Photonic Synapse Research Articles

Physical Reservoir Computing Using Tellurium-Based Gate-Tunable Artificial Photonic Synapses.

We report tellurium (Te) thin-film-based artificial photonic synapses and their application to physical reservoir computing (PRC). The Te-based artificial photonic synapses were fabricated by using sputtered Te thin films and spray-coated MXene (Ti3C2) electrodes. A thorough investigation of the field-dependent persistent photoconductivity (PPC) of the Te channel revealed that the relaxation speed of the transient photocurrent depended on the gate bias. Utilizing the PPC property, the Te device served as an excellent photonic synapse under light pulse stimulus, exhibiting multiple synaptic characteristics such as excitatory postsynaptic current and paired-pulse facilitation, as well as highly linear potentiation-depression characteristics; a simulation-based study further confirmed the effectiveness of the device. Most importantly, by exploiting the nonlinear and fading memory characteristics of the Te photonic synapse, we demonstrate two advanced examples of PRC. In classifying handwritten digits, our system carried out successful digit recognition without binarization or another simplification process with reduced computational cost compared to conventional systems. To solve second-order nonlinear equations, we introduce the strategy of utilizing historical nodes. The combination of historical nodes and the gate-tunable responses of the photonic synapses, which provide an enriched reservoir state, yielded excellent prediction accuracy. Overall, this work will offer an understanding of Te-based optoelectronic devices and their synergetic integration with neuromorphic devices and PRC.

Photonic Synapse of CrSBr/PtS2 Transistor for Neuromorphic Computing and Light Decoding

Abstract Field effect transistors based on 2D layered material have gained significant potential in emerging technologies, such as neuromorphic computing and ultrafast memory response for artificial intelligence applications. This study proposes a facile approach to fabricate an optoelectronic artificial synapse for neuromorphic computing and light‐decoding information system by utilizing the 2D heterostructure of CrSBr/PtS2 to overcome circuit complexity. The CrSBr layer serves as a trapping layer, while PtS2, mounted on top of CrSBr, acts as a channel layer. PtS2 exhibits n‐type semiconductor behavior with a hysteresis that varies with the thickness of the underlying CrSBr layer. The heterostructure device, featuring a 96.3 nm thick CrSBr layer, exhibited a large memory window of 11.9 V when the gate voltage is swept from −10 V to +10 V. Various synaptic behaviors are effectively demonstrated, including paired‐pulse facilitation, excitatory postsynaptic current, optical spike number and intensity‐dependent plasticity using laser light at a wavelength of 365 nm. The device achieves 26 distinct output signals depending on the intensity of the incident laser light, ranging from 10 to 385 mW cm−2, enabling its applications for light‐decoded information security systems. Thus, the investigation presents a unique approach to artificial intelligence and cybersecurity systems.

In-Situ Fabricated Transparent Flexible Nanowire Device with Wavelength-Regulated Dual-Function of Photodetector and Photonic Synapse.

Integrating the dual functionalities of a photodetector and photonic synapse into a single device is challenging due to their conflicting requirements for photocurrent decay rates. This study addresses this issue by seamlessly depositing transparent indium tin oxide (ITO) electrodes onto self-oriented copper hexadecafluoro-phthalocyanine (F16CuPc) nanowires growing horizontally along hot-stamped periodic nanogrooves on a transparent flexible polyimide plastic film. This in-situ-fabricated device achieves bending-stable dual functionalities through wavelength regulation while maintaining high transparency and flexibility. Upon exposure to 450-850 nm light, the device exhibits a rapid and sensitive photoresponse with excellent bending stability, making it ideal for optical sensing in both visible and near-infrared spectra. More importantly, the device exhibits a bending-stable excitation postsynaptic current when exposed to light spikes below 405 nm. This enables the successful emulation of various biological synaptic functionalities, including paired-pulse facilitation, spike-number-dependent plasticity, spike-duration-dependent plasticity, spike-rating-dependent plasticity, configurable plasticity between short-term plasticity and long-term plasticity, and memory learning capabilities. Utilizing this device in an artificial neural network achieves a recognition rate of 95% after 57 training epochs. Its ability to switch between photodetection and synaptic modes by adjusting the light wavelength marks a significant advancement in the field of multifunctional flexible electronics based on nanowire arrays.

