Download files
Abstract
This thesis presents the details of a blended quantum molecular dynamics scheme, utilizing both the canonical (NVT) and isobaric-isothermal (NPT) ensembles, where the number of atoms (N), temperature (T) and either volume (V) or pressure (P) are considered constant. These two (2) schemes have been combined in order to not only circumvent the need for any experimental input data ordinarily required for simulations involving the strictly canonical ensemble, but also minimize the relatively high extra computational cost imposed by the increased cutoff energy necessary to avoid the so-called Pulay stress error while performing isobaric-isothermal ab initio molecular dynamics simulations using a plane-wave basis set.
We compare the results of the blended and canonical schemes via the simulated quench of the damage-tolerant Zr61Ti2Cu25Al12 (ZT1) alloy to below its glass transition temperature, and extend the demonstration of this newly proposed scheme via the simulated quench of an additional ten (10) closely-related alloys from the same ZrTiCuAl family.
There were subtle differences in structural evolution between the two schemes. Notably, the blended scheme generates a more efficiently packed structure, which is feasibly permitted due to the volume changes that transpire as a result of incorporating a parallel NPT component. Further, while the blended scheme obviates the need for any experimental
input data, it is shown that the starting volume is not particularly critical and the computed final density of the alloy is within 1% of the reported experimental values which are either 6.43 or 6.50g.cm-3, depending on the source.
Furthermore, the blended scheme demonstrated a greater degree of transformation in the coordination number compared with the canonical, which remains relatively static. The final value of 12.97 obtained from the blended scheme is closer to the ideal of 13.33 as per close-packing theory, a feature that appears to be related to the evolution of a more complex family of voronoi polyhedra relative to the icosahedral dominant motifs present in the canonical scheme.
Additionally, in both schemes, the partial coordination number in the Zr species, being the primary constituent, demonstrates a plateau in its evolution, but which commences in the blended scheme at a temperature approximately 300K closer to the reported glass transition temperature.
We have been able to calculate elastic constants with reasonable accuracy in line with QMD simulations that engage GGA/LDA, and have observed the signature of glass transition in our kinetic analysis. In our structural investigations, we have shown that the topological evolution is largely complete by the time the glass transition is reached. We have found that aluminium could possibly be used as a ’tuning’ tool to control the degree of disclinated five-fold symmetry via tetrahedral packing.
Our findings suggest that the blended scheme may be a more reliable and an optimally efficient method for implementing ab initio molecular dynamics in the simulation of complex alloys.