Bio-inspired artificial synaptic transistors: evolution from innovative basic units to system integration

Xin Wang ^a, Yixin Ran ^a, Xiaoqian Li ^b, Xinsu Qin ^a, Wanlong Lu ^a, Yuanwei Zhu ^a and Guanghao Lu *^a
aFrontier Institute of Science and Technology, State Key Laboratory of Electrical Insulation and Power Equipment, Xi'an Jiaotong University, Xi'an 710054, P. R. China. E-mail: guanghao.lu@mail.xjtu.edu.cn
^bShandong Technology Center of Nanodevices and Integration, School of Microelectronics, Shandong University, Jinan, Shandong Province 250100, P. R. China

First published on 15th May 2023

---

---

Xin Wang

Xin Wang is currently a PhD candidate in Xi’an Jiaotong University (XJTU), China. She received her bachelor's degree in 2018 from Taiyuan University of Science and Technology and MS degree in 2021 from Zhengzhou University. Her research interests focus on the preparation and application of flexible, low power consumption and multi-function intelligent synapse devices, memory devices and logic circuit.

Yixin Ran

Yixin Ran is currently pursuing his MS degree at the Institute of Frontier Science and Technology, Xi'an Jiaotong University, China. He received his BSc in Materials Science and Engineering from Hunan University, China, in 2017. His research interests focus on the study of optical and electrical distributions in organic semiconductor thin films and the preparation of high-performance multifunctional organic semiconductor optoelectronic devices.

Xiaoqian Li

Li Xiaoqian is currently a PhD candidate at the School of Microelectronics of Shandong University, majoring in microelectronics and solid-state electronics. She received her BS degree from Qufu Normal University (2018). She obtained an MS degree in materials science and engineering from Zhengzhou University (2021). Her research interests are the fabrication and developments of artificial synapses and neurons by means of diverse functional material systems.

Xinsu Qin

Xinsu Qin is currently pursuing her MS degree at the School of Chemistry, Xi'an Jiaotong University, China. She received her BSc in Forestry Products and Chemical Engineering from Northeast Forestry University, China, in 2016. Her research interests focus on the preparation of organic blend thin film devices and high-performance multifunctional organic semiconductor optoelectronic devices.

Wanlong Lu

Wanlong Lu is a PhD candidate at Xi‘an Jiaotong university under the direction of Dr Guanghao Lu. His study is centered on developing new organic semiconductor materials for high performance photoelectric devices.

Yuanwei Zhu

Dr Yuanwei Zhu is currently an associate professor at the school of electrical engineering in Xi’an Jiaotong University (XJTU), China. He received the PhD degree in electrical engineering from XJTU in 2018. His research interests include dielectrics, electrical insulating materials, organic semiconductors and their applications in power equipment and electronic devices. He can be reached at zhuyuanwei@.xjtu.edu.cn.

Guanghao Lu

Guanghao Lu, PhD, professor. He received his bachelor's degree in 2004 from Nanjing University and PhD degree in 2010 from Changchun Institute of Applied Chemistry, Chinese Academy of Sciences. After postdoctoral research in Germany and afterwards in the USA from 2010 to 2014, he joined Frontier Institute of Science and Technology, Xi'an Jiaotong University. The research of his group focuses on polymer thin film semiconductors and electronics.

---

1. Introduction

The concept of mimicking brain computing using neuro-morphology was put forward in the late 1980s, motivated by the human brain which has an efficient information processing ability.^1,2 Unlike von Neumann-based storage separation computing architecture, biological nervous systems possess efficient parallel calculations, ultralow power consumption and high fault tolerance superiority.^3–6 In a typical case, AlphaGo, built with only 1200 central processing units (CPUs) and 180 graphics processors (GPUs), could generate hundreds of thousands of watts of power. Biologically, the difference is the power consumption as low as 20 watts when processing information by the brain which contains 10¹¹ neurons and 10¹⁵ synapses. Similar to biological synapses, artificial synaptic devices that can simultaneously perform information processing and storage functions are considered a promising way to achieve brain-like computing. Therefore, there is a high demand for brain-like chip technology constructed by artificial synaptic devices to promote the evolution of artificial intelligence, sensor interconnection, human-computer interaction and efficient storage computing of explosive information.^7–9 Individual intelligent synaptic devices and sensing interconnect systems are expected to open the door of new-generation intelligent neurorobotics.^10–12

