Skip navigation
Please use this identifier to cite or link to this item:
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorCar, Roberto-
dc.contributor.authorKo, Hsin-Yu-
dc.contributor.otherChemistry Department-
dc.description.abstractHybrid density functional theory (DFT) is widely used to obtain a semi-quantitative understanding of the electronic structures of isolated molecular clusters; however, it has limited applicability to large molecules and complex condensed-phase materials due to its high computational cost. To overcome this difficulty, a linear-scaling algorithm based on maximally localized Wannier functions (MLWF) can be employed. In the first part of this thesis, we present a detailed discussion on the theory, real-space implementation, and performance of this algorithm for enabling large-scale condensed-phase hybrid DFT based ab initio molecular dynamics (AIMD) simulations under realistic isobaric-isothermal (NpT) conditions. For the theory aspect, we discuss how the MLWF-based linear-scaling approach can be integrated into the Car-Parrinello AIMD framework. For our implementation (named exx module), we introduced several features to enable massive parallelization using hybrid MPI/OpenMP technologies, including custom data distribution scheme for MLWFs, static load balancing algorithm, and reusable proto-subdomains to exploit the localization in MLWF representation. Based on performance test using realistic condensed-phase liquid water systems (H2O)64, (H2O)128, and (H2O)256, we find the exx module to be quite efficient and scalable, hence making a hybrid DFT based AIMD simulation feasible. In the second part, we apply the exx module to study molecular crystals and liquids. The first application is to simulate the ice Ih, II, and III phases using AIMD at the dispersion-inclusive hybrid DFT level at their experimental triple point (238 K, 2.1 kbar). In the second application, we exploit the computational efficiency and scalability of our exx module in conjunction with leadership level supercomputers to perform path-integral AIMD (PI-AIMD) simulations on liquid water (300 K) and ice Ih (273 K) both under ambient pressure (1.0 bar) at dispersion-inclusive hybrid DFT level. We identify this level of theory provides a quite accurate description to water. From the resulting trajectory, we also find that the nuclear quantum fluctuation with autoprotolysis-type distortion promotes orbital localization. In the third application, we explore how anharmonicity, nuclear quantum effects (NQE), many-body dispersion interactions, and Pauli repulsion influence thermal properties of dispersion-bound molecular crystals using pyridine and similar organic molecular crystals.-
dc.publisherPrinceton, NJ : Princeton University-
dc.relation.isformatofThe Mudd Manuscript Library retains one bound copy of each dissertation. Search for these copies in the library's main catalog: <a href=> </a>-
dc.subjectdensity functional theory-
dc.subjectlinear-scaling algorithm-
dc.subjectliquid water-
dc.subjectmolecular crystal polymorphism-
dc.subjectthermal expansion-
dc.subject.classificationComputational chemistry-
dc.subject.classificationMaterials Science-
dc.subject.classificationMolecular chemistry-
dc.titleFirst-Principles Study on the Structural and Thermal Properties of Molecular Crystals and Liquids-
dc.typeAcademic dissertations (Ph.D.)-
Appears in Collections:Chemistry

Files in This Item:
File Description SizeFormat 
Ko_princeton_0181D_12876.pdf16.83 MBAdobe PDFView/Download

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.