JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 PAGES 1723–1739 1997 Mineralogical Evidence for Two Magmatic Stages in the Evolution of an Extremely Fractionated P-rich Rare-metal Granite: the Podlesı´ Stock, Krusˇne´ Hory, Czech Republic ˘´ ´ KAREL BREITER1∗, JIRI FRYDA1, REIMAR SELTMANN2 AND RAINER THOMAS2 ´ CZECH GEOLOGICAL SURVEY PRAHA, GEOLOGICKA 6, CZ-15200 PRAHA, CZECH REPUBLIC 1 2 GEOFORSCHUNGSZENTRUM POTSDAM, TELEGRAFENBERG A50, D-14473 POTSDAM, GERMANY RECEIVED JANUARY 1997; ACCEPTED AUGUST 1997 The Podlesı´ granite stock in the western Krusˇne´ Hory (Erzgebirge) Mountains, Czech Republic, represents an extremely fractionated, strongly peraluminous, F- and P-rich, rare-metal granite system of Late Variscan age. The stock, studied in drill core of 300 m length, is formed by albite–protolithionite–topaz granite (‘stock granite’, depth 50–300 m) and shows geochemical and textural zoning. The stock granite is rich in P (~0·5 wt %), F (0·5–1·2 wt %), Rb (~1000 ppm), Li (500–1000 ppm) and Cs (100–150 ppm), and poor in Ti, Mg, Fe, Ca, Sr, Ba, Zr and rare earth elements (REE). Within the uppermost part of the stock (depth 57–115 m), the stock granite is intercalated with albite–zinnwaldite–topaz ‘dyke granite’ layers of 0·5–7 m thickness and a few thin flat dykes of pegmatite. The dyke granite and pegmatite are even more enriched in P (~1 wt % P2O5), F (1·0–1·5 wt %) and Rb (up to 3000 ppm), and are also rich in Nb (up to 100 ppm), and Ta (up to 50 ppm). The chemical data for the rock-forming minerals show distinct differences between the stock granite and the dyke granite and pegmatite. No chemical zoning of the rock-forming minerals, either from the upper, rapidly cooled, or from the lower, slowly cooled parts of the stock granite was observed. These rock-forming minerals formed during only one stage of crystallization from a parental melt with moderate contents of P, F, and Li (0·5 wt % P2O5 in K-feldspar, 5–7 wt % F and 3 wt % Li2O in Li-mica). In contrast, alkali feldspars and zinnwaldite from the dyke granite and pegmatite show well-developed chemical zoning. The rims of K-feldspars are strongly enriched in phosphorus (up to 2 wt % P2O5), and rims of zinnwaldite grains are enriched in fluorine (8–9 wt % F). Both zinnwaldite and topaz from the dyke granite The behaviour of phosphorus in silicic melt and its effect on melt structure and properties have been the subject of intensive study during the last two decades (Mysen et al., 1981; London, 1992; Johannes & Holtz, 1996). ∗Corresponding author. Telephone and fax: 00420-2-5817390. e-mail: [email protected] Oxford University Press 1997 have the OH–F sites fully occupied by fluorine. Also, accessory phosphates were formed during late magmatic crystallization of the strongly P, F-enriched residual melt. Thus, the mineralogical data reflect the presence of two crystallization events within the dyke granite and pegmatite. The presence of two melts, parental and residual, the latter strongly enriched in phosphorus, has been also confirmed by study of melt inclusions in quartz from samples of the stock granite in its dyke-bearing uppermost part. Mineral, melt inclusion and whole-rock chemical data from all rock types provide evidence for two stages of granite evolution, namely crystallization from parental and residual melts. This means that, in addition to the granites in Beauvoir (France) and Argemela (Portugal), the Podlesı´ stock is another well-documented example of a two-stage evolution of a P-rich granite system. KEY WORDS: alkali feldspars; Li-rich micas; granites; Czech Republic INTRODUCTION JOURNAL OF PETROLOGY VOLUME 38 Phosphorus lowers solidus temperature, promotes expansion of the liquidus field of quartz at the expanse of albite, and increases the solubility of Al and high field strength elements (HFSE) in melt (London, 1995). Such enrichment of phosphorus and other volatiles in granitic melts may result from: (1) fractional crystallization (Yin et al., 1995; Charoy & Noronha, 1996), (2) liquid immiscibility (liquation) (Roedder, 1984, 1992; Frezotti, 1992; Raimbault et al., 1995; Seltmann et al., 1997; Zaraisky et al., 1997), (3) filter pressing (London, 1992), and/or (4) postmagmatic metasomatism with hydrothermal element redistribution (Nova´k et al., 1996). During magma ascent or in situ, the early crystallized minerals and/or more evolved portion of the melt can be separated from the primary melt. Both early crystals and segregations of more evolved melt may have undergone individual evolution paths different from the main portion of the original magma. The aim of this study is to assess the potential role for each of these processes in the generation of an extremely phosphorus-enriched Variscan rare-metal granite from Podlesı´ in the western Krusˇne´ Hory (Erzgebirge) Mountains of the Czech Republic. Because only a small number of such P-rich systems is known world-wide (Kudrin et al., 1994; Raimbault et al., 1995; Yin et al., 1995; Charoy & Noronha, 1996) and only limited experimental data exist (see London, 1995), the Podlesı´ system represents an excellent, unique case story of magmatic evolution with only minor alteration. Because of the existence of a quarry and a number of deep drillholes a complete geological section is accessible, allowing us to model magmatic–hydrothermal processes. GEOLOGICAL SETTING The Krusˇne´ Hory (Erzgebirge) is one of the classical metallogenic provinces (Fo¨rster & Tischendorf, 1994; Breiter & Seltmann, 1995), located near the southern margin of the Saxothuringian zone at the northwestern edge of the Bohemian Massif in the Central European Variscides. Its extent of about 10 000 km2 is determined partially by Upper Carboniferous plutons that intrude Upper Proterozoic to Lower Palaeozoic metamorphic sequences. The plutons are composed of many spatially subdivided multiphase intrusions. The late rare-metal enriched intrusions were emplaced into high crustal levels and are ˇcharacterized by stock-, cupola- and ridge-shaped apices (Stemprok & Seltmann, 1994). The geochemical characteristics enable us to distinguish two groups of highly evolved granites (Breiter et al., 1991; Fo¨rster et al., 1996): (1) a weakly peraluminous F-rich and P-poor type related to high-level stocks mostly associated with breccia pipes (Gottesberg, Altenberg, Sadisdorf, Cı´novec, Krupka) [A-type granites in the sense of Loiselle & Wones NUMBER 12 DECEMBER 1997 Fig. 1. Generalized geological map of western Krusˇne´ Hory with granite distribution. (1979)], and (2) a strongly peraluminous F- and P-rich type related to multiple intruded plutons (Ehrenfriedersdorf, Kra´sno, Podlesı´) ([in the sense of Chappel & White, 1974)]. Tin and tungsten ores (greisen type) are associated with both types of rare-metal granites and occur mostly along narrow contact zones with the country rocks. One of these mineralized areas is the ancient mining district of Hornı´ Blatna´–Podlesı´, situated in the western part of the Krusˇne´ Hory Mountains within a pile of Cambro-Ordovician, mostly pelitic, greenschist-facies metasediments (Fig. 1). The metasediments