Memristive Junctions Research Articles

Elucidation of Switching Mechanisms in Memristive Junctions Integrating a Iron(II)‐Ter Pyridine Diazoted Complex

AbstractAn original way of elaborating vertical metal/molecules/metal memristive junctions through diazonium electrografting of the organic layer and inkjet‐printed top electrodes is reported here. The molecule of interest is a FeII coordination complex with ter‐pyridine ligands, having a diazonium anchoring group. The resulting junction exhibits a memristive behavior characterized by a high ON/OFF ratio and plasticity property. Through the application of advanced techniques such as UV–vis and Raman time‐resolved spectroelectrochemistry, the study demonstrates the significant role of switchable azo bonds derived from diazo electrografting in memristive behavior.

Exploiting device-level non-idealities for adversarial attacks on ReRAM-based neural networks

Resistive memory (ReRAM) or memristor devices offer the prospect of more efficient computing. While memristors have been used for a variety of computing systems, their usage has gained significant popularity in the domain of deep learning. Weight matrices in deep neural networks can be mapped to crossbar architectures with memristive junctions, generally resulting in superior performance and energy efficiency. However, the nascent nature of ReRAM technology is directly associated with the presence of inherent non-idealities in the ReRAM devices currently available. Deep neural networks have already been shown to be susceptible to adversarial attacks, often by targeting vulnerabilities in the networks’ internal representation of input data. In this paper, we explore the causal relationship between device-level non-idealities in ReRAM devices and the classification performance of memristor-based neural network accelerators. Specifically, our aim is to generate images which bypass adversarial defense mechanisms in software neural networks but trigger non-trivial performance discrepancies in ReRAM-based neural networks. To this end, we have proposed a framework to generate adversarial images in the hypervolume between the two decision boundaries, thereby leveraging non-ideal device behavior for performance detriment. We employ state-of-the-art tools in explainable artificial intelligence to characterize our adversarial image samples, and derive a new metric to quantify susceptibility to adversarial attacks at the pixel and device-levels.

Avalanches and edge-of-chaos learning in neuromorphic nanowire networks

The brain’s efficient information processing is enabled by the interplay between its neuro-synaptic elements and complex network structure. This work reports on the neuromorphic dynamics of nanowire networks (NWNs), a unique brain-inspired system with synapse-like memristive junctions embedded within a recurrent neural network-like structure. Simulation and experiment elucidate how collective memristive switching gives rise to long-range transport pathways, drastically altering the network’s global state via a discontinuous phase transition. The spatio-temporal properties of switching dynamics are found to be consistent with avalanches displaying power-law size and life-time distributions, with exponents obeying the crackling noise relationship, thus satisfying criteria for criticality, as observed in cortical neuronal cultures. Furthermore, NWNs adaptively respond to time varying stimuli, exhibiting diverse dynamics tunable from order to chaos. Dynamical states at the edge-of-chaos are found to optimise information processing for increasingly complex learning tasks. Overall, these results reveal a rich repertoire of emergent, collective neural-like dynamics in NWNs, thus demonstrating the potential for a neuromorphic advantage in information processing.

Collective capacitive and memristive responses in random nanowire networks: Emergence of critical connectivity pathways

Random nanowire networks (NWNs) are promising synthetic architectures for non-volatile memory devices and hardware-based neuromorphic applications due to their history-dependent responses, recurrent connectivity, and neurosynaptic-like behaviors. Such brain-like functions occur due to emergent resistive switching phenomena taking place in the interwire junctions which are viewed as memristive systems; they operate as smart analogue switches whose resistance depends on the history of the input voltage/current. We successfully demonstrated that NWNs made with a particular class of memristive junctions can exhibit a highly-selective conduction mechanism which uses the lowest-energy connectivity path in the network identified as the “winner-takes-all” state. However, these complex networks do not always behave in the same fashion; in the limit of sufficiently low input currents (preceding this selective conduction regime), the system behaves as a leakage capacitive network and its electrical activation is driven by cascades of breakdown-based activation events involving binary capacitive transitions. Understanding these two regimes is crucial to establish the potential of these materials for neuromorphics, and for this, we present two computational modelling schemes designed to describe the capacitive and memristive responses of NWNs interrogated adiabatically by voltage/current sources. Our results are corroborated by experimental evidence that reveal the fine electrical properties of NWN materials in their respective formation (capacitive) and conducting (memristive) stages.

Solution-Processed Memristive Junctions Used in a Threshold Indicator

Inkjet-patterned memristive metal/oxide/metal structures were characterized to infer ionic and electronic transport parameters such as mobility and ion distribution. The conductance change with respect to voltage and time was measured for an individual crossbar junction and for junctions that are connected in series. It was found that continuous dopant redistribution during voltage scans led to a peak in conductance. The mutable conductance of memristive devices was utilized to demonstrate a threshold indicator, in which the crossbar junctions were connected with a piezo sensor input and an electrophoretic display output. The memristive circuit would switch the color of display pixels, depending on the number of input pulses sensed by the piezo. The threshold indicator demonstrates that memristive junctions can be used as integral control switches. It is composed of passive circuit components and does not require an external battery.

Redox-controlled memristive switching in the junctions employing Ti reactive electrodes

We have proposed a kind of memristive device based on the junctions employing Ti as the reactive electrodes. The role of electrically-derived redox of Ti in such memristive switching is shown. The structural and chemical evidence of the electrically-derived oxidation is presented by TEM and XPS experiment, respectively. Due to the redox of the top electrode Ti and the consequent drift of oxygen vacancies, the device shows two distinct resistance states under a sweeping voltage loading. ON state is controlled by tunneling process, while OFF state is controlled by Schottky emission conductive mechanism. The failure behaviors of such memristive junctions are also discussed. In the light of the redox principle, we demonstrate that the devices could be recovered by loading a long electrical reduction treatment.

Radiation Hardness of ${\rm TiO}_{2}$ Memristive Junctions

Semiconducting TiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> displays non-volatile multi-state, hysteretic behavior in its I-V characteristics that can be exploited as a memory material in a memristive device. We exposed memristive TiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> devices in the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">on</i> and <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">off</i> resistance states to 45 Mrad(Si) of ~1-MeV gamma radiation and 23 Mrad(Si) of 941-MeV Bi-ions under zero bias conditions and none of the devices were degraded. These results suggest that TiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> memristive devices are good candidates for radiation hard electronics for aerospace.

Morphological and electrical changes in TiO2 memristive devices induced by electroforming and switching

AbstractCombining delamination technique with conductive AFM, we have been able to reveal spatially resolved morphology and conductance changes in TiO2 memristive junctions after electroforming and switching. Being able to distinguish between effects caused by electroforming and switching, respectively, we could demonstrate that electroforming results in the creation of localized conductance channels induced by oxygen evolution while subsequent resistive switching causes an additional conducting structure next to the forming spot. We observe that the lateral extent of this structure depends on the number of switching cycles indicating an ongoing breaking of existing and creation of neighbouring current channels during subsequent switching. (© 2010 WILEY‐VCH Verlag GmbH &amp; Co. KGaA, Weinheim)