Presently, very little is known about the lower crust of the Harz Mountains forming part of the Rhenohercynian Zone of the Variscan Orogen after Kossmat (1927; Fig. 1a). The Variscan basement of the Harz Mountains mainly consists of very low-grade Paleozoic metasedimentary rocks, which were predominately deposited between the Early Devonian and Early Carboniferous in marine settings from different sources (Wachendorf et al. 1995; Huckriede et al. 2004; Schwab and Hüneke 2008; Schwab and Ehling 2008; Linnemann et al. 2023), and intensely folded during the Variscan orogeny at < 330 Ma (Schwab 2008, and references therein). High-grade metamorphic rocks are exposed only in the Ecker Gneiss complex (Fig. 1b), which was long considered to represent an exhumed sliver of Cadomian basement beneath the Harz Mountains. This interpretation, however, has been disproved by Geisler et al. (2005) and more recently by Linnemann et al. (2023), showing that the protoliths of metasedimentary rocks of the Ecker Gneiss complex were supplied from Baltica, and deposited during the Devonian at < 410 Ma (410 ± 10; 414 ± 6; 406 ± 12 Ma: U–Pb ages of the youngest detrital zircon grains from three samples). Subsequently, during the Variscan orogeny, the Ecker Gneiss complex was affected by a granulite-facies metamorphic overprint (Appel et al. 2019).

Fig. 1
figure 1

a Position of the Harz Mountains in Germany. b Simplified geological map of the Harz Mountains with geochronological data of plutonic and volcanic rocks. Geochronological data: (1) U–Pb zircon (Baumann et al. 1991), (2) U–Pb zircon (Linnemann et al. 2023), (3) Ar–Ar of biotite and sanidine (Lippolt and Hess 1996), (4) K–Ar of biotite and muscovite (Goll et al. 1998), (5) Ar–Ar whole rock (Goll in von Seckendorff 2012), (6) Pb–Pb zircon (Obst et al. 2001)

During the Late Carboniferous to Early Permian, the Paleozoic rock units of the Harz Mountains were intruded by mafic and felsic plutonic rocks, comprising the Harzburg gabbro, the Oker, Brocken and Ramberg granites at ca. 300 Ma (Baumann et al. 1991; Goll et al. 1998; Linnemann et al. 2023), but also by N–S, NW–SE and E–W trending dikes of similar age (Fig. 1b). In the northeastern Harz Mountains, low-grade sedimentary rocks of Devonian age, the Wissenbach slates, were intruded by E–W trending, steep dipping dikes of the Bode dike system, prior to emplacement of the Ramberg pluton, as indicated by field relationships (Schust 1958). This dike system is made up by a variety of subvolcanic rocks comprising kersantite, fine-grained granite porphyry, microgranodiorite and microgranite (Tietz 1995, 1996), and locally contains abundant xenoliths up to metre size, comprising metamorphic rocks and granitoids. Presently, nothing is known about the composition of the granitoid rocks, their geotectonic setting and intrusion ages, i.e. whether the granitoid protoliths intruded during the pre-, syn- or post-Variscan evolution. In this context, we note that pre-Variscan and Variscan granitoids are nowhere exposed within the Rhenohercynian realm of Germany, neither in the Harz Mountains nor in the Rhenish Massif (Fig. 1a). In this study, we present the first set of data of granite xenoliths from the Bode dike system, comprising petrographic observations, whole-rock geochemical data, results of zircon U–Pb dating and Hf isotope analyses. These data in combination will provide new information about the composition, origin and evolution of the deeper crust of the eastern Harz Mountains. By comparison with data from adjacent geological units, these will also place new constraints on the pre- and post-Variscan evolution in Central Europe.

Geological setting

The Bode dike system is exposed in the northeastern part of then Harz Mountains, close to the city of Thale, mostly west of the Ramberg pluton (Fig. 1b). It comprises two subparallel E–W trending dikes, a southern branch made up by the kersantite of Treseburg, and a northern branch dominated by granite porphyry (Figs. 1, 2). The northern branch, our study area, is well exposed along the northern slope of the Bode stream valley (Fig. 2), where it transects Wissenbach slate of middle Devonian age (Eifelian), intercalated by numerous diabase sills and dikes.