are intruded by a suite of peraluminous granites of the Eibenstock– Nejdek pluton. The granites are traditionally divided into the older intrusive complex (‘normal’, OIC), and the younger intrusive complex (‘tin specialized’, YIC) (Laube, 1876; Breiter & Seltmann, 1995). The youngest, most fractionated members of the suite are stocks of albite–Limica–topaz granites, e.g. Podlesı´, which typically have marginal pegmatites (stockscheider) and are commonly variably greisenized. GEOLOGY AND PETROGRAPHY OF ´ THE PODLESI STOCK The Podlesı´ granite system represents the most fractionated part of the S-type Eibenstock–Nejdek pluton. The Podlesı´ stock (200 m diameter), intrudes phyllites and biotite granite of the Eibenstock–Nejdek pluton (Fig. 2). The phyllite envelope of the granite (in the uppermost 50 m of the studied cores) is strongly altered to protolithionite–topaz hornfelses and is crosscut by numerous steep topaz–albite–zinnwaldite–quartz veinlets, accompanied by greisenization and tourmalinization 1724 BREITER et al. EVOLUTION OF A P-RICH GRANITE Fig. 2. Geological cross-section of the Podlesı´ granite stock. of the surrounding rocks. Steep veinlets of silicified aplites are also abundant. The aureole of hydrothermal alteration (greisens) is limited to regions just outside the contact with the granite stock. The internal fabric of the granite cupola has been studied in three boreholes by the Czech Geological Survey (Lhotsky´ et al., 1988). The contact of the granite with the phyllites is sharp, bordered by a marginal pegmatite zone of 40 cm thickness. The upper part of the cupola (depth 50–115 m in borehole PTP-1) consists principally of reddish, medium- to fine-grained albite– protolithionite–topaz granite (‘stock granite’) with flat layers of albite–zinnwaldite–topaz granite (‘dyke granite’). Adjacent to the contacts of both granite types are several pegmatite dykes (Fig. 3). At greater depth, mediumgrained stock granite has been encountered. Petrographically, the reddish stock granite is equigranular and fine to medium grained (0·2–1·0 mm). Albite appears as euhedral lamellar tablets. K-feldspar has mostly short prismatic subhedral, non-perthitized grains. Long prisms are twinned and euhedral. Quartz grains are isometric and anhedral. The mica is protolithionite [in the sense of Weiss et al. (1993)]. Topaz consists of euhedral to subhedral grains and makes up 3–4 vol. % of the rock. Among accessories, apatite prevails (Table 1). The white dyke granite is fine grained (0·1–0·5 mm). Quartz, K-feldspar and albite grains are mostly anhedral; Fig. 3. Pegmatite dykes penetrating along, or near, the contacts of the stock granite and dyke granite. (Detail from the quarry, thickness of the pegmatite dykes is exaggerated.) both feldspars are found also as short subhedral prisms. Borders of larger feldspar grains have been often leached and replaced by late quartz, albite and topaz. Grains of quartz, K-feldspar and topaz are markedly zoned. The mica is zinnwaldite, and forms subhedral flakes. Large topaz crystals (0·3–0·5 mm) are subhedral to euhedral; the small grains are interstitial. Abundance of topaz 1725 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997 Table 1: Modal composition (in vol. %) of granitoids from the Podlesı´ stock (borehole PTP-1) Sample: 2702 2653 2665 2674 581 Rock: S D S D S Depth (m): 60·5 80·0 89·5 103·5 300·0 Quartz 29·7 33·5 34·3 31·2 32·1 K-feldspar 40·8 33·4 35·3 38·1 30·3 Albite 17·9 17·8 17·7 18·6 26·6 Topaz 3·6 5·8 3·8 3·4 3 Mica 6·8 7·5 7·9 7·9 6·4 Apatite 0·9 0·8 0·7 0·7 0·4 Childrenite 0 0·8 0 0 0 Zircon 0·1 0 0·1 0 0·1 Sericite 0·1 0 0·1 0 0·2 Fluorite 0 0 0 0 0·2 Chlorite 0·1 0·2 0 0 0·1 99·8 99·9 99·9 99·4 Total 100 Rock types: S, stock granite; D, dyke granite. Remark: Pegmatite is too coarse-grained for modal analysis. reaches 3·5–6 vol. %. The most common accessory minerals are phosphates—apatite and childrenite. The upper contact of the ‘dykes’ is sharp and the lower contact is often indistinct (diffuse). The pegmatite consists of relatively large (up to 3 cm × 1 cm) hypidiomorphic columns of partly kaolinitized K-feldspar with a fine-grained granitic groundmass composed of anhedral quartz, K-feldspar, albite and zinnwaldite. Topaz is common. Contacts of pegmatite dykes are in all cases sharp. The similarity in whole-rock chemistry and mineralogy of all three above-mentioned rocks suggests that they are comagmatic at places where melt and exsolved magmatic vapour phase have been undergoing continuous re-equilibration. WHOLE-ROCK CHEMISTRY All granite types from the Podlesı´ cupola are subsolvus, peraluminous, strongly fractionated alkali feldspar leucogranites. The stock granite is strongly peraluminous (A/CNK 1·15–1·25). In comparison with common Ca-poor granites (Clarke, 1992), the stock granite is poor in Ca, Fe, Mg, Sr, Ba, Zr, Sc, V and Pb, and strongly enriched in incompatible elements such as Li, Rb, Cs, Sn, Nb and W. The rock is rich in fluorine and phosphorus; it contains 0·6–1·8 wt % F and 0·4–0·8 wt % P2O5 (Table 2). The high degree of magmatic fractionation is demonstrated also by low K/Rb and Zr/Hf ratios (22–35 and 12–20, respectively) and high U/Th ratio (4–7). Still more evolved dyke granite and pegmatite are relatively depleted in Si, Zr, Sn, W and rare earth elements (REE), and enriched in Al (A/CNK 1·2–1·4), P (0·6–1·5 wt % P2O5), F (1·4–2·4 wt %), Na, Rb, Li, Nb and Ta. The ratios of K/Rb (14–20) and Zr/Hf (9–13) are lower than in the stock granite. All rocks are rich in phosphorus, which is mainly present in the alkali feldspars. P shows a positive correlation with F, Al, Li, Rb, Nb, Ta and peraluminosity (expressed as A/CNK), and negative correlation with Si, Zr and Sn (Fig. 4). There is no correlation among P and Na, K and Sr. High degree of fractionation is documented also by unusually high concentrations of rare metals. Nb and Ta are preferentially concentrated in the more peraluminous dyke granite and pegmatite (50–95 ppm Nb and 30–55 ppm Ta) compared with the stock granite (25–50 and 10–25 ppm, respectively). On the other hand, contents of Sn and W are definitely higher in the stock granite (10–50 ppm Sn, 20–80 ppm W) than in the dyke granite and pegmatite (5–20 ppm Sn and 35–60 ppm W). The REE contents are generally very low. The chondrite-normalized patterns are relatively flat (Ce/YbCN 4–12) having prominent negative Eu anomalies. Both dyke granite and pegmatite are even more depleted in bulk REE, showing remarkably developed lanthanide tetrad effect (Fig. 5). The stock granite within the dyke-rich zone (78–120 m depth in borehole PTP-1) is also enriched relative to its roof and bottom zones in Al, P, F, Mn, Li, Rb, Cs, Nb and Ta, and depleted in Si, Zr and REE. Nevertheless, the differences between the stock granite, the dyke granite and pegmatite are clearly distinguishable (Fig. 6). 