Introduction

The role of volatiles in the origin of granitic pegmatites and their particular textures is emphasized by numerous authors and confirmed by experiments (e.g., London 2008, 2009; Nabelek et al. 2010, and references therein). The significance of volatiles is also documented by the abundance of the minerals with H2O, F, B, P (e.g., micas, topaz, fluorite, tourmaline, apatite, primary Fe,Mn,Ca-phosphates) in most granitic pegmatites. In contrast, volatile-bearing minerals are very rare in the Ruda nad Moravou pegmatite, which is dominated by nominally anhydrous phases (plagioclase, quartz, K-feldspar, pyroxene, garnet). Of the phases containing OH or H2O, epidote-group minerals are the most abundant along with grossular garnet, yet occur only in subordinate amounts; biotite, fluorapatite and betafite/pyrochlore are very rare (Novák and Gadas 2010). Hence, this pegmatite exhibits numerous features indicative of low activity of volatiles.

Grossular garnet is a typical accessory to major mineral present in almost all textural-paragenetic (sub)units of the Ruda nad Moravou pegmatite (Novák et al. 1985; Novák and Gadas 2010). It is developed in crystals and aggregates distinct in size, shape and chemical composition. Consequently, it represents a suitable object for the study of compositional trends of major to trace elements including the behavior of H2O and F throughout the primary crystallization of this nominally “anhydrous” pegmatite characterized by the particular chemical system Na2O-CaO-Al2O3-SiO2>(K2O-FeO-MgO-H2O) and rapid crystallization. Detailed paragenetic, compositional (Electron Microprobe, Laser Ablation Inductively Coupled Plasma Mass Spectrometry, Cathodoluminiscence) and spectroscopic (Infrared and Fourier Transform Infrared Spectroscopy) studies were performed to find out compositional trends of major to trace elements in garnet as well as to reveal the activity of volatiles (mainly H2O) during the pegmatite evolution and compare our results with garnets from granitic pegmatites (e.g., Arredondo et al. 2001). The role of crystal-structural constraints on the incorporation of H2O, F and minor to trace elements into the crystal structure of grossular-rich garnet will also be discussed. Mineral abbreviations from Whitney and Ewans (2010) are applied in the text and images.

Internal structure, mineral assemblages, compositional trends and P-T-X-conditions

The anorthite- and grossular-bearing leucotonalitic pegmatite cuts serpentinized lherzolite, which belongs to the Staré Město Unit. Massive, greyish green to black rodingites (Di>Hbl>Grt,Tr,Phl] commonly form dikes, up to several dm thick, which cut the serpentinized lherzolite close to the pegmatite body. The relationship of the pegmatite dike and rodingite dikes is not clear due to poor outcrops. This geological sequence is interpreted to constitute the lower crust in a Cambro-Ordovician rift affected by Carboniferous metamorphism (Franke and Želažniewicz 2000; Štípská et al. 2001).

A narrow zone of the fine-grained contact zone (Amp>Di>Pl), up to ~10–20 cm thick, replacing host altered serpentinite, with lobate contact with the outermost part of pegmatite, is developed along the contact with host serpentinite. The zoned pegmatite, ~3–4 m thick and ~20–30 m long (Novák and Gadas 2010), consists from the contact inwards of a highly heterogeneous (i) transitional unit which forms a zone, up to 1 m thick, with a sharp contact to the inner coarse-grained unit. The transitional unit is composed of three texturally and compositionally distinct subunits (Fig. 1)—a minor grossular subunit (Grs>>Di>Qz~Pl) typically adjacent to the contact, the volumetrically dominant fine- to medium-grained leucocratic subunit (Pl>Qz>>Di~Ep>Grs), and small and rather rare coarse-grained nests (up to 30 cm in diameter) belonging to the diopside subunit (Pl>Qz~Di) randomly distributed within the leucocratic subunit. (ii) The coarse-grained unit (Pl>Qz>>Grs~Ep) consisting of dominant plagioclase and interstitial quartz is intimately associated with minor graphic intergrowths of pale brown grossular+quartz. It forms a well-defined zone, up to ~30 cm thick, and its contact with the inner graphic unit is typically transitional. (iii) The graphic unit (Pl+Qz>Kfs+Qz), ~10–15 cm thick, consists of coarse-grained graphic intergrowths of dominant plagioclase+quartz and minor K-feldspar+quartz in the inner parts. It gradually evolves to the volumetrically minor blocky unit (iv) built of euhedral crystals of abundant plagioclase and rare K-feldspar (deep green amazonite), both up to 10 cm in size, enclosed in massive quartz. The innermost unit is represented by a (v) quartz core, which attains a thickness of up to ~50 cm. It contains rare mostly colorless crystals of grossular, typically closely adjacent to the blocky unit (Novák et al. 1985; Novák and Gadas 2010). Gradational contacts between the individual textural-paragenetic (sub)units are typical except for the contact of leucocratic subunit (of the transitional unit) and coarse-grained unit (Fig. 1).