Fig. 2
figure 2

Field relationships of the composite northern Bode dike in the Bode stream valley. a Geologic map with sample sites, and b cross section (modified after Tietz 1996). The Bode dike system results from three phases of magma injection: (1) margins made up of granite porphyry, (2) microgranodiorite, and (3) microgranite

The northern branch is 3–8 m wide, about 12 km long, and hosts several stages of magma injections (Fig. 2). The first stage is reflected by a dense zone of fine-grained granite porphyry, which is free of xenoliths and commonly forms the margin of the dike, both hanging and footwall against the surrounding Wissenbach slates. Subsequently, the same dike was injected in its centre part by melts of dacitic and rhyolitic composition, both rich in xenoliths (Tietz 1995, 1996). In the field, both phases show a microgranitic structure, and can be distinguished from each other by their different colours, with the granodioritic phase showing a dark, and the microgranitic phase a bright colour. Sharp and gradational contacts between microgranite and microdiorite suggest a fast sequence of injection, and locally point to magma mingling either within the dike and/or immediately prior to magma injection (Tietz 1995, 1996). Composite dikes, as described above, only occur in the area west of the Ramberg pluton over a length of almost 600 m, and microgranitic and microgranodioritic varieties with xenoliths just over 350 m along strike along the northern slope of the Bode stream valley, north of the cliff “Blaue Klippe” (Fig. 2).

Presently, no radiometric ages are available for the Bode dike system. Field relationships indicate that it is younger than the surrounding Wissenbach slates (Eifelian) and older than the Ramberg pluton (Schust 1958; Fig. 1b). The age of the Ramberg pluton is constrained by results of K–Ar dating of biotite (296 ± 7 Ma) and K-feldspar (300 ± 7 Ma) presented by Goll et al. (1998), and a recently published U–Pb zircon age of 295 ± 5 Ma (LA-ICP-MS; Linnemann et al. 2023). We note that all these ages are significantly older than a zircon U–Pb age of 283 ± 3 Ma (also LA-ICP-MS), previously presented by Zech et al. (2010).

The xenolith suite included in the microgranite and microdiorite varieties of the northern Bode dike comprises metasedimentary rocks, dioritoids (autoliths), amphibolites, paragneisses, micaschists and granitic rocks (Tietz 1996; this study). Leucogranite xenoliths up to cubic metre in size occur abundantly in the microgranodioritic intrusion (Figs. 2 and 3).

Fig. 3
figure 3

Field relationships of xenoliths in the Bode dike. Coarse-grained granite xenolith of ca. 25 cm size, and small quartzite xenolith in a grey microgranodioritic matrix

Sample description

The granite xenoliths investigated in this study were all sampled from the microgranodioritic variety of the Bode dike exposed in several cliffs along the northern slope of the Bode stream valley (Fig. 2). The granitoids are mostly leucocratic, and affected by a different degree of ductile and brittle deformation, showing all transitions from non-deformed granite to orthogneiss (Figs. 3, 4). Based on thin section observations, two types of granite can be distinguished: (1) granitoids with ductile deformation features (samples GK1, GK7, GK8), and (2) undeformed granites with granophyric textures (samples GK2 and GK5; Fig. 4). The granitoids are medium- to course-grained, commonly show a foliation (more or less pronounced) and distinct mylonitic textures, reflected by dynamically recrystallized quartz ribbons surrounding feldspar-rich domains dominated by perthitic K-feldspar and minor plagioclase (Fig. 4a, b). Samples with granophyric textures (GK2 and GK5) are characterised by euhedral quartz phenocrystals, surrounded by a relatively fined-grained matrix dominated by granophyric K-feldspar-quartz intergrowths, locally surrounding plagioclase phenocrystals (Fig. 4c, d).

Fig. 4
figure 4

Scans and photomicrographs (crossed nicols) of granite xenoliths with different structures from the Bode dike: a, b deformed granitoid, sample GK7; c, d granophyric granite, sample GK5. a, b Sample GK7 is a leucocratic, well-foliated granite gneiss, transect by mylonite zones with dynamically recrystallized quartz. It mainly consists of quartz (Qtz) and perthitic K-feldspar (perth Kfsp), (c, d) Sample GK-7 contains euhedral quartz phenocrystals (Qtz) within a fine-grained matrix dominated by K-feldspar-quartz granophyric intergrowths

Analytical techniques

U–Pb dating and zircon imaging

Zircon grains were recovered from five granite xenoliths (GK1, GK2, GK5, GK7, GK8). Sample GK1 was processed and analysed at University Göttingen (UG), and the other four samples at Karlsruhe Institute of Technology (KIT). The samples were crushed with Jaw crusher and steel disc mill to grain sizes < 500 µm, and the heavy mineral fraction enriched by panning (only at KIT) or by heavy liquids and magnetic separation at UG. Finally, zircon grains were manually picked from concentrates under ethanol, mounted on double-sided tape. At KIT, the zircon grains were sputtered with Au for 10 s, and imaged for their morphologies by back-scattered electron (BSE) microscopy using a TESCAN VEGA2 scanning electron microscope (at the Department of Petrology). Representative zircon images are shown in Fig. 5. Subsequently, the same grains were mounted with epoxy and grinded to expose their centre parts (also done at UG). The polished grains were imaged again by BSE at KIT, and CL (cathodoluminescence) at UG, to get information about their internal zoning (Fig. 5 and ESM—electronic supplement material). Based on these images, the most pristine zircon grains/domains were selected for in situ U–Pb dating and Hf isotope analyses.

Fig. 5
figure 5

Back-scattered electron images of zircon grains found in granophyric granite (sample GK5), and granitoids (samples GK7 and GK8). Images show zircon morphologies (numbers 100, 111, 211 etc. represent face indices), and internal structures (images with circles, representing laser spot positions), xo-xenocryst zircon. Numbers beneath zircon images represent no. of analyses and 206Pb/238U age. Note that in sample GK5, two groups of zircon can be distinguished, based on typologies and ages (for details see text)

Uranium–Pb analyses were performed by laser ablation–sector field–inductively coupled plasma–mass spectrometry (LA-SF-ICP-MS) at UG and KIT. Detailed information about analytical conditions in each lab is presented in ESM (Table S1). In each lab, zircon grains of unknown age were analysed together with reference zircon (for details see ESM: Table S2). At KIT, all raw data were corrected offline using an in-house MS Excel© spreadsheet program (Gerdes and Zeh 2006, 2009). A common Pb correction based on the interference and background-corrected 204Pb signal, and a model Pb composition were applied (Stacey and Kramers 1975). At UG, data processing was carried out with the software URANOS of Dunkl et al. (2008). The results of measurements of unknowns and reference zircon grains are presented in ESM (Table S2). Concordia diagrams were plotted by means of the software ISOPLOT 3.75 (Ludwig 2012).

Hf isotope analyses

Lutetium-Hf isotope analysis was carried out with a Resolution M-50 193 nm ArF Excimer laser system coupled to a Thermo Scientific multicollector (MC)-SF-ICP-MS (Neptune Plus) at FIERCE Frankfurt am Main, Germany. The analytical protocols used are the same as described in detail by Gerdes and Zeh (2006), and Zeh and Gerdes (2012; and reference therein). Detailed operating conditions for Lu–Hf isotope analyses and results of standard measurements are presented in ESM (Tables S1 and S3). Multiple measurements of reference zircon GJ1 and Temora-1 during the analytical session yielded 176Hf/177Hf ratios of 0.2820002 ± 0.000022 (2σ S.D.), and 0.282701 ± 0.000046 (2σ S.D.) respectively, in agreement with published values (Woodhead and Hergt 2005). For calculation of the epsilon Hf (εHft), the chondritic uniform reservoir (CHUR) was used as recommend by Bouvier et al. (2008); 176Lu/177Hf and 176Hf/177Hf of 0.0336 and 0.282785, respectively, and a decay constant of 1.867 × 10−11 (Scherer et al. 2001; Söderlund et al. 2004). All two-stage Hf model ages (TDM) were calculated by applying 176Hf/177Hf = 0.283181 ± 0.00023 (n = 46) and 176Lu/177Hf = 0.038055 for the depleted mantle (DM) evolutionary line (average MORB composition of Atlantic and Indian Oceans of Chauvel and Blichert-Toft 2001), resulting in a depleted mantle (DM) evolutionary line ranging from + 14 (today) to zero (at 4.56 Ga). Crustal evolutionary trends were modelled by applying 176Lu/177Hf = 0.0113 for continental crust (average of Taylor and McLennan 1985, and Wedepohl 1995). For all zircon xenocrysts, initial 176Hf/177Hf, εHft and TDM were calculated using the 207Pb/206Pb ages obtained for the respective zircon domains (ESM: Table S2).

