Berger, 1999 estructuras y reologia en migmatitas con cordierita Alemania

Berger, 1999 estructuras y reologia en migmatitas con cordierita Alemania

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: JOURNAL OF PETROLOGY VOLUME 40 NUMBER 11 PAGES 1699–1719 1999 Structures and Melt Fractions as Indicators of Rheology in Cordierite-Bearing Migmatites of the Bayerische Wald (Variscan Belt, Germany) ALFONS BERGER∗ AND ANGELIKA KALT MINERALOGISCHES INSTITUT, UNIVERSITY OF HEIDELBERG, IM NEUENHEIMER FELD 236, D-69120 HEIDELBERG, GERMANY RECEIVED OCTOBER 1, 1998; REVISED TYPESCRIPT ACCEPTED MAY 21, 1999 The relation between melt volume fraction, melt segregation and deformation mechanisms of minerals in migmatites, and the controlling effects of all these factors on the bulk mechanical behaviour, were investigated using rocks from the Bayerische Wald (Bohemian Massif, Variscan belt, Germany). Biotite dehydration melting at 800–850°C and 0·5–0·7 GPa was the migmatite-forming process in this area. Four migmatite types were distinguished that alternate on the scale of several decimetres to several tens of metres. Type MIG1 is massive and undeformed. Types MIG2 and MIG3 are both stromatic (leucosome–mesosome interlayering), but differ in the degree of deformation. Type MIG4 has an interlayering of melanosome and leucosome and is strongly deformed. Melt volume fractions found by volume estimation of pure-melt leucosomes on outcrop, hand-specimen and thin-section scales are 20–40 vol. % and coincide fairly well with melt volume fractions produced in dehydration melting experiments with similar bulk compositions at the relevant pressure and temperature conditions (20–30 vol. %). The degree of melt removal from mesosomes and melanosomes (melt segregation) increases in the order MIG1–MIG3–MIG2–MIG4 and is controlled by melt volume fraction and by strain partitioning between the different types as a result of strength contrasts. The degree of melt segregation controls the formation of microstructures in the four migmatite types. In migmatites with no or only little melt segregation (MIG1 and MIG3), cordierite microstructures are indicative of growth within interconnected layers of melt. In migmatites with considerable melt segregation (MIG2 and MIG4), cordierite shows evidence of intracrystalline plasticity, indicating deformation within a load-bearing framework of minerals. Biotite microstructures in all migmatite types indicate passive rotation within ∗Corresponding author. Present address: Mineralogisch-petrographisches Institut, University of Bern, Baltzerstr. 1, CH-3012 Bern, Switzerland. Fax: ++41 31 6314843. e-mail: berger@mpi.unibe.ch a melt, but the degree of their shape-preferred orientation increases with increasing melt segregation and differential stress. The microstructures suggest that deformation mechanisms and hence the bulk mechanical behaviour of migmatites change in space and time during partial melting. The observed complex interplay of melt volume fraction, melt segregation, bulk strength contrasts, mechanical properties of different minerals and time cannot be described by a single flow law. Detailed mapping of migmatite areas, along with microstructural observations, deformation experiments considering heterogeneous melt distribution and numerical models that integrate different flow laws for various stages of partial melting, may serve to derive quantitative models for the bulk mechanical behaviour of crustal sections in the future. KEY WORDS: deformation mechanisms; melt fractions; microstructures; migmatites INTRODUCTION The mechanical behaviour of migmatites plays a major role during deformation of the crust in the course of continental collision and needs to be considered as a key factor when modelling the tectonic evolution of orogenic belts. Several migmatite areas show convincing evidence © Oxford University Press 1999 JOURNAL OF PETROLOGY VOLUME 40 for dehydration melting being the dominant migmatiteforming process (e.g. unequivocal reaction textures, pressure and temperature estimates above the dry solidus for the rocks concerned, igneous textures and compositions of leucosomes). Whereas the mechanical properties of pure solids on the one hand and pure melts on the other are rather well known, the mechanical behaviour of partially molten systems as mixtures of the two is only poorly understood. Experimental studies (e.g. Arzi, 1978; Van der Molen & Paterson, 1979; Dell’Angelo & Tullis, 1988; Rutter & Neumann, 1995) and theoretical considerations (Vigneresse et al., 1996) have been conducted to describe and quantify the mechanical behaviour of partially molten systems. Early experimental work stressed the key role of melt fraction. Arzi (1978) established the concept of the rheologically critical melt percentage (RCMP). When the RCMP was overstepped, a drastic drop in bulk strength was observed. More sophisticated theoretical considerations based on percolation theory were presented by Vigneresse et al. (1996). These workers distinguished melting systems (e.g. migmatites) from crystallizing systems (e.g. granites). Their considerations resulted in two critical melt percentage values for migmatites, one of them, the melt escape threshold (MET), playing a similar role to that of the RCMP in older concepts. The mechanical behaviour of partially molten rocks must depend on the mechanical properties of melt and minerals, on their volume proportions and on their interaction. Under experimental conditions, strain rates are very high, leading to fracturing of the minerals in the presence of melt (e.g. Rutter & Neumann, 1995). In natural migmatites, deformation mechanisms may be entirely different. Strain rates are substantially lower, favouring processes such as diffusion or dislocation creep, which are probably suppressed in experiments (e.g. Paterson, 1987). The lack of data on the deformation mechanisms, the mechanical behaviour of minerals and their interaction with melt in partially molten rocks implies that experimental or theoretical concepts cannot readily be applied to natural migmatites. In the latter, the experimentally or theoretically predicted drop in bulk strength (Arzi, 1978; Van der Molen & Paterson, 1979) need not necessarily occur at the RCMP or MET or at any other fixed melt volume fraction (Rutter & Neumann, 1995). Moreover, in migmatites, the magnitude of the strength drop at any threshold value may be lower (Barboza & Bergantz, 1998). Therefore, data on the mechanical behaviour of minerals during partial melting and deformation of natural migmatites need to be gained. This is particularly important for migmatites with low melt volume fractions, where the minerals may control the bulk mechanical behaviour. NUMBER 11 NOVEMBER 1999 The structural evolution of rocks deforming during partial melting can be described by assuming migmatites to be a two-phase system consisting of mechanically weak melt and mechanically strong minerals. Handy (1994) discussed solid-state, viscous two-phase systems and the microstructures controlling their bulk mechanical behaviour. He distinguished between a load-bearing framework (LBF) in which the weak phase occupies pockets between grains of the strong phase and a structure of interconnected weak layers (IWL). Handy (1990, 1994) concluded that coalescence of the weak phase (melt in migmatites) develops as a function of melt volume fraction, but is also dependent on stress distribution and viscous strength contrasts between the two phases. Whereas viscosities of pure melts and glasses have been determined by various methods (e.g. Dingwell et al., 1996; Johannes & Holtz, 1996), the viscosities of minerals during partial melting are nearly unknown at natural conditions. This in turn is due to the lack of knowledge on the deformation mechanisms of minerals (e.g. fracturing, dislocation creep, granular flow) in the presence of melt. This paper presents results of a study on the mechanical behaviour of a natural migmatite in the context of the theoretical and experimental concepts outlined above. The aims are (1) to gain information on the mechanical behaviour of minerals present during partial melting, (2) to quantify melt volume fractions and constrain melt segregation and (3) to discuss the effects of these factors on the bulk mechanical behaviour. The Bayerische Wald (Bohemian Massif, Variscan belt) is suited for this investigation for several reasons: (1) migmatites are exposed within a large area, implying that the results will be representative of an entire crustal level rather than representing local phenomena. (2) Reaction textures, the results of thermobarometry and inferences about fluids (Kalt et al., 1999) indicate biotite dehydration melting in the absence of an aqueous fluid to be the dominant migmatite-forming process. Therefore, melt fractions can be estimated by comparison with dehydration melting experiments. (3) Detailed determination of melt-forming and other reactions (Kalt et al., 1999) allows for the distinction of minerals occurring with melt from those formed earlier or later, which is a prerequisite for studying the mechanical behaviour of solid phases during partial melting. (4) Different migmatites in terms of structure, bulk composition, mode and melt fraction occur in rather restricted areas, which allows a variety of structures to be observed and interpreted. (5) The migmatites of the Bayerische Wald are largely devoid of post-migmatite overprint. In this paper, we describe the mechanical behaviour of minerals during partial melting and deformation by using their shape-preferred orientation (SPO; biotite, garnet) and their crystallographic preferred orientation 1700 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD (CPO; cordierite) and by interpreting microstructures in leucosome, mesosome and melanosome. Melt volume fractions are estimated by comparison with those obtained during dehydration melting experiments in compositionally similar systems at the pressure–temperature (P–T ) conditions of interest, by leucosome volume fractions and by modal constraints. Then, the effects of the