et al., 2022).

A total of 143 samples from the Kupfergrube and Zinnklüfte sites were selected to provide systematic transects from proximal to distal portions of the magmatic-hydrothermal system at the Sadisdorf prospect (Fig. 2). The samples include greisen- (microgranite- and pegmatite-hosted) and vein-style mineralization. Details of the location and descriptions of the samples are provided in Appendix Table A1. Most of the samples used in this study (74 samples) were selected from 12 drill core intercepts (labeled by number of drill core plus two-digit year; Fig. 2) from the Geological Survey of the State of Saxony (LfULG). These drill cores originated from several exploration campaigns of the Sowjetisch-Deutsche Aktiengesellschaft Wismut (Soviet-German Trading Company) in the late 20th century. The drill core samples were complemented by 24 samples (SD-DL samples; Fig. 2A) collected from surface exposures and old mining pits; an additional 24 samples (five-digit numbers; Fig. 2) were taken from the geoscientific collections of the TU Bergakademie Freiberg (TUBAF) and the archives of the LfULG. Comprehensive descriptions of the historical sample collection allow for their precise localization with a geospatial uncertainty of less than 20 m.

A total of 134 polished thin sections were produced at the Helmholtz Institute Freiberg for Resource Technology (HIF). The mineralogical relationships and textures were determined using a Carl Zeiss Axio Imager M1m polarized light microscope (transmitted and reflected light) equipped with an AxioCam MRc5 camera for documentation. Optical microscopy was complemented by scanning electron microscopy (SEM) and mineral liberation analysis MLA), and cathodoluminescence microscopy. The SEM was performed using an FEI Quanta 600 MLA-FEG at the Department of Mineralogy at the TUBAF and an FEI Quanta 650F MLA-FEG at the HIF.

The bulk chemistry of 20 granitic rock samples of was analyzed by Activation Laboratories Ltd. (Actlabs, Ancaster, Canada). The analyzed drill core samples include feldspar-altered microgranite and rhyolitic dike. Core samples were complemented with pulp samples from two drill holes (BH 21–86 and 20–86) that intersect the syenogranite (G1), monzogranite (G2–3), feldspathized microgranite (G4), and late rhyolitic dike (G5) rock units. Samples were fused with lithium borate, followed by weak nitric acid dissolution, and were analyzed for major and trace elements using ICP-MS, except for Cl and F, which were analyzed by instrumental neutron activation and by ion-selective electrode potentiometry, respectively. Additional measurement and processing parameters, as well as information on analyzed reference material, are provided in Appendix Table A2.

A subset of 24 samples containing suitable fluid inclusions (FIs) was prepared as doubly-polished thick (150–250 µm) sections at the HIF. Microthermometry was done using an Olympus BX53 microscope equipped with a QImaging Retiga 2000R CCD HD camera and a Linkam THMSG600 heating-freezing stage at TUBAF. FIAs were classified as primary (p), secondary (s), pseudosecondary (ps), and clusters (c) based on the criteria of Goldstein et al. (2003). To ensure reproducibility, the eutectic (T_e; i.e., the temperature of observed first ice melting), melting (T_m), and homogenization (T_h) temperatures were measured three times, respectively, with an estimated uncertainty of ±0.1°C for temperatures below 20°C and ±0.3°C for temperatures above 20°C. Phase transitions were observed during heating at a rate of 1°C/min. The reported results represent the average of the three measurements. The heating-freezing stage was calibrated regularly using synthetic H₂O and H₂O-CO₂ fluid inclusion standards. Fluid inclusions displaying evidence of postentrapment modification, metastable behavior, or anomalous T_h variations (>30°C) for the same liquid-vapor fractions were excluded from the dataset.

Most of the analyzed FIs show a range of T_e between –23.0° and –16.8°C, thus their compositions are best described by the H₂O-NaCl-(±KCl) system. Salinities were calculated using T_m(ice) and the Excel spreadsheet of Steele-MacInnis et al. (2012). The salinity of halite-bearing inclusions was calculated from their T_m(halite) and total homogenization temperature (T_{h tot}) using the equations of Sterner et al. (1988), whereas the equations of Bodnar (1993) were used for two-phase FI T_m(ice). Some FIs had eutectic temperatures between –52° and –50°C, indicating the presence of CaCl₂; their salinities were calculated as recommended by Steele-MacInnis et al. (2011). Most CO₂-bearing FIs contain relatively small amounts of CO₂, low enough to not show a (visually distinct) liquid CO₂ phase at room temperature, but high enough to be detected by Raman spectroscopy, suggesting CO₂ contents <1.5 mol % (Hedenquist and Henley, 1985; Diamond, 2001). A small proportion of FIs display clathrate melting upon heating. Salinities of CO_2-bearing inclusions were calculated assuming an H₂O-NaCl system and the equations of Bodnar (1993). Thus, the reported salinities should be considered as maximum values because the formation of clathrate consumes H₂O, yielding higher salinities for the remaining fluid (Hedenquist and Henley, 1985).

Uncertainty associated with the available emplacement depth estimates (1.5–2.6 km; Thomas, 1994) and with potential transient pressure states (see discussion) prevents an accurate pressure correction of homogenization temperature. Therefore, temperatures that are presented and discussed in this contribution are uncorrected. For reference purposes, pressure-corrected values for lithostatic and hydrostatic regimes at the maximum depth of emplacement (Bodnar and Vityk, 1994; Steele-MacInnis et al., 2012) are collated for all FIs in Appendix Table A4.

A total of 24 fluid inclusions from five samples (eight FIAs) were analyzed qualitatively using a Horiba Jobin Yvon LabRam HR800 Raman spectrometer coupled with an Olympus BX41 microscope at the HIF. All phases (solid, liquid, and vapor) were analyzed separately, if possible. Fluid phase and included mineral spectra measuring parameters (i.e., acquisition time, spectral grading, and accumulation), were optimized through XYZ profiles across each inclusion to maximize the target phase/host signal ratio. Measurements were performed using 472- and 632-nm lasers. For details of the acquisition parameters and results, see Appendix Table A5. Phase identification was done by comparing measured spectra with spectra from a reference library (Kingma and Hemley, 1994; Frezzotti et al., 2012, and references therein).

Three stages of alteration were recognized within the G4 granite (Fig. 3A-D): (1) sodic-potassic alteration (feldspathization), (2) greisen alteration, and (3) sericite-fluorite alteration. Figure 4 summarizes the paragenesis of these different alteration types.

Fresh microgranite: All microgranite (G4) samples exhibit some degree of metasomatic overprint. Their magmatic mineralogy and textures are therefore only partly preserved. The Sadisdorf G4 microgranite contains primary quartz (Qz I) phenocrysts that commonly exhibit snowball textures with growth zoning (Fig. 5A) enclosed by tabular plagioclase (Pl I) crystals, and minor K-feldspar (Kfs I) phenocrysts. Topaz occurs as fine-grained inclusions in quartz and albite (Fig. 5A) or as individual medium-grained euhedral crystals (Fig. 5A) and is considered magmatic in origin (Toz I). Tabular biotite (Bt) and muscovite (Ms I) crystals occur interstitially. Accessory minerals include fine-grained zircon, ilmenite, rutile, monazite, as well as inclusions of uraninite and columbite-tantalite in mica.

Sodic-potassic alteration: This type of alteration is characterized by the pervasive subsolidus replacement of the primary mineral assemblage from the microgranite by albite and less abundant K-feldspar. It occurs as discrete zones in the apical portions of G4 (0–200 m from the contact to the host rock) but was mostly intersected at depth in drill core (Fig. 2B, C).

Albite (Ab II) replaced and overgrew on igneous K-feldspar (Kfs I; Fig. 5B), quartz (Qz I), topaz (Toz I) and mica (Mic I). Igneous anorthite-poor plagioclase (Pl I) was albitized. In some samples, albite is overgrown by abundant late K-feldspar (Kfs II; Fig. 5C). The degree of feldspathization (both sodic and potassic) is variable. Within pervasively altered zones of the G4 microgranite, feldspar-dominated assemblages are common (~85 vol % albite + K-feldspar) with only a few remnants of the original igneous (e.g., Qz I, Pl I, Kfs I, Mic I, or Toz I) and metasomatic (e.g., Ab II) mineral assemblages (Fig. 5B, C).

Greisen alteration and mineralization: Greisen alteration is characterized by the replacement of feldspar and K-mica by quartz (Qz II), topaz (Toz II), and Li-bearing micas (Mic II). Greisen occurs as irregular bodies within the G4 microgranite (Fig. 3A-C; endogreisen), as a pervasive alteration affecting the fine-grained granitic matrix of the magmatic-hydrothermal breccia (Fig. 3E), localized within the pegmatitic carapace (Fig. 3D), and as alteration halos surrounding vein-hosted mineralization in the proximal exocontact of the G4 intrusion.

Greisen alteration is mainly recognized in the upper 200 m of the G4 microgranite (Fig. 2B, C). The degree of greisen alteration varies greatly, with the most intense alteration occurring in the upper 30 m of the intrusion. Greisen alteration zones show both transitional and sharp contacts and are commonly replaced feldspathized microgranite at depth.

The modal abundances of greisen minerals vary locally resulting in the following lithological subtypes based on the nomenclature proposed by Kühne et al. (1972): (1) quartz-mica-topaz greisen (Fig. 3C), (2) quartz-poor mica-topaz greisen (Fig. 3A), (3) quartz greisen, and (4) rare quartz-poor topaz greisen (Fig. 3B).

The quartz-mica-topaz greisen is the predominant subtype of massive greisen and consists of anhedral quartz (46–75 vol %) and coarse Li-Fe mica (33–64 vol %) with lesser anhedral to euhedral topaz (5–22 vol %) that partially replaced or overgrew earlier magmatic minerals (e.g., biotite, plagioclase, topaz, K-feldspar, or quartz) or, in the deeper portions of the system, albite from sodic-potassic assemblages (Fig. 5D). In comparison, quartz-poor mica-topaz greisen alteration contains abundant Li-Fe mica (up to 75 vol %; Figs. 3D, 5E), topaz (~10–15 vol %), and minor medium- to fine-grained (up to 2 mm) quartz. Quartz greisen is characterized by the (almost) complete replacement of feldspars and micas by quartz and overgrowths of hydrothermal quartz (Qz II; Fig. 5F, G) on magmatic quartz phenocrysts (Qz I). Rare quartz-poor topaz greisen (Fig. 3E) occurs sparsely in the upper 50 m of the G4 granite and contains abundant euhedral topaz (35–40 vol %) with inclusion-rich (mineral and fluid) cores. Topaz crystals are overgrown by minor anhedral quartz (~30 vol %).

The Li-Sn-(W) mineralization within the endocontact is associated with either (1) greisen or (2) the pegmatitic carapace in the apical portion of the G4 intrusion. Although greisen alteration extends to a depth of at least 200 m from the apical contact, Sn-W mineralization is mainly confined to the upper 50 m of the G4 intrusion. Endogreisen-hosted mineralization is marked by abundant Li rich mica with disseminated aggregates of euhedral cassiterite (up to 3 mm in size; Fig. 5K) or the replacement of albite by cassiterite (Fig. 5J) and minor wolframite (0.1–2 mm) intergrown with quartz and topaz. In contrast, quartz-poor topaz greisen is enriched in wolframite (Fig. 5L) with respect to cassiterite (e.g., sample SS15221). Here, wolframite is paragenetically older than cassiterite and contains small inclusions of molybdenite.

Pegmatite-hosted mineralization is spatially restricted to the apical parts of G4 where the pegmatite reaches its maximum thickness (~5 m). There, irregularly shaped or tabular wolframite crystals (millimeter to centimeter sized) overgrow very coarse crystalline quartz. Abundant cassiterite postdates wolframite and forms either small (up to 1.5 mm) subhedral-euhedral crystals or thin (150 µm thick) veinlets that cross-cut wolframite crystals. If present, fine-grained (<100 µm) arsenopyrite postdates both wolframite and cassiterite.

The magmatic-hydrothermal breccia body above the G4 intrusion contact was intensely and repeatedly altered (Fig. 3D); a distinction between pervasive or very dense vein-style greisen alteration is thus not possible. The alteration mineralogy is usually fine-grained and dominated by Li-Fe mica with some late Fe oxides (Fig. 3D). Greisen alteration mainly replaced the fine-grained K-feldspar and plagioclase of the granitic matrix (i.e., fine-grained equivalent of the G4), but also replaced breccia clasts (i.e., gneiss, syenogranite, and monzogranite; Fig. 3D). In contrast to the microgranite, the greisen-altered breccia does not host significant Li-Sn-(W) mineralization. Instead, the bulk of the mineralization occurs in clast-hosted veinlets (i.e., prebreccia) or crosscut the breccia.

Sericite-fluorite alteration: This paragenetically late alteration type consists of fine-grained sericite accompanied by variable proportions of fluorite ± clay minerals (mostly kaolinite) ± apatite (Fig. 5H, I). Sericite-fluorite alteration is commonly related to veins or cavities (Fig. 5I); it does not form larger, pervasive bodies. Fine-grained sericite mainly replaced relict feldspars, Li-Fe mica (Fig. 5H), and to a lesser extent topaz. Early sericite was followed by fluorite ± apatite and late clay minerals (Fig. 5I), which in quartz-poor topaz greisen predominate at the expense of sericite. Sericite-fluorite is the youngest hydrothermal assemblage recognized in the Sadisdorf prospect. Minor sulfides (e.g., sphalerite, galena, chalcopyrite, and/or pyrite) may be associated with the sericite and fluorite alteration. In topaz-rich greisen, native Bi aggregates (partly replaced by bismuthinite) accompany sulfide minerals (Fig. 5L).

A total of 20 samples, covering the different intrusive rock types (G1–G5), were analyzed (Table 2). Like other granitic intrusions in the eastern Erzgebirge (e.g., Zinnwald, Altenberg; Tischendorf and Förster, 1990; Förster et al., 1995; Breiter et al., 1999; Breiter, 2012), the magmatic sequence at Sadisdorf is characterized by peraluminous granites that are commonly alkali depleted. Although they may appear macroscopically unaltered, their alkali signatures (Fig. 6A) imply that many of the analyzed samples have undergone either greisen or sodic-potassic alteration. Alteration did not affect G5 samples that, owing to their primary K-feldspar-rich composition (Winter et al., 2008), have high K₂O contents that are assumed to be primary.

