Physchem.cz

Application of cutting edge machine-learning based methods in support of quantum chemistry calculations for enhanced sampling and analysis of catalytic porous materials (zeolites, MOFs, silicate derivatives) - towards bottom-up design of functional porous nanomaterials. Training, optimization and use of class-specific, reactive ML interatomic forcefields, combined with spectral characterization (NMR, FTIR, quasi-elastic neutron scattering) and reactive modelling on realistic systems, in close collaboration with experimental synthesis, spectroscopy and catalysis: towards the solution of grand challenges in the field. Including-

a. Heteroatom distribution, speciation and host-guest dynamics in realistic zeolites
b. Defect formation, modulation and engineering in zeolites
c. Mechanistic analysis of zeolite synthesis and hydrolysis

a. Investigation of heteroatom distribution in zeolites - application of unbiased global structure optimization techniques, combined with enhanced dynamical sampling (both unbiased molecular dynamics and biased dynamics for diffusion/reaction processes) in industrially important complex zeolites. Determination of the role of conditions (temperature, defects, heteroatom coupling, solvent loading) on the energetics and spectral signatures. Development and use of ML tensorial property prediction models for full spectrum calculations and characterisation of samples. Analysis of acidity, solvent dynamics and structure-function relationships in zeolites. Verification alongside high-level quantum chemical calculations (RPA, CCSD, MP2) in collaboration with experts.

Relevant Publications:
1. Digital Discovery, 2025,10.1039/D4DD00306C [GL1]
2. Faraday Discussions, 2025,10.1039/d4fd00100a
3. Chemical Science, 2023,10.1039/D3SC02492J

b. Defect engineering - the overall goal is to tailor the formation, concentration and distribution of defects to control properties relevant to catalysis, including diffusivity inside the pore system, pinning of co-catalytic species (e.g. metal centres) and driving selective reactions. This goal requires an atomistic understanding of the formation processes of defects, the conditions which generate them, the mechanisms of their formation, the nature of the defects themselves, and means by which they affect catalytic properties. The selected candidate will utilise (and improve) ML-driven dimensionality reduction schemes for automatic reaction pathway generation and investigate reactive processes on zeolites/MOFs - beyond the limits of human chemical intuition.

Relevant Publications:
1. Chemistry of Materials, 2021, 10.1021/acs.chemmater.1c02799
2. Angewandte Chemie, 2023,https://doi.org/10.1002/anie.202306183

c. Zeolite synthesis and hydrolysis - the atomistic, mechanistic processes in zeolite synthesis are incompletely understood. A combination of direct dynamical simulation and kinetic modelling, alongside X-ray, NMR and neutron scattering experiments will aim to reproduce, under operando modelling conditions, the early stages of zeolite synthesis, alongside related systems (silica surfaces/gels) and the hydrolysis process that is used extensively in the ADOR scheme, for targeted generation of new zeolites from selected inter-zeolite structural transitions.

Relevant Publications:
1. Nature Communications, 2024, http://dx.doi.org/10.1038/s41467-024-48609-2
2. npj Computational Materials, 2022, https://doi.org/10.1038/s41524-022-00865-w

Collaborations
1. QENS - A. O'Malley (Bath)
2. Enhanced Sampling of in Porous Materials - V. van Speybroeck (Gent)
3. Defect Engineering and Characterization - J.A. van Bokhoven (ETH), M. Shamzhy/ J. Cejka (CU)
4. Synthesis and NMR - R. Morris and S. Ashbrook (St Andrews)
5. Novel Synthetic Pathways - M. Moliner (ITQ Valencia)
6. High level Quantum Chemical Calculations - J. Sauer (Berlin)
7. Zeolite synthesis modelling, ML methods - R. Gomez-Bombarelli (MIT)