Date of Award
5-14-2024
Degree Name
Doctor of Philosophy (PhD)
Degree Type
Dissertation
Abstract
The transition from a liquid to a crystal or a glass is a fundamental process that determines many properties of the final product. Although numerous experiments have been conducted to investigate this transition, the atomistic scale of such behavior remains elusive. The properties of a high-temperature liquid could play an important role during solidification. For example, it has been argued that the ordering of the structure factor in a liquid reveals the glass-forming ability. In this dissertation, computational methods were used to study the structure, dynamics, nucleation properties, and glass-forming ability of metallic liquids and silica glasses. Several Embedded Atom Method (EAM) potentials were investigated and compared with experimental results. The motion of atoms during nucleation was studied, and a unique behavior, namely cooperative motion, was found. The structure factor of metallic liquids was studied and compared with experimental results. A novel technique combining molecular dynamics and reverse Monte Carlo was used to investigate the structural evolution of BaO·2SiO2 glass and the development of machine learning neural network potential. In total, six projects are presented in this dissertation. In the first project, the CuZrAl EAM potential was carefully examined against our experimental results such as the structure factor, radial pair distribution, density, glass transition temperature, and cooperative temperature. Excellent agreement was found. Subsequent dynamical studies of the viscosity and the diffusion coefficient revealed that the breakdown of the Stokes-Einstein relation was caused by the decoupling of the copper element. In the second project, the EAM potential of PdSi and the Angular Dependent Potentials (ADP) of AuSi were compared to experimental data. The correspondence between the experimental data and the EAM results is poor in Pd82Si18 and extremely poor in Au81Si19. A Curie-Weiss type behavior was found in the first peak height of x-ray structure factor in Pd77Cu6Si17 metallic liquids. In the third project, the motion of atoms during attachment and detachment in the nucleation process was carefully studied. It was found that attachment or detachment was achieved via a group of cooperative atoms instead of the individually diffusive type of jump assumed in the Classical Nucleation Theory (CNT). This behavior was found for the first time. In the fourth project, the nucleation parameters obtained from CNT and the Mean First Passage Time (MFPT) were compared. It was found that the CNT results had a better agreement with the experimental data than the MFPT results. The cooperative motion was investigated with the kinetic properties and the local potential energy. Additionally, the cooperative motion was found in the non-classical nucleation pathway, namely coalescence, suggesting that it might be a pervasive phenomenon in the metallic systems. In the fifth project, the structure of the BaO·2SiO2 glass was studied via X-ray and neutron scattering measurement. It was found that the first peak height of the X-ray structure factor increased with increasing annealing time, while the neutron structure factor did not. The Force-Enhanced Atomic Refinement (FEAR), which combines molecular dynamics and reverse Monte Carlo, was used to investigate this phenomenon. The higher mobility of the Ba atoms indicated that the ordering found from X-ray scattering results is due to the high mobility of Ba as it is the dominant source. The ring statistics revealed that the smaller ring size tend to align with the crystal during the annealing process. Additionally, a machine learning neural network potential for PdSi was proposed. Instead of the general input dataset obtained from density functional theory or molecular dynamics, the dataset fitting from experiments using FEAR was also provided, which increased the quality of this potential. In the sixth project, inelastic neutron scattering measurements were made of Cu49Zr45Al6 metallic liquids. The distinct Van Hove function time was obtained and combined with molecular dynamics simulations, demonstrating that the local excitation of the liquid above the cooperative temperature governs the viscosity. This insight highlights the critical role of thermal fluctuations and atomic-level interactions in determining the macroscopic properties of metallic liquids, providing a deeper understanding of the mechanisms driving viscosity and glass formation.
Language
English (en)
Chair
Kenneth Kelton