Search

The processing of multicolor noisy images in visual neuromorphic devices requires selective absorption at specific wavelengths; however, it is difficult to achieve this because the spectral absorption range of the device is affected by the type of material. Surprisingly, the absorption range of perovskite materials can be adjusted by doping. Herein, a CdCl2co-doped CsPbBr3nanocrystal-based photosensitive synaptic transistor (PST) is reported. By decreasing the doping concentration, the response of the PST to short-wavelength light is gradually enhanced, and even weak light of 40 μW·cm–2can be detected. Benefiting from the excellent color selectivity of the PST device, the device array is applied to feature extraction of target blue items and removal of red and green noise, which results in the recognition accuracy of 95% for the noisy MNIST data set. This work provides new ideas for the application of novel transistors integrating sensors and storage computing.

As reported, neuromorphic devices have been widely applied in many fields. However, there are few reports on the application of neuromorphic devices in the field of display. Intelligent displays will inevitably be a requirement for high-performance displays in the future. In order to achieve intelligent display, neuromorphic devices must be applied in low-power, high-resolution display fields. Therefore, in this work, we propose a novel global driving strategy (GDS), where the display driving circuit is developed by applying neuromorphic devices (neurons and synapses) to the circuit. It combines the advantages of neurons and synapses, achieving a low-power, high-resolution display while reducing the row circuit area in the driving circuit by half and increasing circuit integration. In addition, based on the strategy, we built a ${4}\times {4}$ array circuit simulation model for display.

Display circuits play a pivotal role in visual information interaction, but traditional 2T1C display circuits suffer from the separation of storage and computation. Neuromorphic display combines the artificial synapse and light-emitting units into the pixel circuit, empowering the display with smart interaction capability. In this work, we built a novel ambient light adaptive neuromorphic display platform (ALAND) by combining a display driver circuit with the optoelectronic synaptic transistor. The optoelectronic synaptic transistor can perform multiple conduction states under the control of light and electric stimulation, including a 4-bit optical modulation grayscale. Moreover, it can make adaptive decisions according to various ambient lights. This work provides a potentially valuable novel strategy for a more comfortable and clearer visual information interaction.

Inspired by the retina, artificial optoelectronic synapses have groundbreaking potential for machine vision. The field-effect transistor is a crucial platform for optoelectronic synapses that is highly sensitive to external stimuli and can modulate conductivity. On the basis of the decent optical absorption, perovskite materials have been widely employed for constructing optoelectronic synaptic transistors. However, the reported optoelectronic synaptic transistors focus on the static processing of independent stimuli at different moments, while the natural visual information consists of temporal signals. Here, we report CsPbBrI2nanowire-based optoelectronic synaptic transistors to study the dynamic responses of artificial synaptic transistors to time-varying visual information for the first time. Moreover, on the basis of the dynamic synaptic behavior, a hardware system with an accuracy of 85% is built to the trajectory of moving objects. This work offers a new way to develop artificial optoelectronic synapses for the construction of dynamic machine vision systems.

In recent years, artificial synaptic transistors with excellent bionic and energy efficiency have attracted wide attention. However, reports about the synaptic transistors with time-varying input signals are scarce. In this letter, we study the dynamic responses of artificial synaptic transistors to time-varying signals for the first time. We find that the last stimulus has the most significant effect on synaptic weight under the low-voltage stimulus. When considering the presence of a large voltage stimulus in the time-varying input, the closer the time period of the large voltage is to the post-pulse time period, the greater the synaptic weight. Furthermore, we find that the relationship between the time-varying input signal with four pulses and its output result follows the $2\times2$ multiply-accumulate (MAC) operation rule. This means that the MAC rule has been validated at the individual device level for the first time. This work provides a new temporal synaptic performance and a new testing approach, laying the foundation for the development of neuromorphic computing with temporal signals.

Artificial photonic synapses offer an efficient solution for overcoming the von Neumann bottleneck in data storage and processing, providing advantages over electrical synapses by eliminating the bandwidth-connection-density tradeoff and exhibiting low power consumption. Perovskite quantum dots (QDs) have garnered significant attention in artificial photonic synapses due to their facile synthesis and favorable optoelectronic properties. However, challenges such as limited carrier mobility and nonlinearity impede their performance in neuromorphic applications. In this study, CsPbBr3-attached MXene nanostructures (CsPbBr3-MXene), in-situgrowth of CsPbBr3QDs on MXene nanosheets, were proposed as the light-absorbing layer of a synaptic phototransistor. The heterostructure formed by CsPbBr3and MXene enhances photocurrent generation. Comparative analyses between CsPbBr3-MXene synapse transistor and that containing only CsPbBr3revealed a 24.6% higher excitatory postsynaptic current (EPSC) in the CsPbBr3-MXene one under identical light pulse stimulation. Following calculations and comparisons, the linearity exhibited significant improvement, decreasing from 4.586 to 1.099. Furthermore, the recognition accuracy in handwritten digit classification notably increased, rising from 86.13% to 92.05%. Moreover, the F1 score in edge detection had improvement, advancing from 0.8165 to 0.9065, approaching closer to 1. These improvements have demonstrated substantial assistance in the field of neural computing.

Photoelectric synaptic transistors integrate optical sensing and synaptic functions into a single device, which has significant advantages in neuromorphic computing for visual information, recognition, memory, and processing. However, the weight updating of existing photoelectric synapses is predominantly based on separate utilization of light and electrical stimuli to regulate synaptic excitation and inhibition. This approach significantly restricts the processing speed and application scenarios of devices. In this work, we propose bipolar synaptic organic/inorganic heterojunction transistor (BSOIHT) that can effectively simulate bidirectional (excitatory/inhibitory) synaptic behavior under light stimulation. Furthermore, by changing the position of electrode contacts and the metals of source and drain electrodes, carrier injection of the transistor is significantly improved with reduced synaptic event power consumption down to 2.4 fJ. Moreover, the BSOIHTs are adopted to build the neuromorphic vision system, which effectively facilitates image preprocessing and substantially enhances the recognition accuracy from 44.93% to 87.01%. This paper provides new avenues for the construction of energy-efficient artificial vision systems.

