Role played by phonon drag on accuracy of MD simulations of nanowires due to deficiently selected strain rates

Abstract

The literature contains numerous articles devoted to examining the mechanical behavior of nanowires (NWs) using molecular dynamics simulations. Many of these investigations have selected improper strain rates leading to erroneous results concerning ductile–brittle transition. In this study, we tested this hypothesis and proved that such transition in the material behavior existed due to the improper selection of strain rates which eventually changes the propagation velocity of phonons in the conducted atomistic simulations. In the current study, we subjected gold nanowires (Au NWs) with a diameter of 100 Å and lengths ranging from 25 to 1000 Å to varied strain rates. Specifically, we examined the effect of the rate of deformation of the NW upon its mechanical behaviour by dividing its length into several stations along its entire length to capture the strain distribution in each segment along that length. Five orders of magnitudes of strain rates were applied in our work for studying the influence of rate of deformation on the strain distribution along the NW length. The results of our molecular dynamics simulations show that smaller strain rates were necessary for modeling relatively long (> 150 Å) NWs to ensure the transmission of the applied loads through the entire NW length to suppress phonon drag effect. On the other hand, relatively short (< 25 Å) NWs experience large variations in the axial strain along the NW length; with smaller strains near the ends and higher strains at the middle section. As a result, relatively short NWs exhibit higher elastic moduli than longer ones and the NW length’s effect diminishes at lengths exceeding 150 Å. Location of necking, under the application of higher strain rate, shifts away from the loading end of NW towards its middle portion with the decrease in the NW length due to the phonon drag. The slope of the stress–strain curves was found to significantly depend on the NW length, and thus, using the same strain rate over a large range of NW lengths will lead to erroneous results.