Highly uniform organic nanowire synaptic arrays with excellent performance for associative memory

Synaptic devices of optoelectronic nature, which combine light-detection and data-storage capabilities, hold significant promise in the realm of neuromorphic computing. They are especially beneficial for the processing of visual data and the execution of intricate cognitive functions akin to learning, memory retention, and logical reasoning. However, current research is mostly confined to the level of individual devices, with corresponding studies on synaptic arrays being relatively scarce. Type II heterojunctions are widely used in optoelectronics. Bandgap engineering enables efficient separation and trapping of photogenerated excitons in selected materials, reducing carrier recombination. This prolongs carrier retention in the channel, delaying photocurrent decay, crucial for artificial optoelectronic synapses. We designed an organic nanowire/perovskite Type II heterojunction where CsPbBr3 perovskite films and TIPS nanowires serve as the photosensitive and channel layers, respectively. The composite synaptic array was successfully fabricated by in-situ growth of TIPS nanowires on the surface of perovskite thin films. It was further used to fabricate a photonic synapse array that exhibits characteristic photocurrent and stable optical response. Achieving high-performance photonic synapses was facilitated by characteristics such as high mobility and high on/off ratios, stemming from the formation of complete single-crystal nanowires on the perovskite films. Exhibiting synaptic behaviors such as photo-induced enhancements and paired-pulse facilitation, the device enabled a visual sensing system with a 5 × 5 pixel array for simulating memory processes and restoring associative memories. A novel photocurrent model was established for understanding device characteristics. This work provides valuable insights for future utilization of composite organic nanowire arrays in simulating brain-like computations and realizing large-scale intelligent visual sensing systems.

High‐Entropy Perovskite Oxide Photonic Synapses

AbstractHigh‐entropy oxides are garnering increased attention due to their unique properties engineered by chemical disorders. Chemical disorders can cause random potential spatial fluctuations and the resulting anomalous photoconductivity, while they have not yet been exploited in high‐entropy oxides for usefulness. In this work, persistent photoconductivity (PPC) behavior in high‐entropy perovskite Ba(Zr0.2Sn0.2Ti0.2Hf0.2Nb0.2)O3 thin films is reported and used it as a basis to develop artificial photonic synapses. This photonic synapse can emulate distinctive biological synaptic features such as excitatory postsynaptic current, paired‐pulse facilitation, transition from short‐term memory to long‐term memory, and learning‐forgetting processes. This photonic synapse can also implement the basic “AND” and “OR” logic operations. In addition, a 5×5 photonic synaptic array is built to mimic the spatiotemporal sensing and memory functions of a visual sensing‐memory system, exhibiting the improved accuracy of dynamic machine vision classification tasks as compared to traditional visual sensing systems. This study highlights the potential of high‐entropy oxides for application in neuromorphic computing and dynamic machine vision systems.

Retinomorphic Photonic Synapses for Mimicking Ultraviolet Radiation Sensing and Damage Imaging

AbstractPhotonic synapses integrate sensing and processing within a single device, making them promising candidates for emulating the visual perception of a biological retina. Most retinomorphic photonic synapses achieve image pre‐processing by differentially responding to red, green, and blue (RGB). However, the impact of invisible ultraviolet (UV) light on the retina remains insufficiently explored, hindering the perception of the full spectrum by retinomorphic devices. Herein, a photonic synapse is constructed with strong UV absorption of Poly[(9,9‐dioctylfluorenyl‐2,7‐diyl)‐co‐(4,4′‐(N‐(4‐sec‐butylphenyl) diphenylamine)] (TFB) as photosensitive layer. The device exhibits synaptic properties of long‐term memory (LTM) and short‐term memory (STM) at different UV radiation stages. Extracting the electrical conductance of the device under UV radiation generates patterns of different resolutions, facilitating the simulation of retinal damage's impact on object perception. Moreover, an electronic eye system, incorporating the photonic synapse under UV radiation, successfully replicates pupil constriction and eye‐closing behaviors akin to those observed in a real biological eye. This result underscores the authenticity of retinomorphic devices, expending a novel concept for the future development of visuomorphic computing.

