Introduction

Zirconium (Z = 40) and hafnium (Z = 72) are elements of group 4b of the Periodic Table, have similar ionic radii (0.84 and 0.83 Å, respectively, in eightfold coordination, Shannon 1976) and are quadrivalent under all geological circumstances. Consequently, the crystal chemical properties of both elements are very similar, and the Zr/Hf value remains constant during the majority of geological processes (Hanchar and Hoskin 2003). Wang et al. (2010) summarized literature data for 2200 electron probe analyses of granitic zircon and found a median of 1.48 wt% HfO2. They also reported that the Zr/Hf value (by weight) is in the range of 37–38, i.e., slightly higher than the value proposed for the whole Solar System (Zr/Hf = 34.1 ± 0.3 by weight) by Patzer et al. (2010). Pupin (2000) reported the results of 9000 electron probe analyses of magmatic zircon and found a decrease in the Zr/Hf median value from 60–70 in mantle-derived plagiogranites, hypersolvus alkali granites and alkali syenites, to 40–45 in hybrid calc-alkaline granitoids, and finally to 35–37 in peraluminous granites and migmatites.

Belousova et al. (2002) provided 656 LA-ICP MS and electron probe analyses of zircon from rocks ranging from kimberlites through basalts and granitoids to alkali syenites and found 0.7–2.7 wt% HfO2 with an enormous variation in each rock type. Nevertheless, the medians were found to increase from syenites and basalts (0.70 and 0.72 wt%, respectively) through lamproites, alkali syenites and carbonatites (1.17, 1.22 and 1.23 wt%, respectively) to granitoids and kimberlites (1.51 and 1.55 wt% HfO2, respectively). Rubatto (2002) reported approximately 9000–12,000 ppm Hf (ca. 1.1–1.4 wt% HfO2) in zircons from granulites and eclogites. Grimes et al. (2015) found 7000–17,000 ppm Hf (ca. 0.8–2.0 wt% HfO2) in zircons from mid-ocean ridge basalts, continental arcs and post-collisional granitoids, without a statistically significant difference between particular tectono-magmatic sources. All these data indicate that the Zr/Hf values in zircon from the upper crust resemble those of the whole Solar System, while zircon from the majority of mantle-derived rock is Hf-depleted. Some increase in the Hf content in zircon along with the fractionation of a composite granitoid pluton was noted from the McMurry Meadow pluton, California (Sawka 1988), the Sweetwater Wash pluton, California (Wark and Miller 1993), Boggy Plain pluton (Hoskin et al. 2000), and Spirit Mts. batholith, Nevada (Claiborne et al. 2006).

In contrast to the above mentioned rock types, a strong enrichment in Hf – thus a decrease in the Zr/Hf values in zircon – was noted in many fractionated granites and pegmatites (Tab. 1). Cheng et al. (1992) reported zircon with up to 18.7 wt% HfO2, i.e., Zr/Hf = 2.2 by weight, from the Beauvoir granite, France. Wang et al. (2000) found a decrease in the Zr/Hf values in zircon from 30–95 in biotite I-type granites through 7–52 in A-type alkali granites to 3.9–38 in arfvedsonite granites (with a max. 12.37 wt% HfO2) in Laoshan, eastern China. In Cornwall, the Zr/Hf values in zircon decreased from 50–110 in biotite and tourmaline granites to 6–80 in the most evolved topaz granites (Meldon Aplite and Megiliggar Rocks, Breiter et al. 2016). A strong increase in the Hf contents in zircons in fractionated rare metal granite systems was noted from Suzhou (A-type, China, Wang et al. 1996), Yichun granite (S-type, China, Huang et al. 2002), Beauvoir (S-type granite) and Cínovec (A-type, Czech Republic/Germany, Breiter and Škoda 2012). An increase in the HfO2 contents in hafnian zircon during the evolution of the Tanco pegmatite was reported by Van Lichtervelde et al. (2009). An evolution from Hf-rich zircon to hafnon, HfSiO4, was reported from LCT-type pegmatite of Koktokay, China (Yin et al. 2013).

A decrease in the Zr/Hf ratios was registered not only within a magmatic suite but also within individual zircon crystals. Hf enrichment of the zircon rim compared to its core was reported for example from several granite plutons in Japan (Hoshino et al. 2010) and from granites in Elba, Italy (Gagnevin et al. 2010). It was confirmed as a common feature of zircon crystals from some A-type rare-metal granites (Breiter et al. 2014).

In the whole rock, the proposed Zr and Hf abundances increased from 3.6 and 0.106 ppm in the C1 chondrite (Lodders 2010) through 10.8 and 0.3 ppm in the mantle (Palme and O'Neill 2004) to 190 and 5.8 ppm in the upper continental crust (Taylor and McLennan 1995), respectively. This, with respect to a high uncertainty of all these values, gives a nearly constant Zr/Hf value in the range of 35–40. In other words, there is very small, if any, Zr/Hf fractionation among primordial Earth, mantle and common crustal rocks including calc-alkaline granites.

It was a general consensus that the whole-rock Zr/Hf values remain stable in comagmatic suites (Lyakhovich and Shevaleevskii 1962; Kosterin et al. 1963; Broska et al. 1990). This is probably valid in basic, intermediate and poorly evolved granitic rocks. For example, in the Bohemian Massif, the majority of Paleozoic granitoids including Cambrian/Ordovician granites and orthogneisses, Variscan I-type granites of the Central Bohemian pluton, Variscan two-mica granites of the Moldanubian (South Bohemian) pluton and Variscan S-type biotite granites in the Erzgebirge, show Zr/Hf values over a narrow range of 24–35 (with several outliers to 41, unpublished authorsʼ data), i.e., similar to, or only slightly lower than, the Solar value of 34. In contrast, a systematic decrease in Zr/Hf values during fractionation was reported from rare-metal granite suites (RMG, Table 1), such as the Beauvoir granite (from 17 to 6.4, Raimbault et al. 1995), Fichtelgebirge pluton, Germany (38 → 25, Irber 1999), Podlesí stock in Western Erzgebirge (from 21 to 9, Breiter et al. 2005), Landʼs End pluton, Cornwall (from 41 to 19, Müller et al. 2006) and Akchatau, Kazakhstan and Kukulbei, Transbaikalia (both Zaraisky et al. 2009). Similar processes should be expected in complex pegmatites but appropriate whole-rock data are missing.

Table 1 HfO2 contents and Zr/Hf values of zircon from strongly fractionated granites/pegmatites and Zr/Hf values of their host rocks (if available)
Full size table

The origin of Zr/Hf fractionation in RMG and pegmatites was explained by magmatic fractionation (e.g., Wang et al. 1996, 2000). Later, Kempe et al. (2000) questioned magmatic fractionation and proposed the origin of Hf-rich zircon from F-rich fluids associated with albitization. Furthermore, Johan and Johan (2005) explained the origin of the Hf-enriched rims of zircon crystals from Cínovec as a product of postmagmatic fluid-related fluorination. This view was supported by experimentally constrained fluid–melt distribution coefficients for Zr and Hf (London et al. 1988), which indicated a higher activity of Hf in fluids associated with a peraluminous melt. Nevertheless, Linnen and Keppler (2002) experimentally demonstrated a systematic decrease in Zr/Hf values in fractionated metaluminous and peraluminous melts and an increase in Hf contents in crystallizing zircon, i.e., the possibility of fully magmatic origin of Hf-rich zircon. Finally, Zaraisky et al. (2009) experimentally determined a difference in the Zr/Hf value between zircon and metaluminous and peraluminous melts with different contents of fluorine as follows: all experiments revealed that Zr/Hfzircon > Zr/Hfmelt, i.e., the residual melt (and later population/zones of zircon) will be systematically enriched in Hf.

Altogether, there exist satisfactory literature accounts regarding Hf contents and Zr/Hf ratios in both mineral zircon and whole-rock granitoids, and the Zr/Hf value is considered to be a reliable indicator of the evolution of fractionated magmatic suites (Hoskin and Schaltegger 2003; Zaraisky et al. 2009). However, surprisingly, there is only poor knowledge about the relationship between the Zr/Hf ratios in zircon and those in its parental rock. Only Claiborne et al. (2006) reported a notable decrease in the whole-rock (WR) and zircon Zr/Hf values during the evolution of the Spirit Mountain batholith, Nevada; the Zr/Hf values in zircon were systematically higher than those in the host granite. Later Erdmann et al. (2013) mentioned two coexisting populations of zircon in the Nunavut granite in Canada: the older population had a higher Zr/Hf ratio and the younger population had a lower Zr/Hf ratio than the parental rock.

We decided to fill this gap in this article, aiming (i) to define the evolution of HfO2 contents and Zr/Hf ratios in fractionated granitic intrusive and volcanic suites, (ii) to compare the Zr/Hf values in zircon and its parental rocks of different geochemical affiliations and grades of fractionation; and (iii) to evaluate the possible use of Zr/Hf values for the interpretation of rare-metal granite systems. The Teplice caldera in the Eastern Erzgebirge seems to be an ideal object for such a study; here, two deep boreholes allow to study vertical evolution of zircon and its parental rock in two volcanic suites (S-type Schönfeld rhyolite and dacite followed by A-type Teplice rhyolite) and a suite of rare-metal Cínovec granite (A-type). Preliminary data on zircon from the Teplice rhyolite (Breiter and Škoda 2009) were complemented by new zircon and WR data from all volcanic rocks (Table 2). Vertical variations in the composition of zircon in granites from borehole CS-1 have been reported by Breiter and Škoda (2012), and WR data have been reported by Breiter et al. (in review). Here, we extend our attention to zircons from the greisen and minor, but genetically important, granite facies (Table 3) and their comparison with the whole-rock Zr/Hf values.

Table 2 Stratigraphy of the Teplice caldera fill (borehole Mi-4, according to Breiter et al. 2001) with a list of the studied zircon samples
Full size table
Table 3 A vertical section through the Cínovec granite cupola (borehole CS-1, according to Breiter et al. in review) with a list of the zircon samples studied
Full size table

Geological setting and samples

Late Variscan granites and associated acidic volcanic rocks of the Krušné Hory/Erzgebirge form a ca. 80 km long NE–SW- oriented belt, which intruded into the Variscan crystalline complex (mainly Neoproterozoic to Early Paleozoic metapelites and metagranitoids) of the Saxo-Thuringian domain in the NW part of the Bohemian Massif (Hoth et al. 1995; Förster et al. 1999; Cháb et al. 2010; Linnemann and Romer 2010) in a relatively short period from ~330 to 310 Ma (Förster and Romer 2010; Ackerman et al. 2016). The following two types of magma were generated and emplaced synchronously during this event (Breiter 2012): (1) strongly peraluminous P-rich (S-type) melts dominating the western and central parts of the Erzgebirge, and (2) slightly peraluminous P-poor melts (A-type), forming both intrusive and volcanic rocks mainly in the eastern part of the area. Both magmatic suites culminated with strongly fractionated subvolcanic granite intrusions followed by Sn + W ± Li-Nb-Ta mineralization of the greisen type.

The dominant geological structure of the Eastern Erzgebirge is the Teplice caldera with a size of ​​approximately 40 × 20 km (Fig. 1). A detailed knowledge of its stratigraphy (Breiter 1997; Breiter et al. 2001 and references therein) permits the establishment of the sequence of Variscan magmatic events in the Eastern Erzgebirge as follows:

  1. Step 1–

    intrusion of peraluminous biotite granites (Fláje and Telnice plutons in Bohemia, Niederbobritsch in Saxony);

  2. Step 2–

    eruption of peraluminous rhyolite and dacite lavas and tuffs with sedimentary intercalations (so-called Schönfeld Unit) at Schönfeld (Saxony) and Mikulov (Bohemia);

  3. Step 3–

    three stages of sub-aluminous rhyolite volcanism of the A type (“Teplice rhyolite”), gradually passing from explosive to effusive in nature;

  4. Step 4–

    collapse of the rhyolite volcano and creation of the Teplice caldera;

  5. Step 5–

    intrusion of ring dykes of granite porphyry;

  6. Step 6–

    intrusion of biotite granites of the A-type without hydrothermal activity (Preiselberg body in Bohemia, Schellerhau body in Saxony);

  7. Step 7–

    intrusion of strongly fractionated ore-bearing granite of the A-type in several pulses (biotite and zinnwaldite granites at Cínovec and several other localities).

Fig. 1
figure 1

Geological sketch of the Teplice caldera

Full size image

Two deep boreholes realized between 1960 and 1985 provide a good opportunity to study the magmatic evolution of volcanic and subvolcanic rocks of the caldera and vertical differentiation of the post-caldera rare-metal granites. Borehole Mi-4 at the village of Mikulov (Figs. 1 and 2) was 1200 m deep and penetrated the volcanic fill of the caldera in the interval 0–924.5 m. The oldest member of the caldera fill, the Schönfeld Unit, consists of basal rhyolite (BR, 71–73 wt% SiO2) at the depths ranging from 924.5 to 868.3 m and of dacite (DC, 61–67 wt% SiO2) at the depths ranging from 868.3 to 609.6 m. It can be classified as a slightly peraluminous calc-alkaline (S-type) rock similar to common granitoids of the “older intrusive complex” of the Erzgebirge (sensu Tischendorf 1989). The basal rhyolite shows a slight increase in Si and a decrease in Ti, Mg, Nb, Y and REE in upwards direction, while the dacite shows an opposite trend.

Fig. 2
figure 2

Simplified sections through the caldera fill in borehole Mi-4, and through the Cínovec granite cupola in borehole CS-1

Full size image

The subaluminous (A-type, 75–77 wt% SiO2) “Teplice rhyolite” extends from the depth of 609.6 m to the actual surface. Within this profile, three distinct units were recognized – TR1 at a depth of 609.6–490.6 m, TR2 at a depth of 490.6–191.4 m, and TR3 at a depth of 191.4 m to the actual surface. All TR units show reversed vertical differentiation of trace elements, i.e., starting with relatively Rb-, Nb-, Th-, and Y-rich and Ti-, Sr-, Zr-, and Ce-poor basal parts, passing gradually upwards to a less acidic Rb-, Nb-, Th-, and Y-poor and Ti-, Sr-, Zr-, and Ce-rich upper part (details in Breiter 1997). Ring dykes of granite porphyry (Fig. 1, not recorded in borehole Mi-4) represent the latest cumulate-rich portion of rhyolite magma (Breiter 1997; Müller et al. 2005). Twelve samples of zircon and 29 whole-rock samples were analyzed from this borehole (Table 2). Additionally, three samples of granite porphyry from the outcrops of the ring dykes were analyzed.

Borehole CS-1 (Figs. 1 and 2) was drilled to a depth of 1596 m in the center of the Cínovec granite cupola from 1960 to 1962 to study the structure of the pluton subjacent to the actually mined Sn-W deposit (Štemprok and Šulcek 1969). Zinnwaldite granite sensu lato (hereafter ZiG) was shown to exist in several textural and mineral varieties to the depth of 735 m, and biotite granite (BtG) continues at greater depths (Table 3). Geochemically, the Cínovec BtG is evolved, slightly peraluminous P-poor A-type granite, while the comagmatic ZiG is even more strongly enriched in F, Li, Rb, Zr, Th, HREE, Sc, Sn, W, Nb and Ta, and depleted in P, Ti, Mg, Ca, Ba and Sr. Whereas the BtG is chemically generally homogeneous, the ZiG underwent a strong differentiation with upwards increasing contents of LILE and rare elements (Breiter 2012). Twenty samples of zircon from borehole CS-1 and another 6 from the southern part of the Cínovec cupola were analyzed (Table 3). For the evaluation of the whole-rock Zr/Hf values, 254 chemical analyses from a previous study (Breiter et al. in review) were exploited.

Analytical methods

Zircon grains were studied in polished thin sections. Back-scattered electron (BSE) images were obtained on a TESCAN scanning electron microscope housed in the Institute of Geology CAS in Praha before the study of internal zoning of the individual zircon grains and their relative position to rock-forming minerals (Fig. 3).

Fig. 3
figure 3

BSE photomicrographs of zircon: a, zircon with a patchy core and a slightly zoned rim, basal rhyolite of the Schönfeld Unit, sample 3219, borehole Mi-4, depth of 897 m; b, an irregular, partly resorbed zircon grain with an inclusion of glassy matrix, sample 3212, dacite of the Schönfeld Unit, borehole Mi-4, depth of 637 m; c, columnar zoned zircon crystal, sample 3212, dacite of the Schönfeld Unit, borehole Mi-4, depth of 637 m; d, an oscillatory zoned zircon crystal, sample 3200, TR2 unit of the Teplice rhyolite, borehole Mi-4, depth of 234 m; e, a zircon grain with an irregular, patchy core and Hf-enriched rim, sample 3205, TR2 unit of the Teplice rhyolite, borehole Mi-4, depth of 443 m; f, a zircon grain with an irregular patchy core and Th, U, Y-enriched rim, sample 3208, TR1 unit of the Teplice rhyolite, borehole Mi-4, depth of 513 m; g, a slightly zoned crystal with inclusions of silicates, sample 3531, granite porphyry, outcrop at Loučná; h, two zircon crystals with mixed patchy and regular zoning, sample 4692, biotite granite, borehole CS-1, depth of 988 m; i, irregularly zoned zircon crystal (right up) in association with Ca, Th-fluorocarbonate (white), both rimmed by an Fe-oxide aggregate, sample 4692, biotite granite, borehole CS-1, depth of 988 m; j, irregularly patchy zoned Hf-rich zircon grain (bright areas are enriched in Y), sample 4672, zinnwaldite granite of the canopy, borehole CS-1, depth of 24.5 m; k, colander-like zircon crystal with a xenotime rim, zinnwaldite granite, sample 4688, borehole CS-1, depth of 559 m; l, a columnar zircon crystal with a xenotime epitaxis, quartz-zinnwaldite greisen, sample 4971, borehole CS-1, depth of 130 m; m, a colander-like zircon grain (dark gray) surrounded by zoned xenotime, quartz-zinnwaldite greisen, sample South 7, abandoned mine of Cínovec-south; n, zircon grains, some of them zoned, in association with scheelite (Sch), quartz-zinnwaldite greisen, sample CS1/10, borehole CS-1, depth of 53 m; o, a zircon crystal associated with cassiterite crystals (Cas) and a Ce-fluorocarbonate grain (CeF), quartz-zinnwaldite greisen, sample 4971, borehole CS-1, depth of 130 m. All scale bars are 50 μm

Full size image

Elemental abundances of W, P, As, Nb, Ta, Si, Ti, Zr, Hf, Th, U, Y, La, Ce, Pr, Nd, Sm, Gd, Dy, Er, Yb, Al, Sc, Bi, Mn, Fe, Ca, Pb, Mg, S, and F in zircon were determined using a CAMECA SX100 electron probe microanalyser (EPMA, Masaryk University and Czech Geological Survey, Brno) in wavelength dispersion mode. Zircon was analyzed at an accelerating voltage and beam current of 15 keV and 40 nA, respectively, and with a beam diameter ranging from 1 to 2 μm. The following reference materials were used: Ca – fluoroapatite, La – LaPO4, Ce–CePO4, Pr – PrPO4, Nd – NdPO4, Sm – SmPO4, Gd – GdPO4, Dy – DyPO4, Er – ErPO4, Yb – YbPO4, U – metallic U, Th – CaTh(PO4)2, Pb – vanadinite, Zr and Si – zircon, As – lammerite, Y – YAG, Hf – metallic Hf, Mn – Mn2SiO4, Fe – almandine, Al – gahnite, S – SrSO4, F – topaz, Sc – ScVO4, Ti – titanite, Nb – columbite from Ivigtut, Ta – CrTa2O6, W – metallic W, Bi – metallic Bi, Mg – MgAl2O4. Raw data were processed using the X-Phi matrix correction routine of Merlet (1994). Empirically determined correction factors were applied to the overlapping X-ray lines. Empirical formulae of zircon were calculated on the basis of 4 atoms of oxygen in a formula unit (4 O apfu).

The WR major element analyses (wet technique) were performed at the Czech Geological Survey, Praha. The weights of the analyzed samples were 2–4 kg. Replicate analyses of the international reference material (JG-3 granodiorite; Geological Survey of Japan) yielded an average error (1 σ) of ±1% with respect to the recommended values (Govindaraju 1994). Trace elements were determined by ICP mass spectrometry following lithium metaborate/tetraborate fusion of a 0.2 g sample in the laboratory of ACME, Vancouver, Canada. (Details can be found at http://acmelab.com/.)