The flexible organic phototransistors (OPTs) are crucial for next-generation wearable systems for applications where large mechanical deformation is involved. However, most of the reported OPTs utilizing the field-effect transistor (FET) architecture suffer from undesired mechanical flexibility and limited performance due to their interfacial charge transport and inherently low transconductance; moreover, their π-conjugated semiconductor polymers that serve as channels lack specific healing sites, making it difficult to intrinsically heal themselves. Herein, a more flexible and high-performance OPT with enhanced interfacial charge transport vianovel volumetric channel and strong healing capability is developed for the first time. This OPT utilizes an organic electrochemical transistor architecture that consists of intrinsically healing conducting polymer/hydrogel composite films with three-dimensional volumetric channels. Such devices not only efficiently restore their mechanical and electrical performance in 100 ms after undergoing severe damage but also exhibit excellent mechanical flexibility without obviously degraded performance. More importantly, the self-healing OPTs exhibit high performance with a responsivity as high as 1.01 × 105A W−1, detectivity of 1.75 × 1012Jones, and high external quantum efficiency of 3.03 × 104%, higher than those of the majority of the reported FET-based OPTs. All of these results indicate that these novel and intrinsically self-healing OPTs with volumetric channels are ideal for use in next-generation wearable devices.

The extensive application of increasingly sophisticated artificial intelligence in life has promoted the artificial morality (AM) issue. The establishment and implementation of artificial ethics for robots are usually solved by passive program instructions, while active realization at the hardware level remains challenging. Here, inspired by cognitive psychology and neurophysiology, a typical AM device is demonstrated. The first of the three laws of robotics is realized in this device by judging good and evil and solving moral dilemmas. The device exhibits the three states of ego, id, and superego and the six humanities of the categorical imperative, instinctive disregard, instinctive impulse, moral deontology, utilitarianism, and egoism. The origin of morality is revealed in terms of electronics, which is an adversarial collaboration between the subconscious and the conscious. This work provides a practicable path for the consciousness generation and moral formation of artificial intelligence in the future.

The large-scale fabrication and patterning of artificial perceptual systems are vital for the development of bionic systems. Traditional patterning processes are often constrained by the use of mask versions, which are not only expensive but also challenging to produce on a large scale. Inkjet printing technology with maskless patterning capability is well suited for current demands for large-scale preparation and patterning. However, the application of printing techniques is typically confined to the production of simple devices or the preparation of patterned active layers within more complex devices. In this study, we successfully fabricated fully inkjet-printed indium gallium zinc oxide (IGZO) memristor arrays to mimic artificial nociceptor (pain receptors). By integrating a layer of silver into the all metal-oxide indium tin oxide (ITO)/IGZO/ITO memristor, we achieved stable threshold switching characteristics, a large current switching ratio of 105, excellent switching durability of 104cycles scans, and excellent spatial uniformity. We investigated the Ag-based conductive filament conduction mode and mimic LIF characteristics (leaky, integrate and fire), demonstrating the potential of the memristor as an artificial neuron. Lastly, we successfully implemented artificial nociceptor, including “threshold fire”, “relaxation”, “non-adaptation”, and “sensitization”, leveraging the stable threshold switching properties of the device. Our work demonstrates the significant potential of inkjet printing technology in the realization of bionic systems.

Biological nervous system is a complex multifunctional learning system. Our brain can receive, judge, and remember different signals from different sensory organs, such as hearing, touch, taste, and so on. Constructing such an artificial perceptual memory system is critical for biomimetic and real-time neuromorphic computing applications. Here, an artificial tactile perception system with the functions of perception, memory, and feedback is realized by utilizing a triboelectric nanogenerator (TENG) with a polymer light-emitting electrochemical cell (PLEC). The TENG functions as a tactile receptor to convert pressure signals into electrical signals, and the PLEC functions as a light-emitting electrochemical artificial synapse (LEEAS) with the dual output of photoelectric signal, in which the optical signals are used to reflect the strength of the pressure and the electrical signals are used to achieve different memory levels. Interestingly, the artificial tactile system successfully simulates the stress response and pain memory of organisms to different levels of stress signals, which is quite similar to the actual biological perception system and is expected to build artificial skin.

To date, the preparation of high-performance stretchable and elastic semiconductors remains challenging yet urgently demanded by the stretchable electronics field. Herein, we design a multiblock polyurethane elastomer matrix PBTTT-b-HTPB by incorporating crystalline oligothiophene and flexible polyolefin blocks to blend with conjugated polymers for high-mobility semiconductor nanofilms with enhanced stretchability and elasticity. The compatibility between the matrix and the conjugated polymer is found to play the key role in manipulating the vertical and lateral phase separation structure, hence the electrical and mechanical performance of the resulting semiconducting blend films. Though five representative p-type conjugated polymers (PCDTBT, TDPP-Se, PffBT4T-DT, PBTTT and IDTBT) all form vertical continuous structures, as confirmed by film-depth-dependent light absorption spectra, only the former four thermodynamically compatible polymers can generate well-dispersed lateral structures for improved mechanical properties without compromising electrical performance. In particular, the mobility of TDPP-Se/PBTTT-b-HTPB (1:3 by weight) nanofilms reaches 2.20 cm2V–1s–1in thin-film transistors, which is among the highest values for stretchable semiconductors characterized by conventional rigid devices so far. In addition, top mechanical performance with a fracture strain of 446 ± 35% and an elastic recovery higher than 90% in the strain range of 100–150% is realized at the same time. Notably, the excellent elasticity enabled the nanofilm’s long cycling life of up to 5000 cycles at a 100% strain, which is by far the longest cycling life at such a large strain, demonstrating its potential for practical application for wearable electronics.

Flexible organic synaptic transistors (FOSTs) have attracted considerable attention owing to their flexibility, biocompatibility, ease of processing, and reduced complexity. However, FOSTs rarely maintain the mechanical stability of their synaptic properties while meeting the device deformation requirements. Here, we experimentally found that bending deformation had a greater influence on the synaptic performance (i.e., the excitatory postsynaptic current (EPSC) value) of FOSTs than on the on-state current. Moreover, through formula derivation, we proved that the density of bending-induced defect states generated near the channel considerably influences the synaptic performance. We propose a general approach to tune the stable segment of the device using an encapsulation layer. The EPSC value of the ordinary FOSTs without a regulated stable segment decreased by nearly 1.5–2 orders of magnitude after bending. In contrast, the designed flexible synaptic device exhibited relatively stable EPSC. Moreover, the designed FOST exhibited stable paired-pulse facilitation, long-term potentiation, and optical synaptic performance. Furthermore, neuromorphic computational simulations based on our device before and after 500 bending cycles were performed using a handwritten artificial neural network. The device showed stable recognition accuracy after 50 learning cycles (91.55% in the initial state and 90.43% after 500 bending cycles). The successful application of a stable segment in flexible synaptic transistors provides a convenient and simple idea for fabricating flexible neuromorphic electronics with mechanical stability.

