Molecular dynamics simulation of nanofilament breakage in neuromorphic nanoparticle networks

Neuromorphic computing systems may be the future of computing and cluster-based networks are a promising architecture for the realization of these systems. The creation and dissolution of synapses between the clusters are of great importance for their function. In this work, we model the thermal breakage of a gold nanofilament located between two gold nanoparticles via molecular dynamics simulations to study on the mechanisms of neuromorphic nanoparticle-based devices. We employ simulations of Au nanowires of different lengths (20-80 angstrom), widths (4-8 angstrom) and shapes connecting two Au-1415 nanoparticles (NPs) and monitor the evolution of the system via a detailed structural identification analysis. We found that atoms of the nanofilament gradually aggregate towards the clusters, causing the middle of wire to gradually thin and then break. Most of the system remains crystalline during this process but the center is molten. The terminal NPs increase the melting point of the NWs by fixing the middle wire and act as recrystallization areas. We report a strong dependence on the width of the NWs, but also their length and structure. These results may serve as guidelines for the realization of cluster-based neuromorphic computing systems.

Automated summary

This research investigates the mechanisms of neuromorphic nanoparticle-based devices through molecular dynamics simulations. The study reveals that atoms of the nanofilament aggregate towards the clusters, causing the middle of the wire to thin and break. The terminal nanoparticles fix the wire and act as recrystallization areas. Additionally, the width, length, and structure of the nanowires greatly influence the performance of the system.