Machine Learning Interatomic Potential-Enabled Discovery of Chlorofullerenes

Area of Science:

• Computational Chemistry
• Materials Science
• Nanotechnology

Background:

• Chlorination stabilizes fullerenes by altering carbon hybridization and relieving strain.
• Predicting structures of chlorinated fullerenes is challenging due to isomer diversity and varied chlorination sites.

Purpose of the Study:

• To develop a machine learning potential (MLP) for the carbon-chlorine (C-Cl) system with density functional theory (DFT) accuracy.
• To enable high-throughput screening of chlorinated fullerene structures.
• To establish correlations between parent fullerenes and their chlorinated derivatives.

Main Methods:

• Development of a DFT-accurate machine learning potential for the C-Cl system.
• Application of the MLP for high-throughput screening of chlorinated fullerenes.
• Analysis of structural features and stability of chlorinated fullerene derivatives.

Main Results:

• The developed MLP accurately reproduces experimentally identified chlorofullerenes.
• The study establishes correlations between parent fullerenes and their chlorinated forms.
• Theoretical insights into the formation principles of chlorofullerenes are provided.

Conclusions:

• The DFT-accurate MLP is a powerful tool for predicting chlorinated fullerene structures.
• The findings facilitate understanding the structure-stability relationships in chlorofullerenes.
• This work offers theoretical guidance for the synthesis and application of functionalized fullerenes.