Fig. 1
Fig. 1
Full size image

Idealized cross-section through the southeastern segment of the pegmatite body and the mineral assemblages in the individual (sub)units with the relative content of major phases. Selected accessory minerals, the plagioclase composition and garnet labeling (G1–G5) are included. Sharp textural contacts are marked as vertical solid lines; dashed vertical lines mark transitional contacts. The pegmatite occurrence within the Czech Republic and the Bohemian Massif is marked with star at the top right. (Slightly modified from Novák and Gadas 2010)

White to grayish plagioclase forms anhedral to euhedral grains, 0.1–2 mm to 10 cm in size, commonly zoned with An-enriched cores. Its chemical composition evolves from An98–46 in the transitional unit, through An82-39 in the coarse-grained unit to An49–20 in the graphic and blocky units (Fig. 1). Subhedral to euhedral elongated crystals of pale to green diopside up to ~5 cm long, with XMg = 97–82, occur almost exclusively in the contact unit, diopside subunit and grossular subunit. Dark green grains of epidote-group minerals (epidote-clinozoisite-dissakisite-allanite) vary from 0.1 mm to 2 cm in size. Similarly as diopside, they are present chiefly in the outer units—the transitional unit and the outer part of the coarse-grained unit (Fig. 1). The accessory minerals include common euhedral crystals of zircon with variable habit, up to 1 cm in size, rare baddeleyite, U-rich zirconolite, and gittinsite. Accessory titanite in several paragenetic and morphological types is also quite common, whereas betafite-pyrochlore, pyrochlore, fluorapatite, monazite-(Ce), Ti-rich hematite, biotite and rutile were identified as very rare primary phases. Plagioclase, diopside, and zircon show evident compositional evolution from the contact inwards, decreasing Ca/Na, Mg/Fe, and/or Zr/Hf, respectively, which imply a fractional crystallization (Novák and Gadas 2010). No simple compositional trends were recognized in epidote-group minerals except for decreasing ∑REE inwards. Contamination from the host serpentinite is indicated by the elevated contents of Mg and Cr in diopside and epidote-group minerals in the contact zone and diopside subunit (Novák and Gadas 2010). Several subsolidus alteration phases with common albite, clinozoisite, sericite, and Ba-rich feldspars, and rare actinolite, harmotome, chabazite-Ca, armenite, chlorite, prehnite, pumpellyite-(Al), niobian titanite and Nb,Ta,Ti-oxide minerals (niobian rutile, fersmite) were found on thin late veins and vein-like aggregates cutting early plagioclase, grossular, quartz and K-feldspar.

The paucity of minerals containing H2O, F, or P and absence of micas in the pegmatite suggest low activities of these constituents. Minerals containing CO2, B, Li or Be are absent. This implies a pegmatitic melt poor in H2O and other volatiles, which crystallized sequentially, inwards from the contact with the host rocks. The zoned internal structure, typical pegmatitic textures and fractionation trends in some minerals indicate moderate to strong undercooling of the pegmatitic melt (London 2008). The mineral assemblages and chemical composition of the minerals suggest the PT conditions of the initial pegmatite melt at P > ~0.5–0.8 GPa and Tmax ~700–780 °C. However, two-feldspar thermometry (Fuhrman and Lindsley 1988; Elkins and Grove 1990) applied for the pair plagioclase + non-perthitic K-feldspar from the graphic unit yielded significantly lower temperatures, 525–422 °C and 547–481 °C (Fuhrman and Lindsley 1988) and 529–429 °C and 543–464 °C (Elkins and Grove 1990), for P = 0.5 and 0.8 GPa, respectively. They support a strong undercooling of the pegmatitic melt and rapid crystallization. The mineral assemblages in late subsolidus veinlets including e.g., prehnite, clinozoisite, chlorite and pumpellyite-(Al) crystallized at T < ~300–350 °C (e.g., Frey et al. 1991); zeolites indicate lower T. The Ruda nad Moravou pegmatite represents a unique type of non-granitic, quartz-rich and volatiles-poor pegmatite—leucotonalitic to quartz leucogabbroic in bulk composition—with significant early Ca-enrichment and minor in-situ Mg-contamination from the host serpentinite (Novák and Gadas 2010).

Methods

EMPA

The chemical analyses of garnets and other minerals were carried out on a CAMECA SX-100 electron microprobe (Joint Laboratory of Electron Microscopy and Microanalysis, Department of Geological Sciences, Masaryk University, Brno and Czech Geological Survey, Brno) operating in wavelength-dispersion mode at the following conditions: an accelerating voltage of 15 kV, beam current of 10–20 nA, spot diameter of 1 μm. Peak counting times were 10–20 s for major elements and 20–40 s for minor and trace elements while half-length counting times were used for backgrounds. Analytical lines (Kα) in the following standards were used: Andradite (Ca, Fe), spessartine (Si, Mn), sanidine (Al), albite (Na), hornblende (Ti), spinel (Mg), chromite (Cr), fluorapatite (P), topaz (F), YAG (Y). The data reduction was done using PAP Φ(ρZ) corrections (Pouchou and Pichoir 1985). Mineral formulae were calculated assuming all Fe as ferric and a) 12 anions (O2−,OH−,F−) and b) 12 anions (O2−, F−) to compare the differences between both types of calculation. The proportions of end-members were calculated assuming that all Mg is represented by pyrope, all Mn is represented by spessartine, all Fe is represented by andradite and all Ti is represented by unspecified Ti-grt. Surplus of Ca in grossular-rich garnets was tested with different standards to exclude possible errors during measurement.

LA-ICP-MS

The instrumentation for LA-ICP-MS at the Department of Chemistry, Masaryk University, Brno, consisted of a laser ablation system UP 213 (New Wave, USA) and an ICP-MS spectrometer Agilent 7500 CE (Agilent, Japan). A commercial Q-switched Nd:YAG laser ablation device works at the 5th harmonic frequency which corresponds to the wavelength of 213 nm. The ablation cell was flushed with helium (carrier gas), which transported the laser-induced aerosol to the inductively coupled plasma (1 l/min). For LA-ICP-MS measurements, we used a hole-drilling mode (fixed sample position during laser ablation) with a duration of 75 s for each spot. The laser ablation was performed with a laser spot diameter of 40 μm, laser fluence 24 J cm−2 and repetition rate 10 Hz.

