An Optogenetics‐Inspired Flexible van der Waals Optoelectronic Synapse and its Application to a Convolutional Neural Network

Seo, Seunghwan; Lee, Je‐Jun; Lee, Ryong-Gyu; Kim, Tae Hyung; Park, Sang Yong; Jung, Sooyoung; Lee, Hyun Kyu; Andreev, Maksim; Lee, Kyeong‐Bae; Jung, Kil‐Su; Oh, Seyong; Lee, Hojun; Kim, Ki-Seok; Yeom, Geun Young; Kim, Yong‐Hoon; Park, Jin‐Hong

doi:10.1002/adma.202102980

Cited by 87 publications

(49 citation statements)

References 72 publications

(97 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We developed optoelectronic synaptic transistors with different functional materials and configurations for constructing flexible artificial synapses. Compared with electric synaptic devices, optoelectronic synaptic devices enable a large bandwidth, less interconnection energy loss, higher spatiotemporal resolution, [ 95 ] and ultrafast signal transmission. [ 96,97 ] In the future, more attention should be paid to NIR light‐stimulated transistor‐based synapses.…”

Section: Carrier Transport Mechanisms Of Flexible Synaptic Devicesmentioning

confidence: 99%

Flexible Artificial Synapses Based on Field Effect Transistors: From Materials, Mechanics towards Applications

Liu

Zhang

et al. 2022

Advanced Intelligent Systems

View full text Add to dashboard Cite

Von Neumann-based computation, which is the basis for solving complex and well-structured problems, plays an essential role in commercial computing units such as microcontrollers, microprocessors, and field-programmable gate arrays (FPGAs). However, the efficiency of von Neumann computers inevitably decreases when processing a large amount of data owing to the separate storage and processing units. Additionally, the energy consumption of these computers is high when data are switched frequently between the microprocessor and memory. [1][2][3] An efficient biological computing system, the human brain is capable of handling various complex tasks and consumes energy only on the order of several fJ/spike. Therefore, brain-inspired flexible artificial synapses are widely used in prosthetics, neurorobotics, and health monitoring. [4,5] The synapse is composed of a presynaptic membrane, synaptic cleft, and postsynaptic membrane. The cell membrane potential of the postsynaptic membrane changes in accordance with the action transmitted to the axon terminal of the presynaptic neuronal membrane. Since "neuromorphic electronic systems" were first proposed by Carver Mead in 1990, several synaptic devices based on two-terminal and three-terminal structures have been successfully fabricated to mimic the information processing function of the human brain. [6] Although two-terminal devices, such as memristors and phase-change memories, have low power consumption and simple device structures, [7][8][9] threeterminal transistors are more suitable for synaptic devices because of the advantages of parallel processing and memory functions. Furthermore, they comprise several gates to obtain signals from many sources compared with two-terminal devices. As a typical three-terminal element, the structure of field-effect transistors (FETs) is extremely similar to that of biological synapses (Figure 1a). [10] The gate voltage, V GS, serves as the presynaptic input terminal while the source-drain channel current I SD is used as the post-synaptic output terminal. The transmission of neurotransmitters from the pre-synapse to the post-synapse corresponds to carrier transportation. Synaptic devices with different device configurations have been employed in implementing synaptic functions by controlling the carrier transport (Figure 1b), such as floating-gate transistors, electrolyte-gate transistors, ferroelectric-gate transistors, and optoelectronic transistors. Thus, a clear understanding of the carrier transport mechanisms is the basis for designing artificial synapses with multiple functions.Communication between neurons is facilitated by changing the connection strength of the neurons, which is referred to as synaptic weight. [11] The basic characteristics of synaptic devices include excitatory postsynaptic currents (EPSC) and inhibitory postsynaptic currents (IPSC). These characteristics rely on the signals generated under external activation. [12] Synaptic plasticity is grouped into short-term plasticity (STP) and long-term

show abstract

Section: Carrier Transport Mechanisms Of Flexible Synaptic Devicesmentioning

confidence: 99%

Flexible Artificial Synapses Based on Field Effect Transistors: From Materials, Mechanics towards Applications

Liu

Zhang

et al. 2022

Advanced Intelligent Systems

View full text Add to dashboard Cite

show abstract

“…Inspired by neurotransmission of the brain, memristors (memory + resistor) with an inherent resist switching performance have been recognized as excellent candidates for mimicking both artificial synapses and neurons. − The excellent switching performance matrix, such as long retention (>10 years), stable endurance (>10 10 cycles), small switching energy (<10 pJ), , fast switching speed (<1 ns), , and nanoscale cells (<10 nm) pave the way for memristor to be the next-generation in-memory computing. Memristors are formed by an active layer with the top and bottom electrodes. − The conductance of the memristors can be varied by external stimuli, including external electrical field, , optical illumination, , tactile and auditory sensory systems; − therefore, the multiple analog resistance states can be obtained by historical stimuli. Till now, various materials have been applied for the active layer of the memristor including nitrides, oxides, chalcogenides, MXenes, and polymers .…”

