Second‐Order Memristor Based on All‐Oxide Multiferroic Tunnel Junction for Biorealistic Emulation of Synapses

Abstract

The brain has the ability to learn and evaluate as it receives and registers information. Signals between neurons are transmitted via synapses whose plasticity is modulated by usage. A bio-realistic electrical analog is a secondorder memristor, where the short-term internal dynamics influence the longterm state. It is achieved here with an all-oxide multiferroic tunnel junction: La 0.7 Sr 0.3 MnO 3 / BaTiO 3 / La 0.7 Sr 0.3 MnO 3 . Similar to the modulation of synaptic weight by stimuli, multi-levels of resistance may be encoded by mean of voltage pulses, on long time scale and in correlation with short-term effects. Neuromimetic learning functions are demonstrated: short and long-term potentiation/depression, paired-pulse facilitation/depression, spike-rate-and experience-dependent plasticity. The threshold frequency of pulse trains at which depression changes into potentiation depends on the previous activity, with a sliding effect as in neurobiology. The voltage pulses induce reversible changes of both dielectric polarization and oxygen vacancies distribution, generating transient trapped/detrapped charges at the two interfaces that govern the dynamic response. The resistance levels are determined by the final cationic ordering (the redox Mn 3+ /Mn 4+ ratio) at the two interfacial layers. The stimulation/relaxation dynamics are close to that in biological counterparts. Such memristors can be used in hardware artificial networks for advanced processing/storage of the information.

Effect of Al Concentration on Ferroelectric Properties in HfAlOx‐Based Ferroelectric Tunnel Junction Devices for Neuroinspired Applications

“…Similar to the short‐term memory of the human brain, we demonstrated that the shorter the learning interval of the device, the better remembered what is to be learnt. [ 78–80 ]…”

“…Similar to the short-term memory of the human brain, we demonstrated that the shorter the learning interval of the device, the better remembered what is to be learnt. [78][79][80] Finally, using the device as a physical reservoir, we implemented reservoir computing, which is a computing architecture for the efficient processing of sequential and temporal inputs based on the biological nervous system. [71,72,81,82] Reservoir computing includes a reservoir (which receives input and transforms it high dimensionally) and a readout (which is used for training), as shown in Figure 9a.…”

Improved Resistive Switching with Low-Power Synaptic Behaviors of ZnO/Al2O3 Bilayer Structure

“…In neuroscience, spike-rate-dependent plasticity (SRDP) is another essential synaptic function for modifications of memory learning processes that describes the synaptic plasticity dependency on the spike frequency [ 36 , 37 , 38 , 39 ]. Figure 5 a illustrates the frequency-dependent EPSC response ranges from 5 Hz to 100 Hz with an amplitude of +0.85 V/200 µs in the short-term memory (STM) mode.…”

“…The postsynaptic currents increase (learning process) due to the spike train frequency of 5 Hz and 10 Hz at the first stage, whereas at the second stage at spike train frequency of 2 Hz, the current increment or even decreasing trends (forgetting process) were observed. So, the potentiation and depression of synaptic weights are controlled by the pulse train frequency without the application of opposite polarity voltage, which mimics a bio-synapse [ 37 , 38 ]. Similar behavior was reported by Khanas et al and Yan et al previously [ 38 , 41 ].…”

“…So, the potentiation and depression of synaptic weights are controlled by the pulse train frequency without the application of opposite polarity voltage, which mimics a bio-synapse [ 37 , 38 ]. Similar behavior was reported by Khanas et al and Yan et al previously [ 38 , 41 ]. So, ITO/ZnO/Al 2 O 3 /TaN memristor shows the frequency-dependent synaptic weight change, and the potentiation/depression synaptic function can be emulated depending on the history of the frequency.…”