IR spectroscopy

Doubly polished sections were prepared from each (sub)unit of the pegmatite and examined under the optical microscope using a total magnification of 400×. The garnet grains proved in sectors to be free of inclusions and impurities and therefore suitable for IR measurements. Especially a single-crystal from a small cavity within the grossular subunit was of a gem quality, thus convenient for measurement along a profile. In general, there are few fluid inclusions or epidote-group minerals in the host garnets.

The measurement (transmission mode) in the selected points was done at the Institute of Mineralogy and Crystallography, University of Vienna, using a Hyperion 1000 IR microscope (LN2-cooled MCT detector) attached to a Bruker Tensor 27 IR spectrometer (globar light source, KBr beam splitter, DTGS detector). The measured area was unitarily set to a square of 40 × 40 μm2. Each measurement consisted of 120 separate scans to suppress noise.

All fitting procedures were made using the program “PeakFit” (Jandel Scientific). The use of Voigt-shaped curves produced the best fits. The band position error caused by overlapping bands is about ± 6 cm−1 and the amplitude error of about 15 % in the case of strongly overlapping peaks. A compilation of previous studies was used as a basis to set the number of bands and their wavenumber position in the initial fit estimate (Rossman and Aines 1991; Hösch and Langer 1996; Kolesov and Geiger 2005). The background consisted of a single wide Voigt-shaped peak, compensating for the broad absorption centered roughly at 3,450 cm−1 caused by the presence of water in fluid inclusions and moisture adsorbed along fissures in the sample.

Quantitative water determination

The total integral absorbance (A i tot) of all spectral bands related to OH defects in garnet was used for the calculation of the integrated absorption coefficient α i (cm−2; α i  = A i tot/t, where t is the thickness of the sample in cm). As expressed by Beer’s law, α i is directly related by the integrated molar absorption coefficient ε i [l.(mol−1H2O.cm−2)] to the concentration c (mol.l−1). The water content in weight percent is c H2O = (1.8/D) (α i /ε i ), where D is the density of the mineral (3.6 g/cm3) (Libowitzky and Rossman 1996; Libowitzky and Beran 2004). Varying ε i values for grossular-andradite garnet were determined based on different independent calibration methods (Rossman and Aines 1991; Bell et al. 1995). A pronounced correlation between band wavenumber position and the ε i value was determined by Libowitzky and Rossman (1997) and is used frequently to derive ε i for nominally anhydrous minerals if no independent calibration method is available. In general, if an ε i value based on independent calibration data is available, it should be preferred. The one used here was taken from the work of Maldener et al. (2003) based on NRA measurements on garnet. Owing to the similar chemical composition of all garnet types in the pegmatite at Ruda nad Moravou and the presence of the same absorption bands in all garnet spectra, the use of a single ε i value for all measurements becomes justifiable.

Cathodoluminescence

Two methods were applied for the study of cathodoluminescence in garnets. The device with hot cathode HC2-LM, Simon Neuser, Bochum was used in the CL laboratory, Department of Geological Sciences, Masaryk University, Brno, for monitoring of color changes in the individual grossular types. For detailed study of textural relations and zoning EMP-CL was used at following conditions: an accelerating voltage of 15 kV, beam current of 60 nA on a CAMECA SX-100 electron microprobe (Joint Laboratory of Electron Microscopy and Microanalysis, Department of Geological Sciences, Masaryk University, Brno and Czech Geological Survey, Brno).

The methods used in this study differ significantly in terms of the volume of analyzed material. The electron microprobe has a beam-spot diameter of ~1 μm, whereas the LA-ICP-MS employs a laser spot diameter of ~40 μm and the IR measurements in transmission mode represent the result from a large volume of the sample due to the sample thickness ranging from 300 to 500 μm. The LA-ICP-MS results are difficult to correlate with the very fine-scale zoning evident in BSE images of garnet from the grossular subunit, that is, each method analyzes a different scale of zoning.

