PREFERRED ORIENTATIONS OF MICA BEARING ROCKS DESCRIBED BY TEXTURE COMPONENTS

Up to now Texture investigations of phyllosilicates were restricted to pole density measurements of the (001) basal plane. Because of the tabular grain shape parallel to the basal plane an axial symmetry of the texture is assumed usually. This assumption was proved for four mica bearing samples by calculating the ODF from neutron diffraction data. Beside the mica-texture the textures of the other major fractions like quartz and feldspar were determined to detect correlation between them. The textures are explained by components. Each component is characterised by a preferred orientation or fibre axis, a corresponding spreading and an intensity describing the volume fraction of all crystallites belonging to the component. The presented component method can be applied to any crystal symmetry and to polyphase materials with overlapped pole figures.


INTRODUCTION
Earlier Quantitative Texture Analyses (QTA) of geological materials were mainly restricted to monophase samples or to single grain measurements on the U-stage (Schmid and Casey, 1986, Schmid et al., 1987, Kruhl, 1987, Mainprice and Munch 1993. QTA of polyphase materials was hampered by the high degree of overlapping Bragg reflections and the low symmetry of minerals, for which no algorithm to calculate the Orientation Distribution Function (ODF) was available. Meanwhile QTA is possible for all symmetries and polyphase materials (Helming and Eschner, 1991). Position-sensitive detector (Wenk et al., 1986 and energy-dispersive measuring methods (Feldmann, 1989) allow the recording of more or less complete diffraction patterns, and large sample volumes can be investigated by means of neutron diffraction to obtain better counting statistics. Nevertheless, applications are rare although most rocks are of polyphase composition (Wenk and Pannetier, 1990, Helming, 1993, Leiss et al., 1994. This paper deals with texture determinations of polyphase samples and focuses especially on the determination of mica textures. Based on the assumption, that phyllosilicates reorientated 212 K. HELMING ET AL. due to their tabular grain shape and preferred orientations of crystallographic faces other than the basal plane do not exist. First hints to the development of preferred orientations have been derived from the measurement of the anisotropy of magnetic susceptibility of mica bearing rocks. Texture determinations of this mineral group based on diffraction were restricted to pole density measurements of the (001) basal plane. The development of (001) preferred orientations and its relation to deformation is quite well understood (Weber, 1981, Sintubin, 1994, until recently no attempt to determine preferred orientations of other crystallographic directions has been reported. Applying X-rays for (001) pole density measurements requires special corrections for progressive intensity loss due to the low penetration depth of X-rays, the very small scattering angle and the tabular grain shape of the crystallites (Ullemeyer and Weber, 1994). Pole density measurements by means of neutron diffraction allow the investigation of a much larger sample volume, due to the low absorption of neutrons, special correcting procedures as a consequence of anisotropic grain shape are not required. Nevertheless, the scattered intensities (and the contributions to overlapped Bragg reflections) of phyllosilicates are relatively low with respect to other minerals and large errors of texture determination be assumed.
To avoid the disadvantages mentioned above, determinations of pole figures were performed at the TOF-diffractometer NSHR at the pulsed reactor IBR-2 in Dubna, Russia (Feldmann, 1989). Due to the high resolution of that instrument peak overlappings may be reduced experimentally to a minimum, furthermore, the recording of complete diffraction patterns allows the definition of each desired d-range as an experimental pole figure. Four samples were selected to test the reliability of QTA of polyphase rocks: two muscovite-quartzites from the Taquaral Valley, Brazil (quartz and muscovite, Quade et al., 1994), and two granulites from the Moldanubian Zone, Austria (quartz, plagioclase and biotite, Ullemeyer, 1992).

THE COMPONENT MODEL
A direct numerical description of the ODF f(g) is quite expensive, because it requires three orientation parameters and one intensity value for a large number of discrete orientations gi. The number of parameters decreases, if the orientations are given on a regular grid. Compared to this kind of ODF description, the required texture information for many applications is much more compact, but it may be difficult to find out the parameters, which are really important for ODF approximation. Another disadvantage is the necessary interpolation of ODF-values, which do not match the given grid points. In that case, the approximation of the ODF by simple functions, which can be handled easily on the computer, should be used. Most propagation the series expansion with generalized spherical harmonics (Bunge, 1965) has reached: f(g) cnnwlmn(g).
(1) m,n They form an orthogonal system of basic functions. Besides this numerical advantage mean values of anisotropic properties can be efficiently calculated by means of the corresponding C-coefficients. The basic functions T are periodic in the whole G-space (period length 2r/1), so this type of approximation is suitable to characterize the periodical behaviour of the ODF. If the main features of the texture can be described by a few preferred orientations, the number of C-coefficients can be much larger as the number of parameters, which is really required for texture description. In such cases the ODF-approximation by means of texture components C f(g) F + y_ ICf (g) =' (2) c with F + IC 1 and P (g)dg c=l should be preferred (Wassermann, 1939;Bunge, 1969;Matthies et al., 1989Matthies et al., -1992. A component is described by a model function if(g), which is locally restricted in the G-space, and an intensity Ic, which describes the volume fraction of all crystallites belonging to the component c. The quantity F gives the volume fraction of the crystallites, which are randomly oriented in the sample. It may be understood as the intensity of the only global (not restricted in. the G-space) component used in the model, which is given by f0(g)= for each g G. if(g) has a maximum at a preferred orientation gC and decreases with increasing orientation distance &c, which is defined by the rotation angle & co n co cos cos cos + n sin sin (3) 2 2 2 2 2 of the orientation difference gc [&c, tic] gCg-1, between gC [, nc] and g [co, n]. If the intensity decrease does not depend on the rotation axis tic, the component is called isotropic. An anisotropic component is characterized by a fibre axis yC given in the sample coordinate system KA. The corresponding model function if(g) remains unchanged rotating around yC. Its maxima form a great circle in the orientation space given by c= [co, +yq gC with 0 < co < , which includes the preferred orientation gC. The distribution depends only on the angle between the directions g yC and gCyC.
With a Gaussian standard distribution for the isotropic component (Matthies et al., 1989(Matthies et al., -1992 we  P, (y) Ne scsohcsy) Io (S sin(Oh)Sin(tgy)) (7) with COSOh h. gCyC, cOSOy y. yC In contrast to the series expansion with generalized harmonics (Bunge 1965) the component descriptions is not unique especially for weak textures. So the description of a random distribution with f(g) 1 for each g may be given by F 1 or a finite number of components, which are regularly arranged in,the G-space. On the other hand, the component description allows the controlled restriction of texture information.
As shown below this can be necessary, if the quality or quantity of measured data does not allow precise calculation of the ODF. Taking into account only a small number of components, an estimation of the ODF may be possible at least. Searching for important symmetries or parameter of texture forming processes restrictions on special types of components may be useful. The main purpose of this paper is to investigate the coupling between the texture of mica and its axial morphology. If this coupling is dominant, we should expect textures which can be described by fibre components.

DETERMINATION OF THE COMPONENT PARAMETERS FROM DIFFRACTION DATA
PFs which are measured in a diffraction experiment can be understood as twodimensional projections of the three-dimensional ODF +hllY (8) go describes all orientations with its lattice plane normal h parallel (or antiparallel) to the scattering vector y, which is named sample direction and given with respect to the sample coordinate system. Because of Friedel's law PFs are affected by an additional inversion center allowing only the measurement of reduced PFs 'h(Y)" Especially for low crystal symmetries and multiphase systems Bragg-reflections of lattice planes may have the same or similar lattice distances dhkl) To determine the parameters gpe, ypc, bpc and b by a nonlinear algorithm, first estimates are necessary, which may be obtained manually from the graphical representation of the difference pole figures 6S(yr) d(Yr)-dM(yr) Nd on a PC-Monitor.
This procedure is explained in more detail in Helming and Eschner (1991) and Helming et al. (1994). The calculation of the model pole figures lp follows from (5), (7) and (8).