In addition, inspired by the brain, an ultra-low-power efficient computing system,¹³ scientists are planning to design a low-power brain-like computer with efficient computing ability to break through the limitations of power walls in the post-Moore era.^14,15 To achieve this goal, two mainstream techniques involving complementary metal oxide semiconductor (CMOS) based hardware digital systems and an intelligent synaptic software system have been widely studied.^14,16–18 However, digital circuits based on CMOS technology are limited by Moore's Law to exhibit megawatt-standard energy consumption, which is far more than the human brain (20 W).^19–21 Therefore, new hardware architectures such as coherent nanophotonic circuits and quantum neural network,^22,23 novel neural network integration systems such as self-powered neural devices, zero-gate neural devices,^24,25etc. have been designed to meet the low-power requirements of industry and academia.^26,27

Recently, diverse materials including organic semiconductors,²⁸ oxide semiconductors,²⁹ carbon nanotubes,³⁰ two-dimensional materials,³¹ quantum dots,³² multiple composite materials,³³ and device structures mainly focus on two-terminal³⁴ and three-terminal devices,³⁵ which have been attempted to fabricate a synaptic system. Additionally, an artificial intelligence network is introduced to imitate more complex features with low power consumption.^15,36 Compared with two-terminal devices like memristors, atomic switches and phase change memories,^37,38 three-terminal transistors possess higher fault tolerance, higher signal-to-noise ratio and a greater signal modulation property.^39–41 Mainstream neuronal synaptic transistors cover both vertical structures and planar structures, and the channel length could be regulated more accurately by vertical structure transistors and thus achieve a low power consumption.^42,43 According to the different implementation principles of synaptic transistors, they can be divided into polyelectrolyte,⁴⁴ ion-gel,⁴⁵ floating-gate,⁴⁶ electret,⁴⁷ ferroelectric,⁴⁸ polarized type and heterojunction structure transistors.^49,50 Based on these materials and structures, diverse artificial synapse devices and neural networks (ANNs) are developed for different information transfer and simulation applications. For example, stretchable flexible and wearable synaptic electronic systems implemented by printing technology are being vigorously developed with the advantages of organic semiconductors and carbon nanotubes that can be easily dissolved and printed.^51–54 Using typical printing techniques such as intaglio printing, screen printing, and roll-to-roll printing it is easier to realize high-throughput printing patterns, however, the specific mask and ink viscosity needs to be carefully designed.⁵⁵ Inkjet printing, as a layer-by-layer printing technology without a mask, can not only print multilayer complex patterns but also realize high-resolution patterns more easily.⁸ In addition, inorganic semiconductors possess a higher carrier mobility and can be applied to high-frequency electronic circuits,⁵⁶ while quantum dots and heterojunctions have higher sensitivity to light and are suitable for optical synaptic systems.^57,58 Additionally, some biocompatible synaptic devices especially the organic transistors can be degraded without any pollution and toxicity, which further facilitates the convergence of biotechnology and electronics in recording potential information from neurons.^59,60

Until now, major biological synaptic functions such as excitatory/inhibitory postsynaptic current (EPSC/IPSC), short/long term plasticity (STP/LTP) and paired pulse facilitation (PPF) have been widely simulated by controlling presynaptic stimulation.⁶¹ In addition, a more advanced evolution of synaptic transistors towards complex functions and system integration are extensively studied to enrich the diversity of signal transmission and perception processes.^62–64 Multiple inputs gates represent different environment signals to reflect the changes of surroundings and keep a memory of changing information through intelligent bionic systems. Furthermore, coupling vision system simulations to the human eye, tactile sensing systems to robot movement, auditory nervous system to human ears and the combination of a triboelectric system with muscle movement are currently being researched to reflect real-time dynamic changes in external information.^65–69 However, as the increasing data needs to be stored and recorded in the information age, multiple sensors are required, including acoustic, thermal, optical, electrical, strain and other coupling interconnections. In addition, the design of material systems, efficient calculation, operating power consumption and related working mechanism also need to be further optimized.^70–73 Therefore, the incorporation of sensing, conveying, simulating and feedback of information in single devices or synaptic systems are facing enormous challenges to the realization of related artificial intelligence.

