Research group of sun Qijun, Beijing Institute of nano energy, Chinese Academy of Sciences: piezoelectric modulation of graphene artificial sensing synapse

There are about 1011 neurons and about 1015 synaptic connections in the lead human brain Synaptic structure is the key part of information transmission between neurons and the basic unit of cognitive behavior of human brain Therefore, the development of artificial synaptic devices is of great significance for neural morphological engineering In recent years, brain like neuromorphological devices are becoming an important branch in the field of artificial intelligence and neuromorphology, which will inject new vitality into the development of artificial intelligence in the future At present, there are two main types of artificial synaptic devices reported in the world, i.e two terminal resistive devices and three terminal transistor devices Ionic liquids and ionic gel electrolytes have unique ionic interface coupling characteristics and related interface electrochemical processes They have very strong application prospects in neuromorphological devices and systems In addition, piezoelectric Electronics (NAT Rev Mater 2016, 24, 23) as a new interdisciplinary research field, Academician Wang Zhonglin proposed that piezoelectric potential is used as a grid to control the generation, transmission, separation or recombination of carriers at the interface, which provides a "active" trigger and sensing prototype with corresponding electrical output signals for the coupling effect between piezoelectric polarization caused by mechanical stimulation and semiconductor transmission characteristics At present, the development of piezoelectric electronics is not limited to the interface regulation of metal semiconductor or PN junction by materials with both semiconductor and piezoelectric properties The semiconductor devices driven / controlled by piezoelectric potential can be included in the broader category of piezoelectric electronics On the basis of piezoelectric potential modulation semiconductor devices, sun Qijun research group of Beijing Institute of nano energy and system of Chinese Academy of Sciences has made a new breakthrough in this field Relevant research results were published in adv funct Mater (DOI: 10.1002/adfm.201900959) Introduction to the research group of researcher sun Qijun there are 10 researchers in the research group, including 5 doctoral students and 5 Master students At present, the research group relies on Beijing Institute of nano energy and systems, Chinese Academy of Sciences, with the core technology of nano energy and nano system as the research and development goal, and carries out basic and applied basic research in the fields of piezoelectric / friction potential modulation semiconductor devices, electronic skin sensors, wearable self charging energy packs, etc The representative achievements are as follows: (1) piezoelectric regulated semiconductor devices: regulated channel Fermi level (ACS Nano 2019, 13, 582); regulated Schottky barrier (nano energy 2018, 50, 598); piezoelectric write in nonvolatile memory (ACS Nano 2016, 10, 11037); piezoelectric driven electronic skin (adv mater 2015, 27, 3411) (2) Friction potential regulated semiconductor devices: friction modulated molybdenum sulfide double layer transistor (adv mater 2019, 31, 1806905); friction potential regulated double gate transistor (adv mater 2018, 30, 1705088); direct contact friction electronics electronic skin (ACS Nano 2018, 12, 9381) (3) Electronic skin sensor: self driven multi-stage sensor array (ACS Nano 2018, 12, 254); high sensitive humidity sensor array (adv mater 2017, 29, 1702076); extensible multifunctional all graphene sensor (adv mater 2016, 28, 2601) (4) Piezoelectric / triboelectric potential controlled micro nano devices: piezoelectric / triboelectric self charging intelligent color changing energy package: (adv energy mater 2018, 8, 1800069); piezoelectric / triboelectric high stability nano generator (nano energy 2019, 57, 440) Cutting edge research results: with the rise of artificial intelligence, the piezoelectric modulated graphene artificial sensing synapse is more and more demanding for high-performance distributed computing The classical von Neumann architecture is good at dealing with preset problems, providing accurate calculation and solving structural problems in sequence In an intelligent society, traditional computers can only be forced to preset answers to a large number of human-computer interaction sensing information, and its complexity will increase exponentially Therefore, it is necessary to find an active and event driven information processing method Touch operation is the mainstream interactive means of current intelligent devices According to the idea of processing tactile information with decentralized computing model, sun Qijun's research group of Beijing nano Energy Institute, Chinese Academy of Sciences proposed a graphene human sensing synapse that can adjust the weight through the temporal and spatial characteristics of external strain (Figure 1) The system consists of sensing, transmission and processing units, which can be simply regarded as a sensory nervous system Based on the electric double layer formed at the interface between the ionic gels and graphene, the piezoelectric potential can replace the gate pressure to effectively control the artificial synapses This work