Monolithic 2D Perovskites Enabled Artificial Photonic Synapses for Neuromorphic Vision Sensors.

Neuromorphic visual sensors (NVS) based on photonic synapses hold a significant promise to emulate the human visual system. However, current photonic synapses rely on exquisite engineering of the complex heterogeneous interface to realize learning and memory functions, resulting in high fabrication cost, reduced reliability, high energy consumption and uncompact architecture, severely limiting the up-scaled manufacture, and on-chip integration. Here a photo-memory fundamental based on ion-exciton coupling is innovated to simplify synaptic structure and minimize energy consumption. Due to the intrinsic organic/inorganic interface within the crystal, the photodetector based on monolithic 2D perovskite exhibits a persistent photocurrent lasting about 90s, enabling versatile synaptic functions. The electrical power consumption per synaptic event is estimated to be≈1.45×10-16J, one order of magnitude lower than that in a natural biological system. Proof-of-concept image preprocessing using the neuromorphic vision sensors enabled by photonic synapse demonstrates 4 times enhancement of classification accuracy. Furthermore, getting rid of the artificial neural network, an expectation-based thresholding model is put forward to mimic the human visual system for facial recognition. This conceptual device unveils a new mechanism to simplify synaptic structure, promising the transformation of the NVS and fostering the emergence of next generation neural networks.

Multilevel weight optimization of photonic synapses based on slot-ridge waveguides with multi-block GST

Neuromorphic computing has been a candidate to work like human brains to deal with big data energy-efficiently beyond the von Neumann architecture. As an important part, artificial synapses based on waveguides with phase-change materials (PCMs) play a critical role in neural networks. However, because of demanding conditions of triggering intermediate phase states to achieve multilevel weights it is relatively difficult to obtain accurate multilevel weights. In this work, we designed a photonic synapse based on slot-ridge waveguides with multi-block Ge2Sb2Te5 (GST) to precisely and easily get multilevel weights by triggering the specific number of GST islands between crystalline and amorphous states. The properties of the GST material were analyzed by Raman spectroscopy and spectroscopic ellipsometry and the nonlinearity factor of photonic synaptic weights was optimized for the number of GST islands. The recognition tasks of MNIST and the optical recognition of NIST were performed by multilayer perceptron, and the optimal accuracy of the optical recognition of NIST was 92.7 % for 5 GST islands. The accuracy of MNIST was improved by 2 % for 5 GST islands compared with that of 2 GST islands. Besides, the electro-thermal phase transition conditions of crystallization and reamorphization were also obtained by simulation. This work will open the way to precisely and easily achieve photonic synapses with multilevel weights in the future.

All-Optical Data Processing with Photon-Avalanching Nanocrystalline Photonic Synapse.

Data processing and storage in electronic devices are typically performed as a sequence of elementary binary operations. Alternative approaches, such as neuromorphic or reservoir computing, are rapidly gaining interest where data processing is relatively slow, but can be performed in a more comprehensive way or massively in parallel, like in neuronal circuits. Here, time-domain all-optical information processing capabilities of photon-avalanching (PA) nanoparticles at room temperature are discovered. Demonstrated functionality resembles properties found in neuronal synapses, such as: paired-pulse facilitation and short-term internal memory, in situ plasticity, multiple inputs processing, and all-or-nothing threshold response. The PA-memory-like behavior shows capability of machine-learning-algorithm-free feature extraction and further recognition of 2D patterns with simple 2 input artificial neural network. Additionally, high nonlinearity of luminescence intensity in response to photoexcitation mimics and enhances spike-timing-dependent plasticity that is coherent in nature with the way a sound source is localized in animal neuronal circuits. Not only are yet unexplored fundamental properties of photon-avalanche luminescence kinetics studied, but this approach, combined with recent achievements in photonics, light confinement and guiding, promises all-optical data processing, control, adaptive responsivity, and storage on photonic chips.