To build neuromorphic computing networks equivalent to the human brain, single artificial synaptic devices should exhibit low energy consumption down to femtojoules. However, most existing solutions for implementing low-energy synaptic devices based on an Ohmic contact are complex in structure or require specific materials, which hinder the further development of artificial neural networks. In this study, a Schottky-barrier-regulated organic synaptic transistor (SBROST) was reported. The device performance was improved by introducing the Schottky barrier at the contact interface between the source electrode and the semiconductor, thereby considerably reducing the energy consumption of one synaptic event compared with conventional OSTs with an Ohmic contact. The SBROST can not only reduce the device’s operating voltage and current but also possess a simple structure that can be utilized in different organic synaptic devices. Furthermore, high recognition accuracy at low energy consumption can be achieved by the SBROST. After 100 epochs, the SBROST-based handwritten artificial neural network exhibits excellent recognition accuracy (93.53%), which is close to the ideal accuracy (95.62%). The scheme of introducing a Schottky barrier into synaptic transistors offers a new perspective for constructing brain-like neural computing networks.

In response to the traditional MPC algorithm's difficult parameter adjustment leading to the inability to steer in a timely and accurate manner to avoid collisions, this paper designs an optimisation algorithm based on the simulated annealing algorithm with automatic parameter adjustment MPC. A vehicle dynamics model and a prediction model are established, and the simulated annealing algorithm is used to solve the objective function of the predetermined trajectory and obtain the weight matrix applicable to the prediction model. MPC the optimised controller is used to achieve the steering and collision avoidance trajectory tracking control of the vehicle. The simulation results under two operating conditions of medium and high-speed show that the controller can achieve fast response of vehicle steering and collision avoidance at different speeds and can keep the tracking error within 5%. The controller has the characteristics of timeliness, accuracy and stability.

Artificial tactile sensing system could realize the perception of different types of tactile information. Nowadays, most tactile sensors mainly convert the external stimulation signals into different material conductivity value, and processed by complex external circuits to detect the human physiological signals. In this work, the artificial synaptic device was connected with triboelectric nanogenerator(TENG) to fabricate a tactile sensing system in a simple way. This system could detect the external pressure stimulations and accurately recognize small frequency touch signals of $1-2.5$ Hz. In addition, the complex biological behavior was also simulated, which enlightened significance for the future bionic tactile systems and the field of human-computer interaction.

Over the past many years, array circuits, as an important part of electronic systems, show a wide range of application scenarios. However, the pursuit of high-density array circuits will certainly lead to a steep increase in the area of the array driving circuit. In this article, we present a switching array circuit with an artificial neuron, where the neuron acts as a discrete control center for the array circuit, which can effectively reduce the area of the array driving circuit by 75%. In this work, we found that the value of load resistance can affect the switching of individual devices and control the branch threshold voltage range, thus affecting the information processing capability of the branch. This new paradigm of optimized array circuit driving strategy can be used to tune the set and reset threshold voltage of neurons, while processing and outputting information by using this type of configuration. This work provides a potential platform for promoting the development of electronic devices and future high-density array circuits.

Artificial neurons have good application prospects as efficient processing units in artificial neural morphology. Indiun galliun zinc oxide (IGZO) is widely used in artificial synapses and artificial neurons due to its high reliability. In this study, we propose a neural device based on IGZO memristors fabricated by the solution method, owing to its simple process steps, no vacuum treatment process, rapid film formation, easy doping, and more suitability for low-temperature film deposition on a flexible substrate, as well as compatibility with a thin-film transistor (TFT). The influence of In, Ga, and Zn component ratios on neuron performance is investigated. Indium contributes greatly to the conductivity of the device, gallium inhibits the formation of oxygen vacancies, and zinc has a large impact on the morphology of the film, all of which impact neuron properties. Finally, we successfully simulate the integrate and fire (IF) characteristics of neurons through suitable component ratios and explore the difference in the electrical properties of the devices under different annealing atmospheres. The solution-processed IGZO-based neurons show great potential for neuromorphic systems.

With high efficiency and low energy consumption, bio-inspired artificial neuromorphic systems are regarded as the next generation of computing methods and have attracted tremendous attention in recent years. In bio-inspired artificial neuromorphic systems, multi-functional artificial neurons function as processing units to process complex information with high efficiency. However, the majority of reports on artificial neurons are based on electrical stimulation, whereas light-simulated neurons receive less attention. In this work, an artificial optoelectronic neural device based on an Ag/MXene/SiO2/Si structure is demonstrated. By introducing an oxidized MXene (O-MXene) layer, photoelectric integration can be realized on a single device. Compared to pure electrical stimulation, the synergistic effect of light and electrical stimulation can efficiently accelerate the firing behavior of neurons. In addition, we successfully designed and demonstrated a $64\times64$ sensing array based on optoelectronic neurons to recognize and sharpen the input signal trajectory. The proposed artificial optoelectronic neural device shows great potential for optoelectronic neuromorphic systems and is expected to promote the development of multifunctional, high-performance neuromorphic systems.

Organic synaptic transistors with excellent solution processability and biocompatibility have emerged as artificial electronic synapses. Regular organic synaptic transistors suffer from slight conductance variation and asymmetric conductance tuning, limiting the development of the model perception accuracy of the organic neuromorphic systems. Here, we first develop an electret-based vertical organic synaptic transistor (EVOST) with Mxene as the source electrode. The EVOST achieves linear conductance tuning by leveraging the nanoscale carrier transport channel length and high conductivity of MXene. Moreover, we develop an artificial neural recognition system composed of EVOSTs for recognizing the random images from the database, which achieves outstanding perception accuracy of 94.9%. The EVOST provides an alternative way for neuromorphic computing networks.

