Biological Synaptic Behaviors Research Articles

Colloidal Quantum Dots‐Based Photoelectrochemical‐Type Optoelectronic Synapse

AbstractCurrent artificial neuromorphic systems are mainly constructed using solid‐state optoelectronic synaptic devices, limiting their potential artificial intelligence applications in liquid medium. Here, colloidal CuInGaSe/ZnSe quantum dots (QDs) with environment‐benign feature and broad visible light absorption are prepared to fabricate a photoelectrochemical (PEC)‐type optoelectronic synaptic device operating in aqueous electrolyte, enabling unique biological synaptic behaviors including paired‐pulse facilitation (PPF), short‐term plasticity (STP), long‐term plasticity (LTP) and learning‐forgetting‐relearning process when simulated by optical pulses with various wavelength, pulse number, frequency, power density, and applied voltage. These results demonstrate the feasibility of building PEC‐type optoelectronic synapses using colloidal QDs, paving the way to realize available optoelectronic synaptic devices for prospective underwater artificial intelligence technology.

Brain‐inspired Multimodal Synaptic Memory via Mechano‐photonic Plasticized Asymmetric Ferroelectric Heterostructure

AbstractNeuromorphic devices capable of emulating biological synaptic behaviors are crucial for implementing brain‐like information processing and computing. Emerging 2D ferroelectric neuromorphic devices provide an effective means of updating synaptic weight aside from conventional electrical/optical modulations. Here, by further synergizing with an energy‐efficient synaptic plasticity strategy, a multimodal mechano‐photonic synaptic memory device based on 2D asymmetric ferroelectric heterostructure is presented, which can be modulated by external mechanical behavior and light illumination. By integrating the asymmetric ferroelectric heterostructured field‐effect transistor and a triboelectric nanogenerator, the mechanical displacement‐derived triboelectric potential is ready for gating, programming, and plasticizing the synaptic device, resulting in superior electrical properties of high on/off ratios (> 107), large storage windows (equivalent to ≈95 V), excellent charge retention capability (> 104 s), good endurance (> 103 cycles), and primary synaptic behaviors. Besides, optical illumination can effectively synergize with mechanoplasticity to implement multimodal spatiotemporally correlated dynamic logic. The demonstrated multimodal memory synapse provides a facile and promising strategy for multifunctional sensory memory, interactive neuromorphic devices, and future brain‐like electronics embodying artificial intelligence.

Vacancy-driven resistive switching behavior based on wafer-scale MoSe2 artificial synapses

Recently, researches have revealed that synaptic transistors based on two-dimensional materials (2DMs) have substantial advantages and potential for modeling neural synapses. However, the mechanism of artificial synaptic device with 2DMs as conducting channels remains to be further clarified. Here, high-quality wafer-scale MoSe2 nanosheets were achieved by ALD-CVD process, where further 23 × 23 × 4 synaptic arrays were fabricated in situ. Notably, the outstanding MoSe2 synaptic device exhibits an on/off ratio of 3.6 × 107 with hysteresis ratio of 6.8k and simulates crucial biological synaptic behaviors such as paired pulse facilitation and excitatory postsynaptic currents. The mechanism of Se vacancy induction and horizontal diffusion driven is revealed based on high-resolution transmission electron microscopy characterization with first-principles calculations, which suggests a possible mechanistic explanation for the resistive switching behavior of 2DMs transistors. Artificial synaptic devices with 2DMs based on the working mechanism of vacancy diffusion are capable of satisfying the needs of complex neural network applications.

Effects of voltage schemes on the conductance modulation of artificial synaptic device based on 2D hBN memristor: Its applications for pattern classifications

Artificial synaptic devices utilizing nonvolatile memristors with 2D materials show potential for neuromorphic computing systems due to their intriguing electrical properties, simple structure, and capability to emulate biological synaptic behaviors. Here, we fabricated nonvolatile memristors based on 2D multilayer hBN film and demonstrated their analog memory performance along with biological synaptic characteristics, including paired-pulse facilitation and depression (PPF and PPD), and synaptic plasticity. We utilized different voltage scheme algorithms to analyze the impact of synaptic functions on image classification tasks across various pattern datasets through simulations. Our findings reveal that pattern recognition outcomes varied due to the modulation of synaptic plasticity by the applied voltage schemes. We also investigated the relationship between conductance update variations and classification accuracy to highlight the significance of optimized synaptic plasticity in improving image recognition capabilities in large-scale neural networks. Our experimental results contribute to the development of high-performance neuromorphic computing for diverse pattern classification tasks using artificial synaptic devices based on 2D hBN.

