Dielectric-Engineered High-Speed, Low-Power, Highly Reliable Charge Trap Flash-Based Synaptic Device for Neuromorphic Computing beyond Inference

Abstract

The coming of the big-data era brought a need for power-efficient computing that cannot be realized in the Von Neumann architecture. Neuromorphic computing which is motivated by the human brain can greatly reduce power consumption through matrix multiplication, and a device that mimics a human synapse plays an important role. However, many synaptic devices suffer from limited linearity and symmetry without using incremental step pulse programming (ISPP). In this work, we demonstrated a charge-trap flash (CTF)-based synaptic transistor using trap-level engineered Al 2 O 3 /Ta 2 O 5 /Al 2 O 3 gate stack for successful neuromorphic computing. This novel gate stack provided precise control of the conductance with more than 6 bits. We chose the appropriate bias for highly linear and symmetric modulation of conductance and realized it with very short (25 ns) identical pulses at low voltage, resulting in low power consumption and high reliability. Finally, we achieved high learning accuracy in the training of 60000 MNIST images.

All-Solid-State Synaptic Transistors with Lithium-Ion-Based Electrolytes for Linear Weight Mapping and Update in Neuromorphic Computing Systems

“…Overall, increasing the thickness of the LATP layer improved the linearity characteristics of the conductance modulation, resulting in a higher pattern recognition accuracy and stability in neuromorphic systems with on-chip learning. Finally, to compare the synaptic characteristics of our synaptic transistors with those of other inorganic semiconductor-based synaptic transistor devices, the nonlinearity values and recognition accuracy obtained in this study and recently reported studies are summarized in Table . − Compared with other studies, the proposed device exhibits better linearity characteristics and a high pattern recognition accuracy (94.53%) owing to the optimization of the LATP layer with a high ionic conductivity.…”

Synaptic Characteristics and Neuromorphic Computing Enabled by Oxygen Vacancy Migration Based on Porous In2O3 Electrolyte-Gated Transistors

“…Consequently, the crosstalk problem in crossbar configurations can be avoided, and the degree of linearity of the synaptic weight can be remarkably improved. 10,11 Among the various three-terminal synaptic device structures, electrolyte-gated transistors (EGTs) with a channel composed of a transition metal oxide semiconductor have attracted considerable attention from the scientific community due to their low operating voltages. On top of that, their evolutionary structural design permits the facile implementation of various synaptic functionalities similar to the operation of natural biological systems.…”

“…As a result, the above-mentioned limitation in the two-terminal synaptic device can be effectively overcome. , In addition, synaptic devices with three terminals possess low read and write current values, reduced energy consumption, and quasi-linear weight changes. Consequently, the crosstalk problem in crossbar configurations can be avoided, and the degree of linearity of the synaptic weight can be remarkably improved. , …”

Stretchable, skin‐conformable neuromorphic system for tactile sensory recognizing and encoding

“…Inspired by biological systems, neuromorphic electronics have been developed to rebuild and enhance intelligent functions, 4 such as tactile perception, 5,6 artificial olfactory, 7,8 image recognition, 9,10 auditory communications, 11,12 and neuromorphic computing. 13,14 Encouraged by the urgent need for intelligent robotics and replacement prosthetics, remarkable progress has been made in neuromorphic tactile systems with hierarchical structures for sensing and processing tactile information synergistically (Table S1). 15,16 Tactile sensors, which mimic mechanoreceptors, are constructed by resistive devices and triboelectric nanogenerators (TENGs).…”

“…Neural encoding and learning are performed in the processes of collaborating and handling external information. Inspired by biological systems, neuromorphic electronics have been developed to rebuild and enhance intelligent functions, 4 such as tactile perception, 5,6 artificial olfactory, 7,8 image recognition, 9,10 auditory communications, 11,12 and neuromorphic computing 13,14 …”