Indexado en

Documentos y archivos

Licencia y uso

Citaciones

Cited 1 times in

Altmetrics

Grant support

This work was funded by the Ministry of Science and Innovation (Ref. No. PID2021-122516OB-I00). This publication was created as part of NCCR Catalysis (grant number 180544), a National Centre of Competence in Research funded by the Swiss National Science Foundation. We also acknowledge Generalitat de Catalunya and the European Union under Grant 2020_FI_B 00266, 2023-FI-3-00027, and TotalEnergies (contract reference CT00001052). We thank the Barcelona Supercomputing Center (BSC-RES) for providing generous computational resources.

Análisis de autorías institucional

Compartir

A generalized model for estimating adsorption energies of single atoms on doped carbon materials

Publicado en:Journal Of Materials Chemistry a. 12 (-): 11049-11061 - 2024-03-20 12(-), DOI: 10.1039/d3ta05898k

Autores: Minotaki, Maria G; Geiger, Julian; Ruiz-Ferrando, Andrea; Sabadell-Rendon, Albert; Lopez, Nuria

Afiliaciones

Resumen

Single metal atoms on doped carbons constitute a new class of extremely appealing materials, as they present the best metal utilization for catalysis. However, their stability can be compromised by metal aggregation and the formation of nanoparticles, which often results in reduced activity or even catalyst deactivation. In many cases, the carbon hosts are generated via thermal processes, leading to poorly controlled materials. This causes a structural and compositional diversity that is modeled via indirect procedures and by comparison to a collection of structural models with different compositions. Our aim in this work is to develop a general framework based on machine learning techniques to determine the stability of the different structures against aggregation as nanoparticles. Here, we built machine learning models for the cavities and identified the robust features characterizing the metal-support interaction, considering different heteroatoms in the decorative cavity and single metal atoms. The descriptors presented here are accessible and cost-effective, such as the cavity size, electronegativity of the metal and heteroatoms, different covalent radii, and the metal electronic density. These can then be employed in the search for a mathematical equation that describes the adsorption energy via the Bayesian machine scientist. The algorithm is able to separate coordination, and covalent and ionic contributions expressed by the descriptors. This approach paves the way towards general modeling of single atoms in modified carbons particularly addressing one of the crucial features, stability. Single metal atoms on doped carbons constitute a new class of extremely appealing materials, as they present the best metal utilization for catalysis.

Palabras clave