Whole-rock geochemical analyses

Whole-rock geochemical analyses were carried out on powders obtained by jaw crushing and Achate disc milling from 4 granite samples. For analysis, 0.5 g sample powder was mixed with 5 g SPECTROMELT A 12 (66% di-lithium tetraborate/34% lithium metaborate) and fused at a burner station. Major and minor element analyses on fusion discs were obtained using a Bruker S4 Explorer wavelength-dispersive X-ray fluorescence (XRF) spectrometer at the Laboratory for Environmental and Raw Materials Analysis (LERA), Institute of Applied Geosciences at KIT, Germany. Matrix effects were corrected automatically by the Bruker software. The relative analytical error for major and minor elements is 1% and 1–8%, respectively. Volatile components are reported as loss of ignition (LOI).

Trace elements were analysed subsequently by LA-SF-ICP-MS (193nm ArF Excimer laser, Analyte Excite + , Teledyne Photon Machines) coupled to a Thermo Scientific Element XR instrument at KIT, on the same fusion discs used for the WDXRF analyses. The fusion discs were measured together with reference glasses NIST-SRM612 (primary standard), BIR1, BHVO2 and BCR2G (secondary standards). On each fusion disc, five spots were measured with a diameter of 85 µm alternating with measurements of reference material. The analyses were obtained with Al-H cones using the following instrument parameters (for standards and fusion discs): ablation duration = 25 s on peak after 20 s background measurement, laser repetition rate = 10 Hz, laser fluence = 7 J/cm2; mixed Ar–He–N2 carrier gas (Ar = 0.77 L/min, He cell gas = 0.30 L/min, He cup gas = 0.22 L/min, N2 = 12 mL/min), RF power = 1125 W. The oxide formation rate monitored as 254UO2/238U was < 0.08% during the analytical session. Raw data were processed by means of the software GLITTER (van Achterberg 2000), using SiO2 as an internal standard. The geochemical data of unknown rock samples and reference materials are presented in ESM (Table S4). Values represent the mean of five in situ fusion disc analyses. The errors are commonly < 2% (1σ mean) for most trace elements with concentrations > 1 µg/g.


Geochemical data

Whole-rock geochemical data were obtained from four granite xenoliths (granophyric granite: GK2, GK5; granitoids: GK7, GK8). The results are shown in Figs. 6 and 7. All analysed granites show high silica (SiO2 = 72.5 to 75.1 wt%) and alkali oxide contents (Na2O + K2O = 4.2 to 8.7 wt%). Based on major element composition, the granites are classified as peraluminous to subaluminous, magnesian and calc-alkalic (GK7, GK8) or alkali-calcic (GK2, GK5) in composition (Fig. 6a–c). The trace element patterns of all samples normalised to primitive mantle are similar, and characterised by relative depletion in Ba, Nb, Ti, Sr and P, and enrichment in Pb, ± U and Th (Fig. 6e). The chondrite-normalised REE patterns are also similar. They reveal negative europium anomalies (Eu/Eu*N = 0.27–0.5), enrichment of LREE over HREE [(La/Yb)N = 7.1–13.2] and a moderate fractionation of HREE [(Gd/Yb)N = 1.4–2.3]. In discrimination diagrams for geotectonic setting, all samples plot in the field for volcanic arc granites (Fig. 7a, b).

