Volatile and Nonvolatile Memristive Devices for Neuromorphic Computing

Abstract

(5 of 33)www.advelectronicmat.de state after operation an external stimulation (Figure 2f). As the increasing stimulation, the transition state entering to the metallic state leads to the Mott layer with low resistance state (LRS) (Figure 2g). The insulator to metal transition is in a timescale of femtosecond and picosecond, [69] while from the opposite transition from the metal state to the insulator state Figure 3. Second-order memristor for temporal information simulation. a) Conception of the second-order memristor. Adapted with permission. [51] Copyright 2015, American Chemical Society. b) Schematic of an artificial neuron consisting of dendrites, soma, and axon constructed by the secondorder memristor circuit. c) The temporal summation of excitatory postsynaptic currents (EPSCs) for the frequency-dependency conductance evolution of 2nd memristor. Adapted with permission. [52] Copyright 2018, Wiley. d) The second-order memristor networks consist of 128 inputs and 7 outputs for temporal learning simulation, from up to bottom denotes before training state with random weights and different learned motion speeds. Adapted with permission. [49] Copyright 2017, IEEE. e) Transient temperature evolution with Δt = 1 µs and Δt = 100 ns. Adapted with permission.

Low-voltage ultrafast nonvolatile memory via direct charge injection through a threshold resistive-switching layer

“…Revolutionary memory technologies with ultrahigh speed, ultralong retention, ultrahigh capacity, and ultralow energy consumption based on new principles, new materials, and new structures are highly demanded 4,[6][7][8][9][10][11] . In the past decade, memristors, including nonvolatile resistive switching (RS) and volatile threshold switching (TS) based on filamentary switching [12][13][14][15][16][17] , have attracted wide attention due to their high operation speed and high integration density. However, the relatively large cycle-to-cycle and device-to-device variation of memristors still limit their practical application in memory.…”

Metaplastic and energy-efficient biocompatible graphene artificial synaptic transistors for enhanced accuracy neuromorphic computing

“…However, in learning-related applications such as a continuously updating biological interface, the retention time shown here is sufficient. The most important figures of merit for neural network training and by extension, continuous learning, are low-energy dissipation per update, favorable synaptic plasticity, and good endurance 53,54 . Even in the case of adoption in neural network accelerators, though non-trivial, a solution to short retention time can be to use devices that have favorable synaptic characteristics during training (BLASTs), and to then transfer the weights to memory with long retention time for inference when training is finished 55 .…”

Multi-factor-controlled ReRAM devices and their applications

“…208 In particular, the resistance value of the memristor can be rapidly and reversibly switched under the applied voltage, which makes the neuromorphic computing chip or brain-like chip integrated by the memristor not only perform energy-saving computing, but can also be reprogrammable, which brings unparalleled advantages to neuromorphic computing. 209–212…”