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

Area of Science:

• Materials Science
• Catalysis
• Computational Chemistry

Background:

• Supported metallic nanoparticles are vital in catalysis, but their modeling is complex.
• Traditional methods struggle with nanoparticle sizes relevant to experiments.
• Machine learning (ML) advances allow accurate simulations of complex catalytic systems.

Purpose of the Study:

• To investigate fluorine adsorption and spillover on palladium nanoparticles supported by ceria and silica.
• To utilize ML potentials trained with Density Functional Theory (DFT) data for realistic simulations.
• To understand the role of support material defects and interfaces in catalytic processes.

Main Methods:

• Developed ML potentials using the DeePMD framework, trained on DFT data.
• Performed Molecular Dynamics (MD) simulations of palladium nanoparticles on ceria and silica supports.
• Investigated fluorine atom adsorption and migration pathways on the supported nanoparticles.

Main Results:

• Defects on ceria and Palladium/ceria interfaces significantly influence initial fluorine adsorption.
• Interplay between palladium and ceria, including oxygen migration, drives fluorine spillover from Pd to ceria.
• Silica supports do not facilitate fluorine spillover from palladium nanoparticles.

Conclusions:

• ML potentials offer a powerful tool for simulating complex supported nanoparticle catalysts.
• Ceria's unique properties, including defects and interfaces, are critical for controlling fluorine spillover.
• Support material choice fundamentally impacts catalytic behavior, as demonstrated by the contrast between ceria and silica.