Optical memcapacitor based on MoS2 quantum capacitance

It is imperative to study neuromorphic devices where the behavior of artificial synapses is regulated by external conditions. The quantum capacitance of 2D materials is determined by the local density of states near the Fermi level and usually exhibits a sensitive response to external stimuli. This study explores the use of MoS2 quantum capacitance in a memcapacitor (MoS2 quantum memcapacitor, MQM), emulating biological synapse behavior and showcasing higher sensitivity. Notably, without altering the physical structure, MQM realizes the potentiation and depression of capacitance through light and gas coupling manipulation, establishing a memory effect for the device. This enables MQM to reproduce the learning-forgetting cycle multiple times with good repeatability. Moreover, MQM can simulate various biological synaptic behaviors, encompassing short-term memory to long-term memory conversions, short-term depression to long-term depression conversions, and higher pair-pulse-facilitation behavior. These characteristics offer a fresh direction and possibilities for artificial synaptic device research.

Natural Organic Honey-CNT Memristor Based Artificial Synaptic Devices for Sustainable Neuromorphic System

Neuromorphic computing is considered to have the potential to overcome the limitations of traditional von Neumann architecture due to its high efficiency, low energy consumption, and fault-tolerance. Hardware components that can emulate the synaptic plasticity of neurons, i.e. artificial synaptic devices, are required by neuromorphic systems. New devices have been examined for such components, such as phase-change artificial synapse, ferroelectric artificial synapse, and memristor synapses. Among them, memristor, a two-terminal metal-insulator-metal structure that are analogous to a biological synapse with presynaptic neuron (top electrode), postsynapticneuron (bottom electrode), and synaptic cleft (memristive film), is a promising device technology because of its tunable resistance, scalability, 3D integration compatibility, low power consumption, and relatively high speed. In contrary to inorganic materials such as metal oxides, natural organic materials have attracted interest to form the memristive layer because they are renewable, biodegradable, sustainable, biocompatible, and environmentally friendly. In this paper, honey solution embedded with carbon nanotubes (CNTs) was processed into the memristive layer by a low cost solution-based process, with synaptic plasticity of the final honey-CNT memristors characterized, including forget and relearn, spike-rate-dependent plasticity, spike-voltage-dependent plasticity, short-term to long-term memory transition, paired pulse facilitation, and spatial supra-linear summation behaviors. The successful emulation of these essential biological synaptic behaviors demonstrates the potential of honey-CNT memristors as a viable hardware component in neuromorphic computing systems.

PLD‐Grown Semi‐Insulating Ga2O3 Thin Film‐Based Optoelectronic Artificial Synaptic Devices for Neuromorphic Computing Applications

AbstractDevelopment of artificial opto‐electronic synaptic devices plays a crucial role for the practical implementation of energy‐efficient, parallel processing of human brain. In this article, two terminal inter‐digitated devices are fabricated on Gallium oxide (Ga2O3) thin films grown on sapphire substrates by pulsed laser deposition (PLD) technique to study its ability to mimic biological synaptic behaviors. The layers are found to exhibit long persistent photo‐conductivity (PPC) effect, which is identified to be the key parameter to replicate the behavior of biological synapses. Channel resistance and PPC time constants should also be optimized to improve the efficiency of response and energy consumption of synaptic devices. It has been observed that both conductivity and the PPC decay time of Ga2O3‐films can be controlled by varying oxygen pressure and growth temperature (TG). These devices demonstrate their ability to perform paired pulse facilitation (PPF) at very low applied bias in mV‐range. They can mimic biological synapses showing short‐to‐long‐term memory transition (STM‐to‐LTM) and learning‐forgetting behavior. One of these devices is found to show synaptic behavior with the energy consumption of as low as 71fJ electrical and 21nJ optical per synaptic event. These findings thus strengthen the candidature of Ga2O3 films for the development of next‐generation opto‐electronic neuromorphic devices and systems.

Cu/MgO-based resistive random access memory for neuromorphic applications

Resistive Random Access Memory (ReRAM) is considered to be a suitable candidate for future memories due to its low operating voltage, fast access speed, and the potential to be scaled down to nanometer range for ultra-high-density storage. In addition, its ability to retain multi-level resistance states makes it suitable for neuromorphic computing applications. In this paper, we report the resistive switching performance of Cu/MgO/Pt ReRAM. Repetitive resistive switching transitions with low switching voltages (around 1 V), 102 storage windows, and multi-level memory capabilities have been obtained. Biological synaptic plasticity behavior, such as long-duration potentiation/depression and paired-pulse facilitation, has been realized by the Cu/MgO/Pt ReRAM. The simulation of convolutional neural network for handwritten digit recognition is carried out to evaluate its potential application in neuromorphic systems. Finally, the conduction mechanism of the device is studied, and a resistive switching model based on Cu conducting filaments is proposed according to the dependence of I–V results on temperature and electrode size as well as the element distribution in the device. These findings indicate the potential of Cu/MgO/Pt device as high-performance nonvolatile memories and its utilization in future computer systems and neuromorphic computing.