Herein, recent advances in intelligent synaptic devices and system integration applications are discussed with respect to strategies for designing structures of materials and transistors, and coupling of various multifunctional devices and neuronal systems. Fig. 1 exhibits the main framework of this review. We highlight the effect of innovative structure design in materials and devices on the application of intelligent synaptic transistors in Section 2, multifunctional synaptic device including stretchable, degradable and multi-mode reconfigurable devices in Section 3 and the integration of artificial nervous systems in Section 4. These contents establish a foundation for hardware implementation of neuromorphic transistors and their applications in human vision, robot movement and self-powered devices. Potential mechanism of information transfer and challenges based on neuromorphic systems are also discussed in detail, which might give these devices emotional/operative attributes. Finally, the materials, high-throughput preparation techniques, multifunctional devices and multifunctional simulations commonly used in synapses were are summarized in Fig. 12. It is believed that the development of multimodal intelligent synaptic systems and brain-like chips would open a new chapter in artificial intelligence, wearable and stretchable electronic skin for constructing Internet of Everything and brain-like computing systems with perceptual intelligence.

2. Innovative structures of materials and transistors for smart synapses

Recently, proliferating research about transistor-based smart synapses undoubtedly accelerated the development of neuro-electronics. Among which, the structure of materials and devices shows significant impacts on the synaptic functions. Semiconductor materials used for smart synapse construction such as organic, metal oxides, carbon nanotubes, 2D materials, composite heterojunctions, phase-change materials etc. and devices mainly focus on a planar structure with floating gate, electret, ferroelectric and electrochemical mechanisms, both of which have been summarized carefully in related reviews.^41,80–82 However, there are few reviews on the effects of specific structures of semiconductor materials and devices on synaptic properties. This subsection focuses on establishing the effects of novel designs of semiconductor materials (mainly including nanofibers, nanoparticles, single crystal combine with fiber structuring engineering, nanosheets, and dendritic network etc.) and device structures on synaptic properties. The related materials, structures, and properties of synaptic devices are summarized in Table 1.

---

2.1. Novel design of material structure

Tremendous semiconductor materials have been constructed towards artificial synapses. Three operation mechanisms are generally mastered, carrier induction,³⁵ capture,⁸³ exciton generation and then separation,⁸⁴ which has been widely investigated and summarized in recent reviews.^39,85 The relationship between the special structural design of semiconductor materials (like nanofibers, nanoparticles, and single crystals combined with fiber structuring engineering, nanosheets, and dendritic networks, not the morphology of the material) and synaptic properties is the main subject in this section. Back in 2013, Shi et al.⁸⁶ prepared synaptic transistors with multilevel analogue states with control over the composition in ionic liquid-gated devices. In 2020, benefiting from the low conductivity and the anisotropy of selenium nanosheets, Chai et al.⁸⁷ demonstrated anisotropic and low-energy synaptic transistors (less than 0.1 pJ per spike) based on trigonal selenium (t-Se) nanosheets, which offer the potential to process complex signals. Furthermore, to obtain low-power neuromorphic computing and a high-speed response, nanofiber channels with an ion-blocking layer are introduced to transistor synapses (Fig. 2a).⁸⁸ Thanks to the nanofiber architecture, the synapse shows low power consumption (113 fJ per spike, Fig. 2b) and operates up to 13.5 kHz. Notably, the energy consumption of monosynaptic events was further reduced by the combination of rubrene a single-crystalline active layer and nanoribbon structure (Fig. 2c).⁸⁹Fig. 2d shows the minimum energy consumption of an artificial synapse is 0.29 fJ, which is much lower than biological synapses (Fig. 2e), In addition, the devices can be conformed on the head of a weevil without any crevices (Fig. 2f). Further optimized ultra-low energy-consumption (Fig. 2g) through single crystal and fiber structuring engineering offer meaningful guidance toward flexible low power intelligent neural computing. At another level, improved memory cyclic endurance and successful speech recognition for spatiotemporal iterative learning are realized utilizing dendritic network and neuro-fiber transistors, whose dendritic network structure is more similar to the biological neural network structure (Fig. 2h).⁷⁴ The architecture of electrode microfibers, ion-gel dielectric insulators and polythiophene semiconductors not only provides stable multilevel memory traits but also enhanced cyclic endurance by effectively controlling the ion doping and de-doping process under an increased network contact interface (Fig. 2i). These results supplied a strong demonstration that the design of the material structure in sheets or nanofibers can improve the response characteristics of the smart synapses such as stability, multi-state storage, and especially power consumption reduction benefiting from the low conductivity and better contact interface. Therefore, the design concept of channel materials opens blazing trails for energy-efficient neurocomputing/sensing platforms. These findings further indicate the relationships between the structure of semiconductor materials and corresponding synaptic performance, which can be used to design material structures according to the target synaptic properties in the future. However, some challenges, such as the complex preparation processes of special material structures, high costs, etc., still need to be solved.