mechanical behaviour of the solid phases, of melt volume fraction and of melt segregation on the bulk mechanical behaviour of the migmatite are discussed. METHODS OF STUDY Modes of bulk migmatite, leucosome, mesosome and melanosome were derived from point counting of representative migmatite thin sections. In small leucosomes, modes were determined by image analysis using the public domain NIH-Image program developed at the US National Institutes of Health. Quantitative microstructures (SPO and CPO) were used to infer the deformation mechanisms of minerals during partial melting in migmatites of the Bayerische Wald. Cordierite, biotite and garnet were chosen for these investigations as they occurred with melt (Kalt et al., 1999). The CPO is an indicator for intracrystalline plasticity of minerals (dislocation creep and glide). Intracrystalline plasticity is only displayed by grains deforming within a framework of solid particles, but is not shown by grains that can freely rotate in a weak matrix. The degree of CPO is a measure of the intensity of deformation. The SPO of platy minerals such as biotite can serve to identify grains that were passively rotated in a weak matrix as a result of differential stress (e.g. Jeffrey, 1922; March, 1932). The degree of SPO of minerals in general increases with increasing differential stress. The scale of our microstructural observations is comparatively small (thin-section scale). However, microstructures were studied in several thin sections of a sample, and in several samples of a migmatite type. Therefore, although the observations were made on a small scale, they are representative of larger areas (several decimetres to several tens of metres). The CPO of cordierite was measured with a U-stage. Two of the three indicatrix axes were measured and the third was calculated. For details of the techniques, the reader is referred to Reinhard (1931) and Turner & Weiss (1963). The crystallographic directions [100], [010] and [001] were derived directly from the indicatrix directions n , n and n , respectively (Deer et al., 1992). Statistical factors were used to quantify the degree of CPO, based on statistical methods on spheres (Woodcock, 1977; Watson, 1983). Garnet shapes were measured with particle projection techniques using PAROR (Panozzo, 1983). Biotite shape orientation was investigated by measuring the surface orientation of the crystals. The procedure follows the techniques and programs presented by Panozzo-Heilbronner & Pauli (1994). Projections of particles and surfaces were used to quantify the distributions of orientations, presented as rose diagrams. Additionally, A( )min values were calculated and used to characterize the shape anisotropy of biotite grains according to the procedure described by Panozzo (1983, 1984; see also caption to Fig. 5, below). Melt volume fractions were constrained in two ways. (1) Melt volume fractions obtained during dehydration melting experiments should be representative of those expected in nature, provided that experimental conditions approximate the physical conditions (temperature and pressure) and chemical conditions (bulk composition) of melting in natural migmatites. Therefore, melt volume fractions obtained in biotite dehydration experiments at the appropriate P–T conditions were compiled and used to infer possible melt volume fractions in the migmatites of the Bayerische Wald. (2) Volume fractions of leucosome estimated for migmatites of the Bayerische Wald were used to indicate melt volume fractions. Various attempts have been made to estimate melt volume fractions in natural migmatites from leucosome percentages (e.g. Olson, 1984; Nyman et al., 1995). To translate leucosome volume fractions to melt volume fractions, it must be clear that the leucosomes concerned represent pure former melt. Moreover, leucosome volume fractions may be smaller than the original melt volume fractions as they do not include residual melt that remained in mesosomes and melanosomes. Hence, leucosome volume fractions give only crude approximations of the volume percentage of melt. As a result of these difficulties, additional estimates of melt volume fractions have been made in some studies, using the abundances of minerals produced by balanced meltforming reactions (e.g. muscovite + quartz = K-feldspar + sillimanite + melt; Nyman et al., 1995). However, this approach ignores the possibility that different meltforming and solid–solid reactions involving the same minerals may operate simultaneously in different chemical domains of a rock. Apart from the problems outlined above, the estimation of melt volume fractions from leucosome volume fractions has two other handicaps. (1) Given that the volume of the bulk (precursor) system and the extent of melt movement is unknown, it is not clear on which scale leucosome volume fractions represent the actual former melt volume fractions. (2) The accuracy and statistical significance of the estimates are scale dependent. Larger leucosomes can be seen in outcrops and hand specimens, but smaller leucosomes can only be detected in thin 1701 JOURNAL OF PETROLOGY VOLUME 40 sections. Whereas the investigation of thin sections has the advantage of high resolution (each grain can be attributed to either leucosome or mesosome– melanosome), the studied area is small and thus not necessarily representative. On hand specimen and outcrop scales, resolution is lower, but statistical significance is greater. To overcome these problems, leucosome volume fractions were estimated on three different scales: outcrop scale (by using photographs), hand-specimen scale, and thin-section scale (using a petrographic microscope and a screen, magnification 14·8×). Line drawings of the migmatites on all three scales were produced and the volume fractions of leucosome were calculated using image analysis. GEOLOGICAL SETTING AND P–T EVOLUTION OF THE MIGMATITES The Bayerische Wald is located at the southwestern margin of the Bohemian Massif within the internal (Moldanubian) zone of the Variscan belt (Fig. 1). It comprises a variety of gneisses and migmatites that are intruded by granites. In its central part, the Bayerische Wald is cut by a late, narrow NW–SE trending shear zone (Pfahl zone, Fig. 1). Metamorphism is of the hightemperature low-pressure (HT–LP) type. The metamorphic peak has been dated at 318–322 Ma by concordant U–Pb ages of monazite grain-size fractions (Grauert et al., 1974), whereas single-grain U–Pb monazite dating indicates slightly older ages, namely, 323–326 Ma (Kalt et al., submitted). Regional geology, field relationships and metamorphic evolution of the area have been described by Kalt et al. (1999). The rocks followed a clockwise P–T path, biotite dehydration melting in the absence of an aqueous fluid being a dominant feature of the prograde P–T path. The dehydration melting reactions changed with bulk composition on small scales and produced locally varying modal proportions of cordierite, garnet, spinel, orthopyroxene and melt. Minimum and maximum estimates of peak temperatures (800–850°C and 900°C, respectively) and pressure constraints (0·5–0·7 GPa) emerge from comparison with biotite dehydration experiments in compositionally similar systems [see compilation by Kalt et al. (1999)]. However, peak metamorphic conditions cannot be retrieved from thermobarometry. Na-in-cordierite, garnet– cordierite and garnet–orthopyroxene equilibria indicate re-equilibration on the retrograde part of the P–T path at 770–846°C and 0·44–0·51 GPa. MIGMATITE TYPES On the basis of geometric relationships, modes and microstructures of mesosomes and melanosomes, four NUMBER 11 NOVEMBER 1999 migmatite types (MIG1–MIG4) were distinguished (Fig. 2; Kalt et al., 1999). These types are ‘endmembers’, meaning that intermediate migmatites may also be found. The main structural characteristics of the migmatite types are given in Table 1. Type MIG1 migmatites are massive (Fig. 2a) and display no clear planar structures. Foliation or layering is hardly recognizable or absent in these rocks and a stretching lineation is not present. Leucosomes and mesosomes are only recognizable as centimetre-sized patchy light and dark areas, respectively. Hence, it is questionable whether or not type MIG1 can be termed migmatite in the sense of Ashworth (1985). However, Kalt et al. (1999) showed that all rock types (MIG1–MIG4) underwent partial melting at the same P–T conditions at the same time independent of their final structure. Therefore, a common term for MIG1–MIG4 rocks, namely migmatite, is used for the sake of clarity, although it may not be entirely correct for type MIG1. Type MIG2 migmatites are stromatic and display a weak foliation (Fig. 2b). They are characterized by a leucosome–mesosome interlayering with sharp margins, producing a clear planar structure. Leucosomes and mesosomes range from several centimetres to several decimetres in thickness. The leucosomes are commonly plagioclase bearing (in addition to quartz and K-feldspar). Type MIG3 migmatites are also stromatic, but, in contrast to type MIG2, show a well-developed foliation parallel to the compositional layering, locally a weak stretching lineation and no sharp leucosome–mesosome margins (Fig. 2c). MIG3 leucosomes are either plagioclase bearing or of the K-feldspar- and quartz-dominated type. Type MIG4 migmatites are also stromatic, with an interlayering of sharply bound melanosome and leucosome, and are characterized by a pronounced foliation and stretching lineation (Fig. 2d). Leucosomes and melanosomes range from several centimetres to several decimetres in thickness. The leucosomes are of the K-feldspar- and quartzdominated type. MIG1–MIG4 migmatites may alternate on a scale of decimetres to tens of metres. In places, an outcrop of several metres or tens of metres may consist of a single migmatite type, whereas in other cases, a migmatite type is restricted to portions of an outcrop that are only several decimetres in size [fig. 4 in Kalt et al. (1999)]. Sharp boundaries between different types can be observed, but grading of one migmatite type into another