Machine learning accelerated descriptor design for catalyst discovery in CO2 to methanol conversion - UTU Tutkimustietojärjestelmä

A1 Vertaisarvioitu alkuperäisartikkeli tieteellisessä lehdessä

Machine learning accelerated descriptor design for catalyst discovery in CO2 to methanol conversion

Tekijät: Pisal, Prajwal; Krejčí, Ondřej; Rinke, Patrick

Kustantaja: NATURE PORTFOLIO

Kustannuspaikka: BERLIN

Julkaisuvuosi: 2025

Lehti:: npj Computational Materials

Tietokannassa oleva lehden nimi: NPJ COMPUTATIONAL MATERIALS

Lehden akronyymi: NPJ COMPUT MATER

Artikkelin numero: 213

Vuosikerta: 11

Sivujen määrä: 9

ISSN: 2096-5001

eISSN: 2057-3960

DOI: https://doi.org/10.1038/s41524-025-01664-9

Verkko-osoite: https://www.nature.com/articles/s41524-025-01664-9

Rinnakkaistallenteen osoite: https://research.utu.fi/converis/portal/detail/Publication/499252719

Tiivistelmä

Transforming CO2 into methanol represents a crucial step towards closing the carbon cycle, with thermoreduction technology nearing industrial application. However, obtaining high methanol yields and ensuring the stability of heterocatalysts remain significant challenges. Herein, we present a sophisticated computational framework to accelerate the discovery of thermal heterogeneous catalysts, using machine-learned force fields. We propose a new catalytic descriptor, termed adsorption energy distribution, that aggregates the binding energies for different catalyst facets, binding sites, and adsorbates. The descriptor is versatile and can be adjusted to a specific reaction through careful choice of the key-step reactants and reaction intermediates. By applying unsupervised machine learning and statistical analysis to a dataset comprising nearly 160 metallic alloys, we offer a powerful tool for catalyst discovery. We propose new promising candidates such as ZnRh and ZnPt3, which to our knowledge, have not yet been tested, and discuss their possible advantage in terms of stability.

Ladattava julkaisu

This is an electronic reprint of the original article.
This reprint may differ from the original in pagination and typographic detail. Please cite the original version.

s41524-025-01664-9.pdf

Julkaisussa olevat rahoitustiedot:
This project received funding from the European Union – NextGenerationEU instrument and the Research Council of Finland's AICon project (grant number no. 348179).
Open Access funding enabled and organized by Projekt DEAL.