To evaluate compositional changes related to hydrothermal alteration, selected major elements were plotted against Na₂O (Fig. 6B, C) following the suggestion of Štemprok and Dolejš (2010). Igneous rocks at Sadisdorf have similar contents of Al₂O₃ (13–17 wt %) and SiO₂ (68–76 wt %), except those most intensely altered (Fig. 6B). The least altered granitoid samples contain on average 0.9 wt % F and 0.25 wt % Li₂O, whereas the most pervasively greisenized samples (with quartz-rich greisen being a notable exception) are marked by elevated F and Li₂O contents (Fig. 6C). Sodic alteration was marked by depletions in Li₂O (Fig. 6), F, Fe₂O_3, and P₂O₅ (Table 2), a small depletion in SiO₂ (Fig. 6B), and a slight enrichment in Al₂O₃.

Chondrite-normalized rare-earth-element (REE) profiles (Fig. 6D) show a progressive depletion of REE (particularly light REE) and an increasingly negative Eu anomaly for the intrusive sequence from G1 (Eu/Eu* = 0.18) to G4 (Eu/Eu* = 0.03). Unaltered and altered G4 samples exhibit characteristic concave profiles between Gd and Ho, indicative of the lanthanide tetrad effects (Irber, 1999). This tetrad effect (Fig. 6E) is similar to that for other post-Variscan granitoids in the Erzgebirge (Dolejš and Štemprok, 2001, and references therein). The G4 samples that have experienced intense potassic-sodic or greisen alteration show a REE depletion of more than an order of magnitude (Fig. 6D). Greisen samples show significant (Eu/Eu* = 0.02–0.04) negative Eu anomalies (Fig. 6D). The G5 dikes that postdate magmatic-hydrothermal mineralization have a REE signature similar to G1 (Fig. 6D).

To evaluate magmatic fractionation trends, a selection of trace elements and trace element ratios (this study and data available from the ATC, see App. Table A3) were plotted against TiO₂ (Fig. 7) as a proxy for magmatic differentiation. The highly evolved samples have low Nb/Ta and Y/Ho ratios, irrespective of the degree and style of magmatic-hydrothermal alteration (Fig. 7A, B). The late (commonly fertile) Li bearing intrusive rocks show a distinct trend toward lower Y/Ho and Nb/Ta ratios compared to other igneous units of the ATC (Fig. 7A, B).

Greisenized samples generally have higher concentrations of fluid-mobile elements, such as Sn, W, Zn, and Bi (Fig. 7C-F). These samples appear decoupled from the general magmatic fractionation trend and their less-altered equivalents and, except for the Sadisdorf monzogranites G2 and G3, display an inverse relationship of TiO₂ with W, Zn, and Bi (Fig. 7C-F).

The vein-type mineralization occurs as stockworks or randomly oriented veinlets that are typically only a few millimeters wide. This mineralization occurs mostly around the G4 microgranite, which also hosts some fluorite veins. Based on their dominant minerals, the veins have been subdivided into (Fig. 8): (1) oxide, (2) oxide-sulfide, (3) sulfide, and (4) fluorite-(quartz) types (Table 1). This classification differs slightly from previous classifications to account for spatial variations (e.g., Müller, 1887; Seltmann, 1984). A general paragenetic sequence is shown in Figure 4.

Oxide veins: Oxide veins are typically a few millimeters wide and have curved to wavy shapes. Two subtypes of the oxide vein-type were recognized, namely (1) quartz – Li-Fe mica – wolframite ± molybdenite ± cassiterite ± topaz (Fig. 8A) and (2) quartz – cassiterite – Li-Fe mica. Wolframite-rich veins are more abundant in the upper 40 m from the G4 apex and toward the NW flank of the G4 intrusion. Occasionally, wolframite-rich veins are recognized in gneiss fragments of the magmatic-hydrothermal breccia. These veins, however, are conspicuously absent at the Zinnklüfte site. The wolframite-rich veins are composed of subhedral to euhedral wolframite that together with Li-Fe mica usually line the selvages (Fig. 9A). The quartz varies from anhedral (mostly in molybdenite-rich veins) to euhedral (Fig. 9A) and is usually coeval with or younger than the wolframite. Cassiterite is rare and occurs as small (<250 µm) euhedral crystals overgrown on earlier wolframite (Fig. 9A). Wolframite-rich veins exhibit a greisen-type alteration halo, which consists mainly of fine-grained Li-Fe mica, quartz, and topaz with disseminated wolframite, minor cassiterite, and molybdenite. Whereas wolframite and cassiterite are mainly in the veins, molybdenite occurs commonly in quartz and topaz-rich alteration halos (Fig. 8A). Rarely, molybdenite can be recognized as fine- to coarse-grained (up to 1 cm) vein infill, either coeval with or having postdated quartz.

As occurs with the wolframite-rich veins, cassiterite-rich veins (Figs. 8B, 9B) occur preferentially proximal to the G4 intrusion at the Kupfergrube site (<80 m away from the intrusion contact) and in the SW sector of the Zinnklüfte stockwork. However, the cassiterite-rich veins are laterally more extensive than their wolframite-rich analogues. The cassiterite-rich veins consist of anhedral to euhedral quartz intergrown with minor Li-Fe mica and fine-grained topaz. Cassiterite crystals up to 0.5 cm in diameter (Fig. 9B) overgrew the quartz, mica, and topaz and commonly grew into open space. Greisen-type alteration halos around the cassiterite-bearing veins usually contain fine-grained disseminated cassiterite that selectively replaced albite or K-feldspar together with greisen-related mica, quartz, and topaz (Fig. 9C).

Oxide-sulfide veins: These veins are marked by the presence of sulfides and sulfosalts in addition to cassiterite and/or wolframite. Again, two distinct mineralogical subtypes are recognized: (1) quartz Li-Fe mica-chalcopyrite-cassiterite-sphalerite-galena-tennantite-pyrite ± wolframite ± arsenopyrite ± molybdenite ± emplectite ± topaz veins (chalcopyrite-rich subtype; Fig. 8C), and (2) quartz Li-Fe mica-arsenopyrite-cassiterite-sphalerite-galena-tennantite-pyrite ± molybdenite ± topaz veins (arsenopyrite-rich subtype; 8D). The chalcopyrite-rich veins are restricted to the Kupfergrube site and extend vertically from the surface down to 120 m in depth and laterally no more than 150 m from the G4 apex. In contrast, the arsenopyrite-rich veins occur widely (up to 350 m from the G4 apex and in the central-NE sector of the Zinnklüfte site). In the proximal parts of the prospect (<50 m from the granite contact), the oxide-sulfide veins are often found to crosscut oxide veins.

Chalcopyrite-rich veins are composed of anhedral to rarely subhedral quartz usually intergrown with Li-Fe mica and rare topaz. The quartz and mica are overgrown by aggregates of subhedral cassiterite and rare wolframite that are, in turn, overgrown by sulfide aggregates (Fig. 9E). Chalcopyrite, stannite, emplectite, and bismuthinite are paragenetically the oldest sulfides (Fig. 9D, E) followed by sphalerite, galena, and tennantite. Emplectite and bismuthinite occur preferentially at the contact between early oxides and chalcopyrite (Fig. 9E). Stannite may form as individual crystals but is mostly present as exsolutions in chalcopyrite (Fig. 9D), where cassiterite is locally absent. Fine-grained disseminations of cassiterite, chalcopyrite, sphalerite, galena, arsenopyrite, and pyrite are common in the greisen-type alteration halos around the veins.