can also be found. MIG3 migmatites are the most abundant. They often contain boudins of calc-silicate rocks and gneisses. At the margins of these boudins, MIG3 migmatites usually grade into MIG2 types (Fig. 3). MIG1 migmatites are comparatively rare and form isolated parts within the stromatic migmatite types. MIG4 migmatites are commonly observed as boudins or bands within MIG2 and 1702 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD Fig. 1. Simplified geological map of the Bayerische Wald [modified from Kalt et al. (1999)]. The inset shows the outlines of Variscan basement in Europe [modified after Franke (1989)]. MO, Moldanubian zone; ST, Saxothuringian zone; RH, Rhenohercynian zone; BM, Bohemian Massif. The area displayed in the detail map is marked in black. MIG3 migmatites. Petrographic features, reaction textures and mineral compositions of the four migmatite types have been described by Kalt et al. (1999). MESOSOME AND MELANOSOME MODES Mesosome, melanosome and leucosome modes are given in Table 2. Whole-rock modes are given for MIG1 migmatites, because leucosomes and mesosomes are recognizable only as centimetre-sized patchy light and dark areas, respectively. The lack of leucosomes on outcrop and hand-specimen scale indicates that melt was hardly segregated. Hence, MIG1 migmatites are not depleted in quartz and feldspars. MIG2 mesosomes are strongly depleted in quartz and feldspar. This, along with the presence of sharply bound leucosomes, points to considerable removal of melt from mesosomes and crystallization of melt in leucosomes. In contrast to MIG2 migmatites, MIG3 mesosomes are not considerably depleted in feldspar or quartz (Table 2). The lack of considerable depletion, along with the presence of leucosomes that have no sharp margins, suggests that melt was removed from mesosome and crystallized in leucosome sites, but that some melt stayed in mesosome sites and crystallized there. MIG4 melanosomes have >50 vol. % cordierite and garnet, and are very poor in quartz and feldspars (Table 2). These inferred restitic melanosomes, together with the sharply bound leucosomes, indicate almost complete removal of melt from its site of production. 1703 JOURNAL OF PETROLOGY VOLUME 40 NUMBER 11 NOVEMBER 1999 Table 1: Structural characteristics of the four migmatite types Rock type MIG1 Planar fabrics MIG2 MIG3 MIG4 no or only very weak compositional layering of compositional layering of compositional layering of foliation mesosomes and mesosomes and melanosomes and leucosomes leucosomes and foliation leucosomes and pronounced foliation Stretching lineation none none weak Leucosomes only patchy distribution mainly Pl-bearing Qtz–Kfs leucosomes always present of light and dark areas leucosomes CPO of Crd not present intermediate weak strong SPO of Bt weak intermediate strong strong (Grt + Crd)/(Fsp + Qtz)∗ 0·4 2·3 0·3 3·6–4·2 mainly Qtz–Kfs leucosomes Mineral abbreviations according to Kretz (1983); CPO, crystallographic preferred orientation; SPO, shape-preferred orientation; MIG1–MIG4, migmatite types. (For further explanations, see section on migmatite types.) ∗Modal proportions of the minerals in the mesosomes and melanosomes. MESOSOME AND MELANOSOME MICROSTRUCTURES Massive migmatites (MIG1) Apart from their common massive structure, individual microstructures of MIG1 migmatites differ. Whereas some rocks are almost homogeneous, others show a patchy distribution of centimetre-sized dark and light areas. The light areas commonly contain quartz and feldspar, including some larger feldspar crystals of several millimetres in size. The dark areas are dominated by random biotite grains. Other samples show arcuate trails of biotite around clusters of cordierite, feldspars and quartz. Independent of the bulk structures, cordierite always forms large euhedral to subhedral grains (~1 mm in size) that are generally poor in inclusions and have neither an optically observable SPO nor a measurable CPO (Table 1, Fig. 4). Individual biotite grains show high short/long axial ratios, resulting in rather blocky shapes. They either have a very weak SPO or lack it (Table 1, Fig. 5). The rare garnet grains in MIG1 migmatites have a distinct, but very weak elongation (Table 1, Fig. 6). Stromatic migmatites (MIG2) In the mesosomes, cordierite grains, together with some large feldspar grains, are surrounded by biotite aggregates. In some rocks, biotite laths wrap around large cordierite and feldspar grains. Whereas the biotite aggregates show a common preferred orientation, individual biotite grains show only a weak SPO. On average, biotite in MIG2 mesosomes has an intermediate SPO (Table 1, Fig. 5). Cordierite grains are large and have an intermediate CPO (Table 1, Fig. 4). Cordierite locally has deformation twins. Garnet grains in MIG2 leucosomes are of nearly equant shape, with only a weak or no SPO [see A( )min in Fig. 6]. Foliated stromatic migmatites (MIG3) Cordierite grains are commonly large with elongated shapes and are surrounded by smaller quartz, feldspar and biotite grains. Biotite occurs as small inclusions in cordierite and K-feldspar, but also as aggregates aligned parallel to the foliation. In contrast to those in MIG1 and MIG2 migmatites, the biotite (Table 1, Fig. 5) and cordierite grains show a well-developed SPO parallel to the foliation and compositional layering. The microstructures and CPOs of cordierite in MIG3 migmatites are complicated. In most samples, individual cordierite grains show a strong SPO parallel to the foliation, but no twinning or recrystallization. Cordierite grains in samples 7A45 and 6A32 indicate a weak preferred orientation of [010] and a very weak preferred orientation of the other two crystallographic directions, whereas cordierite in sample 7A76 shows a more pronounced CPO (Fig. 4). Melanosome-bearing migmatites (MIG4) The few quartz grains observed in MIG4 melanosomes exhibit irregular grain boundaries with plagioclase and 1704 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD Fig. 2. Schematic line drawings of the migmatite types on thin-section scale. (a) MIG1 whole rock; random and disseminated cordierite grains in an undeformed matrix of quartz, feldspar and biotite. Some biotite flakes may display a weak foliation. (b) MIG2 migmatite; stromatic migmatite with mesosome–leucosome layering. The mesosome is depleted in quartz and feldspar. Biotite and cordierite in mesosome define a weak foliation. (c) MIG3 migmatite; mesosomes are not depleted in quartz and feldspar; cordierite and biotite define a strong foliation. In the lower part of the drawing a leucosome is shown. (d) MIG4 migmatite; melanosomes with strongly elongated garnet grains; cordierite grains may display core–mantle structures (lower part of the figure); the matrix is dominated by recrystallized cordierite grains. Some pockets of quartz–feldspar aggregates may occur in the melanosomes. always occur as isolated, single grains with rather equant shapes. Internal deformation structures are absent. Plagioclase grains occur in millimetre-sized aggregates between the cordierite-dominated framework. They always have lobate grain boundaries against cordierite. Biotite commonly occurs as inclusions in cordierite or within cordierite-dominated polycrystalline aggregates (see below). Cordierite in MIG4 migmatites is dominated by dynamic recrystallization features (Fig. 2d). Recrystallization produced new grains averaging ~150 m, with straight boundaries. Subgrains occur around old clasts, locally forming core–mantle structures. In some areas, cordierite grains are completely recrystallized, the new grains having equant shapes and uniform grain size. As a result of recrystallization, the formerly included minerals (e.g. sillimanite, spinel) are now located at grain boundaries (Fig. 7). Deformation twinning of cordierite can be observed in a few samples. The deformation twins have a constant orientation with respect to the foliation and stretching lineation. A strong CPO was measured in cordierite grains of MIG4 migmatites (Fig. 4). This CPO is characterized by an orientation of [010] near the pole of the foliation, and [100] near the stretching lineation (Fig. 4). The observed CPO is similar to other reported CPOs of cordierite (Fischer, 1938; Fediuk, 1974; Beer, 1981; Van Roermund & Konert, 1990; Kruhl & Huntemann, 1991). 1705 JOURNAL OF PETROLOGY VOLUME 40 NUMBER 11 NOVEMBER 1999 Fig. 3. Outcrop photograph of a non-migmatitic gneiss boudin within stromatic migmatites. The well-developed planar structure of the leucosome–mesosome layering in the MIG2 migmatite at the boudin margin should be noted. MIG2 mesosomes are indicated by arrows. Massive MIG1 migmatites occur in the same outcrop (see upper right-hand corner). The coin in the lower part of the figure is 2 cm in diameter. Table 2: Modes of the migmatites Sample: BW26 7A51 7A68 BW28 BW28 7A65 7A45 6A32 BW68 BW68 6A31 Type: MIG1 MIG1 MIG1 MIG2 MIG2 MIG3 MIG3 MIG3 MIG4 MIG4 MIG4 6A31 MIG4 WR WR DC MS L MS MS MS ML L ML L Qtz 23·8 15·7 10·8 0·8 24·4 19·2 19·4 21·2 1·03 42·5 — 40·3 Kfs n.d. n.d. n.d. n.d. 33·8 n.d. n.d. n.d. n.d. 57·5 n.d. 54·6 Pl n.d. n.d. n.d. n.d. 30·7 n.d. n.d. n.d. n.d. n.d. n.d. 1·9 Fsp (tot) 39·0 57·5 43·0 17·9 35·5 35·7 35·7 14·8 Bt 14·1 13·5 15·9 22·7 33·6 5·6 23·1 17·7 16·6 4·5 — 21·8 2·1 Grt — — 2·3 0·2 5·5 — 0·1 5·0 1·8 — 2·6 0·2 Crd 22·1 10·8 21·2 42·7 — 21·6 13·8 20·7 55·1 — 56·4 — Ilm 1·1 0·2 0·1 2·5 — — — 0·8 3·0 — 4·1 — Sil 0·5 — — 2·1 — — 10·2 — 19·7 — 1·0 — 100·0 100·0 100·0 100·0 99·9 99·4 96·9 100·0 100·0 100·0 100·0 99·2 P P P P P P P P P I P I Total Method Mineral abbreviations according to Kretz (1983); MIG1–MIG4, migmatite types; WR, whole rock; MS, mesosome; ML, melanosome; L, leucosome; DC, dark area; —, not present; n.d., not determined; P, point counting; I, image analysis. These types of CPO fit very well with slip systems identified by transmission electron microscopy (Van Roermund & Konert, 1990; Skrotzki & Siegesmund, 1993). In these studies, (010) [001] is the main slip system and the CPO measured here and in the cited examples reflects an orientation of easy glide for this slip