Nuclear-magnetic-resonance studies and molecular-dynamics simulations of the structure of sodium fluoroberyllate glasses: Evidence of non-tetrahedrally-coordinated beryllium

${}_{}{}^9{}_{}{}^{}\mathrm{Be}$ and ${}_{}{}^{19}{}_{}{}^{}\mathrm{F}$ NMR techniques are used to gain information about the short-range order in two beryllium fluoride glasses: A simple ${\mathrm{BeF}}_2$ glass and a NaF⋅${2\mathrm{B}\mathrm{e}\mathrm{F}}_2$ glass. The magnitude of the average quadrupole coupling constant for ${}_{}{}^9{}_{}{}^{}\mathrm{Be}$ sites is deduced from the location of the first-order satellites and from low-frequency second-order broadening of the central transition of the ${}_{}{}^9{}_{}{}^{}\mathrm{Be}$ spectrum. The average quadrupole coupling constant for the binary glass was a factor of 4 larger than for ${\mathrm{BeF}}_2$ glass. It is argued that this is consistent with the existence of non-tetrahedrally-coordinated Be ions in the sodium fluoroberyllate glass. The central transition of the ${}_{}{}^9{}_{}{}^{}\mathrm{Be}$ NMR spectrum for the nominally pure ${\mathrm{BeF}}_2$ glass exhibited an anomalous narrow line which may be due to the presence of Be metallic clusters. The linewidth of the ${}_{}{}^{19}{}_{}{}^{}\mathrm{F}$ resonance and the ${}_{}{}^{19}{}_{}{}^{}\mathrm{F}$ spin-lattice relaxation time, $T_1$, of the two glasses were also studied as a function of temperature. These measurements indicate that there is some motion of the fluorines in the binary sample slightly above its transition temperature of 388 K. The two glasses were simulated by methods of molecular dynamics. It was found that whereas in simple ${\mathrm{BeF}}_2$ glass virtually all of the Be ions were tetrahedrally coordinated by four F ions, in multicomponent glasses a large number of Be ions have five F ions in the first coordination shell. This is qualitatively consistent with the experimental observations. Meaningful quantitative predictions of the effective ${}_{}{}^9{}_{}{}^{}\mathrm{Be}$ quadrupole coupling constants for the two glasses failed, however, because the simulated glasses are more disordered than glasses prepared in the laboratory. There is a broad distribution of F–Be–F angles in the simulated ${\mathrm{BeF}}_2$ glass. The calculated electric field gradients at these nonideal tetrahedral sites were comparable to those for the non–fourfold-coordinate Be sites in the alkali fluoroberyllate glass. The limitations of molecular-dynamics simulations for providing angular distributions of ions are discussed.