The arsenopyrite-rich veins contain cassiterite-arsenopyrite-sphalerite-galena-tennantite aggregates intergrown with granular quartz, minor Li-Fe mica, and rare topaz (Fig. 9F). Cassiterite was the earliest ore mineral followed by abundant anhedral arsenopyrite with rare sphalerite inclusions. Tennantite, galena, and pyrite postdate arsenopyrite (Fig. 9F). If present, molybdenite is intergrown with quartz or forms inclusions in arsenopyrite. As is seen with the chalcopyrite-rich veins, the alteration halos contain disseminations of ore minerals such as cassiterite and arsenopyrite.

Sulfide veins: Sulfide vein- and stockwork-type mineralization also forms two distinct mineralogical subtypes: (1) quartz-sericite-arsenopyrite-chalcopyrite-galena-sphalerite-tennantite-pyrite-marcasite ± molybdenite ± matildite ± chlorite ± Li-Fe mica assemblages (chalcopyrite-rich subtype; Fig. 9G), and (2) arsenopyrite-pyrite-marcasite ± chalcopyrite ± quartz assemblages (pyrite-rich subtype; 8F, 9H). Both subtypes occur further from the G4 granite than the oxide and oxide-sulfide veins (>200 m distance from the G4 apex, and in the NE sector of the Zinnklüfte). There are no clear trends in the spatial distribution of the chalcopyrite- and pyrite-rich subtypes. In the proximal and intermediate parts of the prospect (<200 m distance from the G4 apex), both vein subtypes crosscut oxide and oxide-sulfide veins.

Gangue minerals in the chalcopyrite-rich subtype include early comb quartz that is accompanied by coeval chlorite and/or sericite. Arsenopyrite-chalcopyrite-galena-sphalerite-tennantite-pyrite-marcasite ± molybdenite aggregates (Fig. 9G) mostly postdate the gangue minerals. Arsenopyrite is paragenetically the earliest sulfide and is sometimes seen as inclusions within quartz (Fig. 9G), followed by galena and matildite and, in turn, chalcopyrite, tennantite, pyrite-marcasite, and sphalerite. Molybdenite and sphalerite may be present as inclusions in galena-rich sulfide aggregates. Pyrite forms fine-grained layered intergrowths with marcasite (Fig. 9H) in textures similar to those identified as a product of pyrrhotite replacement (Kelly and Turneaure, 1970). In contrast to the chalcopyrite-rich veins, the pyrite-rich subtype is characterized by a lower abundance of quartz with only a few subhedral quartz grains intergrown with massive pyrite and arsenopyrite (Fig. 8F). By analogy with the chalcopyrite-rich veins, pyrite forms layered intergrowths with marcasite. Late chalcopyrite may fill thin cracks or voids in arsenopyrite-pyrite aggregates. Despite the rare occurrence of Li-Fe micas in the veins, alteration halos surrounding both vein types have the same greisen mineralogy as that surrounding the oxide and oxide-sulfide vein types. The halos around the sulfide veins contain disseminated pyrite, arsenopyrite, and chalcopyrite.

Fluorite-(quartz) veins: Thin (<3-mm wide) quartz ± Li-Fe mica veins or fluorite veins and cavity infills abound across the Sadisdorf deposit. They crosscut all previously described vein-types as well as the endogreisen (Figs. 8D, E, 9I). Violet fluorite forms monomineralic veins (Figs. 8D, 9I) or occurs as anhedral to subhedral crystals that line or fill cavities (Fig. 8D). Unlike the Mesozoic fluorite veins (e.g., Burisch et al., 2022), which are also present in the district and elsewhere in the Erzgebirge (e.g., Knöttel; Peterková and Dolejš, 2017; Hub stock; Dolníček et al., 2012), these veins are distinguished by their lack of barite, low FI salinity (see Microthermometry section), and Paleozoic minimum ages (Wolff et al., 2015).

Suitable FIs were identified and analyzed in quartz, topaz, cassiterite, sphalerite and fluorite related to greisen and vein/stockwork style mineralization. Based on their composition (e.g., aqueous vs. carbonic) and the number of phases, five main types of FI were identified: (1) vapor-rich aqueous (VL-type), (2) single-phase aqueous (L-type), (3) two-phase aqueous (LV-type), (4) aqueous-carbonic (C-type), and (5) brine (LVH- and LVHS-types). Figures 10 and 11 provide an overview of the FI types and their typical petrographic relationships with their host minerals.

VL-type: This type of inclusion is characterized by a dominance of the vapor over the liquid at room temperature (liquid fraction: 0.4–0.1; Fig. 10A), the homogenization to vapor, and the presumed absence of other gaseous components (i.e., CO₂ or CH₄). The VL-type inclusions are rarely observed as primary trails; more commonly they form pseudosecondary trails hosted by pre- or syn-ore quartz in the oxide, oxide-sulfide, and sulfide veins (Fig. 10B). Typically, they have irregular to round morphologies and sizes between 5 and 44 µm.

L-type: These inclusions are composed of a single aqueous liquid phase (Fig. 10A). L-type FIs mostly occur in fluorite from fluorite (-quartz) veins where they form pseudosecondary trails and clusters commonly accompanied by LV-type FIs with variable liquid fractions (heterogeneous FIAs). Only two L-type FIs were identified in clusters hosted by pre-ore pegmatitic quartz. Overall, this type of FI is rare. Individual inclusions range in size from 8 to 20 µm and have irregular to round morphologies.

LV-type: LV-type FIs consist dominantly of aqueous liquid and have liquid fractions of 0.9 to 0.6. Typically, LV-type FIs are hosted in pre- and syn-ore quartz as well as cassiterite (Fig. 11A) and sphalerite related to all mineralization styles and associations. They may form (1) primary FIAs along crystal growth zones (Fig. 11B), (2) clusters in the cores of crystals, (3) pseudosecondary trails (Figs. 10B, 11B), or (4) secondary trails (Figs. 10B, 11B). Overall, LV-type FIs represent the most abundant type of FIs in the investigated samples, and they may vary in size (3–90 µm) and shape.

C-type: The C-type or carbonic phase-bearing aqueous FIs are two-phase (V + L) inclusions with liquid fractions between 0.2 and 0.7. Their vapor phase contains variable proportions of CO₂ and CH₄ as identified by Raman spectroscopy. However, the gas concentrations are too low to form a visible CO₂ liquid at room temperature. Occasionally, they contain a small (<0.1 µm) unidentified opaque phase. They are the second-most common FI type and are particularly abundant in the endocontact and proximal exocontact mineralization (<60 m away from the G4 contact). These inclusions occur mainly in the topaz-rich greisen at the rims of mineral inclusion-rich topaz cores (Fig. 11C), suggesting that they are of primary origin. Mineral inclusions within topaz consist mostly of (microcrystalline-) quartz (identified by Raman spectroscopy) with potential contraction vapor bubbles (Fig. 11C) similar to those described by Williamson et al. (1997).

The C-type FIs commonly form isolated clusters—pseudosecondary and secondary trails in greisen-related quartz as well as in the pegmatitic carapace. In the exocontact mineralization, C-type FIs are hosted as isolated clusters, pseudosecondary and secondary trails in pre- or syn-ore quartz in oxide, oxide-sulfide, and sulfide veins. Commonly, these FIs have negative-crystal to elongated morphologies with sizes ranging between 6 and 30 µm.