system. Biotite in MIG4 melanosomes shows a strong SPO (Table 1, Fig. 5) and garnet displays very elongate shapes subparallel to foliation (Fig. 6). IMPLICATIONS OF THE MICROSTRUCTURES IN MESOSOMES AND MELANOSOMES Cordierite In MIG1 migmatites, the presence of large subhedral to euhedral cordierite grains and the absence of cordierite 1706 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD Fig. 4. Crystallographic preferred orientation (CPO) of cordierite grains. Scatter diagrams display [100] (×), [010] (•) and [001] (Α) for each sample. For each migmatite type contoured diagrams of the [100] direction are displayed on the right-hand side, contoured at 1·0, 2·0, ... times uniform. Diagrams are oriented with the lineation (where present) in an E–W direction and the pole of foliation (where present) in the north. (a–c) MIG1 migmatites; CPO is absent for all crystallographic directions. (d) MIG2 migmatite; an intermediate state between random cordierite grains as in MIG1 migmatites and a pronounced CPO as in MIG4 melanosomes. (e–g) MIG3 migmatites; in some samples a weak preferred orientation of [010] direction can be observed, whereas other orientations show a nearly random distribution. (h, i) MIG4 migmatites; strong preferred orientation of all crystallographic directions. CPO and SPO indicate cordierite growth in the presence of a weak matrix and in the absence of differential stress. In MIG3 mesosomes, the shape anisotropy of cordierite indicates differential stress, but the comparatively weak 1707 JOURNAL OF PETROLOGY VOLUME 40 NUMBER 11 NOVEMBER 1999 Fig. 6. Shape-preferred orientation (SPO) of garnet grains in MIG1 migmatite, MIG2 leucosome and MIG4 melanosome presented as rose diagrams (planar structure is oriented horizontally). A( )min values of the particle projection are noted in the lower-right hand corner of the diagrams. [Note the strong preferred orientation of the MIG4 garnets and their low A( )min value, consistent with their flattened shape. For calculation of A( )min values, see caption to Fig. 5.] Fig. 5. Shape-preferred orientation (SPO) of biotite grains displayed as rose diagrams (foliation is oriented horizontally). A( )min values of the surface projection are noted in the lower right-hand corner of the diagrams. A( )min values are calculated from projection lengths of mineral surfaces in thin sections obtained at observation angles between 0 and 180°. The largest observed projection length is set to a value of unity [A( )max]. A( )min values vary betwen zero and unity. They approximate unity in thin sections with no preferred orientation of mineral surfaces and tend to values near zero for thin sections where a very pronounced preferred orientation of mineral surfaces is observed. (Note the weak SPO of biotite in MIG1 migmatites, consistent with the absent or weak macroscopic foliation, and the strong SPO in MIG3 and MIG4 migmatites, consistent with the well-developed mesoscopic planar structure.) CPO, along with the lack of dynamic recrystallization features, excludes dislocation creep or glide as an important deformation mechanism for MIG3 migmatites. Therefore, the SPO of cordierite is probably due to passive rotation in a weak matrix. In MIG2 migmatites, the intermediate CPO of cordierite, along with the presence of deformation twins, indicates a weak contribution of dislocation creep to the deformation. In MIG4 migmatites, the shapes of the recrystallized cordierite grains and the observed subgrains point to subgrain rotation recrystallization. The presence of subgrains, dynamic recrystallization and the measured CPO indicate dislocation creep as a main deformation mechanism for cordierite in MIG4 migmatites. Therefore, MIG2 and MIG4 cordierites must have deformed by intracrystalline plasticity in a framework of minerals. As the inference of intracrystalline plasticity in the presence of considerable melt volume fractions is very important, it must be ensured that the observed microstructures developed in the presence of melt. The possibility that microstructures developed after partial melting can be excluded for several reasons: (1) only cordierite grains that were produced during partial melting (see Kalt et al., 1999) were used for the CPO investigation, whereas cordierite grains formed by subsolidus reactions after melt crystallization (see Kalt et al., 1999) were excluded. (2) The compositions of recrystallized and old cordierite grains are exactly the same. Average Na contents and Fe/Mg ratios in different recrystallized cordierite grains 1708 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD Fig. 7. Microphotograph of dynamically recrystallized cordierite grains in a MIG4 migmatite under crossed polars. In the central and upper parts of the figure, inclusions of spinel and sillimanite are present. The arrow indicates a low-angle grain boundary. All cordierite grains in the figure have a similar CPO (compare Fig. 5). Sil, Sillimanite; Spl, spinel; Pl, plagioclase; Crd, cordierite. indicate high equilibration temperatures of ~800°C (see Kalt et al., 1999). (3) The leucosomes of MIG4 migmatites consist mainly of quartz and K-feldspar. Solid-state deformation after crystallization would produce deformation microstructures in quartz, but the leucosomes are undeformed. (4) In MIG4 migmatites, the minerals stable before partial melting reactions (biotite and sillimanite) show a distinct preferred orientation, even if they occur as inclusions in cordierite or garnet. The SPO of biotite and sillimanite must thus have been produced before the overgrowth of cordierite, which shows the same preferred orientation as the included minerals. Hence, microstructural and petrological evidence indicates that the cordierite microstructures developed in the presence of melt. Garnet All garnet grains are solid products of partial melting reactions and show simple chemical zoning patterns, except for very large garnet grains that may have complex zoning patterns with the inner cores recording a metamorphic stage before partial melting (Kalt et al., 1999). Independent of size and composition, garnet grains show different shapes related to their structural position in various migmatites. Garnet grains in MIG2 leucosomes and MIG1 migmatites show nearly equant shapes with a weak or no SPO (Fig. 6). This suggests that the garnets grew within a weak matrix. Garnet grains in MIG4 melanosomes display very elongate shapes subparallel to foliation. The physical reason for the elongate garnet shapes in MIG4 melanosomes may be intracrystalline plasticity or a process leading to incongruent pressure solution. The aim of this study is not to investigate the deformation mechanisms of garnet; for discussion of garnet deformation the reader is referred to Ji & Martignole (1994), Kretz (1994) and Den Brok & Kruhl (1996). However, the flat garnet grains oriented parallel to foliation in melt-depleted MIG4 melanosomes are inferred by us to result from deformation in a solid framework of minerals. Biotite Biotite grains in migmatites can be either relics of biotite dehydration melting or newly grown during crystallization of melt. Mineral chemistry (e.g. TiO2 contents) and the results of thermometry suggest that biotite-out temperatures were not attained in the Bayerische Wald (Kalt et al., 1999). However, microstructures and petrological considerations indicate partial biotite breakdown in the course of melt-producing reactions. The overall melting reaction biotite + sillimanite ± quartz ± plagioclase = cordierite/garnet/orthopyroxene/spinel + melt ± K-feldspar (Kalt et al., 1999) is at least divariant and thus involves biotite on the reactant and product sides, related mainly by Fe–Mg and Ti net transfer. Crystallization of new biotite from the melt is probably a minor process because of the low Fe–Mg contents of the melt (see leucosome compositions in Table 2). Therefore, most of the investigated biotite grains must 1709 JOURNAL OF PETROLOGY VOLUME 40 NUMBER 11 NOVEMBER 1999 have been stable and acquired their SPOs before and during partial melting. SPO of biotite in MIG1 migmatites is weak or absent (Fig. 5), indicating the absence of differential stress during partial melting. In MIG2 mesosomes biotite shows an intermediate SPO whereas in MIG3 mesosomes and in MIG4 melanosomes the SPO is well developed. As the SPO of biotite indicates stress-induced passive rotation in a weak matrix, this process must have been more effective in MIG3 and MIG4 migmatites than in MIG2 migmatites. LEUCOSOME MODES AND MICROSTRUCTURES Concordant, stromatic leucosomes as well as discordant, irregular leucosomes can occur in the same outcrop. The width of the leucosomes varies between several millimetres and several decimetres. Microstructures and modes of leucosomes differ. Leucosomes dominated by K-feldspar and quartz can be distinguished from those additionally containing considerable amounts of plagioclase. K-feldspar- and quartz-dominated leucosomes are either completely or nearly free of plagioclase (Table 2, Fig. 8). They consist mainly of perthitic K-feldspar and quartz, although in rare cases a few antiperthite grains have been observed. Individual perthite compositions are difficult to measure because of exsolution lamellae of varying widths (<2 m to 50 m). Several analyses in one leucosome resulted in an average composition of Or81·4Ab18·1An0·4. In plagioclase-bearing leucosomes, the plagioclase grains are unzoned, antiperthite is not observed and Fe–Mg minerals (e.g. garnet or cordierite) may occur. Bulk chemical compositions of leucosomes were calculated from modes and feldspar compositions. K-feldspar- and quartz-dominated leucosomes as well as plagioclase-bearing leucosomes have compositions similar to those of melts produced in dehydration experiments in psammitic to pelitic systems (Fig. 8). These melts are silica rich, peraluminous, poor in Fe and Mg, and dominated by K2O and Na2O over CaO. The experimental melts and the plagioclase-bearing leucosomes are more enriched in Na2O than are the K-feldspar- and quartz-dominated leucosomes (Fig. 8). This indicates that in contrast to the