ray study of potassium feldspars from diff erent granitoid types and gneisses of Papuk Mt . ( Slavonia , Croatia )

Potassium feldspars from different granitoids and gneisses of Papuk Mt. (Slavonia, Croatia) have been investigated by X-ray powder diffraction. Diffraction patterns classically observed as well as patterns calculated by Rietveld refi nement were compared and discussed. Triclinicity was calculated according to GOLDSCHMIDT & LAVES (1954) while the structural state of the feldspars was determined using the methods of KROLL & RIBBE (1983) and GODINHO & JALECO (1973). Results showed that the type of potassium feldspar depend on the investigated host-rock, indicating variation in the structural state from orthoclase, intermediate microcline to highly ordered microcline. Potassium feldspar megacrysts in biotite-granodiorites and monzogranites are intermediate microcline or orthoclase, while two-mica monzogranites contain low microcline. Gneisses contain low microcline and orthoclase in Brzaja Creek and low microcline in Djedovica Quarry. Classically observed and digital diffraction patterns calculated by the Rietveld refi nement method produced comparable results and provided a very good correlation of the results obtained by different methods. High triclinicity values of feldspars from investigated granitoid and gneiss samples from Papuk Mt. (Slavonia, Croatia) are in accordance with a high Al content in the T1o site and their fully ordered state indicates a slow(er) cooling-rate. Low triclinicity values, an Al content in T1o site around 0.60 and ordering index smaller than 0.80 can be interpreted as a result of relatively fast(er) cooling which allowed lower ordering of the potassium feldspar.


INTRODUCTION
X-ray powder diffraction is the most common method for determining symmetry and ordering of potassium feldspars.For monoclinic crystals the hkl and hk ¯l refl ections show a single sharp peak on the diffraction pattern.For triclinic crystals the hkl and hk ¯l peaks have different 2θ values and form a double peak.In the case of alkali feldspars, the splitting of the 131 and 13 ¯1, that occur at ~29.9° 2θ CuKα, is and [11 ¯0] directions: named tr [110] and tr [11 ¯0], respectively.Assuming that the angular difference of 2 ¯04 and 060 peaks, in the 29-31° region changes linearly with degree of order, GODINHO & JALECO (1973) defi ned an ordering index (∆ Sm ) based on this difference.
Various potassium feldspar polymorphs and their structural states, from the different granitoid types and gneisses found at Papuk Mt. in Croatia are presented (Fig. 1).Results obtained from the classically observed and digital X-ray diffraction patterns calculated by Rietveld refi nement are compared.For both of these patterns GOLDSCHMIDT & LA VES's (1954) triclinicity and KROLL & RIBBE's (1983) ordering path calculations were undertaken and are discuss ed.Re sults for the T 1 o site occupancy estimated by KROLL & RIBBE's (1983) method, are compared with the results calculated by NEVES & GODINHO's (1995) formulae.Finally, the structural state of these feldspars derived from triclinicity and Al occupancy of tetrahedral sites is discussed.
A short review about the previous study of feldspars in this area can be found in and KOVÁCS KIS et al. (2004) and references within.

MATERIALS AND METHODS
Various types of granitoid and gneiss rock samples (Table 1) were collected from several valleys on Papuk Mt.Sampling localities are shown on the compiled geological map in Figure 1.The potassium feldspars from the aforementioned rocks were differentiated by X-ray powder diffraction measurements.X-ray investigation included measurement and indexing of the X-ray powder diffraction patterns of two kinds of materials: (1) non-magnetic, low density separates of fi ner-grained rocks (a mixture of quartz, alkali feldspar and plagioclase, i.e. felsic components that were impossible to separate from each other by hand-picking under the stereomicroscope) and (2) feldspar megacrysts picked out from four samples that are representative for different granitoid types and localities.