LVH- and LVHS-type: High-salinity inclusions can be subdivided into two groups: (1) three-phase (L + V + S) halite-bearing FIs (LVH-type; Fig. 11A) and (2) polyphase (L + V + S₁ +…+ S₅) FIs with a variety of solids (LVHS-type; Figs. 10A, 11D). The LVH-type FIs contain halite crystals at room temperature (up to 31 vol %) in addition to a liquid and a vapor. They mostly form secondary or, less commonly, pseudosecondary FIAs, and clusters in pre-ore quartz of the oxide (wolframite-rich) and the sulfide veins (Fig. 10A). Some inclusions contain minute (<0.5µm) opaque solids that, owing to their small size, could not be identified by Raman spectroscopy. In contrast, the LVHS-type FIs contain multiple solids (up to five) at room temperature (Figs. 10A, 11D), including halite, sylvite, and calcite, in addition to other unidentified phases. The variable quantities and types of solids per FI within the same FIA suggest that the solids were accidentally trapped (Goldstein et al., 2003; Chi et al., 2021), rather than being the products of postentrapment water loss (e.g., Audétat, 2023; Zhang and Audétat, 2023). The LVHS-type FIs occur exclusively as secondary trails or isolated FIAs hosted by pre-ore greisen quartz as well as oxide and oxide-sulfide veins. They are particularly common in the molybdenite-rich subtype of oxide veins, whereas in the oxide-sulfide veins, LVHS-type FIs are restricted to chalcopyrite-rich veins only. In both vein types, FIAs are spatially related to secondary C-type FI trails (Fig. 11D). Overall, LVH-type FIs are small (6–19 µm) with irregular to subrounded morphologies. LVHS-type FIs may be up to 25 µm in diameter and show irregular to negative-crystal shapes.

A total of 323 FIs from 75 FIAs hosted by greisenized microgranite and pegmatite, and the different types of vein-style mineralization in the Kupfergrube and Zinnklüfte sites were analyzed by microthermometry. The inclusions are representative of the proximal, intermediate, and distal portions of the Kupfergrube site (Fig. 2B, C). The microthermometric analyses from the Zinnklüfte site are restricted to cassiterite-rich and fluorite vein assemblages. The results of the microthermometric analyses are summarized in Table 3 and Figure 12. Measurable eutectic temperatures (–23.0° to –16.0°C of the VL-, L-, LVH-, and LV-types of FIs) suggest that the entrapped fluids can be reasonably approximated by the NaCl-(±KCl)-H₂O system. Conversely, C-type FIs can be described within the NaCl-H₂O-CO₂-CH₄ system owing to the presence of minor quantities of gaseous components. The LVHS-type FIAs were not quantitatively analyzed because of their secondary nature and evidence of accidental entrapment of solids (see discussion). In the following, the microthermometric results are presented according to the mineralization type (greisen, pegmatite, oxide, oxide-sulfide, sulfide, or fluorite vein) to which they are related.

Greisenized microgranite: Primary C-type FIs are hosted in topaz from the topaz-rich greisen subtype (Sample SS15221; Fig. 11C) and are characterized by T_m(car) between –56.9° and –56.6°C, indicating that the vapor is CO₂-dominated and contains small but variable proportions of CH₄ (van den Kerkhof and Thiéry, 2001). The ice-melting temperatures range between –6.5° and –2.6°C. If present, clathrate melted at temperatures ranging between 4.0° and 8.8°C. Owing to the low CO₂ content, CO₂-homogenization was not observed (see discussion). The ice-melting temperatures record maximum salinities of 5.5 to 9.9 wt % NaCl equiv (Fig. 12A). The C-type inclusions have high T_h (421°–321°C; Fig. 12A) and homogenize to vapor.

Pseudosecondary FIs and clusters of FIs in pre-ore quartz from quartz- and mica-rich greisen rock types are exclusively LV-type. Their salinities (1.6–11.5 wt % NaCl equiv; Fig. 12A) and T_h (282°–387°C; Fig. 12A) overlap with those of the C-type inclusions but extend toward lower salinity and T_h values.

Greisenized pegmatite: Quartz-hosted inclusions within the mineralized pegmatitic cap host either C-, LV-, or L-type FIs. The aqueous-carbonic inclusions have similar characteristics to the greisen-hosted C-type inclusions; however, no clathrate melting was observed and the CO₂ and CH₄ components were only identified through Raman analysis. Thus, the concentrations of these gases are assumed to be low (Duan et al., 1992; Diamond, 2001). The ice-melting temperatures range from –4.0° to –6.9°C, corresponding to salinities between 6.4 and 10.3 wt % NaCl equiv (Fig. 12A), and the T_h values are between 383° and 392°C (Fig. 12A). The LV- and L-type FIs in the pegmatitic carapace represent secondary inclusions. These are marked by significantly lower T_h than, but similar salinity to, the pseudosecondary C-type inclusions (Fig. 12A).

Oxide veins: As in the endocontact greisen mineralization, LV-type inclusions are the most common inclusion-type in the oxide veins. The quartz- and cassiterite-hosted LV-type FIAs have a similar composition (salinities between 0.5 and 10.7 wt % NaCl equiv; Fig. 12A) with a wide range of T_h (216°–417°C; Fig. 12A). They invariably homogenize to liquid.

Molybdenite-bearing veins (wolframite-rich subtype) host most of the high-salinity FIAs in the oxide vein-type. The inclusions are either LVH-type (pseudosecondary) or LVHS-type (secondary) and are commonly hosted in pre-ore quartz. The LVH-type inclusions have halite dissolution temperatures between 299° and 425°C, corresponding to salinities between 38.0 and 48.9 wt % NaCl equiv, and T_h(LV→L) values of 328° to 358°C (Fig. 12A). The more saline FIs homogenize by halite dissolution, whereas the lower salinity FIs homogenize by the disappearance of the vapor bubble. The latter inclusions occur in the only observed boiling FIA (FIA 38) where they are associated with VL- and LV-type FIs. The VL-type FIAs are common in the pre-ore quartz of molybdenite-bearing veins; they have high T_h (312°–402°C; Fig. 12A) and low to moderate salinity (0.4–9.6 wt % NaCl equiv; Fig. 12A). Cassiterite-hosted FIs of cassiterite-rich veins are invariably LV-type with a range of T_h between 248° and 379°C (Fig. 12) and low salinity (0.4–10.7 wt % NaCl equiv; Fig. 12).

As at Kupfergrube, the FIAs (LV-type FIs) hosted by pre-ore quartz or cassiterite in veins at Zinnklüfte have low to moderate salinity (0.5–9.9 wt % NaCl equiv; Fig. 12A) and T_h ranging from 302° to 426°C (Fig. 12A). In contrast to the oxide veins at the Kupfergube site, at Zinnklüfte they host abundant C-type FIs with ice-melting temperatures between –0.3° and –4.0°C, corresponding to maximum salinities between 0.5 and 6.5 wt % NaCl equiv (Fig. 12A). Carbon dioxide and methane were only detected in the gaseous phase of these FIs by Raman spectroscopy. The homogenization temperatures (389°–398°C; Fig. 12A) overlap with those of the LV-type inclusions.