Surface strategy design concerning entrapment of chalcogenide onto ETFE substrate and extension to the assembly of memristor

Consideration of semi-conductor quantum dots as the suitable band-gap materials has been an effective way to replace the traditional emissive species such as phosphors, lanthanide edifices or organic building blocks. However, very few reports have been performed for the employment of such chalcogenide dots in brain-excitation devices such as memristors. In this work, the colloidal synthesis for the preparation of AgInS2 quantum dots as well as its optical characteristic has been recorded. Further exploration has been focused on the improvement of resistance switching behavior and a layer of manganese dioxide has been covered onto ITO conductive glass by electrochemical deposition. Following step concerns the incorporation of AgInS2 and the toluene solution of PMMA and the PMMA@AgInS2@MnO2 hybrid layer with optimized uniformity and stability has been afforded. Moreover, such functional device has been used to simulate biological synaptic behavior, which will provide an innovative and efficient route for the development of artificial synaptic functions.

Ultrasoft and High-Adhesion Block Copolymers for Neuromorphic Computing.

The "von Neumann bottleneck" is a formidable challenge in conventional computing, driving exploration into artificial synapses. Organic semiconductor materials show promise but are hindered by issues such as poor adhesion and a high elastic modulus. Here, we combine polyisoindigo-bithiophene (PIID-2T) with grafted poly(dimethylsiloxane) (PDMS) to synthesize the triblock-conjugated polymer (PIID-2T-PDMS). The polymer exhibited substantial enhancements in adhesion (4.8-68.8 nN) and reductions in elastic modulus (1.6-0.58 GPa) while maintaining the electrical characteristics of PIID-2T. The three-terminal organic synaptic transistor (three-terminal p-type organic artificial synapse (TPOAS)), constructed using PIID-2T-PDMS, exhibits an unprecedented analog switching range of 276×, surpassing previous records, and a remarkable memory on-off ratio of 106. Moreover, the device displays outstanding operational stability, retaining 99.6% of its original current after 1600 write-read events in the air. Notably, TPOAS replicates key biological synaptic behaviors, including paired-pulse facilitation (PPF), short-term plasticity (STP), and long-term plasticity (LTP). Simulations using handwritten digital data sets reveal an impressive recognition accuracy of 91.7%. This study presents a polyisoindigo-bithiophene-based block copolymer that offers enhanced adhesion, reduced elastic modulus, and high-performance artificial synapses, paving the way for the next generation of neuromorphic computing systems.

Solution-processed LiGdO solid electrolyte for In2O3 synaptic transistor and its application in neuromorphic computing

Emulating biological synaptic behavior using the electrolyte-gated synaptic transistor (EGST) is promising for neuromorphic applications. The EGSTs based on solid electrolytes exhibit excellent stability and compatibility with semiconductor manufacturing processes....

Synaptic Transistor Arrays Based on PVA/Lignin Composite Electrolyte Films

With the development of information society, the traditional von Neumann-based computing system is facing significant challenges. The search for an intelligent computing system similar to the biological brain would be a very effective solution to the present-day von Neumann bottleneck. Electrolyte-gated transistors (EGTs) have received much attention because they can simulate biological synaptic behavior very effectively. However, large-scale EGTs arrays are still lacking because most of the existing reported EGTs use organic or liquid electrolytes, which poses a significant challenge to the current production methods for manufacturing integration using photolithography. Although synaptic transistor arrays using solid-state electrolytes have the potential for large-scale fabrication, the power consumption required for individual transistors is relatively high. In this work, an electrolyte transistor array (10 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\times$</tex-math> </inline-formula> 10) fabricated by the photolithography process was successfully proposed, where the individual transistors in the array used a composite polyvinyl alcohol (PVA)/lignin electrolyte as the gate dielectric layer. The switching ratio of a single transistor can reach 106, and the maximum gate leakage current is 43.96 pA in the voltage range of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$-$</tex-math> </inline-formula> 3 to 3 V. In addition, the synaptic properties, such as excitatory postsynaptic current (EPSC) and paired-pulse-facilitation (PPF), were successfully achieved. When the pulse duration is 100 ms, the energy consumption of the transistor is 0.63 nJ. The number’s dynamic memory and forgetting functions were also successfully simulated by the artificial synaptic transistor array. This work will provide a useful idea for large-scale array integration of EGTs using organic and liquid electrolytes as gate dielectric layers.

