Date of Award

Spring 2021

Degree Type


Degree Name

Doctor of Philosophy (PhD)


Mathematics and Natural Sciences

Committee Chair

Parthapratim Biswas

Committee Chair School

Mathematics and Natural Sciences

Committee Member 2

Christopher Winstead

Committee Member 2 School

Mathematics and Natural Sciences

Committee Member 3

Khin Maung Maung

Committee Member 3 School

Mathematics and Natural Sciences

Committee Member 4

Sungwook Lee

Committee Member 4 School

Mathematics and Natural Sciences

Committee Member 5

Andrew Sung

Committee Member 5 School

Computing Sciences and Computer Engineering


While conventional approaches to materials modeling made significant contributions and advanced our understanding of materials properties in the past decades, these approaches often cannot be applied to disordered materials (e.g., glasses) for which accurate total-energy functionals or forces are either not available or it is infeasible to employ due to computational complexities associated with modeling disordered solids in the absence of translational symmetry. In this dissertation, a number of information-driven probabilistic methods were developed for the structural determination of a range of materials including disordered solids to transition metal clusters. The ground-state structures of transition-metal clusters of iron, nickel, and copper were determined by a force-biased Monte Carlo method and their structural and electronic properties were studied comparatively via force-biased Monte Carlo and ab initio simulations. The force-biased Monte Carlo approach has shown unambiguously that it can effectively determine the putative ground-state structures of a number of small transition-metal clusters.

For complex amorphous materials, an information-driven probabilistic viewpoint was adopted by posing structural determination of disordered solids as an inferential program and the problem of materials design was addressed as an optimization program, jointly supported by experimental data and information. The hallmark of this new approach is that it can produce atomistic configurations of amorphous solids, which are thermodynamically stable and close to a stable local minimum of a quantum-mechanical total-energy functional. The models have structural, topological, electronic, and vibrational properties comparable to experiments. The data-driven approach presented here for amorphous solids not only can produce overall structural and electronic properties but also the microstructural properties of realistic samples from experiments, such as voids and vacancy-type defects, which cannot be addressed directly using currently available computational methods. Ab initio hydrogen dynamics were simulated inside nanometer-size voids in a-Si within the framework of the density-functional theory and the study revealed that the microstructure of the hydrogen distribution and the morphology of the voids were characterized by the presence of a significant number of monohydride Si–H bonds, along with a few dihydride Si–H2 configurations but not any isolated hydrogen. The study also revealed that a considerable number of total H atoms inside voids can appear as H2 molecules. The densities of the bonded and nonbonded hydrogens are observed to be consistent with those from the infrared and Rutherford backscattering spectrometry measurements.