Understanding Supported Metal Cluster Nanocatalysts via Machine Learning

• Thesis title in Czech: Studium chování kovových nanokatalytických částic na površích akcelerované pomocí strojového učení
• Thesis title in English: Understanding Supported Metal Cluster Nanocatalysts via Machine Learning
• Key words: kvantová chemie, metoda strojového učení, neuronové sítě, katalýza, nanoklastry, nanomateriály, modelování, výpočetní chemie
• English key words: quantum chemistry, machine learning, neural networks, catalysis, nanoclusters, nanomaterials, modelling, computational chemistry
• Academic year of topic announcement: 2022/2023
• Thesis type: dissertation
• Thesis language: angličtina
• Department: Department of Physical and Macromolecular Chemistry (31-260)
• Supervisor: Christopher James Heard, Ph.D.
• Author: Mgr. Tereza Benešová - assigned by the advisor
• Date of registration: 07.10.2022
• Date of assignment: 07.10.2022
• Advisors: Michal Mazur, Ph.D.

Preliminary scope of work

Development and application of machine learning methods towards simulations of catalytically important, zeolite-encapsulated single atom catalysts (SAC) and related sub-nanometre scale catalysts. You will identify catalyst design principles that will inform the next generation of heterogeneous catalytic materials.

a) Develop reactive machine learning potentials for the migration, sintering and metal-framework interactions of small clusters in zeolites and upon two-dimensional oxide supports.
b) Apply biased dynamical simulations to understand the dynamical processes at work in these systems at the atomistic scale.
c) Collaboration with experimentalists to aid in design of robust cluster systems for synthesis of practical sub-nanometre scale catalytic materials.
d) Isolation of controlling factors for cluster/nanoparticle stability, diffusivity, reactivity towards industrially important chemical processes (hydrocarbon oxidation, reduction).
e)modelling of reactive processes directly upon operando models of support nanocatalysts
This project will include the development of computational methods, in particular, neural network-based machine learning potentials. These methods will be applied towards long-time simulations and statistical analysis of SAC binding, growth and migration processes. The successful candidate will gain experience in programming, simulation methods, maintaining local and international collaborations, and presentation at international conferences.

Preliminary scope of work in English

Development and application of machine learning methods towards simulations of catalytically important, zeolite-encapsulated single atom catalysts (SAC) and related sub-nanometre scale catalysts. You will identify catalyst design principles that will inform the next generation of heterogeneous catalytic materials.
a) Develop reactive machine learning potentials for the migration, sintering and metal-framework interactions of small clusters in zeolites and upon two-dimensional oxide supports.
b) Apply biased dynamical simulations to understand the dynamical processes at work in these systems at the atomistic scale.
c) Collaboration with experimentalists to aid in design of robust cluster systems for synthesis of practical sub-nanometre scale catalytic materials.
d) Isolation of controlling factors for cluster/nanoparticle stability, diffusivity, reactivity towards industrially important chemical processes (hydrocarbon oxidation, reduction).
e)modelling of reactive processes directly upon operando models of support nanocatalysts
This project will include the development of computational methods, in particular, neural network-based machine learning potentials. These methods will be applied towards long-time simulations and statistical analysis of SAC binding, growth and migration processes. The successful candidate will gain experience in programming, simulation methods, maintaining local and international collaborations, and presentation at international conferences.