experimental melts and the plagioclasebearing leucosomes, the melts crystallizing as K-feldsparand quartz-dominated leucosomes were not buffered by plagioclase. They must have been either generated from plagioclase-free parts of the migmatite protolith or removed from plagioclase-bearing sources. Apart from the deviation in composition, the two types of leucosomes differ in their microstructures. No preferred mineral orientation can be detected in the K-feldspar- and Fig. 8. Qtz–Ab–Or triangular plot. Normative glass compositions produced in biotite dehydration experiments by Montel & Vielzeuf (1997), Patino Douce & Beard (1995, 1996) at 0·5 GPa, and by Patino ˜ ˜ Douce & Johnston (1991) at 0·7 GPa are shown. Η, Experiments at 850°C; •, experiments up to 900°C. Additionally, the cotectic line in the haplogranitic system at aH2O ~ 0·4 at 0·5 GPa is shown [Holtz et al. (1992); experimental data at lower water activities at 0·5 GPa are not available]. Β, quartz–K-feldspar leucosomes of the Bayerische Wald; Ε, plagioclase-bearing leucosomes. quartz-dominated leucosomes, supporting the contention that the latter were probably completely molten during emplacement. In contrast, the large, unzoned, euhedral plagioclase grains in the plagioclase-bearing leucosomes exhibit a well-developed SPO (Fig. 9) subparallel to the leucosome margins. The SPO of those grains, along with the lack of solid-state deformation features in the small matrix grains, indicates flow of a crystal mush (e.g. Benn & Allard, 1989; Paterson et al., 1989). Therefore, plagioclase-bearing leucosomes were probably not completely molten at the time of emplacement. MELT VOLUME FRACTIONS Results of dehydration melting experiments The migmatites of the Bayerische Wald experienced minimum conditions of ~800°C and 0·5 GPa and maximum temperatures of ~900°C during fluid-absent biotite dehydration melting (Kalt et al., 1999). Melt volume fractions produced in biotite dehydration melting experiments conducted at 0·5–0·7 GPa and 800 and 900°C may hence be used to infer possible melt volume fractions in the migmatites. However, the determination of very small melt fractions at temperatures just above the solidus in the experiments (~800°C) is fraught with large uncertainties and most biotite dehydration melting experiments start at temperatures above 800°C. Therefore, 1710 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD Fig. 9. Shape-preferred orientation (SPO) of leucosome minerals in plagioclase-bearing leucosomes presented as rose diagrams of the surface orientations. (a) Small grains of plagioclase, quartz and K-feldspar show no SPO. (b) Selected large euhedral plagioclase grains display a pronounced SPO. the melt volume fractions obtained in experiments at 850 and 900°C are used for comparison here (Fig. 10), the results for 850°C being more representative for the studied samples. The melt fractions produced in the experiments are controlled by complex interactions between various chemical factors. High modal fractions of biotite (Le Breton & Thompson, 1988) and muscovite, Al2SiO5 and plagioclase (Gardien et al., 1995) are thought to increase melt fractions below 900°C considerably. Low XMg values are also considered to favour large melt fractions (Patino ˜ Douce & Beard, 1996; Stevens et al., 1997). Bulk TiO2 contents should also influence the melting behaviour, as they control biotite breakdown (e.g. Stevens et al., 1997). Mainly partitioned into biotite, high TiO2 contents stabilize the latter to higher temperatures and thus limit the melt fraction. The production of cordierite by dehydration melting reactions also seems to influence melt production. Stevens et al. (1997) found that high modal abundances of cordierite (in Mg- and Al-rich systems at low pressures) limit melt production. This is probably due to the fact that cordierite may incorporate part of the limited amount of water released during biotite breakdown, which could otherwise be partitioned into a larger fraction of melt. Figure 10 shows a compilation of melt fractions obtained in biotite dehydration melting experiments at 0·5 GPa in relation to bulk XMg and Al2O3. Although it must be stressed that the compiled experiments were performed under different physical and chemical conditions and that hence variables other than bulk XMg and Al2O3 may influence melt fractions, the latter two seem to exert a major control on partial melting. At 850°C, there is a negative correlation between XMg values and melt fraction, except for compositions A and B (Stevens et al., 1997), which produce very high melt fractions. Apart from A and B, compositions with high Al2O3 contents yield more melt at a given P–T condition than Al2O3poor compositions at any given XMg value. At 900°C, with melt fractions being generally large, the effects of XMg and Al2O3 are not clearly distinguishable. From Fig. 10, no systematic influence of bulk TiO2 contents on melt fractions can be depicted. The compilation shows that natural metapelites and metagreywackes undergoing biotite dehydration melting at 850°C and 0·5 GPa should produce between 10 and 40 vol. % melt unless they are extremely Mg rich. Bulk modes of the migmatite types from the Bayerische Wald and hence bulk chemistry of the protoliths do not deviate dramatically from each other. XMg values calculated from modes (Table 2) and from chemical analyses are ~0·4–0·5 for all migmatite types. Bulk Al2O3 contents can be estimated from modes. They are lowest in MIG1 and MIG3 migmatites, as is evident from their very low modal sillimanite and cordierite contents. MIG2 and MIG4 migmatites are characterized by high modal cordierite and fairly high modal sillimanite contents, and thus have higher bulk Al2O3 contents. A limited number of chemical analyses yields bulk Al2O3 contents between 13 and 17 wt % [the problem of finding representative sample sizes for bulk chemical analyses or determination of modes in coarse-grained and heterogeneous migmatites has been discussed by Kalt et al. (1999)]. The differences in bulk Al2O3 contents at largely identical XMg values suggest that slightly higher melt volume fractions could have been produced in MIG2 and MIG4 migmatites when compared with MIG1 and MIG3 migmatites, but all melt volume fractions should be within 20–30%. Leucosome volume fractions Leucosome volume fractions were estimated on thinsection, hand-specimen and outcrop scale. Only samples containing mainly quartz- and K-feldspar-dominated leucosomes were used, because only those were probably pure melts at the time of emplacement, as indicated by their nearly cotectic compositions and their microstructures. Samples with mainly plagioclase-dominated leucosomes were excluded from the investigations, because of the unknown percentage of solid phases at the time of emplacement. However, with estimations on outcrop scale, plagioclase-bearing leucosomes could not always be entirely excluded. MIG1 migmatites were generally excluded from the volume fraction estimates, as there is no indication of melt segregation. Moreover, MIG2 migmatites are comparatively rare. Therefore, the procedure is restricted to MIG3 and MIG4 migmatites, 1711 JOURNAL OF PETROLOGY VOLUME 40 NUMBER 11 NOVEMBER 1999 Fig. 10. Compilation of melt fractions produced during dehydration melting experiments at 0·5 GPa at 800°C and 900°C. PDJ91, Patino ˜ Douce & Johnston (1991); MV97, Montel & Vielzeuf (1997), SBG, starting composition of Patino Douce & Beard (1995); SFAG, SMAG, starting ˜ compositions of Patino Douce & Beard (1996); A, B, C, NB, AS, BS, CS, NBS, starting compositions of Stevens et al. (1997). Melt fractions are ˜ given in vol. %. For SBG, SMAG and SFAG the vol. % are recalculated from the original data given in wt % (using the following densities: 3 quartz 2·65 g/cm , plagioclase 2·66 g/cm3, biotite 3·0 g/cm3, garnet 4·11 g/cm3, orthopyroxene 3·6 g/cm3, ilmenite 4·7 g/cm3, magnetite 5·2 g/cm3 and melt 2·5 g/cm3). The dashed lines indicating melt fractions <10, 20 and 30 vol. % were graphically fitted to the data shown in the plot, excluding results with compositions A and B. (For further information see the section on melt volume fractions.) and even further to those occurrences in which variable leucosome size allows for observations on at least two scales. Leucosome volume fractions estimated for the few appropriate MIG3 and MIG4 migmatite samples on three scales (outcrop, hand-specimen, thin-section) are listed in Table 3. The obtained leucosome volume fractions range from 20 to 49%. The most reliable leucosome volume data are obtained from MIG4 migmatites (e.g. samples 6A31, BW83/86, Table 3). In these samples, any melt remaining in the melanosomes can be neglected because of the strongly restitic character of the latter, and the leucosomes were probably pure melts. Therefore, the estimated leucosome volume fractions should represent the actual former melt volume fractions within the selected areas. Table 3 shows that for each MIG4 migmatite, the leucosome values estimated on different scales coincide fairly well. On the other hand, these values vary strongly between different MIG4 occurrences. Sample 6A31 has a small leucosome volume fraction (20–23%) whereas BW83 displays the highest leucosome volume fraction (46–49%). This discrepancy is probably due to the fact that in 6A31 melt was removed from the investigated areas whereas in BW83 it still resides within them or melt was added. In MIG3 migmatites, the leucosome volume fractions estimated on different scales for any given occurrence also show a good coincidence. Leucosome volumes range from 26 to 39%, but the amount of melt possibly trapped in the mesosomes and the amount of restitic minerals in the plagioclase-bearing 1712 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD Table 3: Leucosome volume fractions calculated for selected samples Type Sample Outcrop Hand-specimen Thin section Area (m2)∗ Vol. %† Area (cm2)∗ Vol. %† Area (cm2)∗ Vol. %† MIG4 6A31 — —‡ 62 20 8·1 23 MIG4 7A65 ~1·5 30 35 27 8·0 31 MIG4 BW86/ — —‡ 42 41 69 46 8·1 49 ~0·9 33 24 39 83 MIG3 7A76/ 66 26 — —§ MIG3 66 7A45 — —‡ 57 34 — —§ MIG3 BW27 ~0·5 26 33 26 — — ∗Investigated area. †Volume percent of leucosome. ‡Outcrop scale is not available. §Thin-section scale is not available, because all leucosomes are too large. leucosomes are unknown. Therefore, estimated leucosome volume fractions for MIG3 migmatites are only a crude approximation of melt volume fractions. The rough estimates of melt volume fractions made here on the basis of leucosome volume fractions (20–40%) coincide fairly well with the brackets derived from biotite dehydration melting experiments at the relevant P–T conditions (20–30%). With both approaches, the estimated minimum melt volume fraction in MIG3 and MIG4 migmatites is ~20%. The maximum melt volume fraction is higher when estimated by leucosome volume calculations than when estimated from biotite dehydration melting experiments. Reasons for this deviation may be that chemical factors other than XMg and bulk Al2O3 contents, which were not considered here, effectively led to higher melt volume fractions or that leucosome volume fractions overestimate former melt volume fractions as a result of addition of melt to the migmatites. Although they are fraught with large errors, melt volume fractions estimated with both methods tend towards the maximum values for MIG4 migmatites, which is in line with their restitic melanosomes. CONTROLS ON MELT SEGREGATION For the migmatites from the Bayerische Wald, the degree of quartz and feldspar depletion in mesosomes and melanosomes in combination with the inferred melt volume fractions indicates that bulk composition is an important factor controlling the production of melt. The melt volume fraction produced was probably largest in MIG4 migmatites with very Al2O3-rich bulk compositions. Melt volume fractions could not be estimated from leucosome volume fractions for MIG2 migmatites because of the scarcity of the latter, but the fairly Al2O3rich bulk compositions imply that the melt volume fractions should have been similar to or a little lower than those in MIG4 migmatites. MIG1 and MIG3 migmatites have bulk compositions lower in Al2O3 and should thus have produced lower volume fractions of melt when compared with MIG2 and MIG4 migmatites. This inference is supported by the leucosome volume fractions for MIG3 migmatites that tend to lower values than for MIG4 migmatites. Melt volume fractions in MIG1 and MIG3 migmatites should not deviate considerably from each other, because of similar bulk compositions. However, whereas melt in MIG1 migmatites moved at most a few millimetres to form small light patches within an overall massive structure, a considerable fraction of melt moved out of MIG3 mesosomes to crystallize in leucosomes or at mesosome margins, resulting in a stromatic structure. The different degrees of melt segregation may perhaps be partly explained by lower melt fractions in MIG1 migmatites as a result of slightly lower bulk Al2O3 contents (lower modal abundance of sillimanite; Table 2). However, we contend that the spatial relationship between the two migmatite types and their pre-migmatic structures also exert a control on melt segregation. Microstructures indicate high deformation intensities in MIG3 migmatites (strong SPO of biotite and cordierite and a weak CPO of cordierite in MIG3) but no deformation in MIG1 migmatites (no CPO of cordierite and only weak or no SPO of biotite). Massive MIG1 migmatites occur locally as 1713 JOURNAL OF PETROLOGY VOLUME 40 decimetre- to metre-sized bodies within MIG3 as the most abundant type. MIG1 migmatites have higher bulk strength as a result of their homogeneous structure, which is probably of primary nature. MIG3 migmatites have lower bulk strength as a result of their probably primary heterogeneous structure. Hence, strain is partitioned into the mechanically weaker MIG3 migmatites and this seems to enhance melt segregation. The relation between strain partitioning as a result of strength contrasts and enhanced melt segregation is even more evident from the contacts of MIG3 with MIG2 migmatites. MIG2 migmatites are often observed at the borders to lenses or bands of calc-silicate or gneiss (Fig. 3) that represent strain heterogeneities. Here, MIG3 migmatites grade into type MIG2 with no sharp contacts between the two. The occurrence of material with higher bulk strength (calc-silicate, gneiss) produces local highstress zones at the margins of the calc-silicate and gneiss bodies, resulting in higher strain. At these sites, MIG2 migmatites developed, which are characterized by a higher degree of melt segregation than MIG3 migmatites and by microstructures indicating higher deformation intensities (intermediate CPO of cordierite and SPO of biotite). In outcrops lacking strain heterogeneities, mainly MIG3 migmatites are observed, characterized by lower degrees of melt segregation and deformation. In summary, melt segregation in migmatites of the Bayerische Wald was probably controlled by bulk composition and by strain partitioning as a result of bulk strength contrasts. The latter are probably dependent on pre-migmatitic structures and compositional heterogeneities that result in heterogeneous melt distribution. CONTROLS ON MICROSTRUCTURES The SPO of biotite and garnet, as well as the CPO and SPO of cordierite, indicate different mechanical behaviour of each of these three minerals in the different migmatite types. The change in mechanical behaviour is related to the degree of melt removal (segregation) from the melting sites (mesosomes, melanosomes), expressed by the depletion of these sites in quartz and feldspar. For MIG1, MIG2 and MIG4 migmatites, an obvious trend of increasing CPO of cordierite with decreasing modal feldspar and quartz in the mesosomes and melanosomes can be observed (Fig. 11). A similar relationship for MIG1, MIG2 and MIG4 migmatites holds true for the SPO of biotite [using the A( )min values; see Figs 5 and 11]. These results show that the melt fraction remaining at the melting sites and hence melt segregation control structures and deformation mechanisms of the minerals. In MIG1 migmatites, the euhedral to subhedral shapes and large grain sizes of cordierite, biotite and garnet, and the lack of CPO of cordierite and SPO of biotite NUMBER 11 NOVEMBER 1999 Fig. 11. (a) Concentration parameter for the CPO of the [100] direction of cordierite grains vs the modal abundance of feldspar and quartz. Concentration parameters are calculated from the U-stage data with eigenvalue methods; the degree of CPO increases with increasing values. (Note the strong CPO of cordierite in MIG4 migmatites; also, the inverse relationship between modal abundance of feldspars and quartz and degree of CPO of cordierite.) (b) A( )min values of the biotite SPO vs modal abundance of feldspar and quartz. [Note the similar trend to that in (a), except for MIG3 migmatites (see section on implications of the microstructures in mesosomes and melanosomes for further explanation). For calculation of A( )min values, see caption to Fig. 5.] exclude deformation and indicate melt-controlled growth of cordierite, garnet and biotite. In contrast to the undeformed MIG1 migmatites, MIG3 migmatites show a strong SPO of biotite and cordierite, but the CPO of cordierite is weak and there are no microstructural indications of intracrystalline plasticity (i.e. dynamic recrystallization). Development of the strong SPO of biotite and cordierite, along with the absence of intracrystalline plasticity, can only be explained by passive rotation of solid particles in a weak matrix (e.g. Jeffrey, 1922; March, 1932). If the weak matrix were also solid, deformed matrix minerals would be recognizable, which is not so with MIG3 migmatites. Hence, the weak matrix during 1714 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD the development of the strong SPOs of biotite and cordierite is inferred to have been a melt, as is also evident from petrological data. The presence of melt in MIG3 mesosomes is in line with their very limited depletion in quartz and K-feldspar. Therefore, microstructures in MIG3 mesosomes are mainly melt controlled. Bulk structures, for example, the formation of K-feldspar- and quartz-dominated leucosomes, can be best explained by local melt coalescence during foliation development. With increasing deformation intensities, the connectivity of melt increased. The process corresponds to the development of an IWL in solid rocks (foliation weakening; e.g. Jordan, 1987; Handy, 1994). Therefore, mesostructures in MIG3 migmatites are also melt controlled. A completely different situation is encountered in MIG2 and MIG4 migmatites, where considerable melt segregation has to be inferred. The difference between MIG2 and MIG4 migmatites is a larger fraction of segregated melt in the latter. The strong CPO of cordierite (intermediate in MIG2) and evidence of dynamic recrystallization, as well as the strong SPO of cordierite, biotite and garnet (intermediate in MIG2), indicate dislocation creep of the main minerals in mesosomes and melanosomes of MIG2 and MIG4 migmatites. During melt segregation, a framework (LBF) must have been built up by the coexisting cordierite, biotite and garnet, as a result of the low percentage of melt remaining in mesosomes and melanosomes. Within an LBF, solid particles can only be deformed by a solid-state deformation mechanism, which is suggested by the evidence of intracrystalline plasticity of cordierite in MIG4 migmatites (Figs 4 and 7). Hence, microstructures in mesosomes and melanosomes of MIG2 and MIG4 migmatites are controlled by the mechanical behaviour of individual minerals. In summary, microstructures of the four migmatite types are both melt controlled and mineral controlled. Which process dominates depends on the degree of melt segregation. Melt-controlled behaviour is encountered at migmatite sites that are not considerably depleted in melt. Mineral-controlled behaviour is observed at sites of considerable melt removal. STRUCTURAL EVOLUTION OF THE MIGMATITES As outlined in the two previous sections, the melt volume fraction as a function of bulk composition, the degree of melt segregation