Progress and Challenges for Memtransistors in Neuromorphic Circuits and Systems

Abstract

Due to the increasing importance of artificial intelligence (AI), significant recent effort has been devoted to the development of neuromorphic circuits that seek to emulate the energy‐efficient information processing of the brain. While non‐volatile memory (NVM) based on resistive switches, phase‐change memory, and magnetic tunnel junctions has shown potential for implementing neural networks, additional multi‐terminal device concepts are required for more sophisticated bio‐realistic functions. Of particular interest are memtransistors based on low‐dimensional nanomaterials, which are capable of electrostatically tuning memory and learning behavior at the device level. Herein, a conceptual overview of the memtransistor is provided in the context of neuromorphic circuits. Recent progress is surveyed for memtransistors and related multi‐terminal NVM devices including dual‐gated floating‐gate memories, dual‐gated ferroelectric transistors, and dual‐gated van der Waals heterojunctions. The different materials systems and device architectures are classified based on the degree of control and relative tunability of synaptic behavior, with an emphasis on device concepts that harness the reduced dimensionality, weak electrostatic screening, and phase‐changes properties of nanomaterials. Finally, strategies for achieving wafer‐scale integration of memtransistors and multi‐terminal NVM devices are delineated, with specific attention given to the materials challenges for practical neuromorphic circuits.

Highly Linear and Symmetric Synaptic Memtransistors Based on Polarization Switching in Two‐Dimensional Ferroelectric Semiconductors

“…[14][15][16] In this respect, memtransistor has risen as an outstanding multiterminal memristive device to circumvent existing limitations in memristors. [17][18][19][20] As a hybrid form of memristor and transistor, memtransistor can realize nondestructive reading and fine gate-tunable memristive switching behaviors. Particularly, two-dimensional (2D) polycrystalline MoS 2 based memtransistors have been extensively studied for nonvolatile memory and artificial synapses applications.…”

Higher order neural processing with input-adaptive dynamic weights on MoS2 memtransistor crossbars

“…Most importantly, the dual-gated memtransistor (Lee et al, 2020) enables not only gate-tunable learning, like the single-gated memtransistor (Sangwan et al, 2018), but also permits efficient scaling into a crossbar array configuration by suppression of sneak currents, unlike the singlegated memtransistor. Memtransistor-based spiking neuron implementations were discussed in prior works Yuan et al, 2021;Yan et al, 2021b whereas this paper focuses on higher order deep learning using the devices.…”

Advances in Emerging Photonic Memristive and Memristive‐Like Devices

“…In recent years, many emerging approaches, such as exploiting the phase transition properties of materials and the charge trapping behavior of flash memory, have been implemented to optimize these issues and broaden their application scenarios. [ 23 , 24 , 25 , 26 ] Whether these unique devices can be classified as memristors is debatable at this stage even though they exhibit hysteresis curve behavior similar to memristors. [ 27 , 28 , 29 ] In order to avoid confusion, the name “memristive‐like devices” is preferred to use (memristive devices and memristive‐like devices are strictly distinguished in this review).…”