1726 BREITER et al. EVOLUTION OF A P-RICH GRANITE Table 2: Geochemical bulk rock data of representative samples from the Podlesı´ granite stock (major elements in wt %; trace elements in ppm) Sample: SiO2 TiO2 Al2O3 Fe2O3 FeO MnO MgO CaO Li2O Na2O K 2O P 2 O5 F H2O+ H2O– Total Rb Cs Sr Ba Zr Hf Th U Sn Zn Pb Nb Ta W La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Y 2708 2361 2360 581 74·05 0·07 14·08 0·66 0·45 0·02 0·05 0·42 0·17 3·99 4·31 0·44 1·14 0·76 0·07 100·20 67·70 0·03 18·37 0·85 0·46 0·04 0·03 0·50 0·35 3·43 5·99 0·91 1·38 0·93 0·08 100·48 69·70 0·02 16·44 0·37 0·67 0·04 0·03 0·48 0·35 4·50 3·51 1·64 1·48 1·32 0·15 100·21 73·13 0·04 13·89 0·10 1·0 0·03 0·05 0·44 0·16 3·74 4·65 0·47 0·64 0·84 0·08 99·04 1203 143 19 7·2 42 2·3 5·9 34 22 37 5·2 33 11 44 2·1 5·4 0·71 2·4 0·96 0·014 1·15 0·30 1·8 0·27 0·62 0·082 0·52 0·069 9·2 3200 180 64 7·2 10 2·4 7·8 21 16 76 2·7 66 49 53 0·25 0·53 0·053 0·16 0·075 0·005 0·12 0·026 0·17 0·028 0·07 0·013 0·09 0·013 1·5 1812 116 375 127 51 7·2 5·8 32 10 84 3·7 124 113 89 0·60 1·2 0·26 1·24 0·38 0·058 0·57 0·093 0·50 0·083 0·20 0·027 0·15 0·023 3·4 1094 107 11 6·6 46 2·1 6·3 33 54 54 9·6 29 9 30 2·9 7·2 0·92 3·2 1·05 0·017 1·18 0·30 1·8 0·28 0·69 0·103 0·63 0·081 9·5 Fig. 4. Correlation of P2O5 with Al, Li (all in wt %), Ta and Sn (ppm). ×, stock granite; Ε, dyke granite; Μ, pegmatite. 2708, fine-grained stock granite, upper rapidly cooled zone, borehole PTP-1, depth 55 m; 2361, pegmatite, quarry; 2360, fine-grained dyke granite, quarry; 581, medium-grained stock granite, lower part of the stock, borehole PTP-1, depth 300 m. 1727 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997 Fig. 5. Chondrite-normalized REE patterns. MINERALOGY Alkali feldspars All three rock types contain relatively pure end-members of alkali feldspar: albite with <3% An and K-feldspar (Kfs) with a maximum of 5% Ab and <1% An. All structurally studied Kfs are monoclinic (orthoclases). Triclinic microcline (0·65±5) has been found only in the stock granite immediately above the dyke-rich zone. High phosphorus content is typical of alkali feldspars from all studied granite types (Fry´da & Breiter, 1995). The phosphorus content varies distinctly within the dyke and stock granite (Fig. 7). The P2O5 content in Kfs in the dyke granite is very high (between 0·6 and 1·75 wt %); its distribution shows a distinct zonality (Figs 8 and 9), whereas in the stock granite the overall content is lower (0·3–0·7 wt % P2O5) and no zoning has been observed (Fig. 9). The rubidium content in Kfs, unlike that of phosphorus, is uniform in both rock types, without any zoning (0·15–0·25 wt % Rb in the stock granite, 0·35–0·50 wt % Rb in the dyke granite). The P and Rb contents of albites from all rock types are lower and more uniform. The P2O5 contents in albite in the dyke granite are between 0·2 and 1·0 wt %, Fig. 6. Vertical distribution of selected elements in the Podlesı´ stock: P2O5, Al2O3, SiO2, F (wt %), Rb and Zr (ppm). ×, stock granite; Ε, dyke granite; Μ, pegmatite. whereas concentrations in the stock granite are lower (0·05–0·7 wt % of P2O5). No zoning in P2O5 has been observed in albites from the stock granite. Albites from the dyke granite are zoned, but the rim–core differences are lower than in K-feldspars. The Rb contents in albite 1728 BREITER et al. EVOLUTION OF A P-RICH GRANITE Fig. 7. Histograms of P content-frequency in albite (197 measurements) and K-feldspars (326 measurements). are usually <0·1 wt %. The Ba and Sr contents in all feldspars are below detection limits of electron microprobe analysis (EMPA). Li-rich micas In all samples, micas are represented by Li-rich members, whose chemistry, similar to the K-feldspars, differs Fig. 9. P2O5 and Rb distribution in orthoclase from the dyke granite (2360) and stock granite (2647). Scale on each profile represents 0·1 mm. depending on the host granite type (Table 3). For the dyke granite and adjacent stock granite, the zinnwaldite crystals are also distinctly zoned (Fig. 10), with cores Fig. 8. Cathode-luminescence image of P-rich border zone in Kfs in the dyke granite. 1729 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997 Table 3: Chemical analyses and structural formulae of Li-micas Sample: 2358 2359 2360 2665 2687 Depth: quarry quarry quarry 90 m 300 m Rock: S P D S S SiO2 42·9 46·75 48·3 45·0 42·0 TiO2 Al2O3 Fe2O3 FeO 0·75 20·6 0·16 0·29 19·3 20·0 0·17 0·3 11·7 8·2 12·7 1·07 13·2 0 20·7 0·78 21·8 16·9 MnO 0·08 0·19 0·45 0·24 0·18 MgO 0·5 0 0·18 0·43 0·61 CaO 0·13 0·09 0·16 Li2O 2·95 4·2 4·55 3·64 Na2O 0·22 0·25 0·78 0·47 0·34 K 2O 9·34 9·93 8·9 9·91 9·85 1·05 0·72 Rb2O Cs2O 0·68 n.a. 1·12 1·18 n.a. n.a. 9·02 8·67 n.a. n.a. 2·95 0·12 F 7·71 H 2O 0·26 F=O −3·25 −3·80 −3·65 −3·28 −2·10 Total 97·15 98·95 98·58 98·98 99·11 n.a. 0·25 7·78 0 n.a. 4·98 n.a. Si 6·361 6·694 6·812 6·477 Al4 1·555 1·288 1·158 1·523 6·301 1·611 Ti 0·084 0·017 0·031 0 0·088 Total Z 8 8 8 8 8 Al6 2·04 1·972 2·162 1·996 2·244 Ti 0 0 0 0 0 Fe3 0·119 0 0·018 0·033 0 Fe2 1·637 1·404 0·971 1·535 2·117 Mn 0·01 0·023 0·054 0·029 0·023 Mg 0·111 0 0·38 0·092 0·136 Li 1·76 2·419 2·58 2·109 1·78 Total Y 5·677 5·818 5·82 5·793 6·3 Ca 0·013 0·014 0·004 0 0 Na 0·063 0·069 0·213 0·131 0·099 K 1·767 1·814 1·6 1·82 1·885 Rb 0·065 0·103 0·107 0·097 0·069 1·908 2 1·925 2·049 2·061 Cs Total X O 0·008 20 19·914 20 20 21·453 F 3·616 4·085 3·865 3·544 2·363 OH 0·264 0 0·116 0·424 0·184 2358–2360, wet analyses of mono-mineralic concentrates; 2665 and 2687, EMPA, average of ten points. Li, Rb and Cs by wet analyses of mono-mineralic concentrates. All in wt %, all analyses at the laboratory of the CGS Praha. Rock types: S, stock granite; D, dyke granite; P, pegmatite. n.a., not analysed. Recalculation of structural formulae according to Rieder (1977). 1730 BREITER et al. EVOLUTION OF A P-RICH GRANITE Fig. 10. Microprobe profiles across Li-mica crystals from the cooled roof zone of the stock granite (sample 2647) and from the dyke granite (sample 2360). Contents of all elements are expressed in atoms per formula unit. 1731 JOURNAL OF PETROLOGY VOLUME 38 being enriched in Fe, Mg and Ti, and rims rich in Si and Li. Li contents, calculated using the method of Tindle & Webb (1990), are in overall agreement with analyses by atomic absorption spectroscopy (AAS) of the mica separates. Zoning in Si, Fe, Mg and Ti can be explained by changes of melt chemistry during the zinnwaldite crystallization. In contrast to the above elements, there is no zoning in F and Rb (Fig. 10). Fluorine content in profile across the whole crystals is >8 wt % (Fig. 10), indicating that F atoms nearly completely occupy the OH–F sites. If there had been zoning in Rb content, then it was most probably destroyed by post-crystallization redistribution of Rb atoms, which are only weakly bound in inter-layer sites. Protolithionites [in the sense of Weiss et al. (1993)] from the stock granite outside the dyke-rich zone are poorer in Li and Rb, and are also without any distinct zoning. This character argues for a relatively constant melt composition during their crystallization. The distribution of Cs in micas within the rock types and also within single crystals is irregular, but generally the Cs content is higher in micas from the stock granite, varying between 0·06 and 0·49 wt %. The Cs content in the dyke granite varies between 0·05 and 0·26 wt %. No systematic enrichment or depletion during