(1) The separates were prepared by crushing and wet sieving followed by heavy liquid (bromoform) separation of either the 0.25-0.125mm or 0.125-0.063mm fractions.The light fraction was further cleaned by a Frantz isodynamic magnetic separator (model LI).Separates that contained Kfeldspar, plagioclase and quartz grains, were pulverised in an agate mortar.X-ray powder diffraction patterns were obtained on an analogue Siemens D500 powder diffractometer in the 5° to 65° 2θ range.Instrumental parameters were: CuKα radiation, 40 kV, 20 mA, Ni fi lter, registration at 0.5° 2θ/min goniometer, 1 cm/min chart speed, 2×10 3 sensitivity and 0.1 mm detector aperture.NaCl was used as zero shift internal  standard.Powder patterns were recorded on the paper and the peak positions were determined manually.The potassium feldspar polymorph(s) have been identifi ed with the help of the JCPDS database.UnitCell software (HOLLAND & REDFERN, 1997) was used for unit cell parameters calculation.
(2) Data for megacrysts were collected in 5-70° 2θ ranges on a Siemens D5000 theta-theta diffractometer equipped with graphite secondary beam monochromator.Conditions were: CuKα 1 radiation, 0.02° 2θ step size and 5 seconds counting time per step.The Rietveld analyses were performed by the DBWS-9006 PC program package (YOUNG et al., 1994).
Triclinicity and ordering were calculated for both type of patterns (classical and digital) i.e. for both type of samples (separeates and megacrysts).

RESULTS
The powder diffraction patterns revealed that different polymorphs of feldspar are present in the rocks that differ both by rock type and sampling locality.X-ray powder patterns revealed that monzogranites from Rajčevica and Pakra Creeks (Figs. 1 and 2), granitoids from Brzaja Creek and gneisses from Djedovica Quarry contain only low microcline and low albite.Gneisses from Brzaja Creek, monzogranite from Kišeljevac and Šandrovac Creeks and fi ne to medium-grained granitoid samples from Pakra Creek valley also contain orthoclase.Porphyric and slightly porphyric granodiorites are typical of the Pakra Creek valley.Orthoclase is ubiquitous in these samples.If microcline appears, it is intermediate microcline.The potassium feldspar polymorph in the PPM-9 gneiss sample (Brzaja Creek) is monoclinic orthoclase.
Rietveld refi nement for four feldspar megacrysts produced identical results in determination of feldspar polymorphs as the classical method, although small differences in Δ values were present (Table 1).Observed and calculated diffraction patterns of one of the feldspar megacryst samples (2PPG-5) are shown in Figure 3. Table 2 presents unit cell parameters, t 1 o occupancies and ordering for separates as well as for four potassium feldspar megacrysts.Feldspars from granitoids 2PPG-3, PPG-19 (Pakra Creek) and gneisses PPM-5 and PPM-9 (Brzaja Creek) have lattice parameters that conform to a monoclinic cell.Two grains (patterns a and b) of the 2PPG-6 granitoid sample show initial splitting of peak at 29.79 and 29.75 2°θ (Fig. 2).The splitting is not measurable so they are determined as orthoclase.Feldspars from PPG-4, PPG-8, PPG-20, 2PPG-32, 2PPG-33, HEG-31 and PPG-12 granitoids have a 0 between 8.56 to 8.59Å, b 0 from 12.92 to 13.01Å and c 0 in 7.20-7.22Årange, which correspond to the triclinic symmetry.Feldspar of PPG-18 and previously mentioned 2PPG-32 and HEG-31 samples shows a complicated {131} region with orthoclase and low microcline peaks.These peaks prove a presence of more than one potassium phase i.e. coexistence of pha ses having monoclinic and triclinic symmetries in various proportions within the same sample (Table 2).Overlapping of peaks on patterns (PPG-18, PPG-23 and PP-13/6 samples) meant that only a limited number of peaks could be used for calculation of unit cell parameters.Consequently the results for these samples are doubtful (Table 2, grey colour-coded samples).In contrast, feldspar from the 2PPG-5 sample shows intermediate microcline peaks, but those peaks are sharp with fair intensity and they could be accurately read (Figure 2).Intermediate microcline of the 2PPG-5 sample have a 0 = 8.56, b 0 = 12.96 and c 0 = 7.21Å, with α slightly different from 90°, which also revealed triclinic symmetry (Table 2).
