Implementation of Simple but Powerful Trilayer Oxide-Based Artificial Synapses with a Tailored Bio-Synapse-Like Structure

Abstract

The ultimate aim of artificial synaptic devices is to mimic the features of biological synapses as closely as possible, in particular, its ability of self-adjusting the synaptic weight responding to the external stimulus. In this work, memristors, based on trilayer oxides with a stack structure of TiN/TiON/HfO y /HfO x /TiN, are designed to function as the artificial synapses where intrinsically designed oxygen-deficient HfO x layer, less oxygen-deficient HfO y layer, and TiON layer, imitating the corresponding biological functionality of the pre-synapse, synaptic cleft, and post-synapse, respectively, resemble the features of bio-synapses most closely. Thus, diverse bio-synaptic functions and plasticity, including long-term potentiation and depression, spike-rate-dependent plasticity, spike-timing-dependent plasticity, and metaplasticity, are fulfilled in these devices. Moreover, they exhibit analogue plasticity in both potentiating and depressing, fully emulating the learning protocols of excitation and inhibition in the bio-synapses. The structure and Hf/O distribution of these devices, mimicking the structure and Ca2+ deployment of bio-synapses, are consolidated by the high-resolution transmission electron microscopy and energy-dispersive X-ray spectroscopy, respectively. Powerful bio-realistic behavior, implemented in these simple artificial synaptic devices, make them tailored for neuromorphic hardware applications.

Polarization-perceptual anisotropic two-dimensional ReS2 neuro-transistor with reconfigurable neuromorphic vision

“…37,38 Interestingly, from this figure, the DSS also exhibits a strong polarization dependence relationship. Herein, according to the empirical model of symmetric STDP, the results can be well fitted with the following Gaussian equation 43…”

Hardware Implementation of Neuromorphic Computing Using Large‐Scale Memristor Crossbar Arrays

“…[139] However, most filamentary-type memristors suffer from the nonlinear and/or asymmetric weight update issues. [140][141][142][143][144] This is associated with the evolution process of the CFs that are basically divided into two stages. Specifically, the formation process is controlled by the drift process in the first stage, i.e., the CFs are forming, whereas when the CFs have formed, the thickening of CFs is controlled by the diffusion process.…”

Hybrid oxide brain-inspired neuromorphic devices for hardware implementation of artificial intelligence

“…By confining the Ag filament to the atomic scale, current switching characteristics can be observed, much different from that in thicker switching layer likely due to the significantly different atomic kinetics. By inserting a TiNO x buffer layer [158][159][160], oxide memristors exhibit effectively improved reliability and reduced power consumption. AlNO x [161] and SiO x N y [147] have also been used as the function layer in filamentary oxide memristive devices.…”