crystal growth has been observed. Zinnwaldites and protolithionites are clearly distinguishable in the Si–Al plot (Fig. 11), where they form two well-correlated parallel clusters. The Li-rich zinnwaldites are relatively enriched in Al and depleted in Fe; this corresponds well to the whole-rock chemistry of parental rocks. NUMBER 12 DECEMBER 1997 dyke granite have been found. Topaz in the stock granite contains ~18·5–19·5 wt % F (90–95% of theoretical F saturation); in the dyke granite, topaz contains 20–21 wt % F (95–100% of saturation). Some large euhedral crystals from the dyke zone contain zonally arranged quartz inclusions, but no zoning in F or Si/Al ratio has been observed. Phosphates Apatite of two generations occurs in all rocks, older Mnpoor euhedral crystals, and younger Mn-rich interstitial flakes. Both types are poor in Cl (mostly <0·1 wt % Cl) and moderately F rich (2·5–2·9 wt % F, which represents ~35% of theoretical F saturation). Childrenite–eosphorite, zwiesselite and triphylite have also been found in the dyke granite (Table 4). The most abundant phosphate is the childrenite-rich member of the childrenite–eosphorite group, which builds euhedral short prisms 0·02–0·04 mm across. Their texture suggests primary (late magmatic) origin. Other less abundant phosphates also seem to be primary. Small euhedral monazite crystals occur mainly as inclusions in micas; some of them are surrounded by pleochroic haloes. Monazite is a main REE host in all rock types. No amblygonite was found at Podlesı´, although this mineral is the most common phosphate in Li- and F-rich environments, rare-metal granites and granitic (LCT-type) ˇ pegmatites (Cerny´, 1991; Raimbault et al., 1995). Other accessories Quartz In both stock and dyke granite, quartz occurs in two forms. The older subhedral grains often contain zonally arranged albite inclusions. This snowball texture is typical for highly evolved alkali feldspar granites. The younger interstitial quartz grains often corrode feldspars. Rhythmical layered textures [e.g. comb quartz layers and other unidirectional solidification textures (USTs), Carten et al., 1988; Kirkham & Sinclair, 1988] were not observed in the Podlesı´ system. Topaz Topaz is present in two types (generations). Euhedral to subhedral isometric crystals are enclosed in all rock types and are one of the oldest crystallized minerals. The late interstitial topaz occurs only in the dyke granite and pegmatite. Both topaz types are rich in fluorine: nearly all F–OH sites are occupied by F. Only very small differences in the F content between stock granite and Relative abundances of accessory minerals are shown in Table 5. Among them, the Nb,Ta-oxides ˇare the most important for petrogenetic interpretation (Cerny´, 1991). Small grains of the most common Nb–Ta rutile are mainly inherited in mica flakes, sometimes forming zonally arranged clusters in their inner parts. Columbite, occurring only in the dyke granite, is partially included in micas, and partially interstitial. In the pegmatite, mainly Ta-rich rutile occurs (Breiter & Seltmann, 1995). Observed textures and conversion of Nb-rich into Tarich minerals during evolution of the system are in agreement with typical evolution of sodium- (albite-)-rich ˇ pegmatitic and granitic environments, as shown by Cerny´ (1991). MELT INCLUSIONS The first and essential information which one can obtain from the melt inclusions is the fact that they exist. In the case of the samples from Podlesı´ it is important that we can observe melt inclusions in quartz and topaz, as well 1732 BREITER et al. EVOLUTION OF A P-RICH GRANITE Table 5: Distribution of accessory phases within the granitic rocks of the Podlesı´ stock Fig. 11. Si vs Al plot for Li-rich micas (expressed in atoms per formula unit). Protolithionites from the stock granite (+, top of the stock; Μ, depth 150 m; ×, depth 300 m) are poorer in Si (and Li) and richer in Fe than zinnwaldites from the dyke system (Ε, dyke granite; Φ, pegmatite). Cassiterite Columbite Ilmenorutile U-tantalite U-mikrolite Ixiolite Wolframite Huebnerite Scheelite Rutile Ilmenite Haematite Pyrite Bismuthine Powellite Roosweltite Childrenite Zwiesselite Triphylite Stock Dyke granite granite X XX XX XX X Pegmatite XXX X X X XX X X X X XX X X X XXX XX X X X X XX X X X, rare; XX, common; XXX, abundant. Table 4: Chemical analyses (in wt %) and structural formulae of phosphates Sample: 2651 2360 2669 2669 Mineral: zwieselite triphylite childrenite eosphorite udl 20·61 11·02 udl 22·17 udl 9·74 21·74 udl 21·35 31·41 31·54 85·20 84·37 0·65 0·35 0·31 0·70 0·99 0·96 1·01 1·02 5·00 0·65 5·00 0·31 Na2O CaO FeO MnO MgO Al2O3 TiO2 P 2 O5 F Total Na Ca Fe Mn Mg Al Ti P F O Fe/(Fe+Mn) 4·24 2·42 36·73 26·06 0·18 32·78 98·16 0·42 34·81 0·42 45·60 10·35 95·85 1·21 0·09 1·10 0·79 0·01 1·00 4·50 0·58 6·04 0·05 5·68 4·82 24·00 All analyses by EMPA at the laboratory of the CGS Praha. udl, under detection limit of EMPA. 1733 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997 Table 6: Examples of electron microprobe results (Cameca SX 50 and SX 100, GFZ Potsdam) on F- and/or P-rich melt inclusions in quartz of pegmatites and granites from the Podlesı´ stock, Krusˇne´ Hory (Erzgebirge). Inclusion group: Rock type: P-poor, F-rich ´ Podlesı stock granite P-poor, F-poor ´ Podlesı dyke granite P-rich ´ Podlesı stock granite Main group Ehrenfriedersdorf granites Sample: 2358b 2362 2358a SiO2 66·0±1·4 71·2±1·5 68·2±0·8 TiO2 0·1 0·05 0·1 SnO2 0·02 0·02 n.d. Al2O3 16·9±0·2 13·8±1·1 17·3±0·2 FeO 1·2±0·1 3·4±1·1 0·8±0·02 MnO 0·1 0·2 0·1 MgO d.l. CaO 0·4±0·01 d.l. 0·1 d.l. 0·1±0·01 68·4±1·6 0·03 0·1±0·01 14·6±0·7 0·4±0·04 0·05 d.l. 0·1 Na2O 3·7±0·2 3·2±0·4 2·9±0·2 4·1±0·1 K2O 4·8±0·4 5·0±0·3 4·9±0·4 3·4±0·1 Rb2O∗ 0·6±0·08 0·7±0·1 F 8·9±0·3 3·1±0·9 Cl 0·2±0·01 P2O5 Sum H 2O 0·3±0·05 103·22 (2·8±0·6)† n.d. d.l. 0·1 15 11 d K (lm) 25–40 20–60 0·1±0·03 99·5 (1·3±0·4) n 3·8±0·6 0·2±0·01 2·6±0·4 100·77 0·5±0·04 2·2±0·1 3·0±1·0 98·58 n.d. n.d. 5 166 25–40 20–100 In the last column is given the average composition of melt inclusions in pegmatite quartz from the tin deposit Ehrenfriedersdorf (Central Erzgebirge) for the purpose of comparison (Thomas et al., in preparation). d.l., below the detection limit; n.d., not determined; n, number of determinations; d K, diameter of the analysed homogeneous melt inclusions. ∗Calibrated with natural feldspars (from Etyka–Eastern Transbaikalia) and synthetic Rb-feldspars (prepared by St. Melzer, GFZ Potsdam). †Five measurements. as in K-feldspar. Normally, the melt inclusions in feldspar from the granitic rocks of the Erzgebirge are destroyed by recrystallization and hydrothermal overprinting. Therefore, the appearance of melt inclusions in the Kfeldspar is a strong hint that the system was cooled rapidly, and therefore remained quasi-closed, and that postmagmatic processes were not important. From statistical analyses, we obtained from 170 measurements on different samples from Podlesı´ for C W a value of 7·5±0·8 equiv. wt % water. The extrapolated solidus temperature (Thomas et al., 1996) for the same samples is 610±26°C and represents the