As one of the most common synergies between the exteroceptors and proprioceptors, the synergy between visual and vestibule enables the human brain to judge the state of human motion, which is essential for motion recognition and human self-protection. Hence, in this work, an artificial motion sensory system (AMSS) based on artificial vestibule and visual is developed, which consists of a tribo-nanogenerator (TENG) as a vestibule that can sense rotation and synaptic transistor array as retina. The principle of temporal congruency has been successfully realized by multisensory input. In addition, pattern recognition results show that the accuracy of multisensory integration is more than 15% higher than that of single sensory. Moreover, due to the rotation recognition and visual recognition functions of AMSS, we realized multimodal information recognition including angles and numbers in the spiking correlated neural network (SCNN), and the accuracy rate reached 89.82%. Besides, the rapid self-protection of a human was successfully realized by AMSS in the case of simulated amusement rides, and the reaction time of multiple motion sensory integration is only one-third of that of a single vestibule. The development of AMSS based on the synergy of simulated vision and vestibule will show great potential in neural robot, artificial limbs, and soft electronics.

Photonic synaptic transistors are promising neuromorphic computing systems that are expected to circumvent the intrinsic limitations of von Neumann-based computation. The design and construction of photonic synaptic transistors with a facile fabrication process and high-efficiency information processing ability are highly desired, while it remains a tremendous challenge. Herein, a new approach based on spin coating of a blend of CsPbBr3perovskite quantum dot (QD) and PDVT-10 conjugated polymer is reported for the fabrication of photonic synaptic transistors. The combination of flat surface, outstanding optical absorption, and remarkable charge transporting performance contributes to high-efficiency photon-to-electron conversion for such perovskite-based synapses. High-performance photonic synaptic transistors are thus fabricated with essential synaptic functionalities, including excitatory postsynaptic current (EPSC), paired-pulse facilitation (PPF), and long-term memory. By utilizing the photonic potentiation and electrical depression features, perovskite-based photonic synaptic transistors are also explored for neuromorphic computing simulations, showing high pattern recognition accuracy of up to 89.98%, which is one of the best values reported so far for synaptic transistors used in pattern recognition. This work provides an effective and convenient pathway for fabricating perovskite-based neuromorphic systems with high pattern recognition accuracy.

Interfacial engineering at the dielectric/semiconductor interface is highly crucial for fabricating organic field-effect transistors with high performance. In this study, a bilayer MXene/semiconductor configuration is introduced to fabricate a high-performance n-type transistor, where electrical charges are formed and modulated at the SiO2/semiconductor interface, and MXene nanosheets serve as the primary electrical charge channel due to their high mobility and long lateral size. The electrical performance is optimized by adjusting the degree of connectivity of the MXene nanosheets. The proposed MXene/poly{[N,N′-bis(2-octyl-do-decyl)-naphthalene-1,4,5,8-bis(dicarboximide)-2,6-diyl]-alt-5,5′-(2,2′-bithiophene)} (N2200) transistors show boosted n-type characteristics, including a 100-fold increase in field-effect mobility, a large ON/OFF ratio of 104, and a small subthreshold swing of 0.65 V dec−1, all of which are significantly improved compared with single-layer N2200 transistors. The high performance of the two-dimensional MXene nanochannel is due to its electronegativity and high mobility. The electronegativity significantly enhances electron transfer from N2200 to the MXene channel, where they are efficiently transported along the MXene channel. Interestingly, the MXene/p-type semiconductor transistors show suppressed p-type performance because of the highly negative MXene nanosheets. Additionally, the proposed bilayer MXene/n-type semiconductor configuration shows a good configuration generality and improved performance. These findings demonstrate the feasibility of fabricating high-performance n-type transistors using a bilayer MXene/semiconductor combination.

Structural health monitoring (SHM) of bridges and other building structures has been widely concerned by researchers and engineers. SHM achieved a breakthrough by using convolutional neural networks (CNNs) in deep learning. However, the traditional vision based SHM can not monitor the invisible structural damage. The traditional vibration based SHM needs to convert 1D signals into 2D images, which is easy to lose information in the conversion process, and requires a large scale of hardware circuit. In this paper, a 1D in-situ convolution system (ICS) based on vibration signal is proposed. The SHM is realized by 1D convolution calculation of the structural vibration signal detected by a high-sensitivity triboelectric nanogenerator (TENG) through one in-situ convolution transistor. Based on the natural adaptability of TENG to 1D convolution calculation, and combined with the ability of a single in-situ convolution device to replace more than eight multi conductance devices, ICS can reduce the required SHM computing array area by over 87.5%. Finally, the real-time identification of different positions and different damage levels of the steel bridge was realized by simulating the vibration response of the steel bridge.

Developing multifunctional artificial sensory systems is an important task for constructing future artificial neural networks. A system with multisignal output capability is highly required by the rising demand for high-throughput data processing in the Internet of Things (IoT) society. Here, a novel dual-output artificial tactile sensing (DOATS) system with parallel output of photoelectric signals was proposed. Because of the ionic–electronic coupling mechanism in light-emitting synaptic (LES) devices in the DOATS system, modulating electric current and light emission can coexist through ion accumulation and electron–hole recombination. As a result, the DOATS system can realize the simulation of human tactile information, and the recognition of 16 kinds of fabrics was demonstrated with an accuracy rate of 94.1%. A photoelectric hybrid artificial neural network was proposed, which achieved efficient and accurate multitask operation. The DOATS system proposed in this work is promising for implementing photoelectric hybrid neural network and promoting the development of interactive artificial intelligence.

Developing a high-efficiency neuromorphic hardware network is essential to achieve complex artificial intelligence. Here, for the first time, we propose a multi-neuromorphic functional device based on the 2D material Mxene—a switchable neuronal-synaptic transistor (SNST) that can be programmed to realize synaptic or neuronal function—and break the boundaries between neuronal and synaptic modules for a high-efficiency neuromorphic network, including fabrication efficiency, resource utilization efficiency, and operational efficiency. A neural network composed of multiple SNSTs is designed for authenticity data recognition that can equitably redistribute the neuromorphic hardware sources and adjust the topological structure of the hardware network, improving operational speed and reducing the number of devices in the network. Finally, an SNST-based hardware system is developed for facial recognition with a recognition accuracy of about 80%. This work demonstrates that programmable SNST-based neuromorphic chips with a simple fabrication process, equitable resource utilization, and high operational speed have great potential for highly efficient and accurate neuromorphic networks.