Results

Paragenetic and morphological types of garnet and their mineral assemblages

The outermost massive to fine-grained grossular subunit is composed of 90–98 vol.% of garnet (G1) (Fig. 2a, b, c) associated with thin, subparallel, discontinuous and locally folded layers of diopside grains (Fig. 2b), up to 0.5 mm thick, euhedral pseudomorphs after basic plagioclase, up to 10 mm in size (Fig. 2a), composed of a fine mixture of albite, muscovite and clinozoisite. Minor phases are represented by quartz and plagioclase (Fig. 2c, d). Accessory minerals include rare grains of fluorapatite, up to 200 μm, subhedral and strongly metamictized zircon, up to 50 μm (Fig. 2c), and oscillatory-zoned allanite, up to 60 μm in size. Interstitial epidote fills thin cracks in garnet along with minor albite and K-feldspar. The garnet (G1) usually forms anhedral and isometric grains up to 500 μm, in almost monomineral parts of the grossular subunit. Subhedral crystals of garnet (G1), up to 1 mm, in diameter (Fig. 2d), were also found in the plagioclase, quartz and diopside-enriched portions of the grossular subunit adjacent to the leucocratic subunit and in the outer parts of the leucocratic subunit. Locally, inclusions of Ba-enriched epidote-clinozoisite and quartz as well as small inclusions of zircon, (~5 μm) occur in garnet (Fig. 2d). Cores of the individual grains of the garnet (G1) display orange-colored cathodoluminescence with a complex fabric in EMP-CL (Fig. 2e, f). Irregular clusters of interstitial garnet (G2), up to 1 cm in size (Fig. 3a), are randomly distributed within the leucocratic subunit. Garnet fills irregular spaces between the grains of plagioclase, epidote-clinozoisite, quartz and diopside (Fig. 3b). The coarse-grained unit is characterized by common graphic intergrowths of quartz + pale brown garnet (G3), up to 5 cm large, typically present close to the contact with the transititional unit (leucocratic subunit) (Fig. 3c, d). A fine-grained secondary assemblage of epidote-clinozoisite, albite, quartz and titanite is developed along the contacts of garnet and large grains of plagioclase, whereas the cracks in garnet are commonly filled by epidote and rare Nb-rich titanite. No cathodoluminescence was observed in the garnets (G2) and (G3). Only sporadic anhedral grains of garnet (G4), up to 40 μm, were found in plagioclase from the blocky unit. Owing to their dimension and scarcity, only EMPA data are available. The quartz core contains rare, euhedral, colourless to pale green, semitranslucent and strongly fractured crystals of garnet (G5), up to 3 cm across (Fig. 3e; Novák et al. 1985; Novák and Gadas 2010). They are present typically close to the contact with the blocky unit. Domains enclosing abundant subhedral to euhedral inclusions of U-rich pyrochlore-betafite, up to 30 μm, and scarce zircon and galena, were locally observed within the crystals of garnet (G5). Irregular fissures, up to 0.5 mm thick, filled by quartz, epidote, albite and less common Nb-rich titanite and fersmite (Fig. 3f), are abundant. Homogenous orange-to-red colored cathodoluminescence was observed in the garnet (G5). All garnet types (G1, G3, G5), which form rather large grains or aggregates, typically occur towards the outer contacts of the units hosting them (Fig. 1). All garnet types are isotropic, no indications of anomalous anisotropy were observed.

Fig. 2
Fig. 2
Full size image

Images of garnet (G1) from the grossular subunit: a macrophotograph of euhedral pseudomorph (in the center) after plagioclase in white massive grossular, photo; b folded layers of diopside, BSE; c zoned grains of garnet, BSE; d euhedral crystals of garnet with epidote and quartz inclusions from a plagioclase- and quartz-rich part of the grossular subunit, BSE; complexly zoned garnet e BSE versus f EMP-CL images. The scale bars equal 0.5 mm

Fig. 3
Fig. 3
Full size image

Macrophotographs and BSE images of garnets. a, b interstitial garnet (G2) rich domains from the leucocratic subunit; c, d graphic intergrowths of quartz+garnet (G3) from the coarse-grained unit near the contact with leucocratic subunit; e, f fragment of crystal of garnet (G5) from the quartz core The scale bars equals 0.5 mm for the BSE images

Chemical composition of garnet

Electron microprobe

The chemical composition of garnets (G1 to G5) is more or less close the grossular end-member in all pegmatite (sub)units; elevated contents of Fetot (Fe3+ >> Fe2+) is the main difference along with the small variations in the other components, Mn, Mg, Ti and F (Table 1; Figs. 4 and 5). The Fe3+ contents as well as the chemical heterogeneity (variations of Mg, Mn, Ti) of garnet generally increase from the garnet (G1) in the grossular subunit through Fe-enriched garnets (G2) and (G3) in the leucocratic subunit and coarse-grained unit, respectively, followed by a steep decrease in garnets from the blocky unit (G4) and quartz core (G5) (Fig. 5a).

Table 1 Selected EMP analyses of individual types of garnet from the pegmatite
Full size table
Fig. 4
Fig. 4
Full size image

Plot showing the chemical composition of garnets from the individual pegmatite (sub)units

Fig. 5
Fig. 5
Full size image

Plots showing the contents (mol.%) of andradite (Adr), Ti-garnet (Ti-grt), pyrope (Prp) and spessartine (Sps) components and concentration of F and H2O (as OH) both in apfu in garnets from the individual pegmatite (sub)units (a, b, c, d) and the contents of selected trace elements including REE patterns during the evolution from border zone to quartz core (e, f, g, h) in garnets from the individual (sub)units. The shaded part in h marks area under the detection limits of REE. The same colors are used in Fig. a, b, c, d as in the Fig. 4

Of the five garnet varieties, garnet (G1) shows the most complex compositional pattern in BSE images (Fig. 2c, e). Individual grains are generally composed of irregularly shaped cores (G1a, Fig. 2e) containing 3 subzones, which are distinguishable only using EMP-CL (Fig. 2f) and differ slightly in Fe, Mn, Mg and Ti contents (0–0.064 apfu, 0–0.015 apfu, 0–0.034 apfu and 0–0.007 apfu respectively). The content of F is slightly elevated in these 3 subzones (0.024–0.041 apfu, Table 1). The cores (G1a) are overgrown by Fe-enriched zone (G1b) (0.117–0.155 apfu Fe3+, Fig. 2c, e) of variable thickness (0–100 μm) depleted in Al (1.822–1.860 apfu) and F (0.014–0.024 apfu) and slightly enriched in Mg and Mn (0.017–0.022 apfu and 0.011–0.015 apfu respctively; Figs. 4 and 5a, b, c, Table 1). Very similar zoning and chemical composition were identified in subhedral crystals of garnet (G1) containing inclusions of epidote-clinozoisite. Clusters of the interstitial garnet (G2) are irregularly zoned and moderately heterogeneous with Fe3+ 0.216–0.402 apfu, (up to 20 mol.% of andradite component, Figs. 4 and 5a). Elevated contents of Ti (0.01–0.07 apfu, 0.15–1.25 wt.% TiO2; Fig. 5b, Table 1) correlate weakly with Fe3+. Contents of Mn and Mg are very low, 0.01–0.02 and 0.00–0.03 apfu, respectively (Fig. 5c, Table 1). Garnet (G3) shows high variability in Fe3+ 0.17–0.48 apfu (up to 24 mol.% of andradite component; Figs. 4 and 5a) and elevated contents of Mg 0.01–0.03 apfu, Mn 0.01–0.06 apfu, Ti 0.005–0.09 apfu (Fig. 5b, c, Table 1), and low F (0.01–0.14 apfu; Fig. 5d; Table 1). However, the individual grains are rather homogeneous (Fig. 5d). The chemical compositions of scarce grains of the garnet (G4) and large euhedral grossular crystals (G5) are almost identical and very close to end-member grossular (98–100 mol.% Grs; Figs. 4 and 5a). The contents of Fe3+, Mg, Mn and Ti are negligible (≤0.02 apfu in all cases; Fig. 5b, c). However, the content of F in both (G4 and G5) is elevated (0.059–0.089 apfu; Fig. 5d, Table 1).

