Fluorine spillover for ceria-vs silica-supported palladium nanoparticles: A MD study using machine learning potentials

Aug 4, 2026

Supported metallic nanoparticles play a central role in catalysis, yet predictive modeling is difficult because nanoparticle structure and dynamics, especially at metal–support interfaces, are highly complex and often exceed the reach of traditional ab initio approaches. Machine learning potentials provide near-DFT accuracy for molecular dynamics over experimentally relevant temperatures and timescales. This study examines fluorine-atom adsorption on Pd/ceria and Pd/silica using DeePMD-trained ML potentials from DFT data. Defects at ceria–Pd interfaces enable initial fluorine adsorption, while Pd–ceria coupling and oxygen migration from ceria to Pd drive later-stage spillover to ceria; silica prevents fluorine spillover.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages8

Reading Time20 min

AccessFree to download

Abstract

I. Introduction

II. Developing General Potentials for Heterogeneous Catalysts

A. Training

The Journal of Chemical Physics ARTICLE [pubs.aip.org/aip/jcp](pubs.aip.org/aip/jcp) Fluorine spillover for ceria-vs silica-supported palladium nanoparticles: A MD study using machine learning potentials | Cite as: J. Chem. Phys. 159, 024101 (2023); doi: 10. 1063/5.0147132 Submitted: 19 February 2023 • Accepted: 19 June 2023 •
Published Online: 10 July 2023 | | | | | --- | --- | --- | --- | | Da-Jiang Liu1, a) and James W. Evans1 , 2 | | | | | AFFILIATIONS
1 Division of Chemical and Biological Sciences, Ames National Laboratory—USDOE, Ames, Iowa 50011, USA
2 Department of Physics and Astronomy, Iowa State University, Ames, Iowa 50011, USA
Note: This paper is part of the JCP Special Topic on Machine Learning Hits Molecular Simulations.
a)Author to whom correspondence should [be addressed: dajiang@ameslab.gov](be addressed: dajiang@ameslab.gov) | | | | | ABSTRACT
Supported metallic nanoparticles play a central role in catalysis. However, predictive modeling is particularly challenging due to the structural and dynamic complexity of the nanoparticle and its interface with the support, given that the sizes of interest are often well beyond those accessible via traditional ab initio methods. With recent advances in machine learning, it is now feasible to perform MD simulations with potentials retaining near-density-functional theory (DFT) accuracy, which can elucidate the growth and relaxation of supported metal nanoparticles, as well as reactions on those catalysts, at temperatures and time scales approaching those relevant to experiments. Furthermore, the surfaces of the support materials can also be modeled realistically through simulated annealing to include effects such as defects and amorphous structures. We study the adsorption of fluorine atoms on ceria and silica supported palladium nanoparticles using machine learning potential trained by DFT data using the DeePMD framework. We show defects on ceria and Pd/ceria interfaces are crucial for the initial adsorption of fluorine, while the interplay between Pd and ceria and the reverse oxygen migration from ceria to Pd control spillover of fluorine from Pd to ceria at later stages. In contrast, silica supports do not induce fluorine spillover from Pd particles.
Published under an exclusive license by AIP Publishing. [https://doi.org/10.1063/5.0147132](https://doi.org/10.1063/5.0147132) | | | | I. INTRODUCTION It has long been speculated that catalyst supports often participate actively in heterogeneous catalysis rather than acting as inert bystanders, merely providing a high surface area material on which catalysts are dispersed. One of the ways that support can participate in the reaction is through the spillover of adsorbate species from the metal catalyst to the support.1–3 Theoretical studies of spillover are currently being pursued, but mostly for carbon supports.4–6 To date, studies of metal oxide supports are more limited.5,7 This is due to the inherent complexity of the systems, where the structure of the oxide support, the interface between the catalyst and support, and the shape and structure of the catalyst nanoparticles all provide challenges for modeling. In addition, it is necessary to realistically describe the adsorption and diffusion ofchemisorbed reactant species in these complex systems. Metallic nanoparticles supported on oxides, as well as adsorption and reaction on these nanoparticles, provide a distinctive challenge for computational studies.8 Their sizes are often too large for direct ab initio calculations but too small for more phenomenologically coarse-grained continuum modeling. In addition, the need to treat a combination of transition metals, oxides, and chemical reactions makes the development of effective empirical potentials difficult. Machine learning naturally provides an enticing new approach to the problem, and there are some new developments in applying machine learning to the general probl

FAQs

What systems does the study focus on for fluorine adsorption and spillover?

It investigates adsorption of fluorine atoms on palladium nanoparticles supported on ceria and silica, comparing the two support materials’ effects on fluorine behavior.

Which factors on the Pd/ceria interface are crucial for the initial fluorine adsorption?

Defects on ceria and at Pd/ceria interfaces are identified as crucial for the initial adsorption of fluorine.

Why does fluorine spillover occur on ceria-supported systems but not on silica-supported systems?

On ceria, Pd–ceria interplay and reverse oxygen migration from ceria to Pd control fluorine spillover from Pd to ceria; silica supports do not induce fluorine spillover from Pd particles.

Aug 4, 2026

Upload by Olivia Brown