The topological and geometric analysis of organic crystal systems
The topological and geometric analysis of organic crystal systems
The aim of this study was to use topological descriptors to gain insight into the crystal packing of organic compounds and generate crystal structure landscapes that are representative of the packing motif that might be identified by a crystallographer. These descriptors are applied to both sets of experimental compounds, as might be found in the Cambridge Structural Database, for example, or to the large sets of compounds that might be generated as the output of crystal structure prediction calculations.
The crystal structures of fluorinated benzylideneanilines, polyaromatic hydrocarbons, azapentacenes and the nictotinamide:benzoic acid co-crystal were studied through the lens of a novel topological descriptor. This descriptor is constructed from the persistent homology of a set of molecular centroids and orientation vectors extracted from the crystal structure, the homology being computed on a six dimensional space. We were able to generate crystal structure landscapes that completely separated all known packing classes of fluorinated benzylideneaniline as identified by a subject matter expert. We were also able to completely separate the structures of two classes of nictotinamide:benzoic acid co-crystals that were identified to belong to two funnels on the potential energy landscape corresponding to its known polymorphs. While the azapentacens and polyaromatic hydrocarbons proved more resistant to a full description with persistent homology, we were able to produce landscapes that preserve some trends which are consistent with their canonical packing motifs. We also showcase how crystal structure landscapes can be constructed using supervised dimensionality reduction in the context of some existing high fidelity data with known packing motifs in order to obtain landscape that extenuate these chemically relevant features.
topological data analysis, crystallography, crystal structure prediction, Persistent homology
University of Southampton
Doyle, Jack
903c09b3-0b55-4592-98d7-c9168e1dec4b
20 September 2024
Doyle, Jack
903c09b3-0b55-4592-98d7-c9168e1dec4b
Frey, Jeremy
ba60c559-c4af-44f1-87e6-ce69819bf23f
Doyle, Jack
(2024)
The topological and geometric analysis of organic crystal systems.
University of Southampton, Doctoral Thesis, 214pp.
Record type:
Thesis
(Doctoral)
Abstract
The aim of this study was to use topological descriptors to gain insight into the crystal packing of organic compounds and generate crystal structure landscapes that are representative of the packing motif that might be identified by a crystallographer. These descriptors are applied to both sets of experimental compounds, as might be found in the Cambridge Structural Database, for example, or to the large sets of compounds that might be generated as the output of crystal structure prediction calculations.
The crystal structures of fluorinated benzylideneanilines, polyaromatic hydrocarbons, azapentacenes and the nictotinamide:benzoic acid co-crystal were studied through the lens of a novel topological descriptor. This descriptor is constructed from the persistent homology of a set of molecular centroids and orientation vectors extracted from the crystal structure, the homology being computed on a six dimensional space. We were able to generate crystal structure landscapes that completely separated all known packing classes of fluorinated benzylideneaniline as identified by a subject matter expert. We were also able to completely separate the structures of two classes of nictotinamide:benzoic acid co-crystals that were identified to belong to two funnels on the potential energy landscape corresponding to its known polymorphs. While the azapentacens and polyaromatic hydrocarbons proved more resistant to a full description with persistent homology, we were able to produce landscapes that preserve some trends which are consistent with their canonical packing motifs. We also showcase how crystal structure landscapes can be constructed using supervised dimensionality reduction in the context of some existing high fidelity data with known packing motifs in order to obtain landscape that extenuate these chemically relevant features.
Text
JD_thesis_with_corrections_final
- Version of Record
Text
Final-thesis-submission-Examination-Mr-Jack-Doyle
Restricted to Repository staff only
More information
Published date: 20 September 2024
Keywords:
topological data analysis, crystallography, crystal structure prediction, Persistent homology
Identifiers
Local EPrints ID: 494044
URI: http://eprints.soton.ac.uk/id/eprint/494044
PURE UUID: 9160df9e-055f-4a19-b5a2-f1f61218db02
Catalogue record
Date deposited: 20 Sep 2024 16:36
Last modified: 21 Sep 2024 01:32
Export record
Contributors
Author:
Jack Doyle
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics