Developing methods for simulating reactive condensed-phase chemical systems via quantum dynamics and machine learning
Abstract/Contents
- Abstract
- This thesis explores a variety of algorithmic advances that enable the ab initio simulation of condensed-phase chemical systems at reduced computational cost and at the time and length scales required to statistically converge physical properties. We first employ multiple time scale techniques to obtain nanosecond-scale ab initio molecular dynamics simulations of excess protons in two hydrogen-bonded liquids, imidazole and 1,2,3-triazole, using density functional theory (DFT) at the generalized gradient approximation (GGA) level of accuracy. These liquids transport protons via a structural mechanism similar to that of water, and their chemical motifs are present in systems ranging from biological proton channels to proton exchange membrane fuel cells. Using a time correlation function analysis, we decompose the proton transport mechanism into three first-order processes with distinct characteristic timescales, thereby providing a fuller picture of the associated structural transport mechanism. We also uncover two mechanisms that slow down the rate of proton transfer in 1,2,3-triazole relative to imidazole and thus explain their experimentally observed 10-fold difference in proton conductivity. We then develop machine-learned potentials (MLPs) capable of modeling excess protons and hydroxide ions in liquid water at the GGA and hybrid DFT levels of accuracy and use them to perform multi-nanosecond classical and path integral simulations of proton defects in water at a fraction of the cost of the corresponding ab initio simulations. The transport of proton defects in water is of particular interest because it underlies several important chemical and biological processes. Our results show that the MLPs are able to reproduce ab initio trends and converge properties such as the diffusion coefficients of excess protons and hydroxide ions. We use the multi-nanosecond MLP trajectories to study with unprecedented statistical accuracy the role of hypercoordination in the transport mechanism of the hydroxide ion and provide further evidence for the asymmetry in diffusion between excess protons and hydroxide ions. Finally, we develop a computationally efficient quantum-classical Ehrenfest approach based on pure quantum states for the accurate simulation of linear and nonlinear electronic spectra. We demonstrate the performance of our method by showing that it is able to quantitatively capture the exact linear, 2D electronic spectroscopy, and pump--probe spectra for a Frenkel exciton model in slow bath regimes and is even able to reproduce the main spectral features in fast bath regimes.
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2023; ©2023 |
| Publication date | 2023; 2023 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Atsango, Austin Ojiambo |
|---|---|
| Degree supervisor | Markland, Thomas E |
| Thesis advisor | Markland, Thomas E |
| Thesis advisor | Martinez, Todd J. (Todd Joseph), 1968- |
| Thesis advisor | Rotskoff, Grant |
| Degree committee member | Martinez, Todd J. (Todd Joseph), 1968- |
| Degree committee member | Rotskoff, Grant |
| Associated with | Stanford University, School of Humanities and Sciences |
| Associated with | Stanford University, Department of Chemistry |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Austin Ojiambo Atsango. |
|---|---|
| Note | Submitted to the Department of Chemistry. |
| Thesis | Thesis Ph.D. Stanford University 2023. |
| Location | https://purl.stanford.edu/hh249my3122 |
Access conditions
- Copyright
- © 2023 by Austin Ojiambo Atsango
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...