Relationships between MAS NMR and Mineral Structure

Abstract

<p>²⁹Si magic angle spinning nuclear magnetic resonance (MAS nmr) has been used extensively to study the structure, properties, and reactions of silicate minerals, ceramics and glasses. Despite many attempts to relate chemical shift with crystal structure the factors governing ²⁹Si chemical shift are not weII understood.</p> <p>This study investigates the effect of local structural environment of the silicate tetrahedron on ²⁹Si chemical shift. A simple relationship is found between the atomic positions of the cations bonded to the terminal oxygens of the tetrahedron and ²⁹Si chemical shift, which can be used to calculate chemical shift from atomic positions. A plot of calculated against experimental chemical shift has a correlation coefficient of 0.986 for 124 sets of date from all groups of silicates.</p> <p>²⁹Si chemical shifts calculated from diffraction data, are used to interpret MAS nmr spectra of scapolite, a partially disordered mineral system. Models are devised for cation ordering throughout the solid solution series, in which the charge on the cavities are balanced locally by the requisite number of AlO₄ tetrahedra.</p> <p>²⁹Si, ²⁷AI, ²³Na, ⁷Li and ⁹Be MAS nmr spectroscopy is used to investigate the difference in atomic sites between beryls with slightly different compositions. ⁹Be and ²⁹Si MAS nmr results agree with the presence of domains in the Li-Cs beryls with the structure of low-alkali beryls. Lack of a tetrahedral peak in the ²⁷Al spectra of Li-Cs beryls eliminates the coupled substitution of Li for Al and Al for Be. Peaks in the ²³Na spectra of Li-Cs beryls from NaCI in solution in fluid inclusions show the potential of nmr to study fluid inclusions.</p> <p>Models of theoretical structures based on a symmetrical SiO₄ tetrahedron with AI or Si ligands show that chemical shift is linearly proportional to cation-oxygen distance (r) and exponentially proportional to silicon-oxygen-cation angle (α). The possible ranges of chemical shift for each type of silicate are more extensive than those of Lippmaa et al. (1980).</p>