Ab initio investigations of the thermodynamic and kinetic processes underpinning molecular crystallization: The case of paracetamol
File(s)
Author(s)
Zhou, Huanyu
Type
Thesis
Abstract
As a common form of organic compounds, molecular crystals usually experience complex thermodynamic and kinetic transitions, requiring atomic-scale simulations to offer comprehensive understandings. Since density functional theory (DFT) can be difficult to apply directly to probe such delicate transitions, this research develops efficient and predictive modeling workflows using paracetamol, a pharmaceutical of both commercial and scientific importance, as a prototype system.
The first study evaluates how fundamental DFT approximations, including basis set and Hamiltonian, influence lattice dynamics (LD) based on harmonic approximation (HA). Results on paracetamol forms I (FI) and II (FII) indicate the overall comparable performances of the def2-TZVP basis set to the nearly complete plane-wave basis set, though the constant-volume approximation of HA inherently causes systematic deviations.
To resolve this, the second study introduces a quasi-harmonic approximated (QHA) scheme based on the Grüneisen model. It is identified that Fock exchange improves predictions of thermochemical properties, while atomic basis sets reduce phonon continuity. A novel workflow is then proposed as a practical complement to full QHA LD, that equilibrium volumes obtained with plane-wave basis set and (semi-)local DFT Hamiltonians can be efficiently combined with phonons from atomic basis sets and hybrid exchange Hamiltonians. This enhances the prediction of finite-temperature properties.
The final study examines the growth kinetics of FI from ethanol using the effective screen medium-reference interaction site model (ESM-RISM). It demonstrates that unsaturated hydrogen bonds on slow-growing surfaces increase competition with the solvent, creating significant steric hindrance that dictates regrowth kinetics. In contrast, fast-growing surfaces lack these dangling bonds and yield a uniform solvent structure with low hindrance.
In conclusion, by disentangling various errors underlying ab initio LD and pioneering the use of DFT/ESM-RISM for recrystallization, this research provides highly transferable workflows for the predictive modeling of broad classes of molecular crystals in complex environments.
The first study evaluates how fundamental DFT approximations, including basis set and Hamiltonian, influence lattice dynamics (LD) based on harmonic approximation (HA). Results on paracetamol forms I (FI) and II (FII) indicate the overall comparable performances of the def2-TZVP basis set to the nearly complete plane-wave basis set, though the constant-volume approximation of HA inherently causes systematic deviations.
To resolve this, the second study introduces a quasi-harmonic approximated (QHA) scheme based on the Grüneisen model. It is identified that Fock exchange improves predictions of thermochemical properties, while atomic basis sets reduce phonon continuity. A novel workflow is then proposed as a practical complement to full QHA LD, that equilibrium volumes obtained with plane-wave basis set and (semi-)local DFT Hamiltonians can be efficiently combined with phonons from atomic basis sets and hybrid exchange Hamiltonians. This enhances the prediction of finite-temperature properties.
The final study examines the growth kinetics of FI from ethanol using the effective screen medium-reference interaction site model (ESM-RISM). It demonstrates that unsaturated hydrogen bonds on slow-growing surfaces increase competition with the solvent, creating significant steric hindrance that dictates regrowth kinetics. In contrast, fast-growing surfaces lack these dangling bonds and yield a uniform solvent structure with low hindrance.
In conclusion, by disentangling various errors underlying ab initio LD and pioneering the use of DFT/ESM-RISM for recrystallization, this research provides highly transferable workflows for the predictive modeling of broad classes of molecular crystals in complex environments.
Version
Open Access
Date Issued
2025-10-30
Date Awarded
2026-05-01
Copyright Statement
Attribution-NonCommercial 4.0 International Licence (CC BY-NC)
License URL
Advisor
Harrison, Nicholas M.
Heng, Jerry Y. Y.
Sponsor
Imperial College London
BASF UK
Engineering and Physical Sciences Research Council
Grant Number
EP/X035891/1
Publisher Department
Department of Chemistry
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)