Neuromorphic-like dynamics in percolating copper nanoparticle networks: Synthesis, characterization, and limitations

To overcome the limitations of the conventional Von Neumann architecture, the inspiration from the mammalian brain has led to the development of nanoscale neuromorphic networks. In the present research, the synthesis of copper nanoparticle (NP) networks, produced using gas phase condensation based on high-pressure magnetron sputtering, is shown. To be the constituents of electrically percolating networks that exhibit complex, neuromorphic-like, spiking behavior at very low potentials in the mV range, satisfying well the requirement of low energy consumption, crystalline copper NPs with a necessary core-shell copper-copper oxide structure were synthesized. Characterization of the NPs using both scanning electron microscopy and scanning transmission electron microscopy revealed pristine shape, size, and density control of copper NPs, including their oxide shells and potential aggregation of the NPs. Furthermore, a comparison between previous results on molybdenum NP networks and this work is made to reveal the working principles behind copper NP networks. Performing and analyzing long-term electrical measurements did reveal unwanted characteristics and difficulties of this, and potentially previously researched, percolating NP networks. Our results show the switching dynamics of copper NP networks but also reveal several difficulties behind percolating NP networks for realistic neuromorphic systems.