Section: Introductionmentioning

confidence: 99%

“…Memristors are formed by an active layer with the top and bottom electrodes. 21−25 The conductance of the memristors can be varied by external stimuli, including external electrical field, 23,24 optical illumination, 26,27 tactile and auditory sensory systems; 28−30 therefore, the multiple analog resistance states can be obtained by historical stimuli. Till now, various materials have been applied for the active layer of the memristor including nitrides, 31 oxides, 32 chalcogenides, 33 MXenes, 34 and polymers.…”

Section: Introductionmentioning

confidence: 99%

Multitasking Memristor for High Performance and Ultralow Power Artificial Synaptic Device Application

Yang

Zhang

Huang

et al. 2022

ACS Appl. Electron. Mater.

View full text Add to dashboard Cite

The emergence of in-memory computing has shed light on solving high-power consumption and low computation efficiency of the traditional computers built with von Neumann architecture. Memristor, which exhibits history-dependent conductivity modulation, can simulate the synaptic behaviors in the biological brain. However, it remains as a key challenge to fabricate devices that can demonstrate a wide range of synaptic plasticity and maintain stable switching responses over repetitive operating cycles. Hereby, the memristor made of Au/Ti/TiO 2 /Nb:SrTiO 3 (Nb:STO) heterojunction shows a partial nonvolatile bipolar resistive switching behavior with an initial high on/off switching ratio of ∼10 4 , and "writing" and "erasing" with long endurance across 1.5 × 10 4 cycles. Furthermore, we experimentally developed a single device that possesses a 5-bits (32states) reservoir computing system to recognize the binary patterns. We also demonstrated the multidata storage for neuromorphic computing in a 10 × 10 neuromorphic array to recognize the patterns of multilevel resistance states with an ultralow operation power of 4.1 pJ. In addition, various synaptic dependent plasticity performances, including spike-duration, -interval, and -number dependent plasticity, have been realized. Such an on-demand neuromorphic device exhibits a multitask shifting potential for analog bipolar memory and bistate and multistate neuromorphic networks and paves a way to develop the highly efficient memristor devices.

show abstract

“…Recently, optically stimulated synaptic devices are well-known for wide bandwidth, low power consumption, fast signal propagation speed and efficient interconnect, which contribute to neuromorphic computing and mimicking neural activities [9][10][11][12][13]. For example, optically stimulated synaptic devices have been applied to mimic the memory storage of the brain, such as short term memory (STM) and long term memory (LTM) which are two important components of the Atkinson-Shiffrin memory model [9,14,15].…”

Section: Introductionmentioning

confidence: 99%

“…The well-accepted trapping/detrapping processes can be caused by crystal defects, semiconductor/dielectric interfaces and heterojunctions [9,[33][34][35][36][37]]. Park's group [13] verified that the PPC effect in ReS 2 resulted from sulphur vacancies via density functional theory (DFT) calculations. Furthermore, scanning tunneling spectroscopy and X-ray photoelectron spectroscopy confirmed the dangling bonds [12] and interfaces [38] had introduced the PPC effect, respectively.…”

Section: Introductionmentioning

confidence: 99%

A hippocampus-inspired illumination time-resolved device for neural coding

Liu

et al. 2021

Sci. China Mater.

View full text Add to dashboard Cite

Artificial photonic synapses have set off a new upsurge for mimicking a series of neural activities in recent years. In particular, the investigation of learning and memory behaviors with pressure or emotion and corresponding mechanisms is currently the focus of more attention. Herein, a hippocampus-inspired device based on MoS 2 for illumination time encoding is fabricated, in which the encryption technology is employed for data security. In addition, the pressureinduced memory behaviors with full memory function (memory trace) over time such as encoding, storage and retrieval are demonstrated, resulting from the decreasing positive photocurrent of the MoS 2 devices. The proposed mechanism of the memory effect when exposed to the light is elucidated in detail. Moreover, the effect of stress hormone on memory behavior is displayed via different illumination time periods and light intensities. These results indicate the potential application of MoS 2 devices in artificial neural network.

show abstract

An Optogenetics‐Inspired Flexible van der Waals Optoelectronic Synapse and its Application to a Convolutional Neural Network

Abstract: Research data are not shared.

Cited by 87 publications

References 72 publications

Flexible Artificial Synapses Based on Field Effect Transistors: From Materials, Mechanics towards Applications

Flexible Artificial Synapses Based on Field Effect Transistors: From Materials, Mechanics towards Applications

Multitasking Memristor for High Performance and Ultralow Power Artificial Synaptic Device Application

A hippocampus-inspired illumination time-resolved device for neural coding

Contact Info

Product

Resources

About