Donor–acceptor (D–A) alignment that integrates D–A pairs into the modular and versatile crystalline metal–organic frameworks is a powerful strategy to precisely fabricate multifunctional materials with unique optoelectronic properties and applications at the molecular level. Herein, we reported an unprecedented threefold interpenetrating D–A hybrid framework by incorporating an electron-deficient linear viologen zwitterion into the lead-halide systems. The 1D iodoplumbate nanoribbon and interpenetrating close-packed D–A structure endowed this hitherto unknown semiconductive alignment with the anisotropic conductivity and high stability. When used in a thermal sensor, it presented exceptional electrical properties with a high sensitivity (high thermal index Bof 4671 K) and decent temperature coefficient of resistivity (0.72% °C–1) in wide operational temperature ranges (40–220 °C), which are among the best of the related thermistors. This work develops a pathway to bridge the gaps between hybrid materials and electron devices.

The etching process can serve as an effective top-down approach that facilitates direct construction of tuned patterns, reversely studying growth mechanisms and further inducing unexpected physical properties. Currently, etching behaviors of monolayer transition-metal dichalcogenides (TMDs) have been rarely explored and the intrinsic etching mechanism still needs to be fully elucidated. Here, we demonstrate a facile and controllable oxygen-assisted anisotropic chemical etching of two-dimensional (2D) materials. Taking MoSe2as an example, a series of well-defined etched patterns can be fabricated by precisely modulating the pretreatment time of oxygen plasma and the etching time. Atomic characterization exhibits that the edges of etched patterns are mainly zigzag-terminated. Density functional theory calculations highly agree with the experimental results, indicating an energy preference for zigzag edges. The as-produced MoSe2flakes can be used as part of heterojunctions to display intriguing optoelectronic properties. The field-effect transistor based on as-etched MoSe2flakes and the 2D poly(3-hexylthiophene-2,5-diyl) crystal shows a distinct enhanced ambipolar photoresponse. Our strategy offers a new direction for preparing patterned 2D TMDs with desirable characteristics, opening the door for facilitating the development of optoelectronics.

Neural machine translation, which has an encoder-decoder framework, is considered to be a feasible way for future machine translation. Nevertheless, with the fusion of multiple languages and the continuous emergence of new words, most current neural machine translation systems based on von Neumann’s architecture have seen a substantial increase in the number of devices for the decoder, resulting in high-energy consumption rate. Here, a multilevel photosensitive blending semiconductor optoelectronic synaptic transistor (MOST) with two different trapping mechanisms is firstly demonstrated, which exhibits 8 stable and well distinguishable states and synaptic behaviors such as excitatory postsynaptic current, short-term memory, and long-term memory are successfully mimicked under illumination in the wavelength range of 480–800 nm. More importantly, an optical decoder model based on MOST is successfully fabricated, which is the first application of neuromorphic device in the field of neural machine translation, significantly simplifying the structure of traditional neural machine translation system. Moreover, as a multi-level synaptic device, MOST can further reduce the number of components and simplify the structure of the codec model under light illumination. This work first applies the neuromorphic device to neural machine translation, and proposes a multi-level synaptic transistor as the based cell of decoding module, which would lay the foundation for breaking the bottleneck of machine translation.

Despite recent remarkable progress in multiple synaptic devices, searching for artificial synapses with new functions is still an important task in the construction of artificial neural networks. The parallel output functionality of photoelectric signals in artificial synaptic devices is interesting and desirable as on-chip optoelectronic interconnection technology allows the connections between neurons weighted by current and light. In turn, it provides degrees of freedom and reduces circuit lead density in the design of large-scale neural networks. Hence, for the first time, a light-emitting electrochemical artificial synapse (LEEAS) based on poly[2-methoxy-5-(2′-ethylhexyloxy)-1,4-phenylene vinylene]/poly (ethylene oxide)/lithium salt blends with dual output of photoelectric signals was developed in this study. The electrochemical redox reaction enables the device to achieve synaptic plasticity in biology and emulate the memory enhancement process, high-pass filtering characteristic, and classical Pavlov’s conditioned reflex experiment. In addition, the transient luminescence intensity of the LEEAS induced by identical electric spikes exhibits a synaptic-like potentiation behavior. Owing to the combination of electroluminescence (EL) and synaptic memory behavior, an LEEAS array exhibits a unique image display and storage functions that can memorize displayed images. The LEEAS proposed in this work enriches the diversity of artificial synapses, promoting the diversified design and development of next-generation optoelectronic hybrid artificial neural networks.

Enormous synaptic devices are required to build a parallel, precise, and efficient neural computing system. To further improve the energy efficiency of neuromorphic computing, a single high-density synaptic (HDS) device with multiple nonvolatile synaptic states is suggested to reduce the number of synaptic devices in the neural network, although such a powerful synaptic device is rarely demonstrated. Here, a photoisomerism material, namely, diarylethene, whose energy level varies with the wavelength of illumination is first introduced to construct a powerful HDS device. The multiple synaptic states of the HDS device are intrinsically converted under UV–vis regulation and remain nonvolatile after the removal of illumination. More importantly, the conversion is reconfigurable and reversible under different light conditions, and the synaptic characteristics are comprehensively mimicked in each state. Finally, compared with a two-layer multilayer perceptron (MLP) architecture based on static synaptic devices, the HDS device-based architecture reduces the device number by 16 times to achieve a minimalist neural computing structure. The invention of the HDS device opens up a revolutionary paradigm for the establishment of a brain-like network.

The development of various artificial electronics and machines would explosively increase the amount of information and data, which need to be processed via in-situ remediation. Bioinspired synapse devices can store and process signals in a parallel way, thus improving fault tolerance and decreasing the power consumption of artificial systems. The organic field effect transistor (OFET) is a promising component for bioinspired neuromorphic systems because it is suitable for large-scale integrated circuits and flexible devices. In this review, the organic semiconductor materials, structures and fabrication, and different artificial sensory perception systems functions based on neuromorphic OFET devices are summarized. Subsequently, a summary and challenges of neuromorphic OFET devices are provided. This review presents a detailed introduction to the recent progress of neuromorphic OFET devices from semiconductor materials to perception systems, which would serve as a reference for the development of neuromorphic systems in future bioinspired electronics.