Results

Zircon shapes

Zircon in the Schönfeld rhyolite (Fig. 3a) and dacite (Fig. 3b, c) forms mostly homogeneous or slightly zoned isometric to long columnar crystals 30–120 μm in size, sometimes with inclusions of glassy matrix. Bright inner zones are relatively slightly enriched in Y and U. Occasionally, a very subtle Hf enrichment from the core to the rim was found. Isometric oscillatory zoned grains were found only rarely.

Zircon in the Teplice rhyolite mostly forms euhedral, isometric to shortly columnar crystals 20–100 μm, rarely up to 400 μm in size. In all samples, oscillatory zoned grains (Fig. 3d) are associated with grains showing patchy-zoned (metamictized) cores and more regular rims (Fig. 3e, f). The bright rim of the crystal in Fig. 3f is enriched in Y, Th and U, while bright rim of the crystal in Fig. 3e is slightly enriched in Hf. Zircon in granite porphyry usually forms homogeneous, shortly columnar crystals 40–150 μm in size or their fragments. Occasionally, grains have patchy cores and more regularly zoned rims (Fig. 3g). Some grains contain numerous inclusions of glassy groundmass.

Zircon crystals from the biotite granite from Cínovec are mostly 10–50 μm in size, enclosed in dark mica. They are often distinctly zoned and contain contrasting blebs of exsolved mineral phases close to xenotime and thorite in composition (Fig. 3h). Zircon associated with thorite is surrounded by a secondary halo of iron hydroxides (Fig. 3i). Zircon from the zinnwaldite granite features isometric crystals of similar sizes enclosed in quartz and feldspars, but some of the later crystallized grains occupy interstices between other minerals. Some of the crystals are irregularly zoned in BSE images and/or contain numerous tiny cavities (Fig. 3j). Thin rims of xenotime around zircon crystals (Fig. 3k) are common.

Zircon grains from the greisen, mica-free granite facies and feldspathite feature a maximum size of 120 μm and are texturally similar to those from the zinnwaldite granite. The internal composition is patchy, with domains enriched in Hf, Y and Th. In the few zoned grains, the cores are patchy and heterogeneous, porous and enriched in uranium, whereas the rims are compact, homogeneous and enriched in Hf. Epitaxis and overgrowth of xenotime on zircon is common (Fig. 3l, m). During the postmagmatic stage, zircon was usually stable when in contact with the ore-bearing fluid from which cassiterite and scheelite crystallized (Fig. 3n, o).

Chemical composition of zircon

Most published electron probe analyses of zircon included, besides major elements Si, Zr, and Hf, only the following other elements: U, Th, Y, P (Pupin 2000), and occasionally Ca, Al, Sc and Yb (Cheng et al. 1992; Johan and Johan 2005; Hoshino et al. 2010; Výravský et al. 2017 etc.). Trace element analyses provided by other methods like SHRIMP (Claiborne et al. 2006; Grimes et al. 2015) or LA-ICP-MS (Nardi et al. 2013) did not include major elements Si and Zr: the Zr/Hf values were derived from the assumption of an ideal occupation of the zircon crystal lattice. Data regarding the contents of Nb, Ta, W, As and F are generally sparse (Förster 2006; Van Lichtervelde et al. 2009; Förster et al. 2011; Breiter et al. 2006, 2009, 2014; Nb also Grimes et al. 2015).

We analyzed altogether 31 chemical elements in zircon, but to report all of them is beyond the scope of this paper. Below, we describe Hf contents and Zr/Hf values in detail, while other elements are commented only in the case of their unusually high contents. Complete analyses of typical zircon grains from all studied rocks are listed in Table 4. To meet both mineralogical and geochemical conventions, we refer to the contents of Hf and other minor elements and the Zr/Hf values in apfu in Fig. 4, while in all other figures these data are presented in weight units to be simply compared with the WR values. As a reminder, the solar and whole-Earth Zr/Hf value by weight equals 34 (Patzer et al. 2010).

Table 4 Chemical composition of zircon as determined by EPMA analysis
Full size table
Fig. 4
figure 4

Plots revealing ratios of chemical constituents in the zircon grains studied: a, Zr/Hf versus U; b, Zr/Hf versus Th; c, Zr/Hf versus Y; d, Zr/Hf versus F

Full size image

In the calc-alkaline Schönfeld rhyolite, Hf contents in zircon varied in the interval of 0.75–2.1 wt% HfO2 and the Zr/Hf value between 21 and 54 (median 41) (Fig. 5a), while in the comagmatic dacite 0.9–1.7 wt% HfO2 and Zr/Hf = 26–56 (median 43) were found. The contents of other minor elements in zircon from both rocks are relatively low compared to those in other studied rock types: usually less than 1.5 wt% Y2O3, <1 wt% UO2 and ThO2 (Fig. 5b–d).

Fig. 5
figure 5

Vertical zoning of the chemical composition of zircon in the two boreholes Mi-4 (a–d) and CS-1 (e–h)

Full size image

The chemical composition of zircon from the Teplice rhyolite was much more variable (Fig. 5a–d): 0.9–3.2 (median 1.4) wt% HfO2 and Zr/Hf value by weight 13–57 (mostly between 20 and 40, median 32). Wide differences were found within the individual eruptive units. The Zr/Hf values increase clearly upwards from 31 to 42 within TR1, increase upwards with some variation from 33 to 38 in TR2, and slightly decrease upwards from 29 to 26 in TR3 (all values are medians). In the case of TR3, only the basal part of the whole unit has been preserved, and an unknown part has been eroded. The contents of other minor elements in zircon are higher than those in the Schönfeld unit and highly variable, mostly in the range of 0–5 wt% Y2O5, 0–2 wt% UO2 and ThO2, 0–1.5 wt% P2O5, 0–0.8 wt% F, occasionally up to 0.4 wt% As2O5 and 0.15 wt% Sc2O3. Many grains are obviously metamictized and altered, containing 2–4 (up to 8) wt% of non-formula element oxides Al2O3 + FeO + MnO + CaO. Analytical totals of such grains fall down to 90 wt%.

Granite porphyry representing the latest emplaced magma (remnants after extraction of previous rhyolite magmas of TR1–TR3) contains zircon grains with only 0.8–2.0 (median 1.4) wt% HfO2 and Zr/Hf values of 21–55 (mostly 35–50, median 41). The Zr/Hf ratios in zircon in granite porphyry are thus higher than those in the youngest rhyolite unit TR3. The contents of minor and trace elements are the lowest of all the studied rock types: <0.4 wt% Y2O3, <0.2 wt% ThO2 and UO2. Zircon from granite porphyry never shows metamictization-induced alteration, giving analytical totals >98 wt% in all cases.

Zircon from the Cínovec biotite granite is slightly more evolved than that from the Teplice rhyolite (Fig. 5e–h). It contains 1–4.5 (median 2.4) wt% HfO2 and its Zr/Hf values are in the range of 10–55 (mostly 15–25, median 20). The contents of minor elements are highly variable, mostly in the range of 0.5–2 wt% Y2O3, approximately 1 wt% UO2 and less than 0.5 wt% ThO2, but individual grains/zones containing several wt% of each of these oxides were found. Grains with unusually high Zr/Hf values (>30) are poor in U and Th, giving analytical totals of 97–100 wt%. As such, they probably did not undergo metamictization-induced alteration. Their low Hf contents seem to be a primary magmatic feature.

Zircons from the Cínovec zinnwaldite granite and greisen differ significantly from all the previously described. They are geochemically evolved and enriched not only in Hf but also in rare metals Nb, Ta, W, Bi, and Sc, the non-formula elements Al, Ca, and Fe, and also in fluorine. Another remarkable feature is the strong vertical differentiation of zircon chemistry following vertical fractionation of parental granite within the 735 m deep section. Hf contents in zircon generally slightly increase upwards from 2–6 (median 3.3) wt% HfO2 at the base of the zinnwaldite granite body at the depth of 550–735 m to 3–10 (median 7.0) wt% HfO2 at the depth of 0–100 m; the Zr/Hf values decrease accordingly from 7–21 (median 12) to 3.3–12 (median 5.1). Mica-free granite and feldspathite at the depths of 284–369 m form a local positive deviation from this general trend reaching up to 12 wt% HfO2 (Zr/Hf = 2.8), while partly sericitized granites at the depth of approximately 200 m contain zircon with less than 3.4 wt% HfO2 (Zr/Hf = 9–16). Zircon from greisen bodies located in the uppermost part of the granite cupola differs from zircons from neighboring granites in its slightly lower Hf contents, mostly in the range of 2–5 (exceptionally up to 10, median 3.9) wt% HfO2 and Zr/Hf = 4–11 (median 8).

Minor and trace element contents reach their maxima in zircon from the zinnwaldite granite: 1–5 wt% Y2O3, 1–3 wt% Yb2O3, 1.5–4 wt% UO2, 1–3 wt% ThO2, 1–3.5 wt% P2O5, 0.3–2.5 wt% Sc2O3, 0.5–1 wt% Nb2O5, around 1 wt% Bi2O3, etc. Referenced values represent common contents; occasionally substantially higher contents of all mentioned elements were found. Contents of up to 1 wt% Ta2O5, 4 wt% WO3 or 1.5 wt% As2O5 were found in some grains, too.

Many zircon grains from zinnwaldite granites and greisens have been metamictized and altered. They display low analytical totals (occasionally down to 87 wt%) and contain 2–6 wt% of non-formula oxides of Al, Fe, Mn and Ca. The contents of fluorine in granitic zircon range within 0–2.5 (median 0.6) wt% F; in zircon from greisen, fluorine is ubiquitous, ranging 0.5–2.5 (median 0.7) wt% F. The contents of F generally increase with the decreasing analytical totals.

It should be mentioned that we found many zircon-like grains strongly enriched in P, As, Y, U and/or Th at Cínovec. These should be classified as transitional phases between zircon and xenotime, chernovite, coffinite and thorite (Förster 2006; Breiter et al. 2009; Förster et al. 2011). Such grains are usually relatively poor in Hf and are not discussed in this paper.

Whole-rock Zr and Hf contents

Whole-rock chemical compositions of granites from borehole CS-1, including Zr and Hf contents, have been published by Breiter et al. (in review). Major element and some trace element contents of volcanic rocks from borehole Mi-4 have been published and discussed in detail by Breiter (1997); high-quality Zr and Hf data were acquired for this paper. WR analyses of typical samples are shown in Table 5. Below, we comment only on recent Zr and Hf data obtained by ICP-MS (Fig. 6).

Table 5 Representative results of whole-rock analyses
Full size table
Fig. 6
figure 6

A comparison of Zr/Hf values in zircon grains (medians) and parental rocks along the two boreholes studied: a, Mi-4; b, CS-1. Data from granite porphyry are shown for comparison

Full size image

Within the calc-alkaline Schönfeld Unit, dacite displays slightly higher Zr and Hf contents (169–263 ppm and 4.6–7.4 ppm) than comagmatic rhyolite (105–119 ppm Zr, 3.4–3.7 ppm Hf); the Zr/Hf values increase from rhyolite to dacite from 30–32 to 36–39. A-type Teplice rhyolite contains 90–185 ppm Zr and 4.5–6.6 ppm Hf with relatively lower Zr/Hf values of 17–31. Inspecting the Mi-4 profile in more detail (Fig. 6a), the contents of Zr increase within each of the three eruptive units TR1–TR3 from the base to the top (93 → 149, 97 → 185, 123 → 146 ppm), while Hf contents are generally uniform. The Zr/Hf values thus increase within each eruption unit as follows: 21 → 29, 17 → 31, and 19 → 22.

Within the Cínovec granite cupola, the deeper seated biotite granite contains 120–150 ppm Zr and 6–8 ppm Hf with Zr/Hf values in the range of 18–22. Only a subtle general decrease upward (21 → 19) was encountered but local deviations of 24 and 13 were found. In contrast, the zinnwaldite granite shows a strong vertical differentiation: the early crystallized zinnwaldite microgranite (ZiGm) contains 80–120 ppm Zr; the later crystallized and vertically strongly fractionated zinnwaldite granite, however, shows an upwards decrease in Zr contants from approximately 80 ppm at depths of 600–735 m to approximately 40 near the actual surface. Because the contents of Hf vary insignificantly within the range of 5–8 ppm, Zr/Hf values mimic the Zr contents: they range between 12 and 14 in xenoliths of zinnwaldite microgranite and systematically decrease from 11 at the depth of 735 m to 8 at the depth of 100 m. In the uppermost part of the cupola, in the canopy, the Zr/Hf values equal approximately 4.5 (Fig. 6b). Worth mentioning is the fact that Zr/Hf values in greisens (6–7) equal those in the neighboring zinnwaldite granites (5–8).

Discussion

Zr/Hf values during differentiation of granitic melt

The Zr/Hf value is considered to be a reliable indicator of fractionation of granitic melt (Hanchar and Hoskin 2003; Claiborne et al. 2006; Zaraisky et al. 2009). We decided to test its functionality on the studied magmatic suites from the Teplice caldera (Fig. 7). To better understand the specific environment of A-type magmas, the following two typical peraluminous suites were added into figures: late orogenic peraluminous two-mica granites of the Moldanubian pluton, southern Czech Republic/Austria (Breiter 2016) and a rare-metal zinnwaldite granite from the Podlesí stock, Western Erzgebirge, Czech Republic (Breiter et al. 2005). A comparison of the Zr/Hf values with the most common indicator of fractionation, the K/Rb values (Fig. 7a), shows a good agreement in the general trend shifting from approximately K/Rb = 190 and Zr/Hf = 30–40 to K/Rb = 10 and Zr/Hf = 5–10. The scatter in the values of both ratios among individual samples of the biotite and zinnwaldite granites is caused by the different time of crystallization of major mineral hosts of K, Rb (K-feldspar, mica) and Zr, Hf (zircon), and probably also due to the inhomogeneous distribution of zircon in the rocks.

Fig. 7
figure 7

A comparison of whole-rock Zr/Hf values with that of other common indicators of magma fractionation: a, Zr/Hf versus K/Rb; b, Zr/Hf versus Nb/Ta; c, Zr/Hf versus Rb/Sr. Comparative data from the Moldanubian two-mica granites and zinnwaldite granites from the Podlesí stock are present authorsʼ unpublished data

Full size image

The correlation between the Zr/Hf and Nb/Ta ratios (Fig. 7b) is perfect in the A-type rocks from the caldera (Teplice rhyolite and all Cínovec granites) forming an array from ca. Nb/Ta = 13 and Zr/Hf = 30 to Nb/Ta = 1 and Zr/Hf = 5. We conclude that this perfect positive correlation strongly supports magmatic crystallization of major hosts of all mentioned elements, i.e., of zircon, rutile, columbite and Li-Fe micas. Peraluminous suites form an another, more widely scattered array with a relatively lower grade of Zr/Hf fractionation.

A combination of the Zr/Hf ratio with the Rb/Sr ratio (Fig. 7c) shows a clear negative correlation; all suites form one array beginning at Rb/Sr = 0.5 and Zr/Hf = 35–40, and ending at Rb/Sr = 800 and Zr/Hf = 5–13. Small contents of Sr in zinnwaldite granites (often <10 ppm) can be easily affected by late magmatic to post-magmatic reactions with fluids. This explains the wide scatter of the values in the zinnwaldite granite samples.

Figure 8 illustrates in detail the behavior of Zr and Hf during fractionation of magmatic suites within the Teplice caldera in coordinates Hf vs. Zr and Hf vs. Zr/Hf. This figure also shows common peraluminous granites from the Moldanubicum and the Podlesí stock for comparison. Granites from the Moldanubian pluton show a linear trend with a perfect correlation between Zr and Hf (Zr 240 → 50 ppm, Hf 8 → 1.5 ppm, Fig. 8a) and a gradual decrease in the Zr/Hf value (Fig. 8b) from 35 to 24. Rare-metal granites from the Podlesí stock are generally poor in Hf (1.5–3 ppm), and the decrease in the Zr/Hf values (23 → 9) is largely controlled by the decrease in Zr (50 → 17 ppm Zr). The peraluminous Schönfeld Unit from the Teplice caldera shows a Zr and Hf evolution similar to that in the Moldanubian granites: the decrease in both Zr (184–263 to 105–119 ppm) and Hf (4.6–7.4 to 3.4–3.7 ppm) contents from dacite to basal rhyolite is accompanied by a decrease in the Zr/Hf values from 36–39 to 30–32. Consequently, the moderately fractionated peraluminous suites (Schönfeld volcanic and Moldanubian two-mica granites) evolved via a regular decrease in both Zr and Hf, accompanied by a moderate decrease in the Zr/Hf values. The Podlesí stock of zinnwaldite granite represents water-rich residual melt influenced by explosive degassing and undercooling (Breiter et al. 2005) with a rather chaotic distribution of Zr, Hf and other HFSE (Breiter et al. 2006).

Fig. 8
figure 8

Whole-rock evolution of Zr and Hf contents and the Zr/Hf ratio: a, Zr plotted against Hf; b, Zr/Hf plotted against Hf. Data for the Moldanubian and Podlesí plutons are from Breiter (2016) and Breiter et al. (2005)

Full size image

The behavior of Zr and Hf in the Cínovec A-type granites is quite different: Both patterns show a considerable scatter in the Hf contents in both granites and greisens, which did not change during the entire evolution of the pluton, remaining in the range of 4.5–10.5 ppm Hf (medians 7–8 ppm Hf for all rock types). The decrease in the Zr/Hf values from approximately 22 to 5 is controlled almost exclusively by a depletion in Zr from 170 to 25 ppm. Within each rock type we found a good positive correlation between Zr and Hf (Fig. 8a), which is caused by the varying amount of zircon grains in the individual samples, i.e., m-size inhomogeneity of the zircon distribution in granite.

In the A-type Teplice rhyolite, Zr and Hf contents in the individual samples scattered but the general decrease in Zr/Hf values correlated well with volcanic stratigraphy (Breiter et al. 2001) and other indicators of magmatic fractionation (Fig. 7).

Zr/Hf values of zircon grains and their host rocks

The behavior of likely no other chemical element is linked with a single mineral as is the behavior of zirconium and hafnium linked with mineral zircon. This leads to a simple interpretation that the relationship between Zr and Hf found in zircon is applicable to the behavior of Zr and Hf in granite (magma) and vice versa (Butler and Thompson 1965). Zr and Hf may theoretically be fractionated by amphibole, clinopyroxene, garnet and titanite but the low contents of these minerals and low KD values minimized their real influence (Claiborne et al. 2006). Nevertheless, a detailed analysis of large data sets shows that the relationship between the Hf content (Zr/Hf ratio) in zircon and its parent rock is not straightforward.

Figure 9 shows that most of the analyzed zircon grains from the relatively less fractionated rocks (Schönfeld Unit, Teplice rhyolite and Cínovec biotite granite) yielded higher Zr/Hf values than the parent magmatic rock. Only in the case of the Cínovec zinnwaldite granite and its derivatives (greisens) the means of zircon Zr/Hf values approximately equal the host rock values. In this figure, average values for the granitoid zircon on the Earth (Wang et al. 2010), the area occupied by zircons from common latecollisional peraluminous two-mica granites from the Moldanubicum, Bohemian Massif (Breiter 2016), and for two successive granitoids (monzonite, leucogranite) from the Spirit Mountain batholith, Nevada (Claiborne et al. 2006) are also shown for comparison. The Earth average of granitic zircon and a great majority of zircon grains from the Moldanubicum also display higher Zr/Hf values than the corresponding parent rocks, similar to zircon from the Teplice caldera with an analogous grade of host rock fractionation. Interesting is the evolution of the Spirit Mountain batholith: both analyzed samples show a large dispersion of Zr/Hf values in zircon, but the median of Zr/Hf values from the less fractionated monzonite nearly equals that of the whole-rock values (47 in zircon vs. 45 in the rock). In contrast, all zircon analyses from the more fractionated leucogranite give much higher Zr/Hf values (25–68, median 38) than the whole rock value 20.5 (Claiborne et al. 2006).

