Diffusion of encapsulated metal nanocatalysts via machine learning
Thesis title in Czech: | difúze kovových nanokatalyzátorů se strojovým učením |
---|---|
Thesis title in English: | Diffusion of encapsulated metal nanocatalysts via machine learning |
English key words: | Machine learning, free energy methods, biased molecular dynamics, clusters, nanocatalysts, diffusion, zeolites |
Academic year of topic announcement: | 2022/2023 |
Thesis type: | Bachelor's thesis |
Thesis language: | angličtina |
Department: | Department of Physical and Macromolecular Chemistry (31-260) |
Supervisor: | Christopher James Heard, Ph.D. |
Author: | hidden - assigned and confirmed by the Study Dept. |
Date of registration: | 24.10.2022 |
Date of assignment: | 24.10.2022 |
Confirmed by Study dept. on: | 01.02.2023 |
Advisors: | Dr. rer. nat. Andreas Erlebach |
Guidelines |
The project requires learning of computational techniques ranging from dynamics simulations, to calculation of energetic properties via local structure optimisation. The student will learn programming techniques for manipulating data and running calculations, in addition to learning the use of high performance computing infrastructure. Machine learning methods will be tested and applied, which will be learned in collaboration with the nanomaterials modelling group via internal seminars and teaching programmes. Collaboration with experiment within the faculty is part of the project, as part of ongoing collaborative interests between the supervisor and experimental groups at UK. The student will practice and develop oral and written communication skills via presentations and discussion of progress reports, within the group. |
References |
1) D. Frenkel and B. Smit, Understanding Molecular Simulations, Academic Press, London, 2002. 2) Kaerger, Ruthven and Theodorou, "Diffusion in Nanoporous Materials", Wiley-VCH, 2012 3) Meiwes-Broer, "Metal Clusters at Surfaces", Springer, 2000 4) Rapaport, "The Art of Molecular Dynamics Simulation", Cambridge, 2004 5) R.L. Johnston, "Atomic and Molecular Clusters", Routledge, 2002 6) Schuett et al., "Machine learning meets Quantum Physics", Springer, 2020 Annotation: This literature provides a broad overview of dynamical simulations, zeolites as supports and machine learning methods. |
Preliminary scope of work |
Research Motivation: The stabilization of single atom and small cluster nanocatalysts via confinement within zeolite pores is subject of intense experimental scrutiny. However an atomistic, mechanistic underpinning of the diffusion processes that inevitably occur in these systems is lacking. By unravelling the effect of cluster size, zeolite topology and involvement of defects, optimal conditions for stabilizing confined nanocatalysts may be reached. Research Methods: The project will involve a combination of equilibrium and biased molecular dynamical simulations, calculation and analysis of free energy surfaces, and the application of neural-network-based machine learning methods. Programming (python/bash/PLUMED) will be integral to the completion of the project. Annotation: Intercage, and channel diffusion processes for subnanometre Pt clusters encapsulated into industrially relevant 3D zeolite models, will be investigated as a function of particle size, temperature, zeolite topology and the presence/absence of framework defects. We will determine the appropriate dimensionality reduction schemes for correctly describing cluster diffusion processes, and use them to elucidate the diffusion free energies and pathways for clusters. This work will aim towards a general optimization scheme for cluster stability in siliceous 3D porous materials. |
Preliminary scope of work in English |
Research Motivation: The stabilization of single atom and small cluster nanocatalysts via confinement within zeolite pores is subject of intense experimental scrutiny. However an atomistic, mechanistic underpinning of the diffusion processes that inevitably occur in these systems is lacking. By unravelling the effect of cluster size, zeolite topology and involvement of defects, optimal conditions for stabilizing confined nanocatalysts may be reached. Research Methods: The project will involve a combination of equilibrium and biased molecular dynamical simulations, calculation and analysis of free energy surfaces, and the application of neural-network-based machine learning methods. Programming (python/bash/PLUMED) will be integral to the completion of the project. Annotation: Intercage, and channel diffusion processes for subnanometre Pt clusters encapsulated into industrially relevant 3D zeolite models, will be investigated as a function of particle size, temperature, zeolite topology and the presence/absence of framework defects. We will determine the appropriate dimensionality reduction schemes for correctly describing cluster diffusion processes, and use them to elucidate the diffusion free energies and pathways for clusters. This work will aim towards a general optimization scheme for cluster stability in siliceous 3D porous materials. |