tendency of the system towards formation of highly evolved residual melt. In this paper we confine ourselves to the analytical determination of the bulk composition of melt inclusions in the main rock types (Table 6). The results obtained from melt inclusion studies of the Podlesı´ samples allow the following interpretation (only a few inclusions of suitable size could be studied using EMPA, Table 6). (1) The stock granite represents the parental system. The studied melt inclusions in the stock granite (samples 2358a and 2358b) seem, in synthesis with all available geological, geochemical and mineralogical information, to reflect the formation of two immiscible liquids (residual melts enriched extremely in P2O5 plus some F in sample 2358a vs those enriched extremely in F in sample 2358b). From these two liquids, the parental magmas of the dyke granite (see rock sample 2362; in sample 2360 no inclusions of suitable size could be analysed) and pegmatite (rock samples 2359 and 2361; no EMPA data from melt inclusions) were generated. We are aware that the bulk-magma geochemistry may be different from that of the extremely variable melt inclusion compositions; however, the latter reflects the general evolution of the melt system. 1734 BREITER et al. EVOLUTION OF A P-RICH GRANITE (2) The melt inclusion group represented by sample 2358a with ~2·6 wt % P2O5 and 2·2 wt % F seems to represent the liquid producing the dyke granite, whereas the melt inclusion group represented by sample 2358b with ~8·9 wt % F seems to reflect the pegmatite melt. A number of melt inclusions observed in pegmatite and dyke granite samples characterize melt batches trapped at complicated pegmatitic to superpegmatitic–hydrothermal conditions. DISCUSSION The ‘ore specialized’ granites [in the sense of Tischendorf (1977)] of the Krusˇne´ Hory (Erzgebirge) belong to a group of internationally recognized, strongly fractionated granite plutons accompanied by Sn–W mineralization. Several genetic interpretations and models of granite evolution and granite mineralization have been proposed during the last 60 years. Most of these interpretations emphasize the importance of post-magmatic (autometasomatic) overprinting of granites, particularly in the case of albite–Li-mica–topaz facies (Teuscher, 1936; ˇ Tischendorf, 1977; Stemprok, 1979). The abundance of the late albite has been explained as being caused by interaction between primary more potassic granites and hydrothermal sodium-bearing fluids (Beus & Zalasˇkova, ˇ 1962; Tischendorf, 1989; Stemprok, 1993). Nevertheless, the mineralogy of the Podlesı´ granites suggests a fully magmatic origin of the specialized granites including the Krusˇne´ Hory granite plutons. A significant advantage of the Podlesı´ granite system for the study of fractionation and crystallization of rare metal-specialized granites is that greisenization and other types of hydrothermal alteration are nearly absent. Compared with common stanniferous granites in the western Krusˇne´ Hory (Breiter et al., 1991; Fo¨rster & Tischendorf, 1994) and the Cornubian batholith (Manning & Hill, 1990), the Podlesı´ granites are highly enriched in phosphorus, having >1 wt % P2O5. This makes the Podlesı´ cupola comparable with the P-rich family of raremetal granites (Beauvoir, Raimbault et al., 1995; Yichun, Pollard & Taylor, 1991; Yin et al., 1995; Argemela, Charoy & Noronha, 1991; South Mountain Batholith, ˇ Kontak et al., 1996) and LCT-type pegmatites (Cerny´, 1991). The best-known example of a P-rich granitic suite is Beauvoir, France. There, superposition of two discrete units (intrusions) and their contact relations and reactions have been documented within the stock. According to the model of Raimbault et al. (1995), fractionation of this suite occurred at a depth below the emplacement level. During ascent, the most volatile-enriched part of the melt moved faster than the rest of the crystal-enriched portion of the magma. Therefore, the more evolved melt reached the top of the stock as a true melt before the less evolved crystal mush. Another example of a P-rich system is Argemela (central Portugal, Charoy & Noronha, 1991). There, the ‘initial melt’ is well preserved in feeder channels, and a more evolved rock, containing inherited crystal cores with more fractionated rims and interstitial groundmass, forms the upper part of the stock. The change in melt chemistry between parental and residual environments was rapid, as documented by sharp zonation of single mica crystals [model presented by Charoy et al. (1995)]. Unfortunately, the geological relations between feeder channels and more evolved microgranite within the Argemela stock are unknown. Many features of the Podlesı´ magmatic system are similar to that at Argemela. The Podlesı´ system allows the study of the relations between the parental and evolved rocks in outcrop. Within the Podlesı´ stock, according to field evidence, three rock types occur: stock granite, which forms the principal parts of the stock, and subordinate dyke granite and pegmatite, which are found as flat layers in the stock’s upper part. The stock granite in the uppermost peri-contact part of the stock represents the rapidly cooled portion of initial melt and it preserves the original chemistry and mineralogy. Distinct zoning of some Kfs and micas provides mineralogical evidence for two stages of crystallization in the dyke granite and pegmatite, and only one stage in the stock granite. The older stage of crystallization of the dyke granite, represented by Kfs and mica cores, is nearly the same as that encountered in the stock granite, but the younger stage of the stock granite evolution, represented by Kfs and mica rims, documents a distinctly more evolved environment enriched in phosphorus and fluorine. Theoretically, the younger stage could be magmatic or metasomatic in origin. However, as shown in the model below, we suggest a magmatic origin also for the younger stage. As the melt–fluid K DP is higher than unity (London, 1995), the melt should be richer in P than are coexisting fluids. On the other hand, reaction of feldspars with postmagmatic aqueous fluids should produce a decrease of P content in feldspars and production of secondary phosphates, expressed by the reaction P-rich Kfs + zinnwaldite=P-poor Kfs + muscovite + Fe-phosphates [as documented by Breiter & Siebel (1995)]. The P-rich rims of Kfs in dyke granite and pegmatite argue for principally magmatic crystallization of these rocks from an evolved residual melt (Fry´da & Breiter, 1995). Thus, the dyke granite and pegmatite at Podlesı´, in general, are not a product of metasomatic overprinting of the stock granite, but are products of crystallization of a different, more fractionated melt. Thus, mineralogical 1735 JOURNAL OF PETROLOGY VOLUME 38 data support the existence of two melts within the stock: parental and residual. Two main types of melt inclusions (P poor vs P rich) in quartz from the dyke zone correspond perfectly to geological observations and mineralogical data. The first type of melt is nearly identical with the upper, rapidly cooled zone of