Fig. 6
figure 6

Various classification diagrams for granites based on ac major elements, de trace elements (black dots: granitoids GK7, GK8, grey dots: granophyric granites GK2, GK5). a Shand index plot with discrimination fields after Maniar and Piccoli (1989). The dividing line between I- and S-type granites is after Chappell and White (1974). b FeOt/(FeOt + MgO) versus SiO2 (wt%) classification diagram (Frost et al. 2001). The Fe-number (Fe*) dividing line is after Frost and Frost (2008). c Modified alkali-lime index (Na2O + K2O–CaO in wt%) versus SiO2 (wt%) classification diagram (Frost et al. 2001). d Rare-earth element patterns of granites normalised to C1 chondrite after McDonough and Sun (1995). e Trace element composition of granites normalised to primitive mantle after McDonough and Sun (1995)

Fig. 7
figure 7

Discrimination diagrams a Rb-(Y + Nb) and b Nb-Y for granites after Pearce et al. (1984). (black dots: granitoid samples GK7, GK8; grey dots: granophyric granites GK2, GK5)

Results of zircon U–Pb dating

Zircon grains from five xenoliths could be recovered for U–Pb dating, mostly more than 50 grains per sample, but only 5 grains from sample GK2 (granophyric granite). These five grains gave late Permian to Mesozoic 206Pb/238U ages < 261 Ma (ESM: Table S2), and some analyses show a high degree of discordance (e.g. grain U228), suggesting significant Pb loss after zircon growth. In contrast, zircon grains of the granitoid samples GK7 and GK8 yield concordia ages of 418.9 ± 1.8 Ma (n = 17 out of 27), and 417.7 ± 1.8 Ma (n = 16 out of 25), respectively (Fig. 8; Table 1), which are interpreted to date the time of magma crystallisation; also supported by oscillatory zircon zoning patterns (Fig. 5) and zircon Th/U ratios between 0.2 and 1.4 (ESM: Table S2). In situ U–Pb analyses of magmatic zircon grains/domains of granitoid sample GK1 reveal a relatively wide scatter in 206Pb/238U ages between 416 ± 6 and 368 ± 5 Ma. Assuming a coherent magmatic population, nine out of the 20 analyses provide a mean 206Pb/238U ages of 393 + 4.8/− 1.7 Ma (calculated by “TuffZircAge" algorithm of Ludwig and Mundil 2002; Fig. 8b). A few zircon analyses from the three granitoid samples gave significantly older 207Pb/206Pb ages, ranging from 1451 to 559 Ma in sample GK1, 2652 to 1401 Ma in sample GK7, and from 2920 to 1497 Ma in sample GK8. These grains represent xenocrysts and indicate involvement of older crustal matter of Archean to Proterozoic age during granite formation.

Fig. 8
figure 8

Results of U–Pb dating of zircon from granite xenoliths of the Bode dike represented in Wetherill diagrams (a, c-f)—MSWD—Mean squared weighted deviation, c. + e.—of concordance and equivalence (for details see text)

Table 1 Summary of rock types, zircon ages and Hf isotope compositions

Sample GK5 (granophyric granite) contains a complex zircon population, which can be subdivided in two major groups, based on U–Pb ages and zircon morphologies (Figs. 5 and 8e). Group I comprises euhedral zircon grains dominated by {101} pyramids, which yield Devonian 206Pb/238U ages ranging from 404 ± 6 to 391 ± 7 Ma (similar to zircon ages obtained from the granitoid samples GK7 and GK8; Fig. 8c, d). The three oldest analyses give a concordia age of 402.4 ± 3.6 Ma. The youngest age of 391 ± 7 Ma was obtained from a rim (spots 230 and 231; see Fig. 5), perhaps formed during a somewhat younger magmatic event. At least some of the Devonian zircon grains/domains were affected by serious Pb loss, as indicated by significant differences in 206Pb/238U ages obtained by double analyses of the same zircon zone (e.g. spots 240 and 241 in Fig. 5). Group II zircon grains are dominated by {211} pyramids (more or less corroded), and reveal Late Carboniferous to Early Permian ages (Figs.5 and 8e). The most reliable U–Pb analyses, i.e. from grains with a high degree of concordance (90–100%) yield 206Pb/238U ages between 308 ± 5 and 285 ± 4 Ma (n = 8). Five of these analyses give a concordia age of 296.4 ± 2.3 Ma, which is interpreted to represent magma crystallisation. Ages between c. 390 and c. 296 Ma perhaps result from a creeping, differential Pb loss, which affected the Devonian zircon population during Early Permian magmatism. In summary, combined information from zircon images and U–Pb ages suggests that the granophyric granite GK5 emplaced and crystallised during the Late Carboniferous–Early Permian, and that the parental magma was formed by the re-melting of Devonian granite basement (for more discussion see chapter 5.1: “Crustal evolution beneath the Harz Mountains”).

