The Development of Rigorously Correct, Dynamical Pseudopotentials for Use in Mixed/Classical Molecular Dynamics Simulations in the Condensed Phase
- Author(s): Kahros, Argyris
- Advisor(s): Schwartz, Benjamin J
- et al.
Incorporating quantum mechanics into an atomistic simulation necessarily involves solving the Schroedinger equation. Unfortunately, the computational expense associated with solving this equation scales miserably with the number of included quantum degrees of freedom (DOF). The situation is so dire, in fact, that a molecular dynamics (MD) simulation cannot include more than a small number of quantum DOFs before it becomes computationally intractable. Thus, if one were to simulate a relatively large system, such as one containing several hundred atoms or molecules, it would be unreasonable to attempt to include the effects of all of the electrons associated with all of the components of the system. The mixed quantum/classical (MQC) approach provides a way to circumvent this issue. It involves treating the vast majority of the system classically, which incurs minimal computational expense, and reserves the consideration of quantum mechanical effects for only the few degrees of freedom more directly involved in the chemical phenomenon being studied. For example, if one were to study the bonding of a single diatomic molecule in the gas phase, one could employ a MQC approach by treating the nuclei of the molecule's two atoms classically—including the deeply bound, low-energy electrons that change relatively little—and solving the Schroedinger equation only for the high energy electron(s) directly involved in the bonding of the classical cores. In such a way, one could study the bonding of this molecule in a rigorous fashion while treating only the directly related degrees of freedom quantum mechanically. Pseudopotentials are then responsible for dictating the interactions between the quantum and classical degrees of freedom. As these potentials are the sole link between the quantum and classical DOFs, their proper development is of the utmost importance.
This Thesis is concerned primarily with my work on the development of novel, rigorous and dynamical pseudopotentials for use in mixed quantum/ classical simulations in the condensed phase. The pseudopotentials discussed within are constructed in an ab initio fashion, without the introduction of any empiricism, and are able to exactly reproduce the results of higher level, fully quantum mechanical Hartree-Fock calculations. A recurring theme in the following pages is overcoming the so-called frozen core approximation (FCA). This essentially comes down to creating pseudopotentials that are able to respond in some way to the local molecular environment in a rigorous fashion. The various methods and discussions that are part of this document are presented in the context of two particular systems. The first is the sodium dimer cation molecule, which serves as a proof of concept for the development of coordinate-dependent pseudopotentials and is the subject of Chapters 2 and 3. Next, the hydrated electron—the excess electron in liquid water—is tackled in an effort to address the recent controversy concerning its true structure and is the subject of Chapters 4 and 5. In essence, the work in this Dissertation is concerned with finding new ways to overcome the problem of a lack of infinite computer processing power.