Comparison of the values obtained from the observed and calculated diffraction patterns showed that the results match in case of three samples .For the patterns of the PPG-20 sample there is an obvious discrepancy between the t 1 o values obtained by classical and Rietveld refi nement methods (0.96 and 0.84, respectively).This cannot be explained as uncertainty in reading the peak positions (see Figure 2), and is most probably the result of the small number of peaks that could be used for unit cell parameter calculations.
Ordering index ∆ Sm and ΣAl in the T 1 sites calculated according to NEVES & GODINHO (1995) are also shown in Table 2. Investigated potassium feldspars have ∆ Sm from 0.50 to 1 while ΣAl(T 1 ) range from 79% to 99%.According to the fact that the total Al content of the T 1 sites is 55% in  1987) these estimated values correlate well with those determined from unit cell parameters.

DISCUSSION
Polymorphic modifi cations of alkali feldspars are the result of the Al-Si distribution in the crystal structure.The way in which the Al content changes in each tetrahedral site (T 1 o, T 1 m, T 2 o and T 2 m), from the most disordered (sanidine), to the most ordered state (high triclinic/low microcline), is called the ordering path in alkali feldspars (SMITH, 1974;GRIF FEN, 1992).Ordering is a two-step phenomenon: (1) the migration of Al from the T2 to T1 sites and (2) Al migration from T 1 m, T 2 o and T 2 m to the T 1 o site (STEWART & WRIGHT, 1974).If the feldspar is monoclinic, the probability of fi nding Al at the T 1 o and T 1 m sites is equal (t 1 o = t 1 m).& RIBBE (1983).Therefore the [110] method can be used for determining the (Al-Si) distributions among the non-equivalent tetrahedral sites.If ordering proceeds as far as possible, all of the aluminium is found in T 1 o, therefore t 1 o=1.0;t 1 m=t 2 o=t 2 m=0.0 and alkali feldspar is referred to as maximum (or low) microcline.According to the results presented in Table 2 studied granitoids from Brzaja, Kišeljevac, Šandrovac and Rajčevica Creek valleys and gneisses in Djedovica Quarry and one gneiss sample from Brzaja Creek valley contain highly ordered potassium feldspar, because most of the Al atoms are found in the T 1 o site (Fig. 4).These potassium feldspars have high triclinicity (0.81-0.99;Table 1) and correspond to the highly ordered microcline (low microcline) (SMITH, 1974;GRIFFEN, 1992).Al occupancy in the T 1 o site of Pakra Creek valley granitoid feldspars are between 0.53 and 0.62 (Table 2, Fig. 4); their triclinicity is low (around 0.30 or even 0.0) (Table 1) and they correspond to intermediate microcline or orthoclase.Brzaja Creek valley gneiss feldspars, excluding low microcline in PPM-3 sample, with triclinicity value Δ = 0 revealed to be orthoclase.Classical and diffraction patterns calculated by the Rietveld refi nement method gave comparable results.Obtaining correct cell parameters depends primarily upon accurate measurement of the peak positions and the correct indexing of X-ray powder patterns.
Intermediate microcline or orthoclase from biotite-granodiorites and monzogranites compared with highly ordered, structured, potassium feldspars from two-mica monzogranites indicate differences in the rate of cooling, as the most important factor controlling the ordering of potassium feld-spars.The two-mica granitoids as host rocks, have a eutectic composition and peraluminous character.They are syn-collision granitoids according to HORVAT (2004), HORVAT & BUDA (2004) and their overall triclinic symmetry implies a slow cooling-rate and higher degree of order with the aid of fl uid activity or deformation in the case of the gneiss (BROWN & PARSONS, 1989).