Optically Readable Organic Electrochemical Synaptic Transistors for Neuromorphic Photonic Image Processing.

Optically readable organic synaptic devices have great potential in both artificial intelligence and photonic neuromorphic computing. Herein, a novel optically readable organic electrochemical synaptic transistor (OR-OEST) strategy is first proposed. The electrochemical doping mechanism of the device was systematically investigated, and the basic biological synaptic behaviors that can be read by optical means are successfully achieved. Furthermore, the flexible OR-OESTs are capable of electrically switching the transparency of semiconductor channel materials in a nonvolatile manner, and thus the multilevel memory can be achieved through optical readout. Finally, the OR-OESTs are developed for the preprocessing of photonic images, such as contrast enhancement and denoising, and feeding the processed images into an artificial neural network, achieving a recognition rate of over 90%. Overall, this work provides a new strategy for the implementation of photonic neuromorphic systems.

Ultraviolet-selective organic phototransistors for low-power skin-inspired nociceptor

Skin-inspired nociceptors are particularly important for future neurorobotic technology. In this work, an artificial skin nociceptor based on ultraviolet (UV)-selective C8-BTBT phototransistors was reported. By using polymethyl methacrylate as an interfacial modification layer, a large number of standing crystals appeared in the C8-BTBT film, and the photo-dark current ratio of the devices exceeded the value of 1010. The main biological synaptic behaviors of the devices under UV light stimuli were also thoroughly investigated. Interestingly, the minimum power consumption per synaptic event (2.45 fJ) is comparable to the energy required by the biological synapses. The biological behavior of the skin-inspired nociceptor was further simulated, such as allodynia, hyperalgesia, and the self-protection behavior under UV light stimulation. Finally, as a proof of concept, the skin-inspired nociceptor was proposed to be integrated with the UV-transmittance modulator to protect against excess UV stimulation, which fully demonstrated its potential applications in bionic electronics technology.

Synaptic MoS2 transistors based on charge trapping two-dimensionally confined in Sr2-xCoxNb3O10 nanosheets

Two-dimensional (2D) synaptic devices based on charge trap/de-trap have attracted much attention due to their superior characteristics that is gate tunability, large switching ratio, long-term retention, and distinct synaptic potentiation/depression with low power. Here, we introduce a 2D Sr2-xCoxNb3O10 (SCNO) nanosheet layer between the MoS2 channel and SiO2 substrate because charge trap sites confined in SCNO nanosheet layer are anticipated to precisely control the 2D channel conductance. The hysteretic and synaptic behaviors of MoS2 channel can be controlled by charge trap/de-trap in two-dimensionally confined nanosheet layer with different doping ratio of Co. As the x value for the MoS2/SCNO field effect transistor (FET) increases, the shift of the threshold voltage (Vth) in a transfer curve increases and more linear potentiation/depression curves are induced with lower gate voltage pulse. In addition, paired-pulse facilitation is successfully implemented in a MoS2/SCNO FET. Therefore, our results suggest the feasibility of improved and efficient emulation of biological synaptic behavior in MoS2 transistors in contact with 2D oxide nanosheet layer.

Tunable non-volatile memories based on 2D InSe/h-BN/GaSe heterostructures towards potential multifunctionality.

Floating-gate memories based on two-dimensional van der Waal (2D vdW) heterostructures play an important role in the development of next-generation information technology. The diversity of 2D vdW materials and their heterostructures provides flexibility in the design of novel storage architectures. However, 2D InSe/h-BN/GaSe heterostructures are rarely reported in the field of tunable non-volatile memories, probably due to the quality limitation of materials and complex interfaces from stackings. Herein, a floating-gate 2D InSe/h-BN/GaSe memory with high performance and atmosphere stability is demonstrated. It exhibits both a large ON/OFF current ratio of ∼105 and a good extinction ratio of ∼103, with an estimated maximum storage capacity of 5.1 × 1012 cm-2. Moreover, the storage performance can be regulated by optimizing the thickness of the insulating h-BN layer. Different device configurations have been explored to validate the working mechanism. Furthermore, a simulation of biological synaptic behavior is achieved on the same prototype device. The enhanced non-volatile characteristics enable the exploration of the integrated 2D memory and potential multifunctionality.

Forming-Free Tunable Analog Switching in WOx/TaOx Heterojunction for Emulating Electronic Synapses.