2.2. Novel device structure and working mechanism

In addition to the effect of the material structure on synaptic properties, the transistor structure also plays a crucial role in synaptic properties and the corresponding synaptic function. This section focuses on the exploration of the relationships between transistors structure and synaptic performance, including vertical synaptic transistors, planar synaptic transistor, and multi-gate synaptic transistors. Synaptic transistors with different working mechanisms can be applied in many scenarios, and also provide valuable reference for the development of a new generation of smart electronic devices.

3. Multifunctional synaptic devices

The progression of optoelectronic synapses based on phase-change transistors,¹⁰⁰ floating-gate transistors,⁴⁶ ferroelectric transistors¹⁰¹ and electrochemical transistors¹⁰² have created new breakthroughs toward sophisticated artificial sensing systems analogous to that of humans. However, greater challenges remain in the construction of multifunctional synaptic device at the single device level. This subsection focuses on the review of multifunctional synaptic transistors with stretchable, degradable, multi-mode and reconfigurable functions at the single device level. These features not only enrich the application scenarios of the devices but also alleviate the problems of electronic waste and the difficulty of integrating various functional devices with high density in the More than Moore era. Stretchable electronics are accelerating the integration of human-machine interface and Artificial Intelligence (AI) due to its excellent flexibility and tensile properties. In the beginning, stretchable materials, individual stretchable devices and stretchable electronic circuits were widely studied. Further to the post-Moore era, the advent of the physical limit and the limitation of the separation of storage and computation in the von Neumann architecture promoted booming-research on multifunctional stretchable synaptic electronics. Thus, the development from stretchable materials and individual devices to integrated circuits will be briefly traced. Stretchable individual synaptic devices, multiple arrays, and the integration and simulation of multifunctional synapses will be highlighted.

3.1. Stretchable devices

Humans and animals possess flexible and stretchable synapses that can adapt to all kinds of deformation. Similar to biological synapses, stretchable smart synaptic devices have broader applications such as skin health monitors, neuro-prosthetics and flexible robots.¹⁰³ Therefore, stretchable synaptic devices consisting of stretchable materials are expected to meet the needs of brain-like chips in various deformable situations. Except for intrinsically stretchable materials such as carbon nanotubes, silver nanowires, nanofiber networks, etc. most semiconductor materials have been implemented with molecular design and structural adjustments to improve stretchability. The advancement of strain tolerance is effective but the sacrifice of mobility occurs at the same time. To maintain the mobility during stretching, phase separation induced elasticity (Fig. 4a) is first introduced by employing poly(2,5-bis(2-octyldodecyl)3,6-di(thiophen-2-yl)diketopyrrolo[3,4-c]pyrrole1,4-dione-alt-thieno[3,2-b]thiophen) (DPPT-TT) as a high-mobility semiconducting polymer and polystyrene-block-poly(ethylene-ranbutylene)-block-polystyrene (SEBS) as the soft elastomer (Fig. 4b), and this work contributed to high-performance stretchable semiconductors for stretchable devices.¹⁰⁴ Two years later, highly stretchable organic transistors with high-mobility are produced (Fig. 4c) befitting from the high capacitance (5.5 μF cm⁻²), outstanding transparency (Fig. 4d) and stretchability of ionic thermoplastic polyurethane (i-TPU).¹⁰⁵ In addition, Zhang et al. demonstrated a fully stretchable organic electrochemical transistor combining transfer patterning technology with a pre-stretched PDMS substrate to overcome the challenges of large-area patterning and integration techniques (Fig. 4e).⁵² Recently, stretchable transistor arrays (347 transistors per square centi-meter) enabled skin electronics circuits were fully demonstrated (Fig. 4f).¹⁰⁶Fig. 4g shows the array adheres to a human palm demonstrating a good fit between the skin and circuit. The integration of stretchable materials, stretchable device arrays, and stretchable skin circuits ultimately demonstrates superior performance and stability, laying the foundation for the next generation of more complex stretchable skin electronics. Through the above discussion, basic stretchable transistors have realized the development from performance to large area integration. However, multifunctional smart synaptic electronic systems have become the focus of research in recent years, expecting to overcome the traditional von Neumann bottleneck, and thus, the next section mainly focuses on the development of multifunctional synaptic devices.