Results of Hf isotope analyses

Hafnium isotope compositions were measured on zircon grains of three samples, granophyre granite GK5, and the granitoids GK7 and GK8. Magmatic zircon grains of all three samples show overlapping 176Hf/177Hft ratios (Table 1). These range from 0.282439 ± 0.000043 (sample GK5–group I) to 0.282455 ± 0.000067 (sample GK7), and correspond to εHf410Ma between − 2.4 ± 2.7 and − 3.3 ± 1.4 (or εHf295Ma = − 5.0 ± 1.2), and two-stage Hf model ages between 1.25 ± 0.14 Ga and 1.30 ± 0.06 Ga (all errors are 2 sigma)—(Fig. 9; ESM- Table S3). Significantly lower 176Hf/177Hft ratios between 0.280909 and 0.281935 were only obtained from zircon xenocrysts, corresponding to Hf model ages between 2.17 and 3.29 Ga (ESM).

Fig. 9
figure 9

Age versus εHft diagram showing the results of Lu–Hf isotope analyses of zircon grains from Bode dike xenoliths (red, blue, green circles). Data from the Stavelot–Venn Massif (Willner et al. 2013) and the Ruhla Crystalline Complex (Rögis quartzite: Zeh and Gerdes 2010) are shown for comparison. CHUR- chondritic uniform reservoir, DM- depleted mantle evolution, L/H = 176Lu/177Hf = 0.0113, grey arrow—crust evolutionary trend. TDM-two-stage Hf model age


Crustal evolution beneath the Harz Mountains

Results of zircon U–Pb dating of xenoliths from the Bode dike reveal that the basement of the eastern Harz Mountains, hidden beneath very low-grade sedimentary rocks of middle Devonian age, hosts granitic rocks, which were formed during two stages of magmatic evolution (Table 1). Older granites, affected by a different degree of ductile deformation, emplaced and crystallised during the Early Devonian between 419 and 393 Ma, and younger granites with granophyric textures at ca. 295 Ma during the Late Carboniferous–Early Permian.

Combined results of whole-rock geochemistry and zircon U–Pb–Hf isotope analyses suggest that the magmatic protoliths of the granitoids intruded into a supra-subduction zone (magmatic arc) setting, and underwent assimilation and fractionation during emplacement within the crust. This interpretation is backed by the mildly peraluminous, calc-alkaline character, highly fractionated REE patterns with negative Eu-anomalies (pointing to fractionation in the stability field of plagioclase in relative shallow magma chambers), negative Nb–Sr–Ti and positive Pb anomalies in primitive mantle normalised diagrams (Fig. 6), and by discrimination diagrams for geotectonic setting (Fig. 7). Assimilation of older crust is indicated by zircon xenocrysts with 207Pb/206Pb ages between 2920 and 559 Ma found in three xenoliths (Fig. 5, ESM Table S3), and reworking of older crust by subchondritic εHf400Ma values between − 0.8 and − 4.6. These values could result either from mantle-wedge enrichment by “sediments” during previous subduction (e.g. Laurent and Zeh 2015; Couzinie et al. 2016, and references therein), and/or from assimilation of crustal rocks during magma ascent. Two-stage Hf model ages indicate that the reworked crust was derived from a depleted mantle source at ca. 1.2 Ga, on average (Fig. 9). Finally, we note that zircon grains of the samples GK7 and GK8 show a predominance of {110} over {100} faces (Fig. 5), suggesting crystallisation at relatively low temperatures, mostly < 700 °C according to the classification scheme of Pupin (1980). These temperatures are slightly lower than zircon-in-melt saturation temperatures of 759 to 771 °C, based on whole-rock compositions of the investigated samples and the calibration of Harrison and Watson (1983).

