Reconfigurable Synaptic and Neuronal Functions in a V/VO_x/HfWO_x/Pt Memristor for Nonpolar Spiking Convolutional Neural Network (Adv. Funct. Mater. 23/2022)

A review of memristor: material and structure design, device performance, applications and prospects

“…Wang et al fabricated 2D MXene Ti 3 C 2 -based memristors and the power consumption was 0.35 pJ/bit, as illustrated in Figure 7(c) [ 114 ]. Fu et al fabricated a forming-free V/VO x /HfWO x /Pt memristor and power consumptions for writing and erasing are 0.06 pJ/bit and 0.9 pJ/bit, respectively [ 26 ]. Wang et al proposed an ultra-low power Ta 2 O 5 /Al 2 O 3 bilayer-gate-dielectric synaptic transistor and it showed a low power consumption of 19.9 aJ/bit [ 167 ].…”

“… Memristor: RS materials, structure design, performances, and applications. Reproduced with permission from [ 15 , 16 ] copyright 2021, Wiley, [ 17 ] copyright 2021, American Chemical Society, [ 18 ] copyright 2021, Springer Nature, [ 19 ] copyright 2021, American Chemical Society, [ 20 ] copyright 2018, Royal Society of Chemistry, [ 21 ] copyright 2022, Elsevier, [ 22 ] copyright 2020, Springer Nature, [ 23 ] copyright 2020, American Chemical Society, [ 24 ] copyright 2019, Wiley, [ 25 ] copyright 2022, American Chemical Society, [ 26 ] copyright 2022, Wiley. …”

A bi-functional three-terminal memristor applicable as an artificial synapse and neuron

“…The hardware integration of both artificial synapses and neurons into an SNN still faces enormous challenges in terms of process incompatibility. 9 A developed device that has both synaptic and neuronal functions would be an ideal mitigation. One approach is to leverage complementary metal-oxide-semiconductor (CMOS) transistor technologies, with the realization of neuronal functions via external or parasitic capacitors.…”

“…Despite having a highly scalable footprint, these devices are unable to perform synchronous read/write operations due to the shared read/write path. 7,9,[13][14][15] In addition, due to the underlying physics of the device switching mechanisms, the devices tend to have large spatial and temporal variations of conductance modulation under an external electric field. [16][17][18] To overcome this challenge, a more promising candidate appears to be the three-terminal memristor (3TM), offering decoupled writing and reading paths.…”

Achieving reinforcement learning in a three-active-terminal neuromorphic device based on a 2D vdW ferroelectric material

“…Throughout history, creating an artificial neuromorphic computing system with the capabilities of the human brain has been an essential source of artificial intelligence (AI) innovation. [1][2][3][4][5][6] The brain can flexibly handle diverse tasks in the changing real world due to its three-dimensional and complex architecture, where neurons are connected to each other. Generally, there are 8.3 Â 10 9 neurons and 6.7 Â 10 13 connections in the human cerebral cortex.…”