Correlation between the energetic and thermal properties of C40 fullerene isomers: An accurate machine-learning force field study

Abstract

<p>The isomerisation energies and thermal stabilities of the entire set of C₄₀ fullerene isomers are investigated using molecular dynamics simulations with the recently developed machine-learning-based Gaussian Approximation Potential (GAP-20) force field. Our simulations predict high statistical correlation between the relative isomerisation energy distribution of all C₄₀ isomers and their thermal fragmentation temperatures. The most energetically and thermally stable isomer is the symmetric C₄₀-<i>D</i>₂ isomer and its fragmentation temperature is 3875 ± 25 K. In contrast, the least stable isomer is the <i>D</i>_5d nanorod-shaped isomer which is 146.8 kcal mol⁻¹ higher in energy and decomposes at 2975 ± 25 K, i.e., around 900 K below the most stable isomer. In addition, we show that most isomers lie in the range between 20 and 60 kcal mol⁻¹ above the most stable isomer, and nearly 60% of isomers decompose in the temperature range of 3425–3675 K. These computational results provide valuable insights into the synthesis of C₄₀ fullerenes at high-temperatures.</p>