A critical study of the parameters governing molecular dynamics simulations of nanostructured materials
AR Alian and SA Meguid, COMPUTATIONAL MATERIALS SCIENCE, 153, 183-199 (2018).
Molecular dynamics (MD) simulations have been used extensively over the past two decades to determine the mechanical and physical properties of nanomaterials. However, the discrepancy between the reported results from these atomistic studies shadows the reliability of this computationally efficient technique. This inconsistency is attributed to the misuse and incorrect application of MD as evidenced by the arbitrary use of interatomic potentials, cut-off function parameters, strain rate, time increment, and domain size in the conducted simulations. In this paper, we highlight erroneous simulations by investigating the influence of these parameters on the elastic and fracture properties of nanostructured materials; including carbon nanotubes, graphene, and boron nitride (BN) sheets subject to direct and contact loads. The effect of interatomic potential type was investigated by comparing the predicted properties from AIREBO, Tersofff, CVFF, and ReaxFF potentials with those obtained with experimental and DFT techniques. The cut-off function parameters were also investigated to determine the optimum inner and outer cut-off radii selected to capture the actual physical behavior and avoid the reported strain hardening phenomena. Furthermore, MD simulations with strain rates spanning several orders of magnitudes and time increments ranging from 0.1 to 20 fs were performed to define the maximum allowable parameters for each material and loading scheme. Additionally, graphene and BN sheets with side length up to 500 angstrom were modeled to determine the size and edge effects on the mechanical properties. Finally, a set of parameters is recommended in each investigation to help guiding future atomistic studies obtaining reliable results using the available computational resources.
Return to Publications page