In this work, the sputtered deposited WOx/TaOx switching layer has been studied for resistive random-access memory (RRAM) devices. Gradual SET and RESET behaviors with reliable device-to-device variability were obtained with DC voltage sweep cycling without an electroforming process. The memristor shows uniform switching characteristics, low switching voltages, and a high RON/ROFF ratio (~102). The transition from short-term plasticity (STP) to long-term potentiation (LTP) can be observed by increasing the pulse amplitude and number. Spike-rate-dependent plasticity (SRDP) and paired-pulse facilitation (PPF) learning processes were successfully emulated by sequential pulse trains. By reducing the pulse interval, the synaptic weight change increases due to the residual oxygen vacancy near the conductive filaments (CFs). This work explores mimicking the biological synaptic behavior and further development for next-generation neuromorphic applications.

MoTe2-based low energy consumption artificial synapse for neuromorphic behavior and decimal arithmetic

Two-dimensional material-based memristors have shown attractive research prospects as brain-like devices for neuromorphic computing. Among them, transition metal dichalcogenides–based memristors have proved to be one of the most promising competitors. In this work, a two-dimensional memristor based on MoTe2 nanosheets was fabricated and demonstrated. The experimental results illustrate that the two-terminal synaptic based on the Ag/MoTe2/ITO structure exhibits stable bipolar and non-volatile resistive switching characteristics attributed to the controllable formation and rupturing of silver conductive filaments. The device can be successively modulated by a pulse train with a minimum pulse width of 40 ns. More interestingly, the energy consumption of the device to complete one write event is only 74.2 pJ. In addition, biological synaptic behaviors, such as excitatory postsynaptic current gain properties, long-term potentiation (LTP)/long-term depression, spike-timing-dependent- plasticity, short-term plasticity, long-term potentiation (LTP), paired-pulse facilitation, post-tetanic potentiation, and learning-experimental behaviors were mimicked faithfully. Finally, the decimal arithmetic application was introduced to the device, and it is confirmed that addition and multiplication functions can be performed. Therefore, the artificial synapse based on MoTe2 nanosheets not only exhibits the stable non-volatile resistive switching behavior but also facilitates the development of low-energy consumption neuromorphic computing chips based on transition metal dichalcogenides.

Spatiotemporal Modulation of Plasticity in Multi‐Terminal Tactile Synaptic Transistor

AbstractNeuromorphic system based on artificial synaptic devices is considered as a potential candidate to realize the in‐memory computing and parallel processing of data for overcoming the von Neumann bottleneck. However, to fully imitate the complicated functions of the biological neural networks at the hardware level is still a challenging task. In this work, a multi‐terminal MoS2 synaptic transistor is developed, which not only simulates various biological synaptic behaviors, including paired pulse facilitation (PPF), excitatory/inhibitory post‐synaptic current (EPSC/IPSC), spike‐rate‐dependent plasticity (SRDP), and spike‐timing‐dependent plasticity (STDP), but also can independently mimic the parallel signal processing and transmissions in biological multipolar neurons. By combining the multi‐terminal MoS2 synaptic transistor with the micro‐structured polydimethylsiloxane (PDMS) pressure sensors, an intelligent tactile recognition system is built up, which can realize the spatiotemporal recognition of touch position. Furthermore, with sensor selection, the spatiotemporal modulation of synaptic plasticity and the human learning and forgetting behaviors to the knowledge with different difficulty degrees can be mimicked. This work provides a novel interconnection scheme for simulating signal transmission and processing among neurons, showing broad application prospects of the multi‐terminal MoS2 synaptic transistor in intelligent human–computer interaction and bionic neuromorphic perception systems.

High Conductivity Update Linearity MoS2 Memtransistors Array Based on Lithium‐Ion Modulation

AbstractThe typical nonlinear and asymmetric response of synaptic memristors based on 2D layered materials to positive and negative electrical pulses makes the implementation of accurate deep neural networks very challenging. Here, it has been constructed an integrated circuit (IC) process‐compatible array of modulatable neuromorphic devices based on MoS2 three‐terminal memristors and show that the switching characteristics of this memristor become more gradual and symmetric under electrical excitation, yielding up to 98% linearity of conductance updates in optimized simulations of biological synaptic behavior, achieving long‐term plasticity of 150 s. The linear weight update has better predictability and distinguishability. And all‐solid lithium cobaltate is introduced as an independent modulation channel to obtain a modulatable molybdenum disulfide bandgap by controlling the embedding and disembedding of lithium ions, which effectively reduces the contact potential barrier and more flexibly modulates the artificial synaptic properties based on multiple coupling of light, electricity, and ions. This work lays the foundation for the realization of high‐density and energy‐efficient neuromorphic computing chips.