and the deformation mechanisms of the minerals controlled the structural evolution of the migmatites from the Bayerische Wald. A complex interplay between those factors led to the development of different migmatite types. The possible evolution of these ‘endmember’ types in relation to the controlling factors is schematically outlined in Fig. 12. At the starting point there were two similar bulk compositions with considerable difference only in their Al2O3 contents (MIG1 and MIG3 vs MIG2 and MIG4). This induced larger melt volume fractions during partial melting in MIG2 and MIG4 migmatites compared with MIG1 and MIG3 migmatites. MIG1 bodies had higher bulk strengths during partial melting, as a result of their homogeneous structure and perhaps slightly lower melt volume fractions when compared with the surrounding MIG3 migmatites. During deformation, the resulting strain partitioning into MIG3 migmatites enhanced melt segregation and the development of stromatic structures in MIG3 migmatites but impeded melt segregation and the development of deformation-related structures in MIG1 migmatites. Therefore, all minerals grew fairly euhedral, to large size and without SPO or CPO in MIG1 migmatites. Melt was segregated from MIG3 mesosomes, but because of the lack of strain heterogeneities, a considerable amount of melt stayed in the mesosomes. Therefore, as in MIG1 migmatites, no LBF developed and cordierite and biotite growth was melt controlled, resulting in euhedral to subhedral shapes. However, in contrast to MIG1 migmatites, deformation played a considerable role. The minerals underwent passive rotation in the melt, resulting in a weak cordierite CPO and a strong SPO of biotite. In MIG2 and MIG4 migmatites, the comparatively large melt fractions (and the position of MIG2 migmatites in sites of strain heterogeneities) enhanced melt segregation into leucosomes and thus the development of stromatic structures. Melt removal from the sites of production (mesosomes and melanosomes) was effective to the degree that an LBF of minerals could form. MIG2 and MIG4 migmatites differ in the degree of melt removal from the sites of production and in the degree of deformation (both higher in MIG4). Before melt segregation started, passive rotation of minerals in the melt took place, producing a strong SPO of biotite in both migmatite types. This can also be deduced from aligned biotite inclusions in cordierite. The lower degree of deformation in MIG2 migmatites is indicated by the weak SPO of individual biotite grains and the intermediate CPO of cordierite as opposed to the strong biotite SPO and cordierite CPO in MIG4 migmatites. GENERAL IMPLICATIONS FOR THE MECHANICAL BEHAVIOUR OF MIGMATITES Numerous studies have discussed the contribution of melt volume fraction to the bulk mechanical behaviour of 1715 JOURNAL OF PETROLOGY VOLUME 40 NUMBER 11 NOVEMBER 1999 Fig. 12. Schematic line drawings of the possible structural evolution of the different migmatite types. In all line drawings, only four phases are distinguished for the sake of clarity: (1) quartz and feldspar; (2) biotite as the main reactant of the melt-producing reactions; (3) cordierite as the main solid product of the melt-producing reactions; (4) melt. On the left-hand side possible precursors of the migmatites are shown that differ in composition and structure. At the top and bottom the observed endmember migmatite types (MIG1–MIG4) are shown. In the centre, possible intermediate evolutionary stages are suggested. partially molten rocks. Experiments (e.g. Dell’Angelo & Tullis, 1988) and theoretical models (Vigneresse et al., 1996) predict that rocks with small melt volume fractions (1–5%) should deform mainly by intracrystalline plasticity of the minerals. Rocks with melt volume fractions above 20–25% (MET; Vigneresse et al., 1996) or 30–35% (RCMP; e.g. Arzi, 1978; Van der Molen & Paterson, 1979; Fernandez & Gasquet, 1994) should behave more or less as melts. MIG1–MIG4 migmatites of the Bayerische Wald are natural examples with melt volume fractions in the RCMP or MET range. Although melt volume fractions do not differ dramatically from each 1716 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD other, MIG1–MIG4 migmatite structures indicate very different mechanical behaviour. MIG2 and MIG4 migmatites preserve evidence of mineral-controlled mechanical behaviour whereas MIG1 and MIG3 migmatites display melt-controlled microstructures. This observation implies that there is no simple relationship between melt volume fraction and the mechanical behaviour of migmatites. Several other observations on migmatites from the Bayerische Wald help to shed light on factors that also control the mechanical behaviour of migmatites, but that have not been or cannot be considered in experiments and theoretical models; for example, the effects of melt segregation, melt distribution and time. The obtained results indicate that the mechanical behaviour of migmatites during partial melting depends strongly on the degree of melt segregation. As melt segregation is partly controlled by melt volume fraction, the mechanical behaviour of migmatites will probably change with time as melt volume fractions become larger in the course of prograde metamorphism. As long as the degree of melt segregation is low and much of the melt resides at its production site, IWL structures will dominate. As soon as considerable amounts of melt have been segregated, LBF structures can form in the restites. The results obtained here indicate that melt segregation is favoured along strain heterogeneities occurring at sites of pronounced strength contrasts. The effect of strain localization on melt segregation has been described also from Namaqualand (South Africa) by Kisters et al. (1998) on larger scales. Strain localization results in a heterogeneous distribution of melt and enhances bulk strength contrasts even more with time. The effect of melt distribution on the evolution of shear zones has been discussed by Grujic & Manktelow (1998). In their analogue experiments, shear zone patterns and deformation sites were strongly controlled by melt distribution. As in the migmatites of the Bayerische Wald, the mechanical behaviour and melt distribution changed during melt segregation. In the proposed evolutionary scheme for migmatites of the Bayerische Wald (Fig. 12), we assumed an initial difference in melt volume fraction to be responsible for the strength contrast, implying that all bulk compositions melt at the same temperature and that some compositions produce higher melt volume fractions than others. However, in nature, differences in bulk composition (including fluids) probably result in different solidus temperatures. Therefore, during prograde metamorphism, some parts of a migmatite protolith will start to melt earlier than others and this will also result in bulk strength contrasts. The observation that the bulk mechanical behaviour of migmatites may be controlled by the minerals (e.g. MIG4 migmatites) emphasizes the importance of the mechanical behaviour of individual minerals during deformation and partial melting. Various minerals can be produced in the course of dehydration melting (e.g. cordierite, garnet, orthopyroxene, spinel) depending on bulk composition. Therefore, the results obtained here cannot be applied to natural migmatites in general. More data on the mechanical behaviour of other minerals during partial melting and deformation need to be gained. Apart from time, all of the factors controlling the mechanical behaviour of natural migmatites (melt volume fractions, bulk strength contrasts, degree of melt segregation, melt distribution, minerals) are dependent on the composition and the compositional heterogeneity of the migmatite protolith. Many migmatites have sedimentary or metamorphic precursors that are probably heterogeneous in terms of modes and bulk compositions on various scales. Therefore, as shown in the Bayerische Wald, distinct parts within a forming migmatite complex show different mechanical behaviour during partial melting and deformation. Thus, the mechanical behaviour of an entire crustal section is very hard to assess and cannot be described by a single flow law, even if temperature, pressure and melt volume fractions are known. In the future, the combination of three different approaches may serve to derive quantitative models of the bulk mechanical behaviour of partially molten crustal sections, namely, (1) detailed mapping of migmatite areas, along with qualitative and quantitative microstructural observations, as presented in this study, (2) deformation experiments considering various relevant minerals and heterogeneous melt distribution in space and time, from which flow laws for melt–mineral mixtures in relation to melt segregation can be derived, and (3) numerical models that integrate different flow laws for various stages of partial melting. SUMMARY AND CONCLUSIONS Migmatites of the Bayerische Wald can be divided into four types (MIG1–MIG4) based on geometric relationships, modes and microstructures of mesosomes, melanosomes and leucosomes. Type MIG1 is massive and undeformed. Types MIG2 and MIG3 are both stromatic (leucosome–mesosome interlayering), but differ in the degree of deformation. Type MIG4 has an interlayering of melanosome and leucosome and is strongly deformed. Former melt volume fractions are estimated to be between 20 and 40%, based on comparison with melt fractions obtained in biotite dehydration experiments at the relevant P–T conditions and on volume fraction estimates of K-feldspar- and quartz-dominated leucosomes. Modes indicate that hardly any melt was removed from MIG1 migmatites and MIG3 mesosomes, whereas 1717 JOURNAL OF PETROLOGY VOLUME 40 MIG2 mesosomes and MIG4 melanosomes experienced considerable melt depletion. The degree of melt segregation is not only a function of melt volume fraction but is strongly enhanced by strain heterogeneities related to strength contrasts. Quantitative data on the SPO of biotite and the CPO of cordierite indicate that the deformation mechanisms of the minerals are mainly