The granophyric granite GK5 crystallised at 296.4 ± 2.3 Ma, which is suggested by a concordia age obtained from 5 grains of group II zircon dominated by {211} pyramids (Table 1, Figs. 5 and 8). This age reflects magma crystallisation during the Late Carboniferous–Early Permian, and overlaps within analytical error with an U–Pb zircon age of 295.0 ± 4.5 Ma, recently presented for the Ramberg pluton by Linnemann et al. (2023). This overlap and field relationships indicate that the Bode dike system formed immediately prior to Ramberg pluton emplacement. Zircon grains recovered from the granophyre sample GK2 yields late Permian to Triassic 206Pb/238U ages < 260 Ma. These ages are geologically meaningless, and most likely result from Pb loss. The finding of both Devonian and Permian zircon grains in sample GK5 further suggests that the granophyric granites result from partial melting of Devonian granites at depth during Late Carboniferous–Early Permian heating. This interpretation is also backed by similar normalised trace element patterns (Fig. 6) and overlapping Hf isotope compositions (Table 1, Fig. 9). Partial crust melting was most likely caused by mantle upwelling related to crust extension (Fig. 10), which affected wide parts of Central Europe during the Late Carboniferous–Early Permian (e.g.Ziegler 1990; van Wees et al. 2000; Wilson et al. 2004; McCann et al. 2006 and references therein). We note that the age obtained for the granophyric granite xenoliths overlaps intrusion ages derived for granites, gabbros, and intermediate to felsic dikes of the Harz Mountains (Fig. 1), and for diorites, granites and volcanic rocks of the Thuringian Forest (Zeh and Brätz 2000; Zeh et al. 2000; Lützner et al. 2021).

Fig. 10
figure 10

Model explaining the formation of the granites found in xenoliths of the Bode dike of the eastern Harz Mountains. a Late Silurian to Early Devonian granite emplacement in a magmatic arc system caused by NW-ward subduction of the Rheic Ocean beneath Avalonia already attached to Baltica. b Partial melting of pre-Variscan basement beneath the Harz mountains caused by magmatic underplating during Late Carboniferous to Early Permian extension

Implications for Rheic Ocean closure

The Early Devonian intrusion ages (419–392 Ma) obtained from granitoid xenoliths of the Bodegang dike overlap geochronological data of felsic igneous rocks from the adjacent Mid-German Crystalline Zone, and Northern Phyllite Zone. Our ages are only slightly younger than emplacement ages of 444–423 Ma obtained from orthogneisses of the Ruhla Formation of the Mid-German Crystalline Zone (Brätz 2000; Zeh and Will 2010), which emplaced metasedimentary rocks of Paleozoic age (maximum depositional age at < 435 Ma; Zeh and Gerdes 2010), and from the drill core Saar 1 (Sommermann and Satir 1993). In the Saar 1 drill core, the granite gneisses are overlain by a succession of middle to upper Devonian marine sediments deposited after 380 Ma (Zimmerle et al. 1976), indicating that the Late Silurian–Early Devonian magmatic arc was subsided beneath sea level during the opening of the Rhenohercynian Basin. Overlapping intrusion ages of 413–398 Ma were obtained from several other units of the Mid-German Crystalline Zone, e.g. Central Gneiss Unit of the Ruhla Crystalline Complex (Brätz 2000), Rotgneisses of the Spessart Crystalline Complex (Dombrowski et al. 1995), the eastern Odenwald (Reischmann et al. 2001) and the western Odenwald crystalline complexes (Dörr and Stein 2019). In the western Odenwald Silurian/Devonian granitoids locally intruded granitoids of Cadomian age (566 Ma) as has been shown by Dörr and Stein (2019). Volcanic counterparts of the Silurian–Devonian plutonic rocks, i.e. andesites and rhyolites with an age of ca. 433 Ma have also been described from the Northern Phyllite Zone by Sommermann et al. (1994).

