Networks of silver nanowires appear set to replace expensive indium tin oxide as the transparent conducting electrode material in next generation devices. The success of this approach depends on optimising the material conductivity, which up to now has largely focused on minimising the junction resistance between wires. However, there have been no detailed reports on what the junction resistance is, nor is there a known benchmark for the minimum attainable sheet resistance of an optimised network. In this paper we present junction resistance measurements of individual silver nanowire junctions, producing for the first time a distribution of junction resistance values, and conclusively demonstrating that the junction contribution to the overall resistance can be reduced beyond that of the wires themselves through standard processing techniques. We

Nanowire networks are promising memristive architectures for neuromorphic applications due to their connectivity and neurosynaptic-like behaviours. Here, we demonstrate a self-similar scaling of the conductance of networks and the junctions that comprise them. We show this behavior is an emergent property of any junction-dominated network. A particular class of junctions naturally leads to the emergence of conductance plateaus and a “winner-takes-all” conducting path that spans the entire network, and which we show corresponds to the lowest-energy connectivity path. The memory stored in the conductance state is distributed across the network but encoded in specific connectivity pathways, similar to that found in biological systems. These results are expected to have important implications for development of neuromorphic devices based on reservoir computing.

Graphical abstract

In this work, we introduce a combined experimental and computational approach to describe the conductivity of metallic nanowire networks. Due to their highly disordered nature, these materials are typically described by simplified models in which network junctions control the overall conductivity. Here, we introduce a combined experimental and simulation approach that involves a wire-by-wire junction-byjunction simulation of an actual network. Rather than dealing with computer-generated networks, we use a computational approach that captures the precise spatial distribution of wires from an SEM analysis of a real network. In this way, we fully account for all geometric aspects of the network, i.e. for the properties of the junctions and wire segments. Our model predicts characteristic junction resistances that are smaller than those found by earlier simplified models. The model outputs characteristic values that depend on the detailed connectivity of the network, which can be used to compare the performance of different networks and to predict the optimum performance of any network and its scope for improvement.

Motivated by numerous technological applications, there is current interest in the study of the conductive properties of networks made of randomly dispersed nanowires. The sheet resistance of such networks is normally calculated by numerically evaluating the conductance of a system of resistors but due to disorder and with so many variables to account for, calculations of this type are computationally demanding and may lack mathematical transparency. Here we establish the equivalence between the sheet resistance of disordered networks and that of a regular ordered network. Rather than through a fitting scheme, we provide a recipe to find the effective medium network that captures how the resistance of a nanowire network depends on several different parameters such as wire density, electrode size and electrode separation. Furthermore, the effective medium approach provides a simple way to distinguish the sheet resistance contribution of the junctions from that of the nanowires themselves. The contrast between these two contributions determines the potential to optimize the network performance for a particular application.

Networks of metallic nanowires have the potential to meet the needs of next-generation device technologies that require flexible transparent conductors. At present, there does not exist a first principles model capable of predicting the electro-optical performance of a nanowire network. Here we combine an electrical model derived from fundamental material properties and electrical equations with an optical model based on Mie theory scattering of light by small particles. This approach enables the generation of analogues for any nanowire network and then accurately predicts, without the use of fitting factors, the optical transmittance and sheet resistance of the transparent electrode. Predictions are validated using experimental data from the literature of networks comprised of a wide range of aspect ratios (nanowire length/diameter). The separation of the contributions of the material resistance and the junction resistance allows the effectiveness of post-deposition processing methods to be evaluated and provides a benchmark for the minimum attainable sheet resistance. The predictive power of this model enables a material-by-design approach, whereby suitable systems can be prescribed for targeted technology applications.

In spite of the strong interest in brain-like or neuromorphic computation, relatively few devices have emerged whose neuromorphic behavior is embedded in the hardware itself and not reliant on external programming of synaptic weights. We describe here a neuromorphic device based on a TiO2 nanowire that exhibits an associative memory response to the time correlation between voltage and optical stimuli. Memristive characteristics are also observed with current-voltage sweeps showing hysteresis loops and continuum resistance levels. The nanowire device responds to heterogeneous voltage and optical pulse stimuli with spike-like neuromorphic outputs. Moreover, uncorrelated pulses produce a weak response, consistent with the interaction of coincident pulses with adsorbed and bulk oxygen in the surface depletion region, leading to a nonlinear enhancement in conductance. The strength of this learned enhancement depends on the both the time correlation and number of pulse stimuli, consistent with spike timing dependent plasticity. The nanowire devices presented have neural synapse-like properties that could serve as a building block for neuromorphic computation.

Nonpolar resistive switching (RS), a combination of bipolar and unipolar RS, is demonstrated for the first time in a single nanowire (NW) system. Exploiting Ag@TiO core-shell (CS) NWs synthesized by postgrowth shell formation, the switching mode is controlled by adjusting the current compliance effectively, tailoring the electrical polarity response. We demonstrate ON/OFF ratios of 10 and 10 for bipolar and unipolar modes, respectively. In the bipolar regime, retention times could be controlled up to 10 s, and in the unipolar mode,>10 s was recorded. We show how the unique dual-mode switching behavior is enabled by the defect-rich polycrystalline material structure of the TiO shell and the interaction between the Ag core and the Ag electrodes. These results provide a foundation for engineering nonpolar RS behaviors for memory storage and neuromorphic applications in CSNW structures.