Oxide-sulfide veins: The FIAs analyzed in the oxide-sulfide veins are hosted by pre-ore quartz in both the arsenopyrite- and chalcopyrite-rich subtypes. Pseudosecondary FI trails are common in both subtypes, with only a few primary FIAs in the arsenopyrite-rich subtype. Whereas LV-type FIs are common in both subtypes, VL-type pseudosecondary trails were only found in the chalcopyrite-rich subtype. The LV-type FIs homogenize over a wide range of temperature (202°–404°C; Fig. 12A), and the salinity ranged between 0.5 and 8.7 wt % NaCl equiv (Fig. 12A). Cluster and pseudosecondary VL-type FIs hosted by pre-ore quartz have high T_h values (334°–386°C) and invariably low salinity (0.3–2.9 wt % NaCl equiv; Fig. 12A).

Sulfide veins: Pseudosecondary trails and clusters of FIAs hosted by pre- and syn-ore quartz of the sulfide association mainly comprise LV-type; there are also minor proportions of C-, VL- and LVH-type FIs. The VL-type inclusions hosted in pre- and syn-ore quartz have similar T_h, ranging between 269–389°C and 325–379°C, respectively (Fig. 12A). Their salinities show a similar bimodal distribution of low (1.0–6.0 wt % NaCl equiv; Fig. 12A) and moderate values (7.0–17.0 wt % NaCl equiv; Fig. 12A). Four FIAs hosted in sphalerite form pseudosecondary trails and clusters containing only LV-type FIs. Sphalerite-hosted inclusions have invariably low salinity (<1.0 wt % NaCl equiv; Fig. 12A, B), whereas the range of T_h overlaps with that of the quartz-hosted FIs (333–402°C; Fig. 12A, C).

The C-type FIs in pre- and syn-ore quartz have ice-melting temperatures between –0.2° and –3.0°C. Raman analysis indicates low but variable proportions of CO₂ and CH₄ in the vapor; clathrate melting was not observed. The salinity (a maximum value) ranged from 0.4 to 5.6 wt % NaCl equiv and the T_h ranged between 286° and 384°C (Fig. 12A). Similar T_h (335°–372°C) and salinity (2.2–3.3 wt % NaCl equiv; Fig. 12A) values were observed for VL-type FIs hosted in syn-ore quartz. Only one LVH-type FIA could successfully be measured (Sample 6-85-3) owing to the small size of the FI. Inclusions in this FIA homogenize by halite dissolution at 354° to 430°C, corresponding to salinities between 43.5 and 49.6 wt % NaCl equiv; Fig. 12A). Their homogenization by halite dissolution suggests that they may have experienced postentrapment H₂O loss (Audétat and Günther, 1999; Lecumberri-Sanchez et al., 2012).

Fluorite veins: Pseudosecondary and isolated FIAs in late fluorite veins and cavity infill at the Kupfergrube site are mostly LV-type. The fluorite-hosted FIAs have T_h values ranging from 120° to 270°C (Fig. 12A) and low salinity (0.0–4.3 wt % NaCl equiv; Fig. 12A). The fluorite-hosted LV-type FIs from the Zinnklüfte site show similar T_h (164°–238°C) and salinity (0.0–0.8 wt % NaCl equiv; Fig. 12A) to those from Kupfergrube. Owing to background fluorescence (Dill Pasteris et al., 1988; Frezzotti et al., 2012), it was not possible to conduct Raman analyses of FIs hosted by fluorite.

The petrographic observations, geochemical, and fluid inclusion data reported above serve to constrain the spatial and temporal magmatic-hydrothermal evolution of the Sadisdorf prospect. The following discussion will focus on the Kupfergrube site, which represents a semi-continuous exposure from the assumed granitic source of the metals to the distally emplaced stockwork, compared to the stockwork at the Zinnklüfte site, which is interpreted to be an analogue and possible satellite of the Kupfergrube system.

Whole-rock geochemical data from the igneous rocks at the Kupfergrube site support the previous interpretations of Seltmann (1995) that the sequence of early syenogranite, monzogranite, and alkali-feldspar granite intrusions (G1–G4) at the Sadisdorf prospect record a trend of magmatic differentiation (Figs. 6E, F; 7A, B). This sequence of fractional crystallization is associated with a systematic decrease of Nb/Ta ratios (Fig. 7A) and an increase in the tetrad effect (Fig. 6F). The decreasing Nb/Ta ratios can be explained by either fractional crystallization of mica (Stepanov et al., 2014) or by fluid interaction during differentiation (Ballouard et al., 2016; Alferyeva et al., 2020). Recent experimental work of Alferyeva et al. (2020) has shown that fluid interaction, and particularly the interaction with F-rich melts, can significantly reduce the Nb/Ta ratios. Irber (1999) suggested that the tetrad effect in peraluminous granites may be attributed to magma-aqueous fluid interaction during the waning stage of the magmatic evolution. In contrast, Monecke et al. (2011) have shown that the tetrad effect can develop in hydrothermal systems independent of the magmatic evolution where it is associated with the partitioning between Cl and F complexes of lanthanides during fluid-vapor phase separation (i.e., Cl and F partition into the fluid and the vapor, respectively). Thus, the more pronounced tetrad effect for most of the altered samples (Fig. 6E) might be related to a metasomatic overprint—likely related to the interaction with an F-rich, vapor-dominated fluid—rather than their magmatic evolution.

The Y/Ho ratios show only a slight increase with magmatic evolution of the caldera and the older intrusions (e.g., G1 at Sadisdorf; Fig. 7B), but they remain within the chondritic field (24–34) as defined by Bau (1996). The mineralized alkali-feldspar granite (altered G4 granite), in contrast, is marked by low nonchondritic Y/Ho ratios. Nonchondritic Y/Ho ratios are typically related to the interaction of the rocks with halogen-rich fluids (Bau, 1996) because of the preferential partitioning of Y into F bearing fluids (Veksler et al., 2005). This is consistent with the recognition of the exceptionally high abundance of F rich hydrothermal minerals at the Sadisdorf prospect. In summary, the combination of decoupling of Y/Ho ratios, low Nb/Ta values, and the degree of the tetrad effect suggest interaction of the evolved granitic intrusions with an F rich silicate melt/fluid during the late- and post-magmatic stages.

Petrographic observations and fluid inclusion evidence from the Kupfergrube site are in agreement with the suggestion that both endogreisen- and vein-style mineralization are related to a magmatic-hydrothermal fluid (i.e., high T) exsolved from the G4 intrusion. The G4 granite was emplaced ~300 m below the base of the Teplice rhyolite (Fig. 2C), which is estimated to have had a pre-erosional maximum thickness of 1,200 m (Mlcoch and Skácelová, 2010). This implies a minimum emplacement depth of about 1,500 m (P_lithostatic ≥390 bar) consistent with constraints using fluid inclusions and pressure estimates derived from melt inclusions (Thomas, 1994) and structural arguments (Seltmann, 1984).

Crosscutting field relations, mineral paragenesis, and fluid inclusion evidence support two distinct magmatic-hydrothermal mineralization stages (Fig. 15): (1) main stage, and (2) waning stage. The main stage can be further subdivided into two substages: (A) fluid exsolution and greisenization, and (B) fluid-phase separation.

