Crystal structure and crystal chemistry of fluorannite and its relationships to annite

Abstract This study focuses on the crystal-chemical characterization of fluorannite from the Katugin Ta-Nb deposit, Chitinskaya Oblast’, Kalar Range, Transbaikalia, eastern Siberia, Russia. The chemical formula of this mineral is (K0.960Na0.020Ba0.001)(Fe2+ 2.102Fe3+ 0.425Cr3+ 0.002Mg0.039Li0.085Ti0.210Mn0.057)(Al0.674Si3.326) O10(F1.060OH0.028O0.912). This mica belongs to the 1M polytype (space group C2/m) with layer parameters a = 5.3454(2) Å, b = 9.2607(4) Å, c = 10.2040(5) Å, b = 100.169(3)º. Structure refinement, using anisotropic displacement parameters, converged at R = 0.0384. When compared to annite, fluorannite shows a smaller cell volume (Vfluorannite = 497.19 Å3; Vannite = 505.71 Å3), because of its smaller lateral dimensions and c parameter. Flattening in the plane of the tetrahedral basal oxygen atoms decreases with F content, together with the A–O4 distance (i.e. the distance between interlayer cation A and the octahedral anionic position) due to the reduced repulsion between the interlayer cation and the anion sited in O4.

Natural trioctahedral micas were investigated intensively by single-crystal X-ray diffraction (XRD) methods (Brigatti and Guggenheim, 2002); however, studies on Fe-rich micas are rare. At present, the Fe-bearing micas closest to the annite ideal composition were studied by Brigatti et al. (2000a), who characterized an annite from Pikes Peak (Colorado), and by Redhammer and Roth (2002), who provided crystal structure refinements for an annite from Mont Saint-Hilaire (Québec), first addressed by Lalonde et al. (1996), with relevant tetrahedral Fe 3+ content. Studies on other trioctahedral micas with compositions close to annite, but with relevant Al contents, were also reported by Brigatti et al. (2000b). In addition to natural trioctahedral micas, Roth (2002, 2004) studied a variety of synthetic Fe-rich samples with various octahedral contents, but containing no fluorine. Other structural studies of Fe, OH-rich micas were carried out for Cs-tetraferri-annite (Mellini et al., 1996;Comodi et al., 1999) and for Rb-tetra-ferri-annite (Comodi et al., 2003).
This study attempts to: (1) characterize the crystal structure of F-rich annite from the Katugin Ta-Nb deposit, Russia; (2) compare its crystal chemistry with that of OH-rich annite from the Pikes Peak complex, Colorado, USA; and (3) verify the influence of the F-for-OH substitution on the layer.

Samples
The studied fluorannite was found in the Katugin deposit, Chitinskaya Oblast', Kalar Range, Transbaikalia, eastern Siberia, Russia. The deposit consists of a Ta-, Nb-and REE-bearing alkaline metasomatite, which replaces multistage alkali-REE granites and host-rocks that are generally composed of marble, gneiss or amphibolite. The deposit comprises fine-and mediumgrained quartz-albite-microcline rocks. Ta-Ni minerals, zircon and thorite are widespread along with REE minerals (e.g. gagarinite, yttrofluorite, monazite, bastnäsite and xenotime). The depositional environment consists of deposithosting intrusions and metasomatic deposits which occur along major shear zones connected to intraplate and continental-margin rift and strike-slip faults (Solodov et al., 1987;Kremenetsky et al., 2000). The annite from the Pikes Peak batholith (central Colorado) was used to assess the influence of the anionic position on layer topology. Although the chemical and structural data for this sample had been previously reported (Brigatti et al., 2000a), we collected all the data again using the same experimental set-up used to characterize fluorannite.

