CMOS-compatible neuromorphic devices for neuromorphic perception and computing: a review

Abstract

Neuromorphic computing is a brain-inspired computing paradigm that aims to construct efficient, low-power, and adaptive computing systems by emulating the information processing mechanisms of biological neural systems. At the core of neuromorphic computing are neuromorphic devices that mimic the functions and dynamics of neurons and synapses, enabling the hardware implementation of artificial neural networks. Various types of neuromorphic devices have been proposed based on different physical mechanisms such as resistive switching device and electric-double-layer transistors. These devices have demonstrated a range of neuromorphic functions such as multistate storage, spike-timing-dependent plasticity, and dynamic filtering, etc. To achieve high performance neuromorphic computing systems, it is essential to fabricate neuromorphic devices compatible with complementary metal oxide semiconductor (CMOS) manufacturing process. This improves the device reliability and stability and is favourable to achieve neuromorphic chips with higher integration density and low power consumption. This review summarizes CMOS-compatible neuromorphic devices and discusses their emulation of synaptic and neuronal functions as well as their applications in neuromorphic perception and computing. We highlight challenges and opportunities for further development of CMOS-compatible neuromorphic devices and systems.

Recent Progress of Polymeric Dielectrics in Molybdenum Disulfide-Based Devices

“…Neuromorphic devices are a class of high-efficiency, low-power, and adaptive computing systems built by simulating the information sensing, processing, and memorizing mechanism of biological nervous systems, which lay a solid foundation for breaking through the dilemma of conventional von Neumann architecture. 113 Three-terminal FET is commonly used for basic research on neuromorphic devices, in which the channel resistance is modulated via controlling the gate to simulate the changes in synaptic weights. 114 This process is equivalent to action potentials in biological synapses, causing the release of neurotransmitters, resulting in a change in synaptic connection weights.…”

Preparation of MXene-based hybrids and their application in neuromorphic devices

“…Before the advent of memristor, there were three basic electronic components, namely resistor, capacitor, and inductor [4]. Currently, memristors have been acknowledged as highly promising candidate components in constructing neuromorphic computers, owing to their simple construction, smoothly variable conduction state, extremely low energy consumption, sufficiently high switching rate, and high compatibility with CMOS technology [5][6][7][8][9][10][11]. A memristor can be constructed using a certain resistive material as an insulator layer, which is connected with two metal electrodes to form a metal-insulator-metal architecture.…”

Fabrication and integration of photonic devices for phase-change memory and neuromorphic computing

“…These breakthroughs demonstrate a promising path for manufacturing large-scale silicon photonic-electronic chips based on the rapid MPW prototyping service. By leveraging photonic packaging [16], such as photonic wire bonding [17] and flip-chip die bonding for photonic-electronic co-integration [18], silicon photonic chips can be calibrated, reprogrammed, and trained efficiently by using electronic microcontrollers with high reproducibility for neuromorphic computing [19,20].…”