controlled by the degree of melt segregation. Passive rotation within interconnected weak layers of melt at low degrees of melt segregation (MIG1 and MIG3) changes to deformation within a solid framework of minerals at high degrees of melt segregation (MIG2 and MIG4). The development of microstructures in migmatites of the Bayerische Wald, and thus their mechanical behaviour, is controlled by a complex interplay of melt volume fraction, melt segregation, melt distribution, bulk strength contrasts and mechanical properties of different minerals in a dynamic system. Therefore, the mechanical behaviour of migmatites during partial melting varies in space and time and cannot be modelled by a single flow law. Detailed mapping of migmatite areas, along with microstructural observations, deformation experiments considering heterogeneous melt distribution, and numerical models integrating different flow laws for various stages of partial melting, may serve to derive quantitative models for the bulk mechanical behaviour of a crustal section in the future. ACKNOWLEDGEMENTS Mark Handy, Ron Vernon and Jean-Louis Vigneresse are thanked for detailed and constructive reviews that helped to improve the quality of the paper. Simon Harley is thanked for having taken over the editorial handling of the manuscript and for numerous helpful comments. Thanks go to Rainer Altherr and Ian Fitzsimons for critical comments on the paper and its earlier versions. A.B. was funded by a research fellowship of the Deutsche Forschungsgemeinschaft (DFG), which is gratefully acknowledged. A.K. thanks the DFG for financial support (Ka 1023/2, Ka 1023/3) within the special research programme ‘Orogene Prozesse—ihre Quantifizierung und Simulation am Beispiel der Varisziden’. REFERENCES Arzi, A. A. (1978). Critical phenomena in the rheology of partially molten rocks. Tectonophysics 74, 173–184. Ashworth, J. R. (1985). Migmatites. Glasgow: Blackie. Barboza, S. A. & Bergantz, G. W. (1998). Rheological transitions and the progress of melting of crustal rocks. Earth and Planetary Science Letters 158, 19–29. NUMBER 11 NOVEMBER 1999 Beer, W. W. (1981). Die strukturelle Entwicklung der Metamorphite des Bayerischen Waldes. Ph.D. Thesis, University of Gottingen. ¨ Benn, K. & Allard, B. (1989). Preferred mineral orientations related to magmatic flow in ophiolite layered gabbros. Journal of Petrology 30, 925–946. Deer, W. A., Howie, R. A. & Zussman, J. (1992). An Introduction to the Rock Forming Minerals. Harlow: Longman. Dell’Angelo, L. N. & Tullis, J. (1988). Experimental deformation of partially molten granitic aggregates. Journal of Metamorphic Geology 6, 495–515. Den Brok, B. & Kruhl, J. (1996). Ductility of garnet as an indicator of extremely high temperature deformation: discussion. Journal of Structural Geology 18, 1369–1374. Dingwell, D. B., Hess, K. U. & Knoche, R. (1996). Granite and granitic pegmatite melts: volumes and viscosities. Transactions of the Royal Society of Edinburgh 87, 65–72. Fediuk, F. (1974). Cordieritregelung in Moldanubischen Gneisen. Krystallinikum 10, 79–88. Fernandez, A. N. & Gasquet, R. D. (1994). Relative rheological evolution of chemically contrasted coeval magmas: example of the Tichka plutonic complex (Morocco). Contributions to Mineralogy and Petrology 116, 316–326. ¨ Fischer, G. (1938). Uber das Grundgebirge der Bayerischen Ostmark. ¨ Die Gneise nordlich des Bayerischen Pfahles. Jahrbuch der Preußischen Geologischen Landesanstalt 59, 289–352. Franke, W. (1989). Tectonostratigraphic units in the Variscan belt of Europe. Geological Society of America, Special Papers 230, 67–90. Gardien, V., Thompson, A. B., Grujic, D. & Ulmer, P. (1995). Experimental melting of biotite + plagioclase + quartz ± muscovite assemblages and implications for crustal melting. Journal of Geophysical Research 100, 15581–15591. Grauert, B., Hanny, R. & Soptrajanova, G. (1974). Geochronology of ¨ a polymetamorphic and anatectic gneiss region: the Moldanubicum of the Area Lam–Deggendorf, Eastern Bavaria, Germany. Contributions to Mineralogy and Petrology 45, 37–63. Grujic, D. & Mancktelow, N. S. (1998). Melt bearing shear zones: analogue experiments and comparison with examples from southern Madagascar. Journal of Structural Geology 20, 673–680. Handy, M. R. (1990). The solid-state flow of polymineralic rocks. Journal of Geophysical Research 95, 8647–8661. Handy, M. R. (1994). Flow laws for rocks containing two non-linear viscous phases: a phenomenological approach. Journal of Structural Geology 16, 287–301. Holtz, F., Pichavant, M., Barbey, P. & Johannes, W. (1992). Effects of H2O on liquidus phase relations in the haplogranite system at 2 and 5 kbar. American Mineralogist 77, 1223–1241. Jeffrey, G. B. (1922). The motion of ellipsoidal particles immersed in a viscous fluid. Proceedings of the Royal Society of London, Series A 102, 161–179. Ji, S. & Martignole, J. (1994). Ductility of garnet as an indicator of extremely high temperature deformation. Journal of Structural Geology 16, 985–996. Johannes, W. & Holtz, F. (1996). Petrogenesis and Experimental Petrology of Granitic Rocks. Heidelberg: Springer. Jordan, P. G. (1987). The deformational behaviour of bimineralic limestone–halite aggregates. Tectonophysics 133, 185–198. Kalt, A., Corfu, F. & Wijbrans, J. (submitted). Time calibration of a P–T path from a Variscan high-temperature low-pressure metamorphic complex (Bayerische Wald, Germany), and the detection of inherited monazite. Contributions to Mineralogy and Petrology. Kalt, A., Berger, A. & Blumel, P. (1999). Metamorphic evolution of ¨ cordierite-bearing migmatites of the Bayerische Wald (Variscan Belt, Germany). Journal of Petrology 40, 601–627. 1718 BERGER AND KALT MIGMATITES OF THE BAYERISCHE WALD Kisters, A. F. M., Gibson, R. L., Charlesworth, E. G. & Anhaeuser, C. R. (1998). The role of strain localization in the segregation and ascent of anatectic melts Namaqualand, South Africa. Journal of Structural Geology 20, 229–242. Kretz, R. (1983). Symbols for rock forming minerals. American Mineralogist 68, 277–279. Kretz, R. (1994). Metamorphic Crystallization. New York: John Wiley. Kruhl, J. H. & Huntemann, T. (1991). The structural state of the former lower continental crust in Calabria (S. Italy). Geologische Rundschau 80, 289–302. Le Breton, N. & Thompson, A. B. (1988). Fluid absent (dehydration) melting of biotite in metapelites in the early stages of crustal anatexis. Contributions to Mineralogy and Petrology 99, 226–237. March, A. (1932). Mathematische Theorie der Regelung nach der Korngestalt bei affiner Deformation. Zeitschrift fur Kristallographie 81, ¨ 285–297. Montel, J. M. & Vielzeuf, D. (1997). Partial melting of metagreywackes. Part 2: Composition of minerals and melts. Contributions to Mineralogy and Petrology 128, 176–196. Nyman, M. W., Pattison, D. R. M. & Ghent, E. D. (1995). Melt extraction during formation of sillimanite and K-feldspar migmatites, west of Revelstoke, British Columbia. Journal of Petrology 36, 351–372. Olson, S. N. (1984). Mass balance and mass transfer in migmatites from the Colorado Front Range. Contributions to Mineralogy and Petrology 85, 30–44. Panozzo, R. (1983). The two dimensional analysis of shape fabric using projection. Tectonophysics 95, 279–295. Panozzo, R. (1984). Two dimensional strain from the orientation of lines in plane. Journal of Structural Geology 6, 215–221. Panozzo-Heilbronner, R. & Pauli, C. (1994). Orientation and misorientation imaging integration of misorientation and textural analysis. In: Bunge, H., Siegesmund, S., Skrotzki, W. & Weber, K. (eds) Textures of Geological Materials. Oberursel: DGM Informationsgesellschaft, pp. 147–164. Paterson, M. S. (1987). Problems in the extrapolation of laboratory rheological data. Tectonophysics 133, 33–43. Paterson, S. R., Vernon, R. H. & Tobisch, O. T. (1989). A review of criteria for the identification of magmatic and tectonic foliations in granitoids. Journal of Structural Geology 11, 349–363. Patino Douce, A. E. & Beard, J. S. (1995). Dehydration melting of ˜ biotite gneiss and quartz amphibolite from 3 to 15 kbar. Journal of Petrology 36, 707–738. ˜ Patino Douce, A. E. & Beard, J. S. (1996). Effects of P, f (O2) and Mg/ Fe ratio on dehydration melting of model metagreywackes. Journal of Petrology 37, 999–1024. Patino Douce, A. E. & Johnston, A. D. (1991). Phase equilibria and ˜ melt productivity in the pelitic system: implications for the origin of peraluminous granitoids and aluminous granulites. Contributions to Mineralogy and Petrology 107, 202–218. Reinhard, M. (1931). Universaldrehtischmethoden. Basel: Wepf. Rutter, E. & Neumann, D. H. K. (1995). Experimental deformation of partially molten Westerly granite under fluid absent conditions, with implications for the extraction of granitic magmas. Journal of Geophysical Research 100, 15697–15715. Skrotzki, W. & Siegesmund, S. (1993). Cordierite microstructure and texture in a Moldanubian gneiss. Physics and Chemistry of Minerals 19, 401–408. Stevens, G., Clemens, J. D. & Droop, G. T. R. (1997). Melt production during granulite-facies anatexis: experimental data from ‘primitive’ metasedimentary protoliths. Contributions to Mineralogy and Petrology 128, 352–370. Turner, F. J. & Weiss, L. E. (1963). Structural Analysis of Metamorphic Tectonites. New York: McGraw–Hill. Van der Molen, I. & Paterson, M. S. (1979). Experimental deformation of partially-molten granite. Contributions to Mineralogy and Petrology 70, 299–318. Van Roermund, H. L. M. & Konert, R. J. (1990). Deformation and recrystallization mechanisms in naturally deformed cordierite. Physics and Chemistry of Minerals 17, 52–61. Vigneresse, J.-L., Barbey, P. & Cuney, M. (1996). Rheological transitions during partial melting and crystallization with application to felsic magma segregation and transfer. Journal of Petrology 37, 1579–1600. Watson, G. S. (1983). Statistics on Spheres. New York: John Wiley. Woodcock, N. H. (1977). Specification of fabric shapes using eigenvalue methods. Geological Society of America Bulletin 88, 1231–1236. 1719 ...
View Full Document

Ask a homework question - tutors are online