Fluid exsolution and greisenization (main stage): Fluid inclusion petrography shows that C-type FIs are predominant in the greisen and pegmatitic samples. Their high T_h (334°–421°C; Fig. 12A) and high minimum entrapment pressures (up to 338 bar) indicate that they represent the earliest and hottest part of the hydrothermal system. This is further supported by their salinity (7.0 ± 3.0 wt % NaCl equiv on average) and intermediate density (0.57–0.72 g/cm³), which is typical of fluids exsolved early from granitic magmas (5–10 wt % NaCl; Heinrich, 2005; Audétat et al., 2008). The intermediate density and lack of high-salinity inclusions suggest the presence of single-phase fluids at this stage (Fig. 13A). Based on calculations using the SoWat software from Driesner and Heinrich (2007), it was determined that an aqueous fluid with 7 wt % NaCl would require a minimum pressure of 330 bar to be exsolved as a single phase, which is in excellent agreement with estimates from geologic reconstruction. Based on these pressure constraints and the nature of greisen- and pegmatite-hosted FIs, it can be assumed that a magmatic-hydrothermal, mostly aqueous fluid exsolved at ~600°C (Li-F granite solidus; Štemprok, 1980; Shapovalov et al., 2019) from the deeper parts of the parental magma (Fig. 13). This fluid subsequently ascended through the solidified portions of the G4 granite that had cooled below 500°C (Fig. 13A, B). At these temperatures and estimated pressure conditions, the fluid remained a single-phase as documented by C-type FIs. This fluid interacted strongly with the rocks along its flow path resulting in albitization (sodic alteration) at depth and shallower intense greisenization.

Fluid-rock ratios and progressive greisenization are likely to have been the first-order controls on the final modal mineralogy of the greisen alteration (Halter et al., 1998), whereas the composition of the protolith (Kinnaird, 1985), porosity, and permeability (Launay et al., 2019) seem to represent second-order controls on the reactivity of the precursor rocks during metasomatism. Hence, we propose that the high abundance of quartz-poor mica-topaz greisen alteration in the lower portions of the greisenized zone of the G4 granite was the result of enhanced permeability and reactivity of the albitized rocks that were affected by greisen alteration. Selective replacement of feldspar crystals by quartz, Li-Fe micas, topaz, and cassiterite (Fig. 9C) provides evidence that feldspar-consuming reactions not only promoted formation of greisen minerals but also caused cassiterite precipitation during greisen alteration (Heinrich, 1990; Lehmann, 2021).

Phase separation (main stage): Following initial alteration reactions at lithostatic pressure conditions, the magmatic-hydrothermal system likely experienced a complex transition from a ductile lithostatic to a brittle hydrostatic regime. This is documented by fluid inclusions in quartz related to early wolframite-rich oxide veins with VL-, LVH-, and LV-type FIs, the latter commonly with intermediate density (0.54–0.79 g/cm³). The unusual wavy shapes of these veins indicate that they were emplaced under semiductile or transitional conditions (Gustafson and Hunt, 1975; Sillitoe, 2010). High T_h (291°–418°C; Fig. 12A) and high minimum entrapment pressures (up to 318 bar) are conditions consistent with a transitional ductile-lithostatic to brittle-hydrostatic regime (Fig. 13A, B; Hayba and Ingebritsen, 1997; Fournier, 1999). Numerical modeling (Weis et al., 2012) implies that fluids moving between lithostatic and hydrostatic pressure gradients flow via overpressure-permeability waves. A fluid pressure exceeding lithostatic pressure at the front of the upward moving wave creates permeability, resulting in a pressure drop behind it. Thus, ascending fluids may fluctuate between single-phase and two-phase conditions (Fig. 13A, B). Although direct evidence for boiling is limited to only one analyzed FIA, the bimodal distribution of high salinity (e.g., LVH-type) and abundant low salinity vapor-rich (VL-type) in the fluid inclusion data set, suggest a transition from early high-temperature intermediate salinity and density fluids (i.e., single-phase fluids) to two-phase conditions. The presence of gneiss clasts that contain wolframite-bearing veins in the explosive magmatic breccia pipe, constrain the brecciation event to the early oxide vein stage of the magmatic-hydrothermal evolution, supporting over-pressured fluid conditions during this stage. Despite the different pressure conditions, salinity (3.0–9.4 wt % NaCl equiv), and T_h (291°–418°C) of the intermediate density LV-type FIs, they overlap with those of the greisen- and pegmatite-hosted C-type inclusions and are assumed to represent a continuum between the two. Because Mo is transported dominantly as oxo-hydroxy and Na molybdate complexes (Wood and Samson, 2000; Rempel et al., 2009; Shang et al., 2020) in hot (>300°C) acidic hydrothermal fluids, liquid-vapor phase separation is inferred to have caused precipitation of paragenetically early molybdenite (Candela and Holland, 1984; Seo et al., 2012). In contrast, Wang et al. (2021) have shown that in F-rich hydrothermal systems (such as the Sadisdorf prospect), fluoride complexes are more important for W transport. Therefore, an increased pH through fluid-rock interaction is an effective mechanism for wolframite precipitation. The early paragenetic position of wolframite, typically lining vein selvages, suggests that fluid-rock interaction of F-bearing fluids might be an important control on the precipitation and distribution of wolframite. Cassiterite-rich veins, unlike the wolframite-rich subtype, mostly contain fluid inclusions of lower salinity (1.9 wt % NaCl equiv; Fig. 12C; LV-type FIs hosted by cassiterite). Such low salinity and the evidence of early phase separation suggest that the Sn mineralizing fluids were derived from a contracted, previously separated vapor phase (Audétat et al., 2008). Hydrostatic pressures (147–245 bar) would have allowed a low-density vapor (that exsolved at deeper levels) to contract (Heinrich et al., 2004; Williams-Jones and Heinrich, 2005) and subsequently homogenize with (and dilute) more saline fluids (Fig. 13A, B). The range of salinity (0.9–16.4 wt % NaCl equiv) observed in LV-type FIs in the different vein associations overlaps but is wider than that assumed for the original magmatic fluid (7 ± 4 wt % NaCl equiv). This may indicate variable proportions of re-homogenization of vapor and brine (Klemm et al., 2007; Schirra et al., 2022) and would explain the scarcity of brine FIs (LVH-type).

The spatial distribution of proximal oxide, intermediate oxide-sulfide, and distal sulfide veins (Fig. 14) shows that the proportion of sulfides increased progressively at the expense of oxides along the fluid flow path. Cooling (e.g., Kelly and Turneaure, 1970), fluid mixing (e.g., Fekete et al., 2016), and the availability of S (Audétat et al., 2000) have been suggested as mechanisms controlling oxide-to-sulfide zonation in Sn-W mineralized granites. The general overlap of salinities (Fig. 14) and lack of systematic mixing trends between oxide and sulfide assemblages argues against fluid mixing. Homogenization temperatures in gangue- (Fig. 12A) and ore-hosted (Fig. 12B) FIAs show that the main stage vein mineralization formed at similar temperatures. Thus, the lack of systematic variations with increasing distance to the causative granite intrusion (Fig. 14) suggests that cooling was not the main mechanism responsible for spatial zonation. However, the wide range of homogenization temperatures (ca. 170°C) does not exclude the possibility that cooling played a minor role in the evolution of the system, particularly at the vein scale. Similarly, without direct information on the fluid composition, the role of S cannot be constrained and thus remains speculative.

