Perspectives in Computational Catalysis: Data-Driven and Operando Approaches

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27740%2F25%3A10260378" target="_blank" >RIV/61989100:27740/25:10260378 - isvavai.cz</a>

Výsledek na webu

<a href="https://events.it4i.cz/event/368/" target="_blank" >https://events.it4i.cz/event/368/</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Perspectives in Computational Catalysis: Data-Driven and Operando Approaches

Popis výsledku v původním jazyce

The Nanomaterials Modelling group at Charles University developed atomistic simulation methods combining machine learning with quantum chemistry to study catalytic materials under realistic conditions and to investigate systems of high industrial relevance. Their work produced transferable reactive interatomic potentials for zeolites and oxide catalysts, accelerated rare-event sampling, and established property predictors linking structural features to spectroscopic signatures. Applications of machine-learning interatomic potentials and regression models included studies of confined water in aluminosilicate zeolites, aluminium siting determined by solid-state NMR, and the stabilisation of sub-nanometre bimetallic noble-metal clusters through alloying, oxidation, and defects. The results showed that machine-learning-driven dynamical modelling uncovered unexpected binding modes, reproduced complex experimental NMR spectra with high fidelity, and provided insight into sintering and oxidation processes, while ongoing research further extended these approaches with improved property predictors, rapid potential construction, uncertainty quantification, and large-scale data analysis pipelines.

Název v anglickém jazyce

Perspectives in Computational Catalysis: Data-Driven and Operando Approaches

Popis výsledku anglicky

The Nanomaterials Modelling group at Charles University developed atomistic simulation methods combining machine learning with quantum chemistry to study catalytic materials under realistic conditions and to investigate systems of high industrial relevance. Their work produced transferable reactive interatomic potentials for zeolites and oxide catalysts, accelerated rare-event sampling, and established property predictors linking structural features to spectroscopic signatures. Applications of machine-learning interatomic potentials and regression models included studies of confined water in aluminosilicate zeolites, aluminium siting determined by solid-state NMR, and the stabilisation of sub-nanometre bimetallic noble-metal clusters through alloying, oxidation, and defects. The results showed that machine-learning-driven dynamical modelling uncovered unexpected binding modes, reproduced complex experimental NMR spectra with high fidelity, and provided insight into sintering and oxidation processes, while ongoing research further extended these approaches with improved property predictors, rapid potential construction, uncertainty quantification, and large-scale data analysis pipelines.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

—

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů