In the past half year we had three PhD defenses. Underneath you can find a short summary of their PhD research here at CMM.
Congratulations again, Juul, Aran and Sander!
Juul De Vos
High-Throughput Screening of Covalent Organic Frameworks for Clean Energy Applications – Monday August 26, 2024
Supervisors: prof. Veronique Van Speybroeck, prof. Sven Rogge, prof. Pascal Van Der Voort
Summary
Covalent organic frameworks (COFs) are materials characterized by their nanoscale pores, lightweight building blocks and high stability. Due to this unique combination, they are promising new materials for a wide variety of applications such as gas storage, separation and catalysis. Due to their modular nature, the properties of COFs can be designed at the molecular level. However, this characteristic also results in a huge number of possible COFs, which hampers experimental identification of the best performing COF for a given application. In this PhD research, we accelerated the development of promising COFs using an extensive computational screening. To this end, we first used an automated design algorithm to generate the ReDD-COFFEE database, consisting of 268,687 COFs. This database was characterized to investigate the use of COFs in two sustainable energy applications. After establishing the performance limits of COFs for the storage of natural gas in vehicles, good performing candidates for the capture of post-combustion carbon dioxide were identified. During the final screening, we used machine learning to predict the adsorption properties of all COFs in the database. Our results allow experimental researchers to target promising materials for these applications.
Aran Lamaire
A Computational Understanding of Nuclear Quantum Effects and the Structural Organisation of Confined Water in Nanoporous Materials – Wednesday September 11, 2024
Supervisors: prof. Veronique Van Speybroeck
Summary
One of the driving forces in technological progress is the development of new materials. Thanks to the increasing insight into the elementary structure of matter on an atomic scale, the search for new materials is no longer dependent on chance discoveries, but new materials can be designed and developed in a well-considered manner. For the various challenges in our current society, such as CO2 capture or the sustainable production of basic chemicals, nanoporous materials are a promising class of materials in the search for technological solutions. In order to fully understand the connection between the structure and the functional behavior of these materials, a deep insight is needed at the atomic level, which can be obtained using molecular simulations. In this PhD research, the effect of a widely used approximation in computational modeling is studied, which describes atomic nuclei as classical particles, without taking into account their quantum mechanical properties. The impact of this approximation was not only investigated for the structural and thermal properties of the materials themselves, but also for the presence of water in their porous structure. In this way, the fundamental understanding of these materials and their properties can be further deepened using an accurate computational description with a view to applications such as the extraction of water from the atmosphere.
Sander Vandenhaute
Accelerating Molecular Simulation Using Machine Learning: From Wave Functions to Thermodynamics – Wednesday January 22, 2024
Supervisors: prof. Veronique Van Speybroeck
Summary
The properties of materials and molecules can, in theory, be predicted through explicit simulation of the atomic motion at the nano-scale. For complex materials, the required computational power for this is enormously large. This doctoral research specializes in machine learning methods that improve the accuracy and efficiency of such simulations. The workhorse by which this is achieved are specific neural networks that are capable of learning the stable geometry of molecules and materials based on quantum mechanical reference calculations. The research in this thesis focuses on the efficient training and application of these networks to a diverse set of chemical and physical transformations in nanostructured materials.