Self-driven friction nano-generation synaptic transistor

Abstract

The invention relates to a self-driven friction nano-generation synaptic transistor. The synaptic transistor comprises a friction nano-generator and a synaptic transistor, and further comprises: a substrate; an electrode layer formed on the substrate; a shared intermediate layer formed on the electrode layer; a synaptic transistor active layer, a source electrode and a drain electrode which formedon the shared intermediate layer; and a positive friction layer and a negative friction layer which are formed on the shared intermediate layer. The shared intermediate layer serves as a dielectric layer of the synaptic transistor and an intermediate layer of the friction nano-generator, the output voltage of the friction nano-generator is increased, the synaptic function of the synaptic transistor can be realized, the electrode layer is used as an output electrode of the friction nano-generator and a gate electrode of thesynaptic transistor, is simple in structure, light and flexible, and rubs the positive friction layer or the negative friction layer, and the shared intermediate layer generates pulse voltage, generates excitatory post-synapse current between the source electrode and thedrain electrode without external power supply, and realizes a bionic synaptic transistor function through self-driving.

Description

technical field [0001] The invention relates to a self-driven triboelectric nanometer power generation synapse transistor, which belongs to the technical field of semiconductor devices. Background technique [0002] The demand for big data and efficient human-computer interaction brought by the era of intelligence has brought higher requirements to the processing and storage of complex information. Compared with the low power and small size of the human brain, the current computer system has obvious shortcomings. In recent years, imitating the information processing method of the human brain and developing a neural network-like computing system provides a potential way to efficient computing. Most of the current neural network systems use traditional complementary metal-oxide-semiconductor technology. Although computing functions similar to neural networks can be realized by combining code algorithms, their efficiency is still far lower than that of the human brain. The fun...

AI Technical Summary

Problems solved by technology

However, in recent years, research has mainly realized the integration of bionic synaptic transistors and stimulus acquisition terminals through additional complex circuits and requires additional power supply, which is not conducive to the application of flexible electronic devices, wearable devices and bionic skin.

Images