Abstract
The structural variation of monoclinic C2/m alkali feldspars has been analysed, based on 50 previously-determined structures and 10 new structure refinements including three determinations of the structure of an Or90 orthoclase at temperatures of 600 K, 900 K and 1075 K. The influence of temperature, composition and state of Al,Si order on the overall average tetrahedral bond lengths of these structures is statistically insignificant. Analysis of the structures in terms of the tilts of the tetrahedra shows that the tilts evolve uniformly with unit-cell volume irrespective of whether the volume is changed by temperature or exchange of the extra-framework cation. There is a small but systematic decrease in the distortion of the T1 tetrahedron with increasing unit-cell volume.
Comparisons of the refined structures with the results of geometric modelling show that volume changes are driven by exchange of the extra-framework cation, or by an increase in its thermal vibrations upon heating. The extreme anisotropy of the changes in the unit-cell parameters of monoclinic feldspars is not due to anisotropic interaction of the extra-framework cation with the anions of the framework, but due to the tilting of the tetrahedra controlled in part by O-O interactions. The anisotropy and tilting is not significantly modified by either Al,Si ordering or the distortions of the tetrahedra provided that the latter remain constant. The monoclinic feldspars show a slightly reduced anisotropy of strains as the unit-cell volume increases as a result of a decrease in the angular distortion of the T1 tetrahedron. The general conclusion is drawn that the pattern of anisotropy of the elastic properties (thermal expansion, compressibility, elastic compliances) and the cell-parameter changes of a tetrahedral framework structure with changing extra-framework species are intrinsic to the topology of the framework. The relative insensitivity of the anisotropy of the strains induced by volume changes to distortions of the tetrahedra also means that framework models which incorporate regular tetrahedra can be safely used to predict anisotropy, provided that the tetrahedral distortions do not change. If, as in the monoclinic feldspars, such a model does not reproduce exactly the observed anisotropy of the real structures, then this immediately indicates that there is a significant change in the distortion of the framework tetrahedra.