Abstract

This work focuses on the mechanical response of cubic-diamond nanoparticles of several sizes when subjected to a planar indenter. Three sequential stages were considered, i.e., loading, unloading, and reloading. In the large anisotropic strain regime, standard structure detectors stop identifying atoms as having diamond structures, affecting the ability to detect dislocations. A machine learning-assisted structure detector, MultiSOM, is able to detect a significantly larger number of crystalline diamond atoms and also identify much larger dislocation densities. MultiSOM also detects a distorted diamond phase and directional amorphization, similar to what has been observed for other covalent solids at high strain. After unloading, there is a large elastic recovery and significant amorphization remains. It is remarkable that dislocation density increases during unloading, unlike what happens for most materials, where there are large reductions due to dislocation reactions and surface sinks. This "anomalous" behavior is likely associated with low dislocation mobility in diamond, but also with a large number of junctions, which increases with dislocation density and reduces even further dislocation mobility. The unloaded state includes a dense dislocation network that withstands high-temperature annealing. Analysis of the vibrational density of states (VDOS) during recovery is consistent with significant recovery of the crystalline diamond phase. Reloading of the nanoparticles shows lower strength, without significant dislocation growth.