A probable crystallographic path for the thermal phase transitions in single crystals of KNO₃

A probable crystallographic path for the thermal cycle of phase transformations, II-I-III-II, in KNO₃ single crystals is proposed. It is based on the symmetry of the phases and a least-motion hypothesis that the 'shuffles' of the transformations (so remarkably like martensitic transformations in metals) are the minimum possible readjustments subject to accepted van der Waal distances between chemically non-bonded atoms. Twinning is adequately explained by the alternative paths provided by the symmetry of the parent phase. The observed disorientation of 2°9' in the c(II) axes of the end-phase twinned crystals is calculable by the well known Bowles-Mackenzie matrix method. The calculated value is now found to be 2.5°. Crystallographically an important possibility emerges: the structure of the higher-symmetry phase of a transformation may be predicted from only its unit-cell dimensions and space group if the crystal structure of the low-symmetry phase is known.