Two-dimensional (2D) materials based artificial synapses are important building blocks for the brain-inspired computing systems that are promising in handling large amounts of informational data with high energy-efficiency in the future. However, 2D devices usually rely on deposited or transferred insulators as the dielectric layer, resulting in various challenges in device compatibility and fabrication complexity. Here, we demonstrate a controllable and reliable oxidation process to turn 2D semiconductor HfS2into native oxide, HfOx, which shows good insulating property and clean interface with HfS2. We then incorporate the HfOx/HfS2heterostructure into a flash memory device, achieving a high on/off current ratio of ∼105, a large memory window over 60 V, good endurance, and a long retention time over 103seconds. In particular, the memory device can work as an artificial synapse to emulate basic synaptic functions and feature good linearity and symmetry in conductance change during long-term potentiation/depression processes. A simulated artificial neural network based on our synaptic device achieves a high accuracy of ∼88% in MNIST pattern recognition. Our work provides a simple and effective approach for integrating high-kdielectrics into 2D material-based memory and synaptic devices.

In recent years, much attention has been focused on two-dimensional (2D) material-based synaptic transistor devices because of their inherent advantages of low dimension, simultaneous read–write operation and high efficiency. However, process compatibility and repeatability of these materials are still a big challenge, as well as other issues such as complex transfer process and material selectivity. In this work, synaptic transistors with an ultrathin organic semiconductor layer (down to 7 nm) were obtained by the simple dip-coating process, which exhibited a high current switch ratio up to 106, well off state as low as nearly 10–12A, and low operation voltage of −3 V. Moreover, various synaptic behaviors were successfully simulated including excitatory postsynaptic current, paired pulse facilitation, long-term potentiation, and long-term depression. More importantly, under ultrathin conditions, excellent memory preservation, and linearity of weight update were obtained because of the enhanced effect of defects and improved controllability of the gate voltage on the ultrathin active layer, which led to a pattern recognition rate up to 85%. This is the first work to demonstrate that the pattern recognition rate, a crucial parameter for neuromorphic computing can be significantly improved by reducing the thickness of the channel layer. Hence, these results not only reveal a simple and effective way to improve plasticity and memory retention of the artificial synapse via thickness modulation but also expand the material selection for the 2D artificial synaptic devices.

Organic phototransistors (OPTs) have attracted enormous attention because of their promising applications in sensing, communication, and imaging. Currently, most OPTs reported utilize field-effect transistors (FETs) with relative long channel length which usually has undesired amplification because of their inherent low transconductance originated from their low channel capacitance, limiting the further improvement of performance. Herein, a vertical channel hybrid electrochemical phototransistor with a nanoscale channel and large transconductance (VECPT) is invented for the first time to achieve ultrahigh photoresponsivity along with a fast response speed. Benefiting from the nanoscale channel length and large transconductance, the photo-generated carriers in channel can be efficiently dissociated, transported, and amplified into the enlarged photocurrent output. Therefore, the devices deliver substantially improved optoelectronic performances with a photoresponsivity as high as ≈2.99 × 107A/W, detectivity of ≈1.49 × 1013Jones, and fast-speed response of ≈73 μs under a low voltage of 1 V, which are superior to those of the reported OPTs based on FETs. Moreover, the in situ Kelvin probe microscopy is performed to characterize the surface potential of device systems for better elucidating the photosensing mechanism. Furthermore, taking advantage of its excellent optoelectronic performance, an ultraviolet light monitoring system is constructed by integrating VECPT with a light-emitting diode, which also shows the real-time, high-sensitive, and controllable photoresponse threshold properties. All these results demonstrate the great potential of these electrochemical phototransistors and provide valuable insights into the design of the nanoscale channel length device system for high-performance photodetection.

Organic phototransistors with high sensitivity and responsivity to light irradiance have great potential applications in national defense, meteorology, industrial manufacturing, and medical security. However, undesired dark current and photoresponsivity limit their practical applications. Here, a novel vertical organic phototransistor combined with ferroelectric materials is developed. The device structure has nanometer channel length, which can effectively separate photogenerated carriers and reduce the probability of carrier recombination and defect scattering, thus improving the device performance of phototransistors. Moreover, by inserting the poly(vinylidene fluoride-co-trifluoroethylene) (P(VDF-TrFE)) ferroelectric layer, the Schottky barrier at the interface between the semiconductor and source can be adjusted by the polarization of the external electric field, which can effectively reduce the dark current of the phototransistor to further improve the device performance. Therefore, our phototransistors exhibit a high photoresponsivity of more than 5.7 × 105A/W, an outstanding detectivity of 1.15 × 1018Jones, and an excellent photosensitivity of 5 × 107under 760 nm light illumination, which are better than those of conventional lateral organic phototransistors. This work provides a new approach for the development of high-performance phototransistors, which opens a new pathway for organic phototransistors in practical application.

The dynamic analysis of multi-layer cantilever beam's structure is of great significance to the structural design of MEMS devices. Due to the multi-coupling and non-linearity of the multi-layer cantilever beam, this paper presents a method for modelling and solving multilayer micro-beam. Firstly, the dynamic equation mathematical model of electrostatic multi-layer cantilever beam was established based on the energy principle and fluid film damping effect; second, the dynamic model was transformed to single-layer beam solution model by equivalent parameter method. Thirdly, Galerkin mapping reduction algorithm was used to solve this model. The theoretical computing results and simulation results can provide strong support for the study of the working motion state of the MEMS devices.

Multiple types of synaptic transistors that are capable of processing electrical signals similar to the biological neural system hold enormous potential for application in parallel computing, logic circuits and peripheral detection. However, most of these presented synaptic transistors are confined to a single mode of synaptic plasticity under an electrical stimulus, which tremendously limits efficient memory formation and the multifunctional integration of synaptic transistors. Here, we proposed a bi-mode electrolyte-gated synaptic transistor (BEST) with two dynamic processes, the formation of an electrical double layer (EDL) and electrochemical doping (ECD) by tuning the applied voltages, thereby allowing volatile and non-volatile behavior, which is associated with additional ion doping and nanoscale ionic transport. Benefiting from two controllable dynamic processes, we surprisingly found a third state in the transfer curves besides the “off” and “on” states. Moreover, utilizing this unique property, an artificial nociceptor with multilevel modulation of sensitivity was realized based on our bi-mode device. Finally, a haptic sensory system was constructed to exhibit robotic motion that revealed a unique threshold switching behavior, indicating the applicability to peripheral sensing circuits. Hence, the presented bi-mode synaptic transistor provides promising prospects in achieving multiple-mode integrated devices and simplifying neural circuits, which shows great potential in the development of artificial intelligence.

