CrynodebWith modern high performance calculation, computer simulation provides us with a possible way to analyse and visualize the atoms and molecules of materials within nanometre scale and picoseconds which can not be simply achieved in real-life experiments. Therefore it is a great tool to study the nano-structure formation of materials including crystals as well as glasses.
In this thesis, alumino-silicates including nepheline and its compositional variations have been studied. Melts of them have been simulated and glasses have been derived by melt-quenching procedures. Static structures and dynamic properties have been studied. Specifically, the distribution of mobile metallic ions, which fundamentally affects the properties of glasses, has been simulated in this study to give a better understanding of the roles mobile ions play. In particular, we are interested in whether micro-segregation of metallic ions occurs as it does in silicates . Molecular dynamics simulations of alumino-silicates nepheline (KXNa4−XAl4Si4O16) have been performed to obtain
first the structure of molten nepheline. Five different alkali ratios have been taken to explore the effect of ionic radius and field strength on segregation. Radial Distribution functions, REDOR second moments, Vibrational Densities of States and molecular structure snapshots reveal how alkali ions distribute non-randomly, form channels in between the framework atoms of both molten and glassy nepheline. This nanostructure of nepheline at Tg=1200K is virtually the same as the quenched structure at room temperature, confirmed by the snapshots. When K+ replaces Na+ in the system, it occupies more volume. Both melt viscosity and boson peak intensity in quenched glass showed dramatic increasing due to the percolation channels affected by K+ concentration.
To replicate the dynamics properties of melts, three series of molecular dynamics simulations with varied alumino-silicate composition, each containing different ratios of Al/Si, have been performed including nepheline for which Al/Si = 1. Mean square displacements have been calculated from trajectories at 2000K to derive diffusion constant. The Einstein-Stokes equation and the Eyring equation were then used to derive the viscosity and compared. The Eyring results show a better replication of a fitted Adam-Gibbs model with measured viscosity values.
|Goruchwyliwr||Edwin Flikkema (Goruchwylydd) & George Greaves (Goruchwylydd)|