LA-ICP-MS

The individual types of garnet (G1, G2, G3, G5) show different concentrations of trace-elements (Table 2); however, the zoning in garnet (G1) observed by CL study was ignored. Three distinct patterns were revealed. The first one is characterized by a very good positive correlation of Y, REE, V and Ni with the Fe3+ contents (Fig. 5a, e). Other elements also show a positive correlation with Fe3+ contents but less apparently pronounced (e.g., Zr, Hf, Co, Th, U; Fig. 5f, g). The second pattern is characterized by an evident negative correlation with the Fe3+ contents (Ta, Pb; Fig. 5g). Zinc and particularly Be are the only elements with evident increase of their concentrations from the outermost grossular subunit toward the quartz core (Fig. 5f) independent of Fe contents; this is the third trend. Some trace elements (e.g., Sc, Ga, As) exhibit rather uneven behavior. Very low concentrations of Cr (≤12 ppm but mostly below detection limit) are typical for all garnet types.

Table 2 Concentrations of trace elements (LA-ICP analyses) from individual types of garnet
Full size table

The garnets differ in the REE patterns (Fig. 5h, Table 3). The lowest concentrations of REE + Y are in garnets close to the end-member grossular composition—(G1 from the grossular subunit and G5 from the quartz core). Both have maximum contents of MREE (Yb/Sm = 0.1–0.8, 0–0.35 respectively), show a slight negative Gd anomaly and quite considerable scatter in HREE (Fig. 5h). Garnet (G2) from the transitional unit has the highest amounts of REE with the maximum in MREE (Yb/Sm = 0.45–2.7), slightly decreased HREE contents (Yb/Dy = 0.26–0.38; Fig. 5h) and no negative Gd anomaly (Fig. 5h). A somewhat different but uniform pattern is evident in garnet (G3) from the graphic intergrowths with the maximum in HREE (Yb/Sm = 7–21), a steep decrease towards MREE (Fig. 5h) and a slight negative Gd anomaly with a relatively flat pattern for HREE (Yb/Dy = 0.64–1.2).

Table 3 Selected LA-ICP analyses and concentrations of Y and REE from the individual types of garnet. Detection limits are shown in Fig. 5h
Full size table

IR spectroscopy and H2O contents in garnet

A series of sharp absorption bands in the wavenumber range 3,500–3,700 cm−1 is present in the IR spectra of all the investigated types of grossular garnet. A well developed single crystal of garnet (G1) from the grossular subunit, revealed exceptional optical quality. The central part (G1a), also being richer in H2O, shows well developed bands throughout the entire 3,500–3,700 cm−1 wavenumber region, with dominant absorptions present mainly at higher wavenumbers between 3,674 and 3,600 cm−1 (Fig. 6). Additional smaller bands exist at 3,690 cm−1 and at lower energies (roughly 3,585, 3,561 and 3,531 cm−1). The relative amplitudes of the bands described above change from core to rim (G1b) of the crystal. While the band intensity of the peaks at 3,631 and 3,621 cm−1 increases, the amplitudes of the surrounding bands decrease, as do the amplitudes of all the low-energy bands up to 3,600 cm−1.

Fig. 6
Fig. 6
Full size image

Fitted IR spectra of the different garnet types. The left set of charts shows the character change from the core (G1a) to rim (G1b) of garnet in the grossular subunit. Charts at the right show spectra for interstitial garnet (G2), garnet from graphic intergrowths (G3) and crystals from the quartz core (G5)

Interstitial garnet (G2) from the leucocratic subunit shows a uniform spectral shape with the dominant band at 3,600 cm−1 and several well developed peaks at 3,631, 3,614, 3,585 and 3,552 cm−1. Weaker absorptions at 3,674, 3,654 and 3,531 cm−1 occur as shoulders. Due to the fine size of the interstitial garnet clusters embedded in anorthitic plagioclase and quartz, the OH absorption spectra are often distorted by an extremely broad band centred at 3,450 cm−1 caused by the presence of H2O adsorbed along cracks and intergranular boundaries.

The IR spectra of garnet (G3) from the graphic intergrowths with quartz are nearly identical in shape to those of interstitial garnet (G2), despite the big difference in the overall water content, as will be described below.

A sharp contrast in spectral shape is seen in large, greenish single crystals of grossular (G5) from the quartz core. Bands on the high energies up to 3,700 cm−1 are dominant, while the low-energy absorptions in the 3,600–3,500 cm−1 region are missing. No difference in the overall shape of the spectra could be seen between the individual analytical points.