Porphyric biotite-granodiorite and monzogranite have a peraluminous-metaluminous character and were probably formed post-collision in an uplifted environment (HORVAT, 2004;HORVAT & BUDA, 2004).They were emplaced at shallower levels in the crust; thus cooling was relatively fast which prevented the symmetry inversion and retarded ordering.Rapid cooling seems to be the most probable explanation for preservation of monoclinic structures and the moderate degree of Al-Si ordering of potassium feldspars in these rocks.
Estimation of three different methods for K-feldspar characterization: triclinicity calculations according to GOLD-SCHMIDT & LAVES (1954), structural state determination by the method of KROLL & RIBBE (1983) and the ordering index method introduced by GODINHO & JALECO (1973) once again proved the very good correlation of results obtained by these methods.This fact gives greater weight to conclusions arising from them i.e. makes them more reliable.Simple and relatively fast standard methods for potassium feldspar determination and description produced results that correlate well with those obtained by more accurate but more time-consuming methods.
Triclinicity (Δ) calculations showed that the majority of the potassium feldspars occurring in the Papuk granitoids have values higher than 0.70 on the scale, where a microcline with maximum triclinicity has value of 1.The exception is the intermediate microcline modifi cation that occurs in granodiorites of the Pakra Creek valley with triclinicity around or < 0.50.The lowest value is characteristic for the porphyric granodiorite of Pakra Creek valley (Δ=0.29).The highest values (Δ >0.90) are obtained on gneisses (Table 1).Pairs of triclinicity values obtained by two methods, classical (1) and Rietveld (2) are as follows: 0.81-0.96for Brzaja Creek twomica monzogranite, 0.29-0.33 and 0.32-0.29 for Pakra Creek porphyric granodiorites and 0.70-0.72 for Pakra Creek twomica monzogranite (Tab. 1 and Fig. 2).

Figure 3 :
Figure 3: Observed and calculated diff raction pattern (plus signs and line, respectively) of 2PPG-5 sample feldspar megacryst.The lower curve shows the diff erence between the observed and calculated patterns.Tick marks indicate the positions of the allowed refl ections of microcline (upper) and albite (lower).Calculated triclinicity (Δ) is shown inTable 1 (last column).
If the crystal is completely ordered, in the unit cell there will be 1.0 Al + 3.0 Si along b (KROLL & RIBBE, 1983).During the ordering process, T-O bond lengths vary and O-T-O and T-O-T bond angles are affected (KROLL, 1973).The T-O bond lengths refl ect Al occupancy; therefore the Al occupancy can be estimated by calculating T-O bond lengths.An increase in aluminium occupancy causes an increase in the mean bond length, whereas an increase in silicon occupancy causes their decrease (KROLL, 1973).During Al-Si ordering, triclinic feldspars expand along [110] and contract along [11 ¯0].The translation tr[110] estimates t 1 o+t 2 o+t 2 m and tr [11 ¯0] estimates t 1 m+t 2 o+t 2 m.Because the same amount of Al and Si atoms is present in the unit cell regardless of the structural state (GRIFFEN, 1992), unit cell volume (V) should be a function of composition but not of structural state.If tr[11 ¯0] is plotted versus cell volume (V), t 1 o can be estimated directly from the diagram of KROLL
Results obtained byGOLDSCHMIDT & LAVES (1954),KROLL & RIBBE (1983) andNEVES & GODINHO (1995) methods indicate variation in the structural state, from orthoclase, intermediate microcline to highly ordered microcline in the investigated rock samples.High triclinicity values of feldspars from granitoid and gneiss samples from Pa puk Mt. (Slavonia, Croatia) are in accordance with high Al contents in the T 1 o site and their fully ordered state indicate a slow(er) cooling-rate.Low triclinicity values, Al content in T 1 o site of around 0.60, and an ordering index smaller than 0.80 can be interpreted as a result of relatively fast(er) cooling which allowed the existence of less ordered potassium feldspar.