Organic electrochemical transistors (OECTs) have attracted considerable interests for various applications ranging from biosensors to digital logic circuits and artificial synapses. However, the majority of reported OECTs utilize large channel length up to several or several tens of micrometers, which limits the device performance and leads to low transistor densities. Here, we demonstrate a new design of vertical OECT architecture with a nanoscale channel length down to ∼100 nm. The devices exhibit a high on-state current of over 20 mA under a low bias voltage of 0.5 V, a fast transient response of less than 300 μs, and an extraordinary transconductance up to 68.88 mS, representing a record-high value for OECTs. The excellent electrical performance is attributed to the novel structure with a nanoscale channel length defined by the channel material thickness, which is intrinsically different from that of conventional OECTs with the channel length limited by the lithography resolution. Owing to the low thermal budget, we fabricate flexible devices on a flexible substrate, which exhibit unprecedented endurance characteristics and mechanical robustness after 1000 blending cycles. Furthermore, the proposed device is capable of mimicking biological inhibitory synapses for application in intelligent artificial neural networks. Our work not only pushes the performance limit of OECTs but also opens up a new design of OECTs for high-performance biosensors, digital logic, and neuromorphic devices.

Floating gate transistor photomemory (FGTPM) has been regarded as one of the most prospective nonvolatile photomemory devices because of its compatibility with transistor-based circuits, nondestructive reading, and multilevel storage. Until now, owing to the excellent photoelectric properties, lead-based perovskite nanocrystals (PNCs) have been applied in most of the perovskite-based FGTPM devices and embedded in the polymer matrix as the charge trapping layer. However, the polymer matrix and its solvent would degrade the structure of the PNCs, resulting in the loss of their unique photoresponse ability. In addition, lead-based perovskites have environmental unfriendliness and poor stability. Hence, a novel nonvolatile FGTPM based on oligomeric silica (OS) wrapped lead-free double perovskite Cs2AgBiBr6NCs was demonstrated for the first time. Acting synchronously as the protection layer for the discrete Cs2AgBiBr6NCs and charge tunneling layer for the FGTPM device, the OS layer can achieve controllable thickness by adjusting the process parameters, leading to an adjustment of storage properties with a larger memory window (58 V). Owing to the excellent photoresponse ability of the Cs2AgBiBr6@OS composite layer, the FGTPM device exhibited high-performance with repeatable multilevel nonvolatile photomemory and precise photoresponse ability of wavelength/time/power-dependent photoirradiation without extra gate biasing.

Depending on the storage mechanisms, organic field-effect transistor (OFET) memory is usually divided into floating gate memory, ferroelectric memory, and polymer-electret-based memory. In this work, a new type of nonvolatile OFET memory is proposed by simply blending a p-type semiconductor and a n-type semiconductor without using an extra trapping layer. The results show that the memory window can be effectively modulated by the dopant concentration of the n-type semiconductor. With the addition of a 5% n-type semiconductor, blending devices exhibit a large memory window up to 57.7 V, an ON/OFF current ratio (ION/IOFF) ≈ 105, and a charge retention time of over 10 years, which is comparable or even better than those of most of the traditional OFET memories. The discontinuous n-type semiconductor is set as a charge-trapping center for charge storage due to the quantum well-like organic heterojunctions. The generalization of this method is also investigated in other organic systems. Moreover, the blend devices are also applied to optical memory and show multilevel optical storage, which are further scaled up to 8 × 8 array to map up two-dimensional (2D) optical images with long-term retention and reprogramming characteristic. The results reveal that the novel system design has great potential application in the field of digital image memory and photoelectronic system.

Neuromorphic computing inspired by the neural systems in human brain will overcome the issue of independent information processing and storage. An artificial synaptic device as a basic unit of a neuromorphic computing system can perform signal processing with low power consumption, and exploring different types of synaptic transistors is essential to provide suitable artificial synaptic devices for artificial intelligence. Hence, for the first time, an electret-based synaptic transistor (EST) is presented, which successfully shows synaptic behaviors including excitatory/inhibitory postsynaptic current, paired-pulse facilitation/depression, long-term memory, and high-pass filtering. Moreover, a neuromorphic computing simulation based on our EST is performed using the handwritten artificial neural network, which exhibits an excellent recognition accuracy (85.88%) after 120 learning epochs, higher than most reported organic synaptic transistors and close to the ideal accuracy (92.11%). Such a novel synaptic device enriches the diversity of synaptic transistors, laying the foundation for the diversified development of the next generation of neuromorphic computing systems.

Neuromorphic computation, which emulates the signal process of the human brain, is considered to be a feasible way for future computation. Realization of dynamic modulation of synaptic plasticity and accelerated learning, which could improve the processing capacity and learning ability of artificial synaptic devices, is considered to further improve energy efficiency of neuromorphic computation. Nevertheless, realization of dynamic regulation of synaptic weight without an external regular terminal and the method that could endow artificial synaptic devices with the ability to modulate learning speed have rarely been reported. Furthermore, finding suitable materials to fully mimic the response of photoelectric stimulation is still challenging for photoelectric synapses. Here, a floating gate synaptic transistor based on an L-type ligand-modified all-inorganic CsPbBr3perovskite quantum dots is demonstrated. This work provides first clear experimental evidence that the synaptic plasticity can be dynamically regulated by changing the waveforms of action potential and the environment temperature and both of these parameters originate from and are crucial in higher organisms. Moreover, benefiting from the excellent photoelectric properties and stability of quantum dots, a temperature-facilitated learning process is illustrated by the classical conditioning experiment with and without illumination, and the mechanism of synaptic plasticity is also demonstrated. This work offers a feasible way to realize dynamic modulation of synaptic weight and accelerating the learning process of artificial synapses, which showed great potential in the reduction of energy consumption and improvement of efficiency of future neuromorphic computing.