RESULTS
The TOF diffraction pattern of sample STG (Figure 1) shows the high degree of overlapping of Bragg reflections of a polyphase sample containing low symmetrical constituents. The chosen sample consists of quartz (=40%), plagioclase (--35%), biotite (=7%), K-feldspar (=10%) and accessory minerals. At small d-values (d < 2.7/) the number of reflections and consequently the number of partial and complete overlapping increases rapidly. In most cases an experimental PF encompasses peaks of two or more phases (Table 1), PFs consisting of only one lattice plane or reflections of only one phase are rare.     Tables 2 and 3. A various number of peakand fibre-components was determined. The most interesting features are be discussed below: the mica textures are characterized not only by fibre components with its axes parallel to the basal plane normal, significant peak components were also found (P17316, P16815). The sharpness of both textures in these samples guarantees that the calculated preferred orientations and the corresponding halfwidths are no artifacts. STG RPO: 12 (7-18) MG RPO: 11 (7-16) the random part F of the textures is quite large. It ranges between 27.7% and 77.1% for the two-phase samples and between 65.4% and 94.2% for the three-phase samples. But because of the high degree of superposition of the measured pole figures, F and the component intensities I must be considered only as estimates.
The quality of texture reproduction is described by the global error parameter RP0 (Matthies et al., 1988). RP0values of the textures range between 11% and 19%, the spread for single PFs is sometimes large (Tables 2 and 3, Figure 2). In the Figures 3, 4 and 5 recalculated pole figures of some lattice directions are presented, which are of interest for the kinematic interpretation. The predominant features are" in agreement with the dominance of fibre components, the mica PFs are characterized by axial symmetric intensity distributions, which are more or less modified by the peak components.
in sample P17316 only peak components could be found. In contrast to the other samples the pole density maximum of the (001) basal plane PF deviates from the foliation pole by about 30 and the (100) and (010) directions also display preferred orientations.
the recalculated quartz (0001) and 1120 PFs correspond to commonly observed orientation patterns. They are generally described exclusively by peak components, in the case of sample MG the texture is dominated by a fibre component with its axis parallel to the (011) normal. the recalculated plagioclase PFs are either described by peak components (sample STG) or by one dominant fibre componeht with its axis close to the (001) normal and a number of small peak components (sample MG).

DISCUSSION
The high degree of superposed Bragg reflections has consequences for the quality of texture reproduction of the rock constituents. An important influence may be inferred from the relative intensity contribution qd of a lattice plane to a measured PF, which depends on the scattering behaviour and the volume fraction of the mineral phase in the sample. Quartz generally scatters much more than plagioclase and mica minerals. All superposed PFs with contributions of quartz are dominated by this mineral. Small volume fractions (e.g. mica: typically 5-15% in the considered kind of rock) additionally reduce the intensity contributions to an experimental PF. The overlapped PF q (011) + q (101) + b (006) of sample STG (quartz: 40%, biotite: 7%) may serve as an example. The intensity contribution -biotite qid is 2.6% (Table 1), i.e. biotite causes only a minor modification of the q (011) + q (101) intensity contribution, although a strong preferred orientation of the biotite (006) basal plane is confirmed by independent X-ray measurements (Ullemeyer 1992). In the corresponding PFs of Figure 2 this intensity maximum is located in the centre. It is assumed, that intensity contributions to such a small degree are close to the error magnitude to be expected, and therefore the efficiency of that PF for the determination of the biotite texture is minimal.

CONCLUSIONS
Because of the morphology of phyllosilicate grains, ODF calculations always require diffraction measurements with neutrons possessing a high penetration depth. However in quartzite it is impossible to find separate mica reflections which do not belong to the basal plane. Therefore the error magnitude of the quantities q, N, I, F, calculated by the component method, may be considerable. But if, like in our cases, the investigated textures are not too weak, the method allows an accurate determination of preferred orientations and corresponding halfwidths. Beside the measurement of the anisotropy of magnetic susceptibility this gives another, more precise tool to investigate the behaviour of mica during texture-forming processes. From this point of view our results show uniquely that mica-textures are much more complicated as assumed up to now and they are not axial symmetrical in each case. Texture modifying processes in a mica bearing rock may consequently depend on both, the tabled shape of the grains (axial symmetrical) and the crystal lattice (monoclinic). To describe the interface between these extreme cases rotational, elliptical components (Eschner, 1993, Helming, 1994 will be used in future.
The amount of texture information, that can be determined with the component method, is dependend on the quantity and quality of the measured data. Hence, like shown in the presented multi-phase examples, the determination of the most important texture information is possible starting from 20 pole figures measured by neutron diffraction.