Waning-stage fluid cooling and dilution: The FIAs in the waning-stage fluorite veinlets and cavity infills show distinctively lower temperatures (120°–270°C) than all of the older vein generations, implying that the fluid underwent significant cooling with time. Their salinities (0.9–7.6 wt % NaCl equiv) overlap with those of the FIs of the main stage. However, the fluorite veins contain abundant low-salinity FIs (i.e., seven out of eight FIAs with <2.0 wt % NaCl equiv). Therefore, either simple cooling or cooling accompanied by mixing with dilute meteoric fluids was likely responsible for the formation of the fluorite veinlets (Figs. 13B, 15B). The occurrence of fluorite veinlets across the entire deposit is best attributed to rapid cooling of the ore-forming G4 intrusion and related telescoping of mineral assemblages. With system cooling the isotherms retreated, which moved the magma-fluid interface to greater depth, resulting in a reduced and more dispersed fluid flux (Lamy-Chappuis et al., 2020). As a result of the downward migration of the fluid exsolution interface, distal alteration and vein infill mineral associations were superimposed on initially proximal mineral associations (Fournier, 1999; Scott et al., 2016; Launay et al., 2023). The sericite-fluorite (-kaolinite) alteration that accompanies the late fluorite veins is characteristic of hydrolytic alteration (Hemley et al., 1980; Pirajno, 1992) related to lower temperature and high acidity because of the dissociation of acids at low temperatures (Hedenquist and Taran, 2013).

The NE-SW–trending Zinnklüfte stockwork zone has a vein paragenesis, a mineralogical zonation, and fluid inclusion characteristics similar to those of the Kupfergrube site. However, the alteration assemblage at Zinnklüfte contains more muscovite, and Li-bearing micas are rare. Microthermometric measurements of fluid inclusions in cassiterite-rich veins yielded homogenization temperatures (302°–426°C) similar to those of fluid inclusions in the Kupfergrube area, but the range of salinities is wider (0–10 wt % NaCl equiv). Late-stage fluorite is also present at Zinnklüfte, and it hosts fluid inclusions with characteristics similar to those of the fluorite at Kupfergrube, i.e., low temperature (164°–238°C) and low salinity (<1.0 wt % NaCl equiv). Furthermore, recent U-Pb cassiterite ages (Leopardi et al., 2024) show that the mineralization at both sites was coeval (315–311 Ma) and suggests strongly that they are genetically related.

The eastern Erzgebirge hosts numerous magmatic-hydrothermal (greisen) deposits and prospects, e.g., Sadisdorf, Zinnwald/Cínovec, Altenberg, and Krupka (Fig. 1A). All the greisen systems of the eastern Erzgebirge are thought to be genetically related to the latest magmatic stages of the ATC, yet the individual occurrences show significant differences (Table 4) with respect to their size, style of mineralization (e.g., massive greisen, stockwork, vein, or breccia), and their metal tenor. Bulk-rock TiO₂ contents and trace elements in quartz reveal a regional trend (Table 4) of increasing magmatic differentiation toward the southeast. There, the Sadisdorf prospect (NW) and the Knöttel/Krupka district (SE) intrusions represent the least and the most evolved magmas, respectively (Schilka, 1986; Neßler, 2017; Peterková and Dolejš, 2019; this study).

The geometry of the intrusions and magmatic-hydrothermal mineralization styles vary significantly across the eastern Erzgebirge. Subvertical veins, stockworks, and explosive breccias are prominent features of the Altenberg and Sadisdorf deposits, where preexisting structures (e.g., cataclastic zones and lithological contacts) constrained and controlled the ore system (Seltmann, 1984; Schilka, 1986; Märtens et al., 1992). Conversely, flat-dipping veins, related to either thermal stress during cooling (Jackson et al., 1989; Neßler, 2017) or to emplacement within a compressive tectonic regime (e.g., Foxford et al., 2000), are dominant at the Zinnwald/Cínovec deposit.

At the province scale, the Sadisdorf deposit shows many features similar to those of the Hub stock (Krásno district) in the western Erzgebirge (not in the ATC). Similar TiO₂ and F contents, styles of mineralization (e.g., vein and massive greisen), architecture, mineral paragenesis (e.g., oxide to sulfides and late fluorite), FI homogenization temperatures, salinity, and overall spatial and temporal evolution (e.g., oxide to sulfide and late fluorite), indicate that the fluids, conditions, and processes were comparable in both systems. Dolníček et al. (2012) used fluid inclusion data to conclude that the Hub stock system evolved by boiling of a magmatic-derived fluid and accumulation of a low-density vapor in the apical portion of the intrusion. This conclusion is consistent with our observations at the Sadisdorf prospect (and with Korges et al., 2018, for the Zinnwald deposit), and it highlights the importance of phase separation in shallow Sn-W systems within the Erzgebirge and probably elsewhere.

This study provides insights into the anatomy of a well-preserved Li-Sn-(W-Cu) vein and greisen system and its magmatic-hydrothermal evolution in space and time. The Sadisdorf prospect shares many similarities with other Li-Sn greisen deposits in the (eastern) Erzgebirge, including high F content, regional structural control, and a shallow level of intrusion. Like other shallowly emplaced intrusion-related magmatic-hydrothermal systems, the fluid evolution at the Sadisdorf prospect involved the decompression of a single-phase magmatic-hydrothermal fluid (7.0 ± 4 wt % NaCl equiv and >340°C) during the main stage. However, unlike many other Li-Sn-(W-Cu) mineralized systems, the lack of systematic spatial changes in temperature and salinity of the fluids associated with proximal (oxide) as well as distal (sulfide) mineral assemblages indicate that fluid mixing and cooling were not the main factors controlling zonation. Instead, fluid-rock interaction and chemical changes along the fluid path are considered to have played an important role in this zonation. In contrast, cooling, with or without ingress of meteoric water, was a major factor during the waning stage—controlling the progressive inward overprinting and telescoping of the Sadisdorf mineral system.

We thank J. Richter, M. Lapp, and P. Paschke (LfULG) for providing access and logistical support to drill cores, sample material, and data. We are grateful to C. Kehrer (TUBAF) for access to the geoscientific collection at TUBAF, and we thank R. Würkert and M. Stoll for sample preparation at the HIF. M. Rudolph (HIF), S. Gilbricht (TUBAF), and V. Lüders (GFZ) are thanked for support with Raman spectroscopy, mineral liberation analysis, and fluid inclusion microthermometry. We are thankful for the critical comments from Economic Geology Associate Editor A.E. Williams-Jones, reviewer Pilar Lecumberri-Sánchez, and an anonymous reviewer, which contributed greatly to improving this paper. This project is funded as part of the DFG (Project No. 441189074) priority program Dynamics of Ore Enrichment (DOME; SPP 2238).

Dino Leopardi is a Ph.D. candidate at the TU Bergakademie Freiberg and a research associate at the Helmholtz Institute for Resource Technology, Germany. His Ph.D. project uses fluid inclusions analyses, B isotopes, and mineral geochemistry to understand the evolution of fluids and the processes controlling the distribution of light and mobile elements in granite hosted Li-Sn-W deposits. He received his Bachelor’s degree in geologic engineering from the Universidad Nacional Autónoma de México, Mexico, in 2014. Afterward, he worked as mine geologist for three years before obtaining his M.Sc. degree in bedrock geology from the Lund University, Sweden, in 2020.