Light-stimulated long-term potentiation behavior enhanced in a HfO2/InGaZnO photonic synapse

Photonic-based neuromorphic systems offer numerous advantages, including low crosstalk, high bandwidth, and low power consumption, making them an ideal choice for ultrafast neuromorphic systems. Although these neuromorphic systems promote the utilization of amorphous oxide semiconductors (AOSs) for promising photonic synapses, precise control of their unique persistent photoconductivity to mimic synaptic plasticity using an external light stimulus has rarely been studied. This study investigates the significant enhancement of the long-term potentiation (LTP) behavior in an InGaZnO (IGZO) photonic synapse through facile HfO2 passivation. In HfO2/IGZO synapses, the HfO2 layer is crucial in enhancing the LTP characteristics. The improvement in the persistent photocurrent behavior of the HfO2/IGZO synapse was attributed to the higher activation energy required for the neutralization of ionized oxygen vacancies, facilitating the LTP behavior. The LTP was dynamically modulated depending on external light pulse conditions. A large number of multilevel states (1024 states) were enabled by consecutive light simulations. Furthermore, HfO2/IGZO synapses exhibit different learning capacities depending on the external stimuli configuration. These findings demonstrate our ability to manipulate synaptic functionality in AOS-based artificial photonic synapses, paving the way for multifunctional photonic synapse circuitry. This advancement holds the potential for realizing an artificial vision system in the future.

Nearly Panoramic Neuromorphic Vision with Transparent Photosynapses.

Neuromorphic vision based on photonic synapses has the ability to mimic sensitivity, adaptivity, and sophistication of bio-visual systems. Significant advances in artificial photosynapses are achieved recently. However, conventional photosyanptic devices normally employ opaque metal conductors and vertical device configuration, performing a limited hemispherical field of view. Here, a transparent planar photonic synapse (TPPS) is presented that offers dual-side photosensitive capability for nearly panoramic neuromorphic vision. The TPPS consisting of all two dimensional (2D) carbon-based derivatives exhibits ultra-broadband photodetecting (365-970nm) and ≈360° omnidirectional viewing angle. With its intrinsic persistent photoconductivity effect, the detector possesses bio-synaptic behaviors such as short/long-term memory, experience learning, light adaptation, and a 171% pair-pulse-facilitation index, enabling the synapse array to achieve image recognition enhancement (92%) and moving object detection.

Tin Doping Induced High‐Performance Solution‐Processed Ga2O3 Photosensor toward Neuromorphic Visual System

AbstractGa2O3 is an emerging wide‐bandgap semiconductor with high deep ultraviolet absorption, tunable persistent photoconductivity, and excellent stability toward electric fields, making it a promising component for neuromorphic visual systems (NVSs). However, Ga2O3‐based photosensors with high responsivity and long response decay times are required for efficient NVSs. A solution‐processed doping strategy for fabrication of Ga2O3 is proposed with tin foil as a dopant source. Tin‐doped Ga2O3 (Ga2O3:Sn) photosensors are obtained with ultrahigh responsivity and extremely long response decay times. These behaviors are attributed to substitutional tin and oxygen vacancies that modulate defect‐related hole trapping. High‐performance Ga2O3:Sn photosensors can mimic photonic synaptic behaviors and image pre‐processing functions. NVSs based on a Ga2O3:Sn photonic synapse array perform pattern recognition with an accuracy of 97.3% under an unprecedented low‐light pulse stimuli of 0.5 µW cm−2. This work provides a low‐cost solution‐processed approach to ultrasensitive Ga2O3:Sn NVSs and will facilitate developments in artificial intelligence technology.