Laser irradiation of semiconductors has attracted considerable attention due to its important background in all kinds of applications. Unfortunately, it has rarely been applied to optimize the morphology of the active layer in organic photovoltaics (OPV). This work provided previously unavailable information about the effect of laser annealing (LA) on the morphology and function of OPV and its comparison to thermal annealing (TA). The results showed that, compared with TA, the active layer treated with LA tended to form smaller donor crystals and a face-on structure, which facilitated hole transport. Moreover, small angle neutron scattering (SANS) experiments showed that, compared to TA, acceptor phase separation was suppressed by LA and became weaker with an increase of laser intensity. Furthermore, the maximum interfacial area between the acceptor phase and its surrounding matrix was achieved with proper laser intensity, which is found to be crucial for electron transport, bimolecular recombination, and subsequent device performance. Finally, the generality of femtosecond LA on the optimization device performance of OPV including fullerene and nonfullerene systems was also investigated. The findings in this work shed light on the application of LA for the further improvement of OPV device performance.

Reliably discerning real human faces from fake ones, known as antispoofing, is crucial for facial recognition systems. While neuromorphic systems offer integrated sensing-memory-processing functions, they still struggle with efficient antispoofing techniques. Here we introduce a neuromorphic facial recognition system incorporating multidimensional deep ultraviolet (DUV) optoelectronic synapses to address these challenges. To overcome the complexity and high cost of producing DUV synapses using traditional wide-bandgap semiconductors, we developed a low-temperature (≤70 °C) solution process for fabricating DUV synapses based on PEA2PbBr4/C8-BTBT heterojunction field-effect transistors. This method enables the large-scale (4-in.), uniform, and transparent production of DUV synapses. These devices respond to both DUV and visible light, showing multidimensional features. Leveraging the unique ability of the multidimensional DUV synapse (MDUVS) to discriminate real human skin from artificial materials, we have achieved robust neuromorphic facial recognition with antispoofing capability, successfully identifying genuine human faces with an accuracy exceeding 92%.

In this work, a novel vertical quantum-dot light-emitting transistor (VQLET) based on a vertical organic thin-film transistor is successfully fabricated. Benefiting from the new vertical architecture, the VQLET is able to afford an extremely high current density, which allows most of the organic thin film transistors (OTFT) even with low mobility (for instance, poly(3-hexylthiophene)) to drive a quantum-dot light-emitting diode (QLED), which was previously unavailable. Moreover, the hole injection barrier could be modulated by the additional gate electrode, which precisely optimizes the charge balance in the device, a critical issue in QLED, resulting in the precise control of current density and brightness of the VQLET. The VQLET shows a high performance with a maximum current efficiency of 37 cd/A. Furthermore, integrating OTFT and QLED into a single device, the VQLET features drastic advantages by realizing active matrix quantum-dot light-emitting diodes (AMQLEDs), which significantly reduces the number of transistors and frees the large area fraction occupied by transistors. Hence, these results indicate that the VQLET provides a new strategy for realizing a low-cost, solution-processed, high-performance OTFT-AMQLED for the flat panel display technology. Moreover, the novel design offers a unique method to exquisitely control the charge balance and maximize the efficiency the QLED.

The identifying characteristic of an organic electrochemical transistor (OECT) is the coupling between ionic and electronic charges within the entire volume of the channel. In this work, by taking advantage of the volumetric nature of the OECTs’ response, a novel flexible photodetector is reported for the first time based on all-solid-state OECT with an excellent responsivity of up to 6.7 × 106A/W, detectivity as high as 3.6 × 1013Jones, and a fast response of ∼0.13 s in the visible range, which are superior to those of the majority of the reported organic phototransistors (OPTs) based on field-effect transistors (FETs) and even better than those of FET-based phototransistors with two-dimensional (MoS2and graphene) and perovskite materials. The high performance of the devices was ascribed to the combination of the higher carrier mobility of poly(3,4-ethylenedioxythiophene) doped with poly(styrene sulfonate) (PEDOT:PSS) as a channel and the volumetric nature of the OECTs’ response, and the charge density of the volumetric channel was efficiently modulated by incident light compared to FETs. Moreover, OECT-based OPTs with quantum dots (CdSe/ZnS) as a light sensitizer were characterized under ultraviolet light, and they exhibited excellent photosensitivity, which further verified the superiority of OECT for phototransistors. Furthermore, a flexible image sensor was fabricated for the first time by integrating flexible OECTs-OPTs into a 10 × 10 array, which can clearly identify the target image under a bending state, indicating the great potential of OECTs-OPTs in the application of low-power, ultrasensitive flexible photodetectors and imaging technology.

Biological synapses were the basic structure for information transmit, which also perform key computation functions in neural circuits. Such functionality inspires artificial intelligence to mimic human brain, which has 1015synapses and consumed only ∼1–100 fJ per synaptic event during massively parallel information processing. It thus indicates that artificial synapse devices with low power consumption are highly desired to mimic human brain. In this work, we developed a distinctive structure of self-powered synapse transistor to emulate synapse functions, and the voltage was provided by a triboelectric nanogenerator (TENG) without using additional voltage to generate pre-synapse spike. Moreover, TENG could simultaneously function as tactile sensor to realize self-powered tactile synapse with simple device structure. The essential synaptic functions such as excitatory postsynaptic current (EPSC), paired-pulse facilitation (PPF), dynamic filtering, and short-term plasticity (STP) to long-term plasticity (LTP) were demonstrated in the self-powered synapse transistor devices by touching the TENG. Moreover, the logic gate “AND”, and “OR” were achieved using TENGs as multiple pre-synaptic, and tactile study was also mimicked. The integration of TENG and synapse transistor device has great potentials in the applications of novel e-skin devices for artificial intelligence and human-computer interaction.

Based on the theory and method of MBSE, a model of Steering-By-Wire (SBW) system with mechanical and electrical integration is established. Based on the model, a robust control strategy with H∞ mixed sensitivity is used, and a robust controller is designed. The high order robust controller is poor in real-time and has high cost. Routh, Hankel norm and ISE index reduction method are used to reduce the controller. Through time domain and frequency domain analysis, the effects of various reduction methods on the controller are discussed. Simulation results show that the ISE index reduction method has better reduction effect and has good robust performance and robust stability.