Project: The development and application of oxide-supported metal single atom and small cluster nanocatalysts has seen a surge of interest in recent years, due to high specific activity, and improvements in synthetic and characterization techniques. Nevertheless, in situ characterization, and in the case of catalytic processes, operando characterization remains limited, due to the small size, high fluctionality and complexity of the system. This hinders the goal of knowledge-led, targeted synthesis of novel catalytic nanomaterials for future scientific and industrial purposes. Computational characterization has similarly been hindered by the high computational expense of calculating the electronic structure of metals, a lack of proper sampling of possible structures, the dynamic reconstruction of the system under reactive stress, and the limited realism of model systems.[1] This limitation can be lifted by acceleration of simulations via machine learning methods, which we have developed in recent years.

Overview: The project will combine the development and application of machine learning-based simulation methods to probe the nature of oxide-supported catalytic (sub)-nanoscale metal clusters and single atom catalysis under operando conditions, going beyond traditional limitations of timescale and complexity. This work will involve big data-driven atomistic investigations of dynamical and reactive processes, via unbiased structure sampling techniques, accelerated reactive free energy methods, and experimental characterization, to bridge the materials gap, in combination with experimental support.

The objectives of the work are threefold:

1. Development and refinement of state of the art, multi-elemental equivariant neural network-based interatomic potentials (NNIP), taking into account electronic structure, charge, spin and variable oxidation states on noble mono- and binary metallic and metal-oxo clusters, in addition to relevant surface and environmental conditions, (including water, common poisons, and reactant/product species for selected catalytic processes). This task will include the implementation of active, delta and transfer learning techniques that have been developed within the group in recent years[6-7].
2. Implementation of an ML-driven kinetic modelling pipeline, including self-learning kinetic Monte Carlo, in order to bridge the time gap and connect stable configurations to long time kinetic stability/aging of selected nanocatalysts under operando conditions.
3. Application of trained NNIP (in support of PhD and masters students) towards: i)unbiased global structure elucidation and energy landscape characterization under realistic and reactive environments, for mono and binary metal/metal oxo particles in the single atom to 2 nm size range, ii) dynamical modelling of particle migration, growth and deactivation pathways under reactive stress on complex supports (oxidation/reduction environments), iii) catalytic reaction pathway investigation and tuning for archetypal thermocatalytic oxidation/reduction chemistry (e.g. C₂H₂ partial oxidation, CO₂ reduction) via NNIP free energy simulation techniques. This will be performed alongside direct two-way collaboration with experimental collaborators for tuning of potentials, prediction of properties, screening and characterization of catalytically relevant particles via various techniques (e.g. XPS, LEED-IV, EXAFS, HRTEM, TOF-MS).

Ultimately, the tools developed in this work will allow for an unprecedented degree of atomistic understanding of a valuable and growing class of nanomaterials, along with direct application towards truly rational catalyst design at the atomic level. These tools will be made available for generalization to other nanomaterials for extended implementation in the field. The successful candidate will be able to build on accumulated knowledge base in our group, both with respect to development of machine learning potentials and application of computational methods towards experimental catalytic applications.

Publications of research group related to this topic:

[1] Grajciar, L., Heard, C.J., Bondarenko, A.A., Polynski, M.V., Meeprasert, J., Pidko, E.A., Nachtigall, P. (2018): Towards Operando Computational Modeling in Heterogeneous Catalysis. In: Chemical Society Reviews (IF=40.1), 22, 8307-8348
[2] Hou D., Grajciar L., Nachtigall P., Heard C.J.*, (2020): Origin of the unusual stability of zeolite-encapsulated sub-nanometer platinum. In. ACS Catal. (IF=12.9), 10, 11057-11068
[3] Hou D., Heard C.J.* (2022), Migration of zeolite-encapsulated Pt and Au under reducing environments. In Catal. Sci. Technol. (IF=5.0), 12, 1598-1609
[4] Ament K., Köwitsch N., Hou D., Götsch T., Kröhnert J., Heard C.J., Trunschke A., Lunkenbein T., Armbrüster M., Breu J. (2021): Nanoparticles Supported on Sub‐Nanometer Oxide Films: Scaling Model Systems to Bulk Materials. In: Angewandte Chemie (IF=16.8). 60, 5890-5897
[5] Li A., Zhang Y., Heard C.J., Gołąbek K., Ju X., Čejka J., Mazur M. (2023): Encapsulating metal nanoparticles into a layered zeolite precursor with surface silanol nests enhances sintering resistance. In: Angewandte Chemie (IF=16.8), 62, e202213361
[6] Heard, C.J., Grajciar. L, Erlebach, A.E: (2024) Migration of zeolite-encapsulated subnanometre platinum clusters via reactive neural network potentials. In: Nanoscale (IF=8.3), 16, 8108,8118
[7] Erlebach, A., Šípka, M., Saha, I., Nachtigall, P., Heard, C.J., Grajciar, L., (2024): A reactive neural network framework for water-loaded acidic zeolites. In: Nature Communications (IF = 17.0), doi: 10.1038/s41467-024-48609-2.

Existing collaborations related to the topic:

1. Nanocatalysis group (Prof. Stefan Vajda – Heyrovksy Institute, Czech Academy of Sciences)
2. Collaboration network within COST Action – COSY (COST CA201101)
3. Prof. David Wales (Cambridge)
4. Gareth Parkinson (TU Wien)

Qualifications:

Required - PhD. in Chemistry, Physics, Material Science or a related field; good knowledge of English; experience in programming (ideally Python or similar) and molecular/materials modelling.

Preferred - background in Machine Learning, Statistics, Statistical Mechanics, Quantum Chemistry/Physics, solid state physics; experience with high-performance computing (including GPU accelerated) and Linux.

Salary: co-founding 1000 EUR/month is ensured

Co-founding resources: Department of physical and macromolecular chemistry budget

Department: Department of physical and macromolecular chemistry

Supervisor: Assist. Prof. Christopher Heard

E-mail: heardc@natur.cuni.cz

Phone: +(420) 892 589 332

Position available from: January 1, 2025

Deadline date for applications: 24th July, 2024

Applicants must submit required documents to:  heardc@natur.cuni.cz (project supervisor)

and in a copy to pavla.pouskova@natur.cuni.cz (International Department)