Realize low-power artificial photonic synapse based on (Al,Ga)N nanowire/graphene heterojunction for neuromorphic computing

The fast development of brain-inspired neuromorphic computing systems has stimulated urgent requirements for artificial synapses with low-power consumption. In this work, a photonic synaptic device based on (Al,Ga)N nanowire/graphene heterojunction has been proposed and demonstrated successfully. In the artificial synaptic device, the incident light, the nanowire/graphene heterojunction, and the light-generated carriers play the roles of action potential, pre-synaptic/post-synaptic membrane, and neurotransmitter in a biological synapse, respectively. As a key synaptic function, the paired pulse facilitation index of the photonic synapse can reach 202%, which can be modulated by the interval time between two adjacent light pulses. It is found that the graphene defects, the surface band bending, and the Al vacancies on the surface of (Al,Ga)N nanowires can be the key reasons contributing to the synaptic characteristics of artificial photonic devices. Hence, the dynamic “learning–forgetting” performance of the artificial synaptic device can resemble the “learning–forgetting” behavior of the human brain. Furthermore, the hand-written digits are set up to mimic a typical characteristic of human perceptual learning. After only three training epochs, the simulated network can achieve a high recognition rate of over 90% based on the experimental conductance for long-term potentiation and long-term depression. In supervised learning processes, such few training times are beneficial to reduce energy consumption significantly. Therefore, in the area of neuromorphic computing technology and artificial intelligence systems requiring low-power consumption, this work paves a potential way to develop the optoelectronic synapse based on semiconductor nanowires.

Room-temperature stable CsPbI3 perovskite quantum dots prepared by layer-by-layer assembly for photonic synapse

CsPbX3 (X = I, Br, or Cl) perovskite materials have attracted great interest due to versatile and excellent photoelectric properties, applicable to various devices ranging from memory to solar cells. In particular, the cubic phase of CsPbI3 perovskite has a band gap of 1.73 eV, absorbing most of the visible region. However, despite excellent material properties, it is still difficult to apply to devices due to the unstable crystalline phase of CsPbI3 at room temperature. To overcome this shortcoming, we fabricated a cubic CsPbI3 perovskite quantum dots (CsPbI3-PQDs) thin film using in-situ ligand exchange layer-by-layer self-assembly (LbL SA). These room-temperature stable CsPbI3-PQDs thin films could control not only the thickness in nano-scale, but also the electrical properties. For device application, CsPbI3-PQD was formed on the fluorine doped tin oxide (FTO) substrate by LbL SA method, and W/CsPbI3-PQD/FTO photonic synapse was fabricated and evaluated. This photonic synapse exhibited photo-sensitive current-voltage characteristics under various UV stimuli such as intensity, exposure time, and interval. Furthermore, we successfully emulated the essential synaptic behaviors induced by photonic stimulus including short term plasticity, long term plasticity, and paired pulse facilitation. The CsPbI3-PQD-based device exhibited resistive switching characteristics with Ron/Roff of 103, stable endurance (103 cycles) and retention time (∼ 9 ×103 s). Additionally, we fabricated a heterojunction CsPbI3-PQD/a-IGZO based synaptic device and improved synaptic behaviors under various visible stimuli by enhancing the maintenance capability of photogenerated charge carriers through energy band engineering.

Metal-Oxide Heterojunction Optoelectronic Synapse and Multilevel Memory Devices Enabled by Broad Spectral Photocarrier Modulation.

Broad spectral response and high photoelectric conversion efficiency are key milestones for realizing multifunctional, low-power optoelectronic devices such as artificial synapse and reconfigurable memory devices. Nevertheless, the wide bandgap and narrow spectral response of metal-oxide semiconductors are problematic for efficient metal-oxide optoelectronic devices such as photonic synapse and optical memory devices. Here, a simple titania (TiO2 )/indium-gallium-zinc-oxide (IGZO) heterojunction structure is proposed for efficient multifunctional optoelectronic devices, enabling widen spectral response range and high photoresponsivity. By overlaying a TiO2 film on IGZO, the light absorption range extends to red light, along with enhanced photoresponsivity in the full visible light region. By implementing the TiO2 /IGZO heterojunction structure, various synaptic behaviors are successfully emulated such as short-term memory/long-term memory and paired pulse facilitation. Also, the TiO2 /IGZO synaptic transistor exhibits a recognition rate up to 90.3% in recognizing handwritten digit images. Moreover, by regulating the photocarrier dynamics and retention behavior using gate-bias modulation, a reconfigurable multilevel (≥8 states) memory is demonstrated using visible light.