Stretchable transistors used for synaptic electronics not only mimic excitatory and inhibitory functions similar to the nerves, but also assume paramount importance in the field of soft robots and neuromorphic systems. To date, stretchable transistors and synaptic responses have been successfully developed by using a single synaptic device or a combination of sensing systems. However, the diverse synaptic properties, the motion sensing controlled by synaptic devices, and the controllable multilevel storage states still face challenges due to the cognitive deficiency in materials, technologies, and physical mechanisms.^107–110 This subsection provides an overview from simple stretchable synaptic devices to complex multifunctional synapses. A low voltage stretchable synaptic was realized using scalable route inkjet-printed, which display mobilities as high as 30 cm² V⁻¹ s⁻¹ and working voltages as low as 1 V, thanks to the double layer capacitive in poly (vinylidene fluorideco-hexafluoropropylene) (PVDF-HFP) dielectric (Fig. 5a and b). Furthermore, the simple synaptic behaviour of neurons was imitated as shown in Fig. 5c.⁸ In addition, based on the previous results shown in Fig. 4a, stretchable phototransistors are constructed with hybrid PDPP-TT and SEBS in a weight ratio of 5:5, which shows an outstanding photo-sensing performance even at a strain up to 100%.¹¹¹ Based on worm-like blended films, top-contact transistors were fabricated combining sequential thermal lamination with transfer procedures (Fig. 5d). Fig. 5e is the corresponding rapid and increased switching to X-ray pulse under different dose rates. Additionally, the device displayed outstanding optical transparency both in the original state and under strain up to 100% (Fig. 5f). The transistors were able to recover to the original state even experiencing stretching, twisting and pocking deformations (Fig. 5g). These devices were also sensitive to faint UV light, which is difficult to distinguish by human eyes (Fig. 5h). Furthermore, the combination of stretchable synaptic transistors and ANN simulation enables digital identification. The top-gated synaptic devices are fabricated using a stretchable bilayer semiconductor (Fig. 5i). Even at a 80 mV presynaptic pulse, the device also exhibits memory characteristics and thus low energy consumption. More noteworthy, when the transistors are stretched by 50%, the image recognition accuracy is still maintained at over 90% (Fig. 5j).⁹

With the advancement of research, more complex and versatile stretchable synaptic transistors have been successfully constructed as shown in Fig. 6. Lee et al.⁵¹ reported the organic optoelectronic synapse system by recognizing and converting photoelectric signals into Morse code (Fig. 6b), and furthermore, artificial muscles were controlled through postsynaptic electrical responses (Fig. 6a). This work promotes the development of bioinspired soft electronics and neurologically inspired robotics. In 2019, fully stretchable transistors and integrated synaptic systems were constructed with all rubbery based materials as shown in Fig. 6c, which shows no mechanical damage after stretching up to 50% with 1000 cycles.⁷⁶ Furthermore, the deformable synapse and corresponding synaptic transmission process is shown in Fig. 6d. Multifunctional intelligent synaptic function like the filtering behaviour of a synaptic transistor was further demonstrated (Fig. 6e). This work further combines the rubbery synaptic transistor and neurologically integrated system with multifunctions, which enrich neurological functions and pave the way toward soft machines. Multimode and multifunction electronic devices provide the possibility for a human–machine interface. Zhang et al.¹¹² proposed a stretchable multimodal neuromorphic array including situ temperature-responsive, strain-responsive and excellent self-healing functions (Fig. 6f), which is similar as human skin to sense, deliver and memory external environmental changes. In order to improve the mechanical deformation ability, P3HT nanofibers are used as semiconductors for stretchable organic synaptic transistors. After experiencing large stretching deformation, the synaptic transistors also exhibited excellent mechanical stability. The number of channel carriers and synaptic plasticity are further controlled by adjusting the ions in the ion-gel dielectric layer. Based on this, different decay constants and functionality are simulated (Fig. 6g).¹¹³ One year later, the vertical organic field effect transistor is constructed by the same group, which can effectively reduce the channel length, implement a high current density (5.19 mA cm⁻²), and thus obtain a low working voltage (−0.5 V). Based on this, Guo et al.¹¹⁴ coupled a triboelectric nanogenerator (TENG) and synaptic transistor, which ensures the direct interactions between external information and human machine interface (Fig. 6h). The evolution from the design of stretchable materials, stretchable transistors with simple synaptic functions to abundantly multifunctional stretchable synaptic transistors will undoubtedly promote the development of stretchable electrons in human brain computing. Therefore, the total stretchability of the material, the improvement of performance and its combination with the skin will further promote the development of stretchable synaptic transistors in the future.

