HT-study of the P31c ↔ P63 phase transition in kalsilite, KAlSiO4

Kalsilite, KAlSiO4, occurs as different polymorphs in K-rich silica undersaturated volcanic rocks, and in metamorphic rocks. The low-kalsilite form has space group P63. The structure is an ordered three-dimensional framework of AlO4 and SiO4 tetrahedra forming six-membered rings pointing up (U) and down (D) alternately, in a tridymite-like framework topology (UDUDUD). The rings are di-trigonally distorted, and are stacked along the c direction, sharing the apical O atoms in a staggered configuration [1]. Na-free metamorphic twinned crystal of kalsilite from Punalur district in Kerala, India, has having P31c symmetry [2]: individual sheets of P31c kalsilite are essentially the same as those of the P63 structure but are stacked in an eclipsed manner, with the di-trigonal rings in succeding sheets point in the same directions. Annealing of P31c kalsilite results in the P63 form starting at 473 K and is complete at 773 K [3]. The transition is described as being irreversible and proceeds by tetrahedral rotation such that successive (001) sheets undergo opposite-sense rotations [3]. The process was previously studied by means of ex-situ powder X-ray diffraction of annealed samples. The progress of the P31c ↔ P63 transformation was followed [3] by measuring the intensities of the 111 and 112 reflections: the intensities of reflections hhl type should be sensitive to the structural change, because those with l = even (a-type) would be expected to decrease in intensity, whereas those with l = odd (b-type) should increase from zero (hhl, l = odd are systematically extinct in space group P31c). The presence of the P63 phase was also confirmed by Rietveld refinement and after complete transformation annealing at 773 K for 14 days, by single crystal X-ray diffraction (SCXRD) [3]. We have performed an in-situ high-T SCXRD study of twinned metamorphic kalsilite from the same sample studied by [2] and [3], with the aim of following the structure transformation in-situ and of testing reversibility. We have collected data on lattice parameters each 25 °C steps up to 923 K and the intensity of the (11l) reflections (with l = 1-6). Complete intensity data collections were made at 298, 373, 473, 573, 673, 773, 873 and then at room-T after cooling. Upon heating, the a cell parameter and cell volume increased up to 923 K, while the c lattice decreases showing a faint but neat collapse of ~1 ‰ at 473 K. This feature is accompanied by the increase of intensity of b-type reflections that were extinct, while a-type reflections become progressively weaker, and thus corresponds with the onset of the phase transformation. Structure refinement at 473 K confirmed P63 symmetry and the absence of twinning. The process saturates at 573 K [Ib/Ia=constant] and over such temperature both groups of reflections keep decreasing intensity keeping their averaged ratio constant, while structure keeps P63 symmetry. We ended the annealing experiment at 923 K in order to avoid the P63 ↔ P63mc phase transition [4]. On cooling the intensity of the b-type reflections increases down to 573 K. At that temperature we switched off the furnace to quench the P63 structural state. We collected intensity data during cooling and recorded a gradual decrease of b-type reflection intensity while the c lattice parameter slightly increased. Structure refinement showed the inversion of the structure back to P31c space group, with {0001} twining again present in the same ratio as before annealing. The transformation is accompanied by a change in the bonding scheme of the K atoms while no disorder of Al and Si is observed in the tetrahedra. References. [1] Perrotta, A.J., Smith, J.V. (1965): Min Mag, 35, 588–595. [2] Cellai, D. Bonazzi, P., Carpenter, M.A. (1997): Am Mineral, 82, 276–279; [3] Cellai, D., Gesing, T.M., Wruck, B., Carpenter, M.A (1999): Am Mineral, 84, 1950–1955; [4] Carpenter, M.A., Cellai, D. (1996) Am Mineral, 81, 561–584. This work was supported in part by NSF grant EAR 0738692 to NL Ross and RJ Angel