Isotope–geochemical data reveal that most of the Silurian/Devonian granitoids and felsic volcanics of the Mid-German Crystalline Zone and Northern Phyllite Zone are calc-alkaline, and show typical magmatic arc signatures. Neodym-Sr isotope data further indicate reworking of older crust during granite emplacement (Dombrowski et al. 1995; Brätz 2000). These features are identical to those derived from the Bode dike xenoliths during this study, and suggest that granitoids beneath the eastern Harz Mountains and in the adjacent rock units formed part of a coherent Silurian–Devonian magmatic arc system (for compilations see Franke 2000; Zeh and Will 2010; Dörr and Stein 2019). Hafnium model ages of ca. 1.2 Ga derived from the Bode dike xenoliths are very likely meaningless in a geological context, and rather reflect mixing of mantle melts with crustal matter, comprising pre-Silurian igneous and metasedimentary rocks. This is directly indicated by zircon xenocrysts with Archean to Proterozoic ages found in several Bode dike xenoliths. The source of the zircon xenocrysts is ambiguous. Based on their ages, these could result from the reworking of Cadomian orthogneisses (as described from the Odenwald by Dörr and Stein 2019), but also from (meta)sedimentary rocks deposited at < 565 Ma (206Pb/238U age of the youngest xenocryst in sample GK1). We note that detrital zircon grains with ages of 580 ± 20 Ma are extremely rare (< 2%) in quartzites of presumed Silurian–Devonian protolith age, as exposed in the Ecker Gneiss complex of the Harz Mountains (Geisler et al. 2005; Linnemann et al. 2023), in the Rhenish Massif (Eckelmann et al. 2014), and in the Ruhla Formation of the Mid-German Crystalline Zone (Zeh and Gerdes 2010). However, such grains occur relatively abundant (ca. 20%) in metagreywackes of presumed Neoproterozoic age of the Northern Phyllite Zone (Warstein paragneiss; Dörr and Stein 2019), and in lower Cambrian to Silurian sediments of the Brabant Massif (Linnemann et al. 2012; Willner et al. 2013), having an Avalonia provenance. Thus, the finding of xenocrysts with ages of 570–600 Ma, c. 1.5 Ga and 2.6–2.9 Ga in the Bode dike xenoliths points to the assimilation of Avalonia-derived sediments, which is also supported by overlap in age-Hf isotope data (Fig. 9). We further note that Hf model ages of ca. 1.2 Ga derived from zircon grains in the Bode dike xenoliths overlap well with Nd model ages of 1.03–1.34 Ga estimated for Silurian intrusive and volcanic rocks of the Brabant Massif (Linnemann et al. 2012).

In summary, comparison of zircon age-Hf data and whole-rock isotope–geochemical results suggests that Silurian to early Devonian granitoid rocks constituting the basement beneath the eastern Harz Mountains formed part of a magmatic arc system, which resulted from westward subduction of the Rheic Ocean beneath Avalonia, which at 420–395 Ma was already connected to Baltica (see model in Zeh and Gerdes 2010; Linnemann et al. 2023), and overlain by Baltica-derived clastic sedimentary rocks (Fig. 10a).

During Variscan collision at < 330 Ma, the Rheic Ocean and the Rhenohercynian “back arc” basin became finally closed (see model in Zeh and Gerdes 2010), and the Silurian–Devonian magmatic arc fragmented and overthrust by very low-grade metasedimentary rocks of Paleozoic age (mostly Emsian to Visean), resulting in the NW-vergent fold-and-thrust architecture of the Harz Mountains. As shown by Franke (2000) and Linnemann et al. (2023), the Paleozoic sedimentary units of the Harz Mountains were supplied from very different sources: from Baltica-Avalonia, from Cadomia, and the Late Devonian–Carboniferous flysch from the Mid-German Crystalline Zone. Finally, all rock units became juxtaposed in a giant accretionary prism and were affected by very low-grade metamorphic overprint. The only exceptions are exotic slivers of Variscan high-grade paragneiss with Baltica affinity, the so-called Ecker Gneisses (Fig. 10b). It is likely that the Devonian granites, exposed as xenoliths in the Bode dike, were also deformed during Variscan collision, although there is no direct evidence for it.


  1. 1.

    The Bode dike of the eastern Harz Mountains contains xenoliths of granite composition, in addition to metasedimentary rocks and amphibolites.

  2. 2.

    The granite protoliths were formed during two different magmatic events, as indicated by combined results of petrography, zircon U–Pb dating, Hf isotope analyses and whole-rock geochemistry.

  3. 3.

    Deformed granites are witnesses for the existence of a Late Silurian–Early Devonian magmatic arc system, which resulted from NW-ward subduction of the Rheic Ocean beneath Avalonia-Baltica between 440 and 390 Ma.

  4. 4.

    Granites with granophyric textures result from re-melting of Silurian–Devonian granites, due to mantle upwelling in an extensional regime, which affected wide parts of Central Europe during the Late Carboniferous–Early Permian.