A perovskite-based artificial photonic synapse with visible light modulation and ultralow current for neuromorphic computing

Photonic artificial synapses have become favorable candidates for the basic architecture of neuromorphic computing systems due to their high bandwidth, low power consumption, and low crosstalk. However, work on visible light modulated synaptic devices with biological postsynaptic currents has rarely been reported. Herein, we report a photonic synaptic device operating at ultra-low currents. It employs perovskite CsPbI2Br as the light-absorbing layer and polymethyl methacrylate (PMMA) as the carrier blocking layer. The resulting device exhibits excellent visual cell simulation capability in the visible light range. An ultra-low post-synaptic current of picoampere level can be achieved with visible-light modulation, which is much lower than that of normal photonic synaptic devices. It also exhibited superior memory retention of over 1000 s and low power consumption of 4.25 pJ. Furthermore, the sensory memory (SM), short-term memory (STM), and long-term memory (LTM) are successfully simulated in visible light. An artificial neural network-based neuromorphic computing was also simulated, with the recognition accuracy of MNIST data reaching 95.99% for small digits and 95.34% for large digits. Overall, our work provides promising strategies for low-current photonic neuromorphic applications.

Plastic Photonic Synapse Based on VCSOA for Self-Learning in Photonic Spiking Neural Network

We propose and demonstrate a plastic photonic synapse to achieve all-optical synaptic plasticity based on the potentiated and depressed dynamics in commercially available vertical-cavity semiconductor optical amplifier (VCSOA). The effects of presynaptic and postsynaptic spike powers, bias current, initial wavelength detuning between presynaptic spike and VCSOA on the potentiated and depressed time windows are investigated. The numerical results show that all-optical synaptic plasticity can be achieved successfully based on VCSOA with delayed self-feedback loop. The proof-of-concept experimental results show that plastic photonic synapse can be achieved based on the VCSOA with subsequent iterations. Besides, the curve of VCSOA gain in the potentiated and depressed time windows is similar to that of spike timing dependent plasticity. Furthermore, all-optical unsupervised self-learning and recognition of first spike timing of an input pattern are realized numerically in the photonic neural network including two proposed plastic photonic synapses. The plastic photonic synapse based on a VCSOA is interesting and valuable for the self-learning in all-optical neural networks. Moreover, the proposed plastic photonic synapse offers great potential for dense, high-performance neuromorphic photonic systems.

Ultralow-power consumption photonic synapse transistors based on organic array films fabricated using a particular prepatterned-guided crystallizing strategy

The prepattern-guided crystallization method was used to prepare array thin film transistors. The photonic synapses realized synaptic behaviour with 0.036 fJ per spike and demonstrated many important functions.

Highly uniform photonic synapse arrays based on TIPS-pentacene nanowires/CsPbBr3 heterojunctions

Organic nanowire thin films was fabricated by template imprinting method. The photonic synaptic array simulated the characteristics of artificial light synapses. Based on a composite synaptic array, an artificial vision system was achieved.

On-Chip Photonic Synapses with All-Optical Memory and Neural Network Computation.

Inspired by the human brain, neural network computing was expected to break the bottleneck of traditional computing, but the integrated design still faces great challenges. Here, a readily integrated membrane-system photonic synapse was demonstrated. By pre-pulse training at 1064 nm (cutoff wavelength), the photonic synapse can be regulated both excitatory and inhibitory at tunable wavelengths (1200-2000 nm). Furthermore, more weights and memory functions were shown through the photonic synapse integrated network. Additionally, the digital recognition function of the single-layer perceptron neural network constructed by photonic synapses has been successfully demonstrated. Most of the biological synaptic functions were realized by the photonic synaptic network, and it had the advantages of compact structure, scalable, adjustable wavelength, and so on, which opens up a new idea for the study of the neural synaptic network.