the stock granite. The second one is enriched in Al and P, and depleted in Si and Ca. Both melts are very rich in fluorine. This corresponds well to the degree of F abundance in both topaz and mica in the stock granite and dyke granite (>90 atom. %). The melt-inclusion data for P-poor melts yield 3–9 wt % of fluorine, whereas the whole-rock analyses of the stock granite give only 0·8–1·8% of F; therefore the bulk of the fluorine from this melt had partitioned into an aqueous fluid before crystallization ceased. This is in agreement with experimental melt–fluid K DF p1 ( Johannes & Holtz, 1996). In contrast, the whole-rock F content in the dyke granite (1·4–2·2 wt %) is nearly the same as that in the P-rich melt inclusions (~2·2 wt %), meaning that little fluorine was lost from the P-rich melt of the dyke granite. The high concentration of network modifying cations caused lower viscosity and higher mobility of residual melt (Mysen, 1987; London, 1995; Johannes & Holtz, 1996). A small volume of evolved melt could be produced without any observable changes in composition of the rest of the parental melt. In Kfs and mica crystals, the ratio of volumes of their inner cores (which originated from the initial melt) to their outer rims (which originated from residual melt) is ~1:1. Thus, the actual crystal mush before crystallization of the dyke granite consisted of roughly one-half crystals chemically identical to those in stock granite and onehalf residual melt. The present chemistry of the dyke granite represents an average of these two principal constituents, with some changes that resulted after fluidadmixing during crystallization. The formation of two melts may be explained by: (1) batches or segregations of residual melt trapping crystals inherited from the parental melt, (2) liquid–liquid immiscibility at depth during magma ascent, or (3) in situ crystallization within the stock. In the first two cases, the dyke granite forms true dykes; in the last case it would be better described as ‘magmatic layers’. The diffuse character of some foot-wall contacts of the dykes and lack of steep feeder channels support in situ differentiation and/or only a short injection of dyke granite magma into stock granite that was not fully crystallized. The sharp character of the pegmatite contact with both of the other rocks argues for interpretation of the pegmatite as an injection of a small portion of residual, in this case K-enriched melt. The processes discussed above operated at a depth of ~25–70 m below the contact with phyllites (depth NUMBER 12 DECEMBER 1997 75–120 m in borehole PTP-1). At greater depth, the melt constitution was stable, nearly like the initial melt, as is documented by homogeneous Kfs and mica crystals and whole-rock chemical data in the stock granite of the deeper part of the stock (120–310 m in borehole PTP1). After the magma in the cupola crystallized, the continuation of fluid flow from deeper parts of the intrusion caused moderate Li, F and Rb enrichment of the more permeable stock granite at 93–115 m depth, under the barrier of the cooled roof and below the thicker dyke granite layer, as well as some re-mobilization of weakly bound elements, e.g. Rb and Cs, in mica lattices. CONCLUSIONS The crystallization history of the Podlesı´ granite system can be summarized as follows. The composition of the initial melt was similar to that of the present stock granite. The crystallization of this magma started in the outer upper part of the stock, which cooled quickly and did not suffer subsequent late- and post-magmatic alteration (stock granite at depth 50–78 m in the borehole PTP-1). Here, feldspars and micas show no zoning. Under the rapidly crystallized carapace, in a practically closed system, fluids and volatiles became enriched. Mineralogical and melt inclusion data confirm the presence of three melt types with different crystallization paths: (1) The stock granite, which is nearly identical to those mentioned above. (2) The dyke granite, in which major minerals are strongly zoned, and whole-rock chemistry suggests enrichment in Al, P, F and Li during crystallization. Rapid change in melt chemistry is documented in Kfs by P-rich rims; the change in mica compositions was more gradual. (3) Pegmatite, which is mineralogically similar to the dyke granite, but enriched in K and depleted in Si and Na. Therefore, besides the parental melt, a residual melt mainly enriched in Na, P, F, Li, Rb and Nb, and a pegmatitic melt enriched in K, P, F, Li and Rb existed here. The diffuse character of some foot-wall contacts of the ‘dykes’ supports in situ differentiation of the residual melt or only local migration and injection into a stock granite that was not fully crystallized. The pegmatite dykes represent injections of a small portion of another residual, K-enriched melt. At greater depth the homogeneous stock granite crystallized solely from the parental melt. ACKNOWLEDGEMENTS This manuscript was carefully reviewed by J. Lowenstern, ˇ P. Nabelek and M. Stemprok. We thank V. Sixta, P. Dulski and R. P. Taylor for support with chemical 1736 BREITER et al. EVOLUTION OF A P-RICH GRANITE analyses, and V. V. Shatov for modal analyses. We appreciate the advice and assistance of E. Gantz and P. Jones during SEM-CL and EMPA studies. The manuscript profited from numerous discussions with T. Jarchovsky´, R. P. Taylor, W. D. Sinclair, D. London, R. Martin and others, and their helpful comments. REFERENCES Beus, A. A. & Zalasˇkova, N. E., 1962. High-temperature postmagmatic metasomatism in granitoids (in Russian). Izvestiya Akademiya Nauk SSSR, Seriya Geologicheskaya 14, 13–31. Breiter, K. & Seltmann, R., 1995. Ore mineralizations of the Krusˇne´ hory Mts. (Erzgebirge). Third Biennial SGA Meeting, Excursion Guide. Prague: Czech Geological Survey, 200 pp. Breiter, K. & Siebel, W., 1995. Granitoids of the Rozvadov pluton, Western Bohemia and Oberpfalz. Geologische Rundschau 84, 506–519. Breiter, K., Sokolova´, M. & Sokol, A., 1991. Geochemical specialization of the tin-bearing granitoid massifs of NW Bohemia. Mineralium Deposita 26, 298–306. Carten, R., Walker, B. M., Geraghty, E. P. & Gunow, A. J., 1988. Comparison of field-based studies of the Henderson porphyry molybdenum deposit, Colorado, with experimental and theoretical models of porphyry system. In: Taylor, R. P. & Strong, D. F. (eds) Recent Advances in the Geology of Granite-Related Mineral Deposits. Canadian Institute of Mining and Metallurgy Special Volume 39, 351–366. ˇ Cerny´, P., 1991. Rare element granitic pegmatites. Part I: Anatomy and internal evolution of pegmatite deposits. Geoscience Canada 18, 49–67. Chapell, B. W. & White, A. J. R., 1974. Two contrasting granite types. Pacific Geology 8, 173–174. Charoy, B. & Noronha, F., 1991. The Argemela granite–porphyry (Central Portugal): the subvolcanic expression of a high-fluorine, rare-element pegmatite magma. In: Pagel, L. & Leroy, A. (ed.) Source, Transport and Deposition of Metals. Rotterdam: Balkema, pp. 741–744. Charoy, B. & Noronha, F., 1996. Multistage growth of a rare-element, volatile-rich microgranite at Argemela (Portugal). Journal of Petrology 37, 73–94. Charoy, B., Chaussidon, M., & Noronha, F., 1995. Lithium zonation in white micas from the Argemela microgranite (central Portugal): an in-situ ion-, electron-microprobe and spectroscopic investigation. European Journal of Mineralogy 7, 335–352. Clarke, D. B., 1992. Granitoid Rocks. Topics in the Earth Sciences 7. London: Chapman & Hall, 283 pp. Fo¨rster, H.