3.2. Degradable devices

With the tremendous growth of electronic-waste (e-waste), biodegradable electronics have opened up great opportunities for green and sustainable neuromorphic electronics both in academia and industry. E-Waste not only pollutes the environment but also harms human health. Therefore, the development of green degradable materials and artificial synaptic devices is expected to reduce the harmful effects of e-waste on the environment and humans, and promote the development of green brain-like chips. Furthermore, degradable synaptic transistors also open the doors for “green” brain-computer interfaces and flexible artificial intelligence systems.¹¹⁵ In 2020, Huang et al.⁶⁰ first introduced natural chlorophyll into organic semiconductors to build a multifunctional organic transistor, which can achieve both photodetection and synaptic functions by controlling the gate voltage (Fig. 7a). The introduction of natural chlorophyll offers the possibility for degradable synaptic transistors, however, the degradability of the devices was not further explored as it was constructed on rigid substrates. One year later, the group further compounded chlorophyll-a and carbon nanotubes to construct degradable photonic synaptic transistors shown in Fig. 7b.¹¹⁵ Furthermore, low energy consumption (17.5 fJ per pulse) was realized at a low drain voltage of −0.0001 V. Fig. 7c demonstrates the environmentally friendly properties of the transistor array with good degradability, which provides the possibility to next-generation degradable neuromorphic electronics. Liu et al.⁹⁷ constructed biodegradable organic synaptic transistors (Fig. 7d) using natural neutral polysaccharides from dextran (Fig. 7e) as dielectrics for the proton conduction properties. A glutaraldehyde (GA) crosslinking method was introduced to control the hydroxyl contents and proton conduction, and Fig. 7f shows the reaction principle. The synaptic devices adhered to an uneven brain model indicating the ultra-flexibility and promising potential in future wearable electronics (Fig. 7g). The array of devices shown in Fig. 7h disappeared within 15 seconds of dripping, indicating excellent degradability. Degradable and ultra-flexible synaptic transistors were designed using natural apple pectin as the dielectric.¹¹⁶ Due to the proton-induced electric double layer and thus the high capacitance of pectin, the operating voltage of a single synaptic was reduced to −20 mV as shown in Fig. 7i. Furthermore, it can be degraded in water within 100s after it completes its function, which provides a powerful opportunity to create “green” biological neuromorphic systems. More strikingly, both ultra-flexible and degradable self-supporting organic synaptic transistors for neuromorphic applications are achieved based on natural polysaccharide dielectrics.⁵⁹ The resulting device is as thin as 309 nano-meters, so possesses excellent flexibility and degradability as shown in Fig. 7j. Fig. 7k displays outstanding conformability while adhering to a human brain model and can be degraded in water without any toxic by-products. The above research demonstrates the potential of green sustainable electrons towards artificial neural networks, from the degradability of monolayer materials to the acquisition of self-supporting devices and ultra-flexible degradable properties. However, it still faces challenges from partial degradation to complete degradation.