Chemical composition
Major-element composition of the crystals used for structure refinement were obtained by wavelength-dispersive spectroscopic (WDS) methods using an ARL-SEMQ electron microprobe (EMP) at the Dipartimento di Scienze della Terra, Università di Modena e Reggio Emilia. Analyses were performed with a 15 kV accelerating voltage, 15 nA beam current and a 5À10 mm beam-spot diameter. Different counting times were employed, both at peak and background positions: 10 s for Na, 15 s for K, Si, Fe, Mn, Al, Mg, Ca and Ti, and 40 s for F, as suggested by Signorelli et al. (1999). The following standards were used: fluorite (F); microcline (K); albite (Na); spessartine (Al,Mn); ilmenite (Fe,Ti); clinopyroxene (Si); olivine (Mg). Analysis and data reduction were performed using the Probe software package of Donovan (1995). The (OH) À content of annite was measured by thermogravimetric analysis (TGA) in He gas flow using a Seiko SSC 5200 thermal analyser (heating rate 10ºC/min; gas-flow rate 100 ml/min), equipped with a mass spectrometer (GeneSys ESS, Quadstar 422). A similar approach was unsuitable for fluorannite due to the limited amount of available material.
The compositions reported in Table 1 were obtained by combining the average of at least seven EMP point-analyses with information related to single-crystal structure refinement (i.e. comparing the mean electron count of interlayer and octahedral cation sites obtained from chemical analysis with those from single-crystal structure refinement). Further information taken into consideration for the derivation of the chemical formula includes: (1) (OH) À determination on several annite crystals selected from the same sample containing the crystal used for structure refinement; and (2) information related to Fe-oxidation state as indicated for fluorannite by Shen et al. (2002). The chemical formula is based on O 12ÀxÀyÀz (OH) x F y Cl z .

XRD analysis
Small crystal fragments of fluorannite from the Katugin deposit (crystal size: 0.130 mm6 0.105 mm60.015 mm) and annite from the P i k e s P e a k b a t h o l i t h ( 0 . 1 2 5 m m 6 0.100 mm60.010 mm) were analysed using a Bruker AXS X8 APEX automated diffractometer with a four circle Kappa goniometer, an APEX 4K CCD detector, flat graphite monochromator and Mo-Ka-radiation (l = 0.71073 Å ) from a fine focus sealed tube. Three sets of 12 frames were used for the initial unit-cell determination; each frame measured with 0.5º j rotation and 10 s exposure time. The crystal-detector distance was 40 mm and the collection strategy was optimized by the APEX program suite (Bruker, 2003a). The refined cell parameters are a = 5.3454(2) Å , b = 9.2607(4) Å , c = 10.2040(5) Å , b = 100.169(3)º, V = 497.19 Å 3 for fluorannite, and a = 5.3841(1) Å , b = 9.3259(3) Å , c = 10.2549(3) Å , b = 100.851(1)º, V = 505.71 Å 3 for annite. The whole Ewald sphere (Ô9 h, Ô16 k, Ô18 l) was recorded in the range 4º < y < 41º. A total of 6217 reflections (unique reflections: 1559; R int : 0.0386) and 5685 reflections (unique reflections: 1359; R int : 0.0303) were collected for fluorannite and annite, respectively. A semi-empirical absorption collection based on the intensity of equivalent reflections was applied using the SADABS software (Sheldrick, 2003). The SAINT-IRIX (Bruker, 2003b) package was used for data reduction and unit-cell refinement. Anisotropic crystal-structure refinement was carried out using the SHELX-97 package of programs (Sheldrick, 1997) in the monoclinic space group C2/m with neutral atomic scattering factors and starting from the previously determined atomic coordinates of annite (Brigatti et al., 2000a). These methods finally led to the positional and displacement parameters for all atoms reported in Table 2. Ionized X-ray scattering curves were employed for non-tetrahedral cations, whereas ionized vs. neutral species were used for Si and O (Hawthorne et al., 1995). The final refinement yielded the following agreement factors: R = 0.0384 and R = 0.0398 for fluorannite and annite, respectively. A final calculated difference electron density (DED) map did not reveal significant excess in electron density above the background. Table 3 reports relevant cation-anion bond lengths, the mean electron count at the octahedral and interlayer sites, and selected parameters derived from structure refinement. The observed and calculated structure factors can be requested directly from the authors.  (Table 1).
When compared to annite, fluorannite shows a smaller cell volume (V fluorannite = 497.19 Å 3 ; V annite = 505.71 Å 3 ), because of its smaller lateral dimensions and c parameter (c = 10.2040(5) Å and c = 10.2549(3) Å for fluorannite and annite, respectively). According to Robert et al. (1993), F enters the mica structure in the octahedral anionic site substituting for OH. Hence the reduction in the c parameter can be attributed to the electrostatic interaction of the interlayer cation on the anionic site, which is obviously greater when F substitutes for the OH group. This effect is particularly clear in trioctahedral micas, where OH points directly towards the interlayer cation, rather than being inclined towards the empty octahedral site, as in dioctahedral micas. The interlayer separation also decreases: from 3.375 Å in annite to 3.316 Å in fluorannite (Table 3).
In fluorannite, Si and Al occupy the tetrahedral site in a ratio Si/(Si+Al) = 0.83. The site volume is 2.307(1) A 3 . The hTÀOi mean bond distance (hTÀOi = 1.651 Å ) is slightly shorter than in annite (hTÀOi = 1.660 Å ), but very close to the value measured for fluorphlogopite (hTÀOi = 1.648 Å , Gianfagna et al., 2007). The tetrahedral cation is displaced towards the apical oxygen atom, thus also accounting for a slightly smaller tetrahedralcation-apical-oxygen-atom distance than the tetrahedral-cation-basal-oxygen-atom distances (Table 3). A further consequence is the increase of the O basal -TÀO apical angles (i.e. t parameter = 110.97º). Furthermore, the basal tetrahedral area in fluorannite (3.094 Å 2 ) is appreciably smaller than in annite (3.141 Å 2 ). All these effects can also be related to the large Si content, which, together with the high-charge octahedral cations, contributes to charge-balance the anionic O-for-OH substitutions.
Another effect associated with the same crystal-chemical mechanism occurs in the octahedral sites; the angles formed by [001] with M1ÀO4 and M2ÀO4 bonds [M2ÀO4 [001] ] vary as a function of F content (Fig. 1c). Figure 1d introduces the variation of the angle defined by the octahedral position M2, the anionic position O4 and the other symmetry-dependent M2 cation M2 ÀÔ4 O4 À M2. This angle, which is also observed to decrease with tetrahedral Si content and to increase with octahedral Al content, reflects structural modification associated with AÀO4 distance, which increases with the angle and decreases with the octahedral thickness, as expected. Consequently the F-for-OH substitution, which directly leads to a decrease in the AÀO4 distance (Fig. 1b) is also connected to an increase in octahedral thickness, which is inversely related to AÀO4 (Fig. 2a). This latter effect is consistent with the limited Al content observed in fluorannite and annite, with respect to the annite crystals from peraluminous granites used for comparison. Also the variance of octahedral unshared edges [s(O3ÀO3)], directly related to AÀO4 (Fig. 2b), reflects octahedral heterovalent substitutions (Brigatti and Guggenheim, 2002) and, in our case, mostly Li, Fe 3+ and Al 3+ octahedral contents. Octahedral chemistry plays a significant role in determining octahedral and interlayer structural parameters. In particular, the octahedral Al content could be related to the variance of the distances between the octahedral cations and oxygen atoms [s(MÀO)], with a positive correlation. This evidence further emphasizes the homo-octahedral character of fluorannite, which is octahedral-Al free.
All the observed trends confirm the influence of F content on the overall layer topology, with particular significance for the structural parameters measured along [001]. Furthermore, F, as previously observed, seems to effectively stabilize the trioctahedral structure of the mica, as confirmed by the almost-complete homo-octahe-dral character of fluorannite, as well as promote populations of large octahedral cations, consistently with an increase in octahedral thickness.