Machine learning electron density in sulfur crosslinked carbon nanotubes

John M. Alred, Ksenia V. Bets, Yu Xie, Boris I. Yakobson

Index: 10.1016/j.compscitech.2018.03.035

Full Text: HTML

Abstract

Mechanical strengthening of composite materials that include carbon nanotubes (CNT) requires strong inter-bonding to achieve significant CNT-CNT or CNT-matrix load transfer. The same principle is applicable to the improvement of CNT bundles and calls for covalent crosslinks between individual tubes. In this work, sulfur crosslinks are studied using a combination of density functional theory (DFT) and classical molecular dynamics (MD). Atomic chains of at least two sulfur atoms or more are shown to be stable between both zigzag and armchair CNTs. All types of crosslinked CNTs exhibit significantly improved load transfer. Moreover, sulfur crosslinks show evidence of a cooperative self-healing mechanism allowing for links to rebond once broken leading to sustained load transfer under shear loading. Additionally, a general approach for utilizing machine learning for assessing the ground state electron density is developed and applied to these sulfur crosslinked CNTs.