-J. & Tischendorf, G., 1994. The western Erzgebirge– Vogtland granites: implications to the Hercynian magmatism in the Erzgebirge–Fichtelgebirge anticlinorium. In: Seltmann, R., Ka¨mpf, H. & Mo¨ller, P. (eds) Metallogeny of Collisional Orogens. Prague: Czech Geological Survey, Prague, pp. 35–48. Fo¨rster, H.-J., Seltmann, R. & Tischendorf, G., 1996. Post-collisional A-type silicic magmatism in the Variscan Erzgebirge. In: V. M. Goldschmidt Conference (Heidelberg) Abstracts Volume 1(1), 174. Frezotti, M. L., 1992. Magmatic immiscibility and fluid phase evolution in the Mount Genis granite (southeastern Sardinia, Italy). Geochimica et Cosmochimica Acta 56, 21–33. Fry´da, J. & Breiter, K., 1995. Alkali feldspars as a main phosphorus reservoir in rare-metal granites: three examples from the Bohemian Massif (Czech Republic). Terra Nova 7, 315–320. Johannes, W. & Holtz, F., 1996. Petrogenesis and Experimental Petrology of Granitic Rocks. Minerals and Rocks, 22. Berlin: Springer-Verlag, 335 pp. Kirkham, R. V. & Sinclair, W. D., 1988. Comb quartz layers in felsic intrusions and their relationship to porphyry deposits. In: Taylor, R. P. & Strong, D. F. (eds) Recent Advances in the Geology of GraniteRelated Mineral Deposits. Canadian Institute of Mining and Metallurgy Special Volume 39, 50–71. Kontak, D. J., Martin, R. F. & Richard, L., 1996. Patterns of phosphorus enrichment in alkali feldspar, South Mountain Batholith, Nova Scotia, Canada. European Journal of Mineralogy 8, 805–824. Kudrin, V. S., Stavrov, O. D. & Shuriga, T. N., 1994. A new spodumene type of tantalum-bearing rare-metal granites. Petrology 2, 75–81. Laube, G. C., 1876. Geologie des bo¨hmischen Erzgebirges I. Archiv fu¨r Naturwissenschaftliche Landesdurchforschung, 208 pp. Lhotsky´, P., Breiter, K., Bla´ha, V. & Hrochova´, H., 1988. Economicgeological investigations of Sn-mineralization near Podlesı´ in the western Krusˇne´ hory (in Czech). Internal Report. Prague: Czech Geological Survey. Loiselle, M. C. & Wones, D. R., 1979. Characteristics and origin of anorogenic granites. Geological Society of America, Abstracts with Program 11, 468. London, D., 1992. Phosphorus in S-type magmas: the P2O5 content of feldspars from peraluminous granites, pegmatites, and rhyolites. American Mineralogist 77, 126–145. London, D., 1995. Geochemical features of peraluminous granites, pegmatites, and rhyolites as sources of lithophile metal deposits. In: Thompson, J. F. H. (ed.) Magmas, Fluids, and Ore Deposits. Mineralogical Association of Canada Short Course 23, 175–202. Manning, D. A. C. & Hill, P. I., 1990. The petrogenetic and metallogenetic significance of topaz granite from the southwest England orefield. Geological Society of America, Special Paper 246, 51–69. Mysen, B. O., 1987. Magmatic silicate melts: relations between bulk composition, structure and properties. In: Mysen, B. O. (ed.) Magmatic Processes: Physicochemical Principles. Geochemical Society Special Publication 1, 375–400. Mysen, B. O., Ryerson, F. J. & Virgo, D., 1981. The structural role of phosphorus in silicate melts. American Mineralogist 66, 106–117. ˇ Nova´k, J. K., Pivec, E. & Stemprok, M., 1996. Hydrated iron phosphates in muscovite–albite granite from Waidhaus (Oberpfalz, Germany). Journal of the Czech Geological Society 41, 201–207. Pollard, P. J. & Taylor, R. P., 1991. Petrogenetic and metallogenetic implications of the occurrence of topaz Li-mica granite at the Yichun Ta–Nb–Li mine, Jiangxi Province, south China. In: Pagel, L. & Leroy, A. (eds) Source, Transport and Deposition of Metals. Rotterdam: Balkema, pp. 789–792. Raimbault, L., Cuney, M., Azencott, C., Duthou, J. L. & Joron, J. L., 1995. Geochemical evidence for a multistage magmatic genesis of Ta–Sn–Li mineralization in the granite at Beauvoir, French Massif Central. Economic Geology 90, 548–576. Rieder, M., 1977. Micas: calculation of crystallochemical formulas ´ ´ by a FORTRAN IV computer program. Veˇstnı´k Ustrˇednı´ho Ustavu ´ Geologickeho 52, 333–342. Roedder, E., 1984. Fluid Inclusions. Mineralogical Society of America, Reviews in Mineralogy 12, 644 pp. Roedder, E., 1992. Fluid inclusion evidence for immiscibility in magmatic differentiation. Geochimica et Cosmochimica Acta 56, 5–20. Seltmann, R., Breiter, K., Fry´da, J., Thomas, R., Dulski, P. & Taylor, R. P., 1997. Liquid–liquid immiscibility in the Podlesı´ stock. In: IAVCEI General Assembly, Puerto Vallarta, Mexico, 19–24 January 1997, Volcanic Activity and the Environment, Abstracts. Mexico: Unidad Editorial, p. 102. ˇ Stemprok, M., 1979. Mineralized granites and their origin: review of MAWAM contributions. Episodes 12, 49–53. ˇ Stemprok, M., 1993. Genetic models for metallogenic specialization of tin and tungsten deposits associated with the Krusˇne´ hory/Erzgebirge granite batholith. Mining Resource Geology, Special Issue 15, 373–383. 1737 JOURNAL OF PETROLOGY VOLUME 38 ˇ Stemprok, M. & Seltmann, R., 1994. The metallogeny of the Erzgebirge (Krusˇne´ hory). In: Seltmann, R., Ka¨mpf, H. & Mo¨ller, P. (eds) Metallogeny of Collisional Orogens. Prague: Czech Geological Survey, pp. 61–69. Teuscher, E. O., 1936. Prima¨re Bildungen des granitischen Magmas und seiner Restlo¨sungen im Massiv von Eibenstock–Nejdek. Mineralogische und Petrographische Mitteilungen 47, 211–262. Thomas, R., 1994. Estimation of the viscosity and the water content of silicate melts from melt inclusion data. European Journal of Mineralogy 6, 511–535. Thomas, R., Rhede, D. & Trumbull, R. B., 1996. Microthermometry of volatile-rich silicate melt inclusions in granitic rocks. Zeitschrift fu¨r geologische Wissenschaften 24(3–4), 505–526. Tindle, A. G. & Webb, P. C., 1990. Estimation of lithium contents in trioctahedral micas using microprobe data: application to micas from granitic rocks. European Journal of Mineralogy 2, 595–610. Tischendorf, G., 1977. Geochemical and petrographic characteristic of silicic magmatic rocks associated with rare-element mineralization. ˇ In: Stemprok, M., Burnol, L. & Tischendorf, G. (eds) Metallization Associated with Acid Magmatism, Vol. 2. Prague: Czech Geological Survey, pp. 41–96. Tischendorf, G. (compiler), 1989. Silicic Magmatism and Metallogenesis of the Erzgebirge, Vol. 107. Zentralinstitut fu¨r Physik der Erde Potsdam, pp. 1–316. Weiss, Z., Rieder, M., Smrcˇok, L., Petrˇ´ıcˇek, V. & Bailey, S. W., 1993. Refinement of the crystal structures of two ‘protolithionites’. European Journal of Mineralogy 5, 493–502. Zaraisky, G. P., Seltmann, R., Shatov, V. V., Aksyuk, A. M., Shapovalov, Yu. B. & Chevychelov, V. Yu., 1997. Petrography and geochemistry of Li–F granites and pegmatite–aplite banded rocks from the Orlovka and Etyka tantalum deposits in Eastern Transbaikalia, Russia. In: Papunen, H. (ed.) Mineral Deposits: Research and Exploration. Where do they meet? Proceedings of the 4th Biennial Society for Geology Applied to Mineral Deposits Meeting, Turku, 11–23 August 1997. Rotterdam: Balkema, pp. 695–698. Yin, L., Pollard, P. J., Hu, S. & Taylor, R. G., 1995. Geologic and geochemical characteristics of the Yichun Ta–Nb–Li deposit, Jiangxi Province, South China. Economic Geology 90, 577–585. APPENDIX Methods for microprobe analysis of minerals at the Czech Geological Survey Praha Minerals were analysed using a CAMSCAN 4-90DV electron microscope equipped with LINK eXL and Microspec WDX-3PC X-ray analysers. An accelerating voltage of 15 kV or 20 kV, a beam current of 3 nA, and a counting time of 100 s were used for the energydispersive analysis (system LINK eXL). This system was used for analysis of Si, Ti, Al, Fe, Mg, Mn, Ba, Ca, K, P, Na, Nb, Ta and W, and the detection limits of these elements are <0·1 wt %. The WDX system was used for analysis of F (TAP, 10 kV, 100 nA), Cs (PET, 20 kV, 30 nA), and Rb (TAP, 20 kV, 50 nA). The detection limits for F, Cs and Rb are <0·01 wt %. Natural minerals were used as standards for Si (quartz), Al (corundum), Ca (wollastonite), Ba (celsian), K (adularia; orthoclase— MAC Standard 2726), Na (albite—MAC Standard 2726), NUMBER 12 DECEMBER 1997 P (apatite), Nb (columbite—Standard M42), Ta (columbite—Standard M43), W (wolframite—Standard M45), Cs (pollucite—Standard M47) and F (CaF2). The synthetic phases RbAlSiO4 and CsF3 were used as standards for Rb and Cs. Electron microprobe analyses on melt inclusions Samples in the form of small polished thick-section chips (10 mm × 10 mm × 0·5 mm) were cut from hand samples and heated in thin-walled quartz glass ampoules under slightly oxidizing conditions (Ni/NiO + C) for 20 h by rapidly placing the ampoules in the hot zone of a tubular furnace. For this experiment we used temperatures between 600 and 900°C. However, each sample was heated only at one temperature and only for one time. After heating, the sample was quenched in liquid N2 at about 500°/s to transform the melt in the inclusion to a more or less homogeneous glass usable for microprobe analysis. Microprobe analyses were carried out using the Cameca CAMEBAX-SX50 and -SX100 instruments at the GeoForschungsZentrum Potsdam. The quartz chips containing melt inclusions were, after heating and quenching, polished down until the inclusions were exposed on the surface. Analyses were made in the wavelength-dispersive mode with the following conditions: 15 kV acceleration voltage, 10 nA beam current, spot size of 10–30 lm, 40 s counting time for F using the multilayer PC1 crystal, 60 s for Sn, 40 s for Rb and P, and 20 s for other elements. Synthetic oxides and minerals were used as standards. Rb was calibrated with natural Rb-rich feldspar (from Etyka–Eastern Transbaikalia; provided by R. Seltmann) and synthetic Rb-feldspar (prepared by St. Melzer, GFZ Potsdam). The accuracy of this technique was verified by analysing chips of synthetic hydrous glasses (given by F. Holtz and H. Behrens) and Macusani glass; the agreement between predicted and measured values is very good. Microthermometric (thermokinetic) determination of the volatile contents of melt inclusions Primary rock-forming minerals of granites and pegmatites can trap minute inclusions of the melt from which they are formed. Melt inclusions provide information on pressure, temperature and composition of the melt that were present during geological processes that have long since ceased. In particular, such inclusions contain information about the magmatic concentrations of volatile and semivolatile elements in the melt which are obtainable from 1738 BREITER et al. EVOLUTION OF A P-RICH GRANITE bulk analysis of rock samples. Degassing and recrystallization processes at the final stage of crystallization and postmagmatic–hydrothermal processes may change in part the composition of rock samples, whereas inclusions may preserve genetic information. It is widely accepted that melt inclusions represent trapped samples of melt, and hence their composition provides constraints on the original abundance of volatile and fluxing components (H2O, CO2, F, P and B) in the magma. However, in contrast to extrusive rocks with their glass inclusions, silicate melt inclusions in intrusive rocks are rarely studied. This is partly because they are small and mostly crystallized, and are therefore commonly overlooked or misrecognized. Because the melt inclusions are totally crystallized at room temperature, we have to homogenize these inclusions to glass before any analyses. This procedure must be performed so as to avoid or to assess volatile loss by diffusion. Leakage (i.e. opening of the inclusions) causes rapid and nearly complete loss of water and other volatiles. Leaked inclusions generally fail to homogenize under the experimental conditions (600–1000°C and 20 h). We used a heating–quenching technique that was first presented by Thomas (1994) and further developed by Thomas et al. (in preparation). Using the simple kinetic technique based on the different homogenization behaviour of the crystallized melt inclusions as a function of volatile content, temperature and heating duration (Thomas, 1994), we can obtain values for volatile contents for inclusions within the host mineral which are situated too deep in the sample or are too small for ion probe investigations or Fourier transform IR (FTIR) spectroscopy. The method cannot separate or distinguish the effect of water from that of other volatile, semi-volatile or fluxing components. Therefore, we express the results as ‘equivalent water content’ (C W), the amount of water which would produce the observed kinetic effect. In a first approximation, C W is the sum of H2O, F and P2O5 determined with the electron microprobe. This is analogous to the convention of reporting salinity in fluid inclusions as equiv. wt % NaCl from freezing point depression data. Analysis of whole-rock and mineral concentrates Major elements were analysed in the laboratory of the Czech Geological Survey Praha by standard methods of wet chemistry with detection limits well under the measured contents and standard deviation (SD) <1% relative. F was analysed by ion selective electrode, SD <0·13 wt %; phosphorus by spectrophotometry, SD <5% relative; Rb, Li and Cs by AAS, SD <5% relative. Rb, Nb, Sn and Zn were analysed by X-ray fluorescence in the laboratory of the Czech Geological Survey Praha, with a detection limit of 7 ppm, SD <4% relative (Rb), SD ~10% relative others. Ta and W were analysed by the neutron activation method by Bondar Clegg, Ottawa. Detection limits are 1 ppm (Ta) and 2 ppm (W). Other trace elements were analysed using inductively coupled plasma mass spectroscopy at GFZ Potsdam with the following detection limits: 0·004 ppm (Cs, Pr), 0·006 ppm (La), 0·007 ppm (Ce), 0·008 ppm (Gd, Er), 0·009 ppm (Ho, Lu, Y, U), 0·010 ppm (Sr, Tb, Tm, Yb), 0·011 ppm (Nd, Dy), 0·012 ppm (Eu), 0·014 ppm (Sm), 0·021 ppm (Pb), 0·052 ppm (Ba), 0·13 ppm (Th, Zr), 0·17 ppm (Hf ). 1739
© Copyright 2026 Paperzz