Here we provide a perspective concept of neurohybrid memristive chip based on the combination of living neural networks cultivated in microfluidic/microelectrode system, metal-oxide memristive devices or arrays integrated with mixed-signal CMOS layer to control the analog memristive circuits, process the decoded information, and arrange a feedback stimulation of biological culture as parts of a bidirectional neurointerface. Our main focus is on the state-of-the-art approaches for cultivation and spatial ordering of the network of dissociated hippocampal neuron cells, fabrication of a large-scale cross-bar array of memristive devices tailored using device engineering, resistive state programming, or non-linear dynamics, as well as hardware implementation of spiking neural networks (SNNs) based on the arrays of memristive devices and integrated CMOS electronics. The concept represents an example of a brain-on-chip system belonging to a more general class of memristive neurohybrid systems for a new-generation robotics, artificial intelligence, and personalized medicine, discussed in the framework of the proposed roadmap for the next decade period.

/pdf/neurohybrid-memristive-cmos-integrated-systems-for-3jydyczdtn.pdf

Neurohybrid Memristive CMOS-Integrated Systems for Biosensors and Neuroprosthetics.

We propose a stochastic model for a memristive system by generalizing known approaches and experimental results. We validate our theoretical model by experiments carried out on a memristive device based on Au/Ta/ZrO2(Y)/Ta2O5/TiN/Ti multilayer structure. In the framework of the proposed model we obtain the exact analytic expressions for stationary and nonstationary solutions. We analyze the equilibrium and non-equilibrium steady-state distributions of the internal state variable of the memristive system and study the influence of fluctuations on the resistive switching, including the relaxation time to the steady-state. The relaxation time shows a nonmonotonic dependence, with a minimum, on the intensity of the fluctuations. This paves the way for using the intensity of fluctuations as a control parameter for switching dynamics in memristive devices. N V Agudov et al Nonstationary distributions and relaxation times in a stochastic model of memristor Printed in the UK 024003 JSMTC6 © 2020 IOP Publishing Ltd and SISSA Medialab srl 2020 20 J. Stat. Mech.

/pdf/nonstationary-distributions-and-relaxation-times-in-a-337u7dmjve.pdf

Nonstationary distributions and relaxation times in a stochastic model of memristor

The stochastic resonance phenomenon has been studied experimentally and theoretically for a state-of-art metal-oxide memristive device based on yttria-stabilized zirconium dioxide and tantalum pentoxide, which exhibits bipolar filamentary resistive switching of anionic type The effect of white Gaussian noise superimposed on the sub-threshold sinusoidal driving signal is analyzed through the time series statistics of the resistive switching parameters, the spectral response to a periodic perturbation and the signal-to-noise ratio at the output of the nonlinear system The stabilized resistive switching and the increased memristance response are revealed in the observed regularities at an optimal noise intensity corresponding to the stochastic resonance phenomenon and interpreted using a stochastic memristor model taking into account an external noise source added to the control voltage The obtained results clearly show that noise and fluctuations can play a constructive role in nonlinear memristive systems far from equilibrium

Stochastic resonance in a metal-oxide memristive device

The low-frequency noise in a nanometer-sized virtual memristor consisting of a contact of a conductive atomic force microscope (CAFM) probe to an yttria stabilized zirconia (YSZ) thin film deposited on a conductive substrate is investigated. YSZ is a promising material for the memristor application since it is featured by high oxygen ion mobility, and the oxygen vacancy concentration in YSZ can be controlled by varying the molar fraction of the stabilizing yttrium oxide. Due to the low diameter of the CAFM probe contact to the YSZ film (∼10 nm), we are able to measure the electric current flowing through an individual filament both in the low resistive state (LRS) and in the high resistive state (HRS) of the memristor. Probability density functions (Pdfs) and spectra of the CAFM probe current in both LRS and HRS are measured. The noise in the HRS is found to be featured by nearly the same Pdf and spectrum as the inner noise of the experimental setup. In the LRS, a flicker noise 1/fγ with γ ≈ 1.3 is observed in the low-frequency band (up to 8 kHz), which is attributed to the motion (drift/diffusion) of oxygen ions via oxygen vacancies in the filament. Activation energies of oxygen ion motion determined from the flicker noise spectra are distributed in the range of [0.52; 0.68] eV at 300 K. Knowing these values is of key importance for understanding the mechanisms of the resistive switching in YSZ based memristors as well as for the numerical simulations of memristor devices.