Fig. 9
figure 9

Deviation of Zr/Hf values in zircon from Zr/Hf values in the parental melt. Data for the mean of granitic zircon on the Earth are combined from Wang et al. (2010) and Patzer et al. (2010), data for the Moldanubian pluton are from Breiter (2016). Data from the Spirit Mountain batholith, Nevada (whole scatter of individual zircon analyses and their medians from monzonite and leucogranite) come from Claiborne et al. (2006). The ideal correlation Zr/Hfzircon versus Zr/HfWR = 1 Black line) is shown as visual guide

Full size image

Granitoids do not contain any other real host of Zr and Hf besides zircon. Therefore, average Zr/Hf values in zircon should equal average Zr/Hf values in the whole rock. A possible explanation for the discrepancy shown in Fig. 9 is the presence of thin Hf-enriched rims on zircon crystals. Even a thin rim can be volumetrically significant (e.g., 3 μm rim of a 50 μm cubic core represents 29 vol.% of the grain). This model is, however, not consistent with the detailed measurements by Claiborne et al. (2006): the relatively Hf-enriched zircon rims still showed Zr/Hf values higher than those of the host leucogranite. An another, probably a more realistic, explanation is the existence of a population of very small, late-crystallized Hf-enriched zircon grains. These may be overlooked or, if found, they may not get analyzed due to their small sizes.

Strongly fractionated rocks, here represented by zinnwaldite granites, are Zr-poor, and the zircon saturation occurred locally at the very end of the crystallization of the melt; most zircon grains crystallized in interstices. These grains are small (approximately 20 μm) and not zoned; therefore, their analyses better match the whole-rock composition.

A strong negative effect of peraluminity on zircon and hafnon solubility was experimentally established by Linnen and Keppler (2002). Consequently, Zaraisky et al. (2009) explained experimentally the decrease in the Zr/Hf values during pronounced fractionation of F-enriched rare-metal and pegmatite melts. At 800 °C, 1 kbar, ASI = 1.2 and F content between 1 and 2 wt%, which is well comparable with the condition in the carapace of the Cínovec pluton, they found Zr/Hfzircon ~ 6, while Zr/Hfmelt ~ 4.3. This perfectly coincides with the most fractionated samples of granite from the top of the Cínovec cupola and their zircon (Fig. 9).

Zr/Hf fractionation, a key to correct interpretation of magmatic suite evolution

Variations in whole-rock Zr/Hf values along boreholes Mi-4 and CS-1 in combination with medians of Zr/Hf values in zircon are shown in Fig. 6. This figure can be analyzed from the following two perspectives: (i) conformity of both variables, i.e., similarity of the whole-rock and zircon Zr/Hf values, and (ii) continuous or discontinuous evolution of the Zr/Hf patterns, i.e., general geochemical evolution of the suites. In both cases, the two discussed boreholes provide contrasting examples.

  • Ad (i): At Cínovec (Fig. 6b), a good agreement is observed between the Zr/Hf values in zircon and the whole rock. Anomalously high values were encountered only in zircon grains from two samples of biotite granite. The shape of these zircon grains does not differ from that from other samples with “correct” Zr/Hf values; these samples perhaps reflect an accumulation of early zircon grains crystallized at the beginning of magma solidification. In contrast, almost all zircon grains from the Teplice rhyolite (Fig. 6a) exhibit much higher Zr/Hf values than the whole rock. In this case, larger zircon grains, grown up early in a deep magma reservoir, were probably analyzed. A part of Zr and Hf, still dissolved in the melt at the time of the eruption, were then incorporated into rapidly crystallizing microzircons (<1 μm) or remained dispersed in volcanic glass. The observed differences are consistent with the hypothesis of Erdmann et al. (2013) that zircon grains with higher Zr/Hf values grow from early water-unsaturated melts, while the late zircon grains with low Zr/Hf values crystallized from water-saturated residual melts. This may also explain the origin of some extremely Hf-enriched zircon grains at the top of the granite cupola.

  • Ad (ii): a large number of whole-rock samples from borehole CS-1 at Cínovec (Fig. 6b) sensitively indicates several petrologic boundaries, in particular at depths of 735 m, 522 m and 368 m. Biotite granite is richer in Zr and displays higher Zr/Hf values and a wider local Zr/Hf variability but exhibits no systematic vertical variation; all textural varieties of biotite granite contain zircons of identical chemical compositions. In contrast, zinnwaldite granite shows a distinct vertical Zr/Hf development but it is much less variable in detail than the biotite granite. Upward fractionation of the zinnwaldite granite reduced the Zr/Hf value from 10–11 at a depth of approximately 700 m to approximately 5 in the uppermost 100 m of the borehole. Higher Zr/Hf values at a depth of ca 380–500 m correspond to rafts of zinnwaldite microgranite, which represent the first portion of the solidified melt along the upper contact of the intrusion. This “carapace granite” (sensu Cobbing et al. 1986) was later broken and plunged down into viscous melt (Breiter et al. in rev). Vertical variation in Zr/Hf is in perfect accordance with upwards enrichment in lithophile and ore elements (Li, Rb, Nb, Ta and F, Breiter in rev.).

    Despite of the significantly lower number of samples analyzed from borehole Mi-4, independent vertical/temporal evolution of all three eruption units of the Teplice rhyolite is well constrained (Fig. 6a). The Teplice rhyolite evolved as three subsequent comagmatic magma batches. Each eruption started with geochemically more evolved material (low Zr/Hf) from the top of the reservoir and terminated with cumulate-enriched, geochemically primitive magma portion (high Zr/Hf) from the bottom of the reservoir.

Influence of magmatic fluids on Zr/Hf ratios

London et al. (1988) found experimentally that Dfluid/melt for Hf is higher than that for Zr at conditions relevant to rare-metal granites and pegmatites; they concluded that hydrothermal zircon may be Hf-enriched. Crystallization of Hf-rich zircon rims in A-type rare-metal granites from Na, F-rich fluids during albitization was proposed by Kempe et al. (2004). Particularly at Cínovec, the origin of Hf-rich zircon via metasomatic fluorination was explained by Johan and Johan (2005).

Recent detailed studies of natural systems provided the following different results: while hydrothermal rims on zircon from the Koktokay pegmatite, China, are extremely Hf-enriched (Yin et al. 2013), hydrothermal zircon from Tanco pegmatite, Canada, is Hf-impoverished compared to the foregoing magmatic zircon population (Van Lichtervelde et al. 2009). At Cínovec, an 85 m thick zone of mica-free granite with feldspathite intercalations appears at a depth of 284–369 m in borehole CS-1, i.e., approximately 100 m below the major greisen bodies. Mica-free granite is supposed to be a restite after exsolution of a F, Li, Sn-rich fluid which induced greisenization in the uppermost part of the granite cupola (Breiter et al. in rev). This rock contains zircon with 4–12 wt% HfO2 and Zr/Hf = 3–8, i.e., with a similar or slightly higher Hf content and lower Zr/Hf value than neighboring granites. Greisen bodies located within the uppermost 200 m of the borehole exhibit identical Zr/Hf whole-rock values and contain zircon with a chemical composition similar to that of the surrounding granites.

