|Institution:||University of Pittsburgh|
|Full text PDF:||http://d-scholarship.pitt.edu/29282/1/xun-formatted_1.pdf|
Water is the most important liquid on the earth, yet the physics behind many properties of water is still poorly understood. Particular interesting are the condensed phases of water: ice and liquid water, that both possess anomalous properties. In this thesis, I focus on using different methods, including force fields as well as DFT calculations, to predict several key properties of proton-disordered ice Ih and liquid water. The focus of my study of ice Ih is on its comparable dielectric constant ε_s with liquid water, which directly results from its proton-disorder nature. Predictions of the dielectric constant of ice Ih from pairwise additive force fields fall appreciably below than the experimental values, with significant improvement being achieved by polarizable force fields. I examined the performance of different force fields, and confirmed that the polarizable AMOEBA models with three polarizable sites per molecule outperform polarizable models such as DC97 with a single polarizable site. Since it is difficult to resolve the subtle energetic difference of different proton ordering arrangements in ice Ih with force fields, I studied the energetics of ice Ih, from DFT calculations using the BLYP functional as well as several dispersion-corrected BLYP functionals. As shown in my study, the dispersion-corrected functionals not only give better energy predictions but also get better lattice parameters and equilibrium volumes for the optimized ice Ih unit cells. I also predicted the structural as well as dynamical properties of liquid water from ab initio molecular dynamics simulations with several dispersion-corrected BLYP functionals. The results of calculations all confirmed that including dispersion corrections in functionals is essential to get faster water rotational dynamics and a “softer” structure of liquid water. Finally I parametrized a two-channel dispersion-corrected atom-centered pseudopotential (DCACP2) based on the BLYP functional to correct for the long-range dispersion force for three rare gas elements: helium, neon and argon. By fitting the interaction energy of three homonuclear dimers against CCSD(T) calculations, the resulting DCACP2-BLYP method performs significantly better than the one-channel DCACP approach for the two-body binding energies of the dimers. I also explore the factor responsible for the success of DCACP2 method.