Measurement of the activation energies of oxygen ion diffusion in yttria stabilized zirconia by flicker noise spectroscopy

Multilayer Metal-Oxide Memristive Device with Stabilized Resistive Switching

The breakthrough in electronics and information technology is anticipated by the development of emerging memory and logic devices, artificial neural networks and brain-inspired systems on the basis of memristive nanomaterials represented, in a particular case, by a simple ‘metal–insulator–metal’ (MIM) thin-film structure. The present article is focused on the comparative analysis of MIM devices based on oxides with dominating ionic (ZrOx, HfOx) and covalent (SiOx, GeOx) bonding of various composition and geometry deposited by magnetron sputtering. The studied memristive devices demonstrate reproducible change in their resistance (resistive switching – RS) originated from the formation and rupture of conductive pathways (filaments) in oxide films due to the electric-field-driven migration of oxygen vacancies and / or mobile oxygen ions. It is shown that, for both ionic and covalent oxides under study, the RS behaviour depends only weakly on the oxide film composition and thickness, device geometry (down to a device size of about 20×20 μm2). The devices under study are found to be tolerant to ion irradiation that reproduces the effect of extreme fluences of high-energy protons and fast neutrons. This common behaviour of RS is explained by the localized nature of the redox processes in a nanoscale switching oxide volume. Adaptive (synaptic) change of resistive states of memristive devices is demonstrated under the action of single or repeated electrical pulses, as well as in a simple model of coupled (synchronized) neuron-like generators. It is concluded that the noise-induced phenomena cannot be neglected in the consideration of a memristive device as a nonlinear system. The dynamic response of a memristive device to periodic signals of complex waveform can be predicted and tailored from the viewpoint of stochastic resonance concept. (© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)

Field- and irradiation-induced phenomena in memristive nanomaterials

Design and training principles have been proposed and tested for an artificial neural network based on metal-oxide thin-film nanostructures possessing bipolar resistive switching (memristive) effect. Experimental electronic circuit of neural network is implemented as a double-layer perceptron with a weight matrix composed of 32 memristive devices. The network training algorithm takes into account technological variations of the parameters of memristive devices. Despite the limited size of weight matrix the developed neural network model is scalable and capable of solving nonlinear classification problems. The learning and functionality of the network are demonstrated by using its computer model for the classification of activity propagation directions in simulated neuronal culture.

One-Board Design and Simulation of Double-Layer Perceptron Based on Metal-Oxide Memristive Nanostructures

Construction and training principles have been proposed and tested for an artificial neural network based on metal-oxide thin-film nanostructures possessing bipolar resistive switching (memristive) effect. Experimental electronic circuit of neural network is implemented as a double-layer perceptron with a weight matrix composed of 32 memristive devices. The network training algorithm takes into account technological variations of the parameters of memristive nanostructures. Despite the limited size of weight matrix the developed neural network model is well scalable and capable of solving nonlinear classification problems.

Alexander N. Sharapov

Papers

Field- and irradiation-induced phenomena in memristive nanomaterials

Multilayer Metal-Oxide Memristive Device with Stabilized Resistive Switching

One-Board Design and Simulation of Double-Layer Perceptron Based on Metal-Oxide Memristive Nanostructures

Towards Hardware Implementation of Double-Layer Perceptron Based on Metal-Oxide Memristive Nanostructures.