The strong enrichment in Nb (up to 4 wt% Nb2O5) and Ta (up to 1 wt% Ta2O5) in zircon from Cínovec, much higher than generally reported from zircon (Belousova et al. 2002, Nardi et al. 2013, Grimes et al. 2015), may arise suspicion of strong hydrothermal alteration, i.e., secondary enrichment in Nb, Ta, and potentially other HFSE including Hf. Nevertheless, Van Lichtervelde et al. (2009) found up to 4.7 wt% Ta2O5 in a U, Th-poor, i.e., non-metamict zircon from the Tanco pegmatite and then synthesized zircon with up to 3.7 wt% Ta2O5 in laboratory conditions (Van Lichtervelde et al. 2011). We conclude that the Nb, Ta-rich zircon grains need not have necessarily undergone a strong reaction with fluids and, as such, they retain primary Zr/Hf values. Altogether, no evidence of fluid-related enhancement of Hf has been found.

Possible influence of late fluids on Zr/Hf values in zircon

Zircon grains in the studied granites are usually enriched in radioactive elements U (mostly approximately 1 wt%, occasionally >4 wt% UO2) and Th (often 1–5 wt% ThO2). Some of them exhibit patchy zoning, an enrichment in the non-formula elements Al, Ca, Fe, Mn and Mg, and their analytical totals are reduced to 90–95 wt%. Low analytical totals down to 90 wt% and the uptake of water and non-formula elements into zircon are explained by an alteration of a previously radiation-damaged (i.e. metamictized) material in a fluid-driven reaction (Nasdala et al. 2009). Metamictization and related fluid-driven alteration affected mainly the cores of the studied grains (see typical examples in Fig. 3a, g), which are enriched in U and/or Th, while the rims are sometimes slightly enriched in Hf. Relatively low Hf in the core can be explained via crystallization from the initial melt poor in Hf, or via selective dissolution of Hf after metamictization. Zircon from the Teplice rhyolite is relatively poor in U and Th but also often exhibits features typical for metamict zircon.

We tried to resolve whether a reaction of metamictized zircon with aqueous fluid may alter the Zr/Hf value. Anderson et al. (2008) found the amorphous domain in natural metamict zircon to be relatively enriched in Hf compared to the associated highly crystalline zircon; the primary magmatic or late post-metamict origin of this enrichment could not be determined. The Zr/Hf value in a 619 Ma old zircon from the Eastern Desert, Egypt remained unaffected up to an alpha dose of 13.7 × 1018 events/g but strongly fluctuated above 20 × 1018 α-events/g (Geisler et al. 2003b).

According to the experiments, the Zr/Hf values are generally stable in contact with hot (350–650 °C) CaCl2- and AlCl3-bearing fluids (Geisler et al. 2003a) but a small decrease from 58 to 51 was observed at 175 °C and 1 M HCl-CaCl2 fluid (Geisler et al. 2002). According to Geisler et al. (2003a) and Schmidt et al. (2006), Zr migrates at only very short distances within a single grain after the dissolution of radiation-damaged domains of zircon, being incorporated into recrystallized matter again. Wood (2005) and Linnen and Cuney (2005) agreed that the solubility of Hf in aqueous fluids is higher than that of Zr but no experimental evidence for natural zircon is available. Altogether, it can be only speculated that a relative Hf impoverishment in metamictized zircon cores may be partly caused by selective dissolution in fluid.

According to Ewing et al. (2000), zircon from the Variscan granites (model age 320 Ma) will be actually metamict if it contains more than 0.4 wt% U (0.45 wt% UO2); about 75% of analyzed zircon grains from the Teplice caldera exceed this limit in their U contents, and 13% of analyzed grains exceed 3 wt% UO2. So, the major part of the studied zircon grains should be considered metamict. The content of 0.4 wt% U in 320 Ma old zircon corresponds to a cumulative alpha dose of ~4 × 1018 events/g. As shown in Fig. 10a, substantial contents of non-formula elements are characteristic for zircon grains with an alpha dose of >3 × 1018 α/g, which is in good agreement with the estimate of Ewing et al. (2000). At the same alpha dose, the analytical totals drop to 95–90 wt%. Figure 10b illustrates that the B-site elements (Si accompanied by P as a part of the isostructural “xenotime component”) are released more easily than the A-site elements; this indicates a relatively low mobility of Zr and Hf in damaged zircon. Approximately one fifth of the zircon grains analyzed have experienced an alpha dose exceeding 2 × 1019 α/g, which represents the threshold for possible Zr/Hf decoupling (Geisler et al. 2003b). Nevertheless, Fig. 10c indicates that the low Zr/Hf values in zircon from Cínovec granites are not correlated with the alpha dose. Thus, the low Zr/Hf values were not noticeably changed after metamictization.

Fig. 10
figure 10

Influence of radiation-induced damage (expressed using the alpha dose calculated for a model age of 320 Ma) and subsequent alteration on the chemical composition of zircon: a, The relation of CaO content and alpha dose shows that only zircon with a cumulative dose exceeding 3 × 1018 α/g incorporates non-formula elements like Ca (Al, Fe, Mn, Mg); b, the relation of Si + P (B-site elements) and Zr + Hf + U + Th + Y + REE + Sc (A-site elements) shows that the B-site elements (mainly Si) are more easily released from the zircon lattice than the A-site elements (mainly Zr and Hf); c, the plot of Zr/Hf against alpha dose indicates that these two parameters are not correlated. Not that in b, only elements were considered that are reliably bound in the zircon crystal lattice

Full size image

Conclusions

A detailed study of Hf contents and Zr/Hf ratios in zircon and its parental rock from the Teplice caldera led to the following generalized conclusions:

  • Zr/Hf values in the whole rock may serve as a sensitive indicator of the fractionation of evolved granitic melts (rhyolites, granites, pegmatites), only very weakly influenced by secondary fluid-related processes.

  • Zr/Hf values in individual zircon grains are scattered but their general evolution trend in a rock series is consistent with the evolution of the whole-rock Zr/Hf values.

  • Whole-rock contents of both Zr and Hf during the differentiation of S-type magma equally decrease, being accompanied by a moderate decrease in the Zr/Hf values. In the A-type suites, however, Hf contents remain nearly constant, and a strong decrease in the Zr/Hf values is controlled only by a decrease in Zr contents.

  • Mica-free granite facies at Cínovec, produced after the exsolution of F, Li-rich late magmatic fluids, contain zircon with very low Zr/Hf values; there is no evidence of preferred partitioning of Hf into fluids.

  • Quartz-zinnwaldite greisens show the same Zr/Hf whole-rock values and contain zircon identical with that in the neighboring granites in its chemical composition. There is no indication of an influence of F, Li-rich fluids on zircon during greisenization.

  • Actual results support the hypothesis by Erdmann et al. (2013) that zircon crystallizing from a water-unsaturated melt is Hf-depleted, while zircon crystallizing from a water-saturated melt is identical with the parental melt in its Zr/Hf values or is relatively Hf-enriched compared to the parental melt.

  • The most fractionated samples from the canopy of the A-type Cínovec pluton fit well the experiments by Zaraisky et al. (2009) for a slightly peraluminous, F-rich melt having Zr/Hf ~4.5–5.5 and containing zircon with Zr/Hf median of ~4.5–5.0.