Conventional computation techniques face challenges of deviations in Moore's law and the high‐power consumption of data‐centric computation tasks. Neuromorphic engineering attempts to overcome these issues by taking inspiration from neuron assemblies, ranging from distributed synaptic plasticity through orchestration of oscillator‐like action potential toward avalanche dynamics. Although solid networks of nanoparticles (NPs) are proven to replicate fingerprints of criticality and brain‐like dynamics, the aspect of dynamic spatial reconfigurations in the connectivity of networks remains unexplored. In this work, Ag/poly(ethylene glycol) (PEG) nanofluids are demonstrated as potential systems to mimic the spatio‐temporal reconfiguration of network connections. The nanofluids are prepared by directly loading Ag NPs from the gas aggregation cluster source into liquid PEG. The NPs exhibit a negative zeta potential in PEG; if the potential difference is applied between two electrodes submerged in this nanofluid, the NPs migrate toward the anode, accumulate in its vicinity, and form a conductive path. Spikes of electric current passing through the path are detected, accompanied by resistive switching phenomena, similar to the random switching dynamics in solid NPs networks. The unique behavior of Ag/PEG nanofluids makes them promising for the realization of spatio‐temporal reconfigurations in network topologies with the potential to transition to 3D.
High‐temperature plasmonics deals with optically active nanostructures that can withstand high temperatures. A conventional approach relying on standalone noble metal nanoparticles fails to deliver refractory plasmonic nanomaterials, and an alternative route envisions metal nitrides. The main challenge remains the development of advanced synthesis techniques and the insight into thermal stability under real‐life application conditions. Here, hafnium nitride nanoparticles (HfN NPs) can be produced by gas aggregation using reactive magnetron sputtering, a technique with a small environmental footprint are shown. As‐deposited NPs are of 10 nm mean size and consist of stoichiometric, crystalline fcc HfN. They are characterized by optical absorption below 500 nm caused by interband transitions and in the red/near‐infrared (NIR) region due to intraband transitions and localized surface plasmon resonance (LSPR). The optical response can be engineered by tuning the NP composition as predicted by finite‐difference time‐domain (FDTD) calculations. Going beyond the state‐of‐the‐art, the HfN NP thermal stability is focued under ultrahigh vacuum (UHV) and in air. During UHV annealing to 850 °C, the NPs retain their morphology, chemical and optical properties, which makes them attractive in space mission and other applications. During air annealing to 800 °C, HfN NPs remain stable until 250 °C, which sets a limit for air‐mediated use.
Developing novel memristive systems aims to implement key principles of biological neuron assemblies – plasticity, adaptivity, and self‐organization – into artificial devices for parallel, energy‐efficient computing. Solid‐state memristive devices, such as crossbar arrays and percolated nanoparticle (NP) networks, already demonstrate these properties. However, closer similarity to neural networks is expected from liquid‐state systems, including polymer melts, which remain largely unexplored. Here, the resistive switching in silver/poly(ethylene glycol) (Ag/PEG) nanofluids, prepared by depositing gas‐aggregated Ag NPs into PEGs of varying molecular mass, is investigated. These systems form long‐range conductive NP bridges with reconfigurable resistance states in response to an electric field. The zeta‐potential of Ag NPs and molecular mobility of PEG determine the prevalence of low resistance (ohmic) state, high resistance states (poor conductance) or intermediate transition states governed by space‐charge‐limited conduction or electron tunneling. The occurrence of these states is given by the interparticle gaps, which are determined by the conformation of PEG molecules adsorbed on the NPs. It is presented, for the first time, an equivalent circuit model for the Ag/PEG system. These findings pave the way to adopt polymer melts as matrices for neuromorphic engineering and bio‐inspired electronics.