The presence of trace hydrogen contents in garnet is now well established as the main reason for the occurrence of absorption phenomena within the wavenumber region described above. Many previous papers (Aines and Rossman 1984; Lager et al. 1989; Rossman and Aines 1991; Andrut et al. 2002) focus on obtaining detailed crystal-chemical information regarding the actual character of the various types of OH defects indicated by the absorption bands. By comparing IR spectra of synthetic standards with natural analogues from rodingites, Rossman and Aines (1991) described the well-known hydrogrossular substitution, (SiO4)4−↔(O4H4)4−, causing the occurrence of absorptions at 3,600 and 3,640 cm−1 in pure grossular. Further splitting into the rather large number of bands may be only partly associated with changes in sample chemistry and trace element contents; other types of OH defects are presumed (Aines and Rossman 1984; Rossman and Aines 1991). The purpose of this study is to determine quantitatively the behaviour of volatile components in grossular garnet in the different zones of the pegmatite. Therefore, the actual character of OH defects present in our garnets will not be discussed in detail.

The recalculation of the total integral absorbance values from all acquired spectra to H2O concentration via Beer’s law yielded variable contents of H2O in the individual garnet types. Garnet (G1) shows concentrations one order of magnitude lower; the cores (G1a) with 0.03 wt% and the rims (G1b), with as little as 0.01 wt% H2O, represent the most anhydrous grossular garnet found within the pegmatite. The interstitial garnet (G2) from the leucocratic subunit generally displays uniform contents of H2O of 0.02 wt% comparable with the garnet (G1b) from the grossular subunit. The slightly varying values around 0.22 wt% H2O in garnet (G3) from the graphic intergrowths are one order of magnitude higher than in the two outermost (sub)units. Garnet (G5) from the quartz core with 0.04 to 0.05 wt.% H2O contains about five times less H2O then garnet (G3) from graphic intergrowths.

Fluorine and OH groups have the same charge and atomic radii compatible with O2− (Shannon 1976). If charge needs to be balanced, F and OH anions can play the same role in heterovalent substitutions, effectively “competing” against one another. The marked negative correlation between OH and F contents may suggest their occupying the same structural site. The total (H2O+F) contents are comparable throughout all (sub)units within the pegmatite. Low quantities of accessory elements such as Na, K or P in garnet do not appear to influence the H2O or (OH+F) contents in any way.

The changes in the H2O content along a profile from the border to the centre of the pegmatite body can be closely correlated with the changes in the chemical composition of garnet. As a general rule, the closer its composition corresponds to the grossular end-member, the less H2O (bound as OH) and the more fluorine it contains. Elevated amounts of Ti, Fe, Mn and Mg in the garnet structure are generally accompanied by a significant increase in water concentration. The sum of the volatile components (H2O+F) increases about twice from the outer (sub)units (G1, G2) to the centre (G3, G5) of the pegmatite dike.

Discussion

Morphological and textural evolution of garnet

The garnet from the pegmatite Ruda nad Moravou exhibits complex morphological and textural evolution from the border zone to centre of the pegmatite body: (i) anhedral to subhedral grains (<0.5–1 mm) from the outermost grossular subunit and adjacent leucocratic subunit (G1); (ii) irregular interstitial clusters (<1 cm) in the leucocratic subunit (G2); (iii) graphic intergrowths with quartz, up to 5 cm in diameter, in the coarse-grained unit (G3); (iv) very rare microscopic subhedral grains (~40 μm) in the blocky unit (G4); (v) large euhedral crystals of the garnet, up to 3 cm in size, from the quartz core (G5). Garnet is also apparently concentrated in the outermost parts of the individual (sub)units (Fig. 1) except for the interstitial garnet (G2) and inclusions of (G4), where the textural relations remain unclear. Such variable morphology of the garnet grains and aggregates, and their textures within a single pegmatite body are remarkable. Usually more or less perfectly developed crystals of garnet, varying in size (mm to cm), and their aggregates are the most common (e.g., Kleck and Foord 1999; Arredondo et al. 2001; London 2008).

Irregular interstitial clusters of garnet are relatively exceptional in granitic pegmatites. Garnet+quartz graphic intergrowths are also scarce in simple granitic pegmatites (e.g., Němec 1983; Zhang et al. 2001), and common only in some elbaite-subtype pegmatites (Novák and Povondra 1995). Highly variable morphology and textural relations to the associated minerals (plagioclase, quartz, epidote) imply a distinct position of the garnet in the crystallization sequence within the individual (sub)units, where it may have been very early crystallizing mineral (e.g., G5 in the quartz core) or a late forming but still primary (magmatic) phase (interstitial G2 in the leucocratic subunit). Graphic intergrowths of quartz and other minerals are commonly explained by a rapid crystallization from high-viscosity undercooled melt (London 2008) and graphic intergrowths of garnet+quartz are the earliest ones in the pegmatite followed by plagioclase+quarz and K-feldspar+quartz intergrowths in the adjacent graphic unit. Consequently, the individual types of garnet from Ruda nad Moravou have distinct position in the crystallization sequence within the individual (sub)units.

Compositional evolution of garnet during the pegmatite evolution

Garnets from granitic pegmatites are typically spessartine-almandine solid solution (London 2008). An increase in Mn/(Mn+Fe) from the border (wall) zones to the central parts of the pegmatite bodies is the most typical (e.g., Morgan and London 1999; Kleck and Foord 1999) although reverse evolution also is known (Manning 1983). Minor to trace amounts of the other garnet components—andradite, grossular and pyrope are common but only exceptionally andradite-dominant garnet (Lentz 1992; Breiter et al. 2010), grossular-dominant garnet (Swanson and Veal 2010; Novák and Gadas 2010), and garnet with high contents of Mg, up to 3.73 wt.% MgO (Gordiyenko and Leonova 1976), occur. The parental pegmatites with such unusual garnet compositions are believed to have been contaminated from their host rocks (Breiter et al. 2010; Novák and Gadas 2010) or their geochemical signature is uncertain. The concentrations of Y+REE are mostly low; only some garnets from NYF (or mixed) pegmatites contain up to ~3.5 wt.% ∑Y+HREE (e.g., Wang et al. 2003; Hönig et al. 2010; Grew et al. 2010).

Garnet from the Ruda nad Moravou pegmatite is Ca-rich varying from nearly end-member grossular—(G1) from the grossular subunit, garnet inclusions (G4) from the blocky unit, and large crystals (G5) from the quartz core, to grossular-andradite Grs73–86 Adr11–24—(G2) in the leucocratic subunit, and graphic garnet+quartz intergrowths (G3) from the coarse-grained unit. The individual crystals (grains) exhibit weak (G3) to moderate compositional zoning (G1, G2, G3) or they are almost homogeneous (G4, G5; Figs. 2, 3, 4 and 5a, b, c). The amounts of minor elements (Mn, Fe2+, Mg, Ti) are very low, up to 2 mol.% of spessartine, 2 mol.% almandine, 1 mol% of pyrope and 4 mol.% of Ti-garnet evidently related to high contents of Fe3+ in garnet (Fig. 5a, b, c, Table 1). The behaviour of the most trace elements is more complicated: a very good positive correlation of Y, REE, V and Ni with the Fe3+ contents (Fig. 5a, d); a weak positive correlation with Fe3+ contents (e.g., Zr, Hf, Co, Th, U; Fig. 5e, f); an evident negative correlation to the Fe3+ contents (Ta, Pb; Fig. 5f); an evident increase from the outermost subunit toward the quartz core (Be) disregarding Fe contents (Fig. 5e). Very low concentrations of Cr are typical for all garnet types including the garnet (G1) associated with Cr-enriched diopside in the grossular subunit. Preferential incorporation of some elements into Fe3+-enriched garnet and others into Fe3+-poor one is very likely constrained by crystal-structural factors, when the significant substitution Fe3+ = Al3+ enables incorporation of the elements into the dodecahedral A-site (e.g., REE, Y, Ni), whereas Zr and V likely enter the octhedral B site. The surplus of up to Ca ≤ 3.15 apfu in the garnet (G5) (Table 1) is unusual although near-end member grossular has typically higher Ca ≤ ~3.05 apfu (Dachs et al. 2012). Distinct behaviors of minor and trace elements in Fe3+-rich garnets (G3, G2) and in almost Fe-free garnets (G5, G4, G1) imply the participation of distinct substitutions in the individual garnet groups including incorporation of OH and F also supported by distinct IR spectra (Fig. 6; Dachs et al. 2012).

The most important carriers of REE in the pegmatite, epidote-group minerals (allanite, dissakisite, epidote, clinozoisite), are the most abundant in the leucocratic subunit. In this case, we can expect preferential incorporation of LREE into these minerals and lower concentrations of LREE in the associated garnet (G2), particularly if this interstitial garnet is the latest in the crystallization sequence of this subunit evidently later than associated epidote (Fig. 3b). Surprisingly, the garnet (G2) has the highest concentrations of REE and elevated contents of LREE comparing the other garnet types (Fig. 5h). The garnets (G2) and (G3) have very similar chemical compositions and behavior of trace elements (Fig. 5e, f, g, h), although their parental (sub)units (leucocratic subunit and coarse-grained unit) show the only sharp contact between the individual units within the pegmatite body (Fig. 1). Moreover, they are characterized by the absence or abundance of typical pegmatite textures (e.g., graphic intergrowths, blocky crystals), respectively.

Contents of H2O in garnets

There are relatively few determinations of H2O concentration in garnets from granitic pegmatites (e.g., Arredondo et al. 2001). Some differences exist in the value of εi (integral absorption coefficient) for grossular garnet, representing the most significant source of error in the terms of absolute water contents. Their relative changes in different garnet types, important for our interpretations, remain largely unaffected. Spessartine-almandine from the localities Rutherford No. 2 pegmatite, Amelia, Virginia; the Himalaya Mine, Mesa Grande, California, and the George Ashley Block pegmatite, Pala, California yielded 0.022–0.158, 0.0017–0.034 and 0.0004–0.007 wt.% H2O, respectively. The increase in H2O amounts from the border zone to the center of the pegmatite body is typical. The Himalaya Mine (e.g., Foord 1976) and George Ashley Block pegmatite (Webber et al. 1997; Kleck and Foord 1999) exhibit an LCT signature and the degree of their fractionation varies from a highly evolved lepidolite-subtype pegmatite with pockets lined with crystals of elbaite to complexly zoned but geochemically rather primitive pegmatite. The Rutherford No. 2 pegmatite is distinct; the presence of fergusonite indicates some affinity to the NYF family; however, common muscovite along with accessory apatite, garnet and tourmaline indicate rather an LCT signature with evident features of peraluminous composition. Nevertheless, a high Ca-content in garnet (3.59–0.86 wt.% CaO) also suggests an unusual geochemical situation and points out that the Rutherford No. 2 pegmatite is not an ordinary LCT-family pegmatite and shows some tendency to the NYF types.

