Ferroelastic phase transition in cryolite, Na3AlF6, a mixed fluoride perovskite: High temperature single crystal X-ray diffraction study and symmetry analysis of the transition mechanism

Abstract

Abstract. Cryolite, Na3A1F6[=2Na+(Nag.sAlg.~)F3] is a mixed fluoride perovskite, in which the corner-sharing octahedral framework is formed by alternating [NaF6] and [A1F6] octahedra and the cavities are occupied by Na § ions. At 295 K, it is monoclinic (~ phase), space group P21/n with a=5.4139(7), b=5.6012(5) and c = 7.7769 (8) A and fl = 90.183 (3), Z = 2. A high temperature single crystal X-ray diffraction study in the range 295-900 K indicates a fluctuation-induced first-order phase transition from monoclinic to orthorhombic symmetry at TO~885 K, in contrast to a previous report that it becomes cubic at ,-~ 823 K. The space group of the high temperature 1~ phase is Immm with a = 5.632 (4), b= 5.627 (3) and c= 7.958 (4)/~, Z=2 at 890 K. Above To, the coordination number of the Na § ion in the cavity increases from eight to twelve and the zigzag Nal-A1 octahedral chains parallel to c become straight with the Nal -F--A1 angle = 180 ~ The phase transition is driven by two coupled primary order parameters. The first corresponds to the rotation of the nearly rigid [A1F6] group and transforms according to the F4 + irreducible representation of Immm. Coupled to the [A1F6] rotation is a second primary order parameter corresponding to the displacement of the Na2 + ion in the cavity from its equilibrium position. This order parameter transforms according to the X~ irreducible representation of Immm. Following Immm ~ P21/n phase transition, four equivalent domains of P21/n are determined relative to Immm, which are in an antiphase and/or twin relationship. The abrupt shortening of the octahedral A1-F and Na--F bonds and a sudden change in orientations of the atomic thermal vibration ellipsoids above To indicate a crossover from displacive to an order-disorder mechanism near the transition temperature. The fl phase is interpreted as a dynamic average of four micro-twin and -antiphase domains of the ~ phase. This view is consistent with the entropy of phase transition, ASt .... (11.43 JK-1 mol-1) calculated from heat capacity measurements (Anovitz et al. 1987), which corresponds closely to R ln4 (11.53 JK -1 tool-l), where 4 is the number of domains formed during the phase transition. The dynamic nature of the/~ phase is independently confirmed from a considerable narrowing of the 27A1 nuclear magnetic resonance (NMR) line-shape above To .

High temperature powder diffraction and solid state DFT study of β‐cryolite (Na3AlF6)

“…Nowadays it is however accepted, that the temperature of the cryolite phase transition is 563±2 °C [3], the value being also derived from the thermodynamic measurements. Interestingly, the transition temperatures determined in the powder diffraction studies are close to the recent [3] estimate, ≈ 550 °C [4] and 567 °C [5], respectively, while the value (612 °C) found in the variable temperature single crystal study of a natural cryolite [6] does not agree with any of those established from the thermodynamic experiments [1,3]. Fig.…”

Fast ionic mobility in cryolite studied by quasielastic neutron scattering

“…At low temperatures, cryolite has a monoclinic crystal structure (P2 1 /n). Several high temperature x-ray diffraction experiments showed a transition from the monoclinic to a cubic (F m3m) [8,9] or orthorhombic (Immm) [10] structure around T = 567 • C. This phase transition is accompanied by a sudden increase in the electrical conductivity. A second jump in the conductivity is observed at T = 880 • C, well below the melting point (T m = 1013 • C) [11].…”

Combined control of aluminum bath composition by X‐ray diffraction and X‐ray fluorescence analysis

“…In the cases when alkali metals fluorides are supplemented, it is helpful to use an additional composition characteristic to control electrolysis process (2):…”

“…Crystal structures of a number of bath mineral phases vary because of nonequilibrium crystallization during sampling. [2] This significantly declines the accuracy of CR analysis. For baths that do not contain magnesium fluoride, potassium fluoride and lithium fluoride, the actual accuracy is ΔCR(p = 0.95)~0.04.…”