Impact of chirality on nanotube fracture strain: comprehensive machine learning potential calculations

Abstract

We present a comprehensive study of fracture strains in carbon nanotubes (CNTs) as a function of their chirality using the direct integration of the exchange potential (DIEP) machine learning potential model. By simulating fracture using quasi-static pulling of the CNT structures, we simulated the fracture of 186 CNTs ranging from 476 atoms up to 5616 atoms in size. Our results reveal clear trends in the fracture strains for armchair, zigzag, and chiral CNTs. In particular, armchair nanotubes tend to exhibit localized bond scission near the ends of the supercell, while zigzag nanotubes suffer more distributed bond breakage along their length. This observation is consistent with previous molecular dynamics simulations. We validate the model's performance by comparing its predicted fracture strains with published empirical model. In addition, we highlight specific CNT configurations with notably high or low fracture strains, shedding light on the interplay between chirality, diameter, and defect propagation mechanisms. The demonstrated predictive power of this machine learning potential underscores its promise for large-scale, high-throughput fracture simulations of CNTs and potentially other nanomaterials.