The anorthite- and grossular-bearing pegmatite from Ruda nad Moravou with grossular-dominant garnet is a unique example of a pegmatite compositionally related to leucotonalite or to leucocratic quartz gabbro (Novák and Gadas 2010); hence, a comparison with ordinary granitic pegmatites given above is limited. In spite of paragenetic and compositional differences, the concentrations of H2O in garnet (0.02–0.22 wt.%) are comparable with the Rutherford No. 2 pegmatite (0.022–0.158 wt.%); the Himalaya and George Ashley Block pegmatites yielded about 1 to 2 orders of magnitude lower concentrations. Whether the elevated but still very distinct contents of CaO in garnets from both pegmatites (Ruda nad Moravou and Rutherford No. 2 up to 38.78 and 3.59 wt.% CaO, respectively) have played a role in the incorporation of H2O into the garnet crystal structure remains unclear. However, in the Rutherford No. 2 pegmatite H2O contents are negatively correlated with the concentrations of Ca. Moreover, the mineral assemblages at Ruda nad Moravou with accessory epidote-group minerals as the only nominally H2O-bearing minerals are distinct from the volatiles-enriched assemblages involving muscovite, tourmaline, apatite and fluorite at the Rutherford No. 2 pegmatite.

We also did not find an increase of H2O in garnet from the border zone to the centre of the pegmatite from Ruda nad Moravou, as is typical for the other localities (Arredondo et al. 2001). Instead, elevated contents of H2O in garnet are positively correlated with Fe3+ (andradite component) and negatively correlated with F contents. In summary, the concentrations of H2O in garnet from the Ruda nad Moravou pegmatites are comparable or slightly higher then in the other granitic pegmatites, but the reason, why H2O preferentially enters the garnet structure, seems in part to have been controlled by crystal-chemical factors. Both the activities of H2O and F in the residual melt as well as the crystal structural constraints (elevated content of Fe3+) seem to have played a role. However, the sum of volatile components (H2O + F) increases about twice from the outer (sub)unit to the centre. The H2O concentrations in grossular garnet from Ruda nad Moravou are similar to those in grossular from calc-silicate rocks, e.g., H2O concentrations generally do not exceed ~0.2 wt.% in grossular from high-temperature skarns and upper amphibolite facies rocks (Rossman and Aines 1991; Dachs et al. 2012).

Summary and conclusions

The morphological and compositional evolution in garnet from the Ruda nad Moravou pegmatite is unique and manifested mainly by highly variable morphology of garnet grains including anhedral to euhedral garnet (G1) of the grossular subunit, interstitial garnet (G2) in the leucocratic subunit, graphic intergrowths of quartz+garnet (G3) in the coarse-grained unit, microscopic inclusions of garnet (G4) in the blocky unit, and large euhedral crystals (G5) from the quartz core (Figs. 2 and 3). The marked increase in the amounts of Fe3+, Mn, Mg, Ti and most trace elements from (G1) to (G2) and futher to (G3) may have resulted from fractionation of the melt. Once the residual melt was depleted in these elements, which is evident in the absence of diopside and near absence of epidote-group minerals and titanite within the coarse-grained unit, the next garnet to crystallize, (G4) in the blocky unit and (G5) in the quartz core, are thus close to the end-member grossular. The formation of garnet (G5) was likely controlled by high activity of Ca in the very early stage of the quartz core crystallization at T < ~400–450 °C. Such low temperature did not allow formation of anorthite-rich plagioclase as in the outer zones (Fig. 1). Other low-temperature Ca-rich and Fe-poor silicates (e.g., clinozoisite, prehnite) known from pockets in some pegmatites or as replacement products after plagioclase (Breiter et al. 2010; Dosbaba and Novák 2012) were still not thermodynamically stable. However, why garnet, except for the interstitial garnet (G2), typically occurs in the outer parts of the individual (sub)units (Fig. 1), remains unclear.

Relatively high concentrations of H2O in garnet are comparable with the Rutherford No. 2 pegmatite, Virginia and with some grossular-rich garnets from various high-temperature calc-silicate rocks e.g., skarns (Rossman and Aines 1991). It is quite surprising given the evidence for low water activities in Ruda nad Moravou pegmatite relative to the volatile-rich Rutherford No. 2 pegmatite and H2O-rich fluids producing skarns (Meinert 1992). No hydrogarnet was discovered although low temperature alterations including e.g., clinozoisite, albite, sericite and prehnite are common in the pegmatite. Because almost all H2O-enriched garnets are Ca-rich (grossular-andradite) or at least Ca-enriched (Ca-enriched almandine-spessartine) (Rossman and Aines 1991; Arredondo et al. 2001; Arredondo and Rossman 2002), the incorporation of Ca into the crystal structure of garnets seems to accelerate the incorporation of H2O disregarding its activity in the distinct geological environment—rather “anhydrous” leucotonalitic melt or H2O-rich fluid-generated skarns in the contact envelope of granite plutons. However, the sum of all volatile components (H2O + F) increases about twice from the outer (sub)unit to the centre and implies an increase of their activity throughout of primary crystallization as is typically found in granitic pegmatites (e.g., London 2008).

Partitioning of trace elements into garnet and their concentrations are largely constrained by the crystal structure. High Fe3+ contents in garnets correlate positively with Mn, Mg, Ti, Ni, V, Y, H2O and in part also with Th and U (Fig. 5e, g). Low to negligible amounts of Fe3+ correlate fairly with the concentrations of F, Pb and Ta. Beryllium is the only element showing an increase of several orders of magnitude from the outer grossular subunit to the central quartz core (Fig. 5f). Consequently, the crystal structural constraints seem to evidently dominate over the geochemical ones. This is also supported by the low content of Cr in garnet relative to the associated diopside and dissakisite from the contaminated subunits.