represents a further step in the direction of using active neuromorphological electronic skin for robots and artificial limbs Fig 1. The unique ion response behavior of the biosensing process and the artificial sensing synapse device schematic (source: Adv Funct Mater.) gel and the strong ion / electron coupling phenomenon at the interface of graphene are in line with the demand of neurobionic and biochemical sensing fields The effect between the piezoelectric polarization caused by mechanical stimulation and the transmission characteristics of semiconductors endows the output potential with the spatiotemporal information (strain, quantity, time, etc.) of external pulses, and the ions in the dielectric gather directionally under the influence of the piezoelectric potential generated by the directional array of dipoles, thus regulating the conductivity of graphene channels, as shown in Figure 2 As shown in Figure 2, the artificial synaptic device can respond to the external strain directly The neural network chip based on graphene, combined with the active matrix of generator, can realize the interface function of static and dynamic piezoelectric control and neural morphological mechanical sensory system At the same time, as a family of four materials compatible with photolithography, it is easy to use the existing silicon-based semiconductor CMOS production line to manufacture a new generation of LSI Figure 2 In depth research on the basic performance and working mechanism of transistor devices (source: adv funct Mater.) in the field of biological nerve conduction promotes the further improvement of tactile information processing system imitating neural sensory feedback As a new field of biomimetic neural morphological computing, synaptic electronics has shown a strong momentum of development Synaptic plasticity is the basis of distributed computing for perceptual signals, which is processed in the process of signal transmission according to the weight Synapse is the physical node of biological signal transmission in nerve fiber, which has bidirectional plasticity Its weight can not only be increased to represent a reinforcement learning behavior, but also be suppressed to maintain the overall low power consumption characteristics of the nervous system (Fig 3) The artificial synapse based on the simulation of intelligent piezoelectric transistor is helpful to build the neural morphological interface and realize the deep learning of robot sensing It has the necessary conditions to simulate the peak time sequence dependent plasticity function of the neural system It also has the potential to realize the unsupervised learning, action capture and pattern recognition of artificial intelligence through the pulse neural algorithm Figure 3 Plasticity regulation of artificial synapses (source: adv funct Mater.) multiple sensory feedback can establish a dynamic spatiotemporal logic relationship between neural networks (Figure 4) It is different from the classical sensor in sensory synapse, and it has unique characteristics of neural morphological computation As the basic performance of spatial resolution, the device has different response to the input of different signal sources For different order of stimulus pulse, the response degree of the device is also different, so as to realize the basis of time resolution At the same time, this artificial synapse also shows the logical relationship between compression strain and tensile strain, which is the basic unit of building a more complex and multi-functional large-scale artificial neural network in the future, and the foundation of realizing parallel perceptual computing system Figure 4 Multi sensory feedback based on dynamic spatiotemporal logic (source: adv funct Mater.) this work may pave the way for self driving artificial intelligence and neural robots The research results were published in adv func Mater Under the title of "piezotronic graphene artistic sensor synapse", Chen Youhui, Gao Guoyun and Zhao Jing were the co first authors Based on this work and expansion, sun Qijun's research group has now completed the follow-up more intelligent and systematic work Review of previous reports: researcher sun Qijun and Academician Wang Zhonglin of Beijing Institute of nano energy and systems, Chinese Academy of Sciences: Sun Qijun and Academician Wang Zhonglin of Beijing Institute of nano energy and systems, Chinese Academy of Sciences: a new type of MoS2 friction ion electronic transistor Nowadays, people and scientific research have been paid more and more attention in the economic life China has ushered in the "node of science and technology explosion" Behind the progress of science and technology is the work of countless scientists In the field of chemistry, in the context of the pursuit of innovation driven, international cooperation has been strengthened, the influence of Returned Scholars in the field of R & D has become increasingly prominent, and many excellent research groups have emerged in China For this reason, CBG information adopts the 1 + X reporting mechanism CBG information, chembeangoapp, chembeango official microblog, CBG wechat subscription number and other platforms jointly launch the column of "people and scientific research", approach the domestic representative research groups, pay attention to their research, listen to their stories, record their demeanor, and explore their scientific research spirit