3.3. Multimodal and reconfigurable devices

The ubiquity of information technology leads to the generation of a large amount of data. Especially in the era of data explosion, there is a high demand for multifunctional, multi-mode transistor devices, which have greater storage capacity, information processing capacity, and therefore exhibit more diverse neuronal functionalities. A reconfigurable single-gate artificial synaptic is constructed employing hydrogen-rich silicon nitride film and ambipolar graphene as the gate dielectric and semiconductor, respectively.¹¹⁷ The dynamic reconfigurable synaptic functions are attributed to the ambipolar property of graphene, two kinds of hysteresis induced by the traps and ions in SiNx:H film. Based on this, more complex functions like the modulation of tactile perception have been imitated shown in Fig. 8a. Higher voltages (15 V/−15 V) induce ion movement and modulate operating modes of transistors (excitatory or inhibitory). Smaller gate voltages (5 V/−5 V) induce trapping of holes and electrons, which in turn modulates the channel conductance to achieve different responses to the same gate voltage stimulus. The gate voltage stimulation (5 V/−5 V) represents two states (heat and cold, Fig. 8a-1). When a transistor is in the excitatory mode (Fig. 8a-2), the positive signal (heat stimulus) results in a potentiation behaviour (“Ah, it's not okay!”). The negative signal (cold stimulus) results in a depression behaviour (“Well, it's okay!”). However, if the synapse is set to the inhibitory mode (Fig. 8a-3), the positive signal (heat stimulus) causes depression behaviour. The negative signal (cold stimulus) induces a depression behaviour. Since the charge trapping and polarization show opposite effects on the channel conductance,^118–121 Wang et al.⁷⁷ further utilized electret trapping combination with an electrochemical induction composite dielectric layer to construct an electrochemical-electret coupled organic synapse with reconfigurable facilitation-to-depression switching capabilities under the same presynaptic gating pulses. Fig. 8b is the stacked structure of an organic synapse including tri-layer dielectric, poly(vinylidene fluoride-trifluoroethylene-chlorofluoroethylene) terpolymer [P(VDF-TrFE-CFE)] served as a dielectric to reduce operation voltages, poly(2-vinylnaphthalene) (PVN) served as electret to trap charge, poly[(1-vinylpyrrolidone)-co-(2-ethyldimethylammonioethyl methacrylate ethyl sulfate)] [P(VP-EDMAEMAES)] served as a buffer protective layer and poly(isoindigo-co-bithiophene) (PIID-BT) was used as the channel material. Under a low gate voltage pulse, electrochemical doping dominated the channel conductance, which induced synaptic facilitation, and conversely, when increasing voltage pulse, electret charging is triggered which induced synaptic depression due to the depletion of charges (Fig. 8c).

In addition to controlling channel conductance through charge trapping and induction, bipolar semiconductor materials can operate in different modes to induce reconfigurable synaptic properties. Wei et al.¹²² utilized ambipolar conjugated-polymer poly[2,5-bis(2-decyltetradecyl)pyrrolo[3,4-c]pyrrole-1,4(2H,5H)-dione-alt-5,5′-di(thiophen-2-yl)-2,2′-(E)-1,2-bis(3,4-difluorothien-2-yl)ethene] (PDPP-4FTVT) as a semiconductor constructed reconfigurable intelligent synaptic device (Fig. 8d). The device can respond oppositely to the same gate voltage stimulus signal, which is achieved by changing the operating mode of the ambipolar transistor (Fig. 8e and f, initial and programmed state). When the device is operating in the original state (electron transport), electrons are injected into the insulator layer under positive gate pulses, which result in the decreased PSC (Fig. 8g), but increased PSC are acquired for the programmed transistor (hole transport) under the same positive gate pulses (Fig. 8h). Reconfigurable transistor and multi-mode memory functions are demonstrated by regulating ferroelectric oxygen vacancies distribution dynamics.¹²³Fig. 8i shows the redistribution process of oxygen vacancies in Bi₄Ti₃O₁₂ during the negative gate pulse of −100 V, which results in a typical ferroelectric counter-clockwise hysteresis loop and low SS (54 and 34 mV dec⁻¹ in forward and backward sweeps, Fig. 8j). Fig. 8k shows the corresponding synaptic excitatory (−10 V, 300 ms) and inhibitory properties (+10 V, 300 ms). More strikingly, a hysteresis-free logic transistor is realized possessing a subthreshold swing from 68 down to 46 mV dec⁻¹ (Fig. 8m and n) by modulating the location of oxygen vacancies and pinning the ferroelectric domains through cooling control, as shown in Fig. 8l. These new proof of concepts of functional reconfiguration and muti-operation mode in a single-gate field-effect transistor is of great advantage in future multifunctional synaptic system applications.

4. Integration of artificial nervous systems

Giving the artificial nervous system the ability to sense the environment further enhances our ability to perceive the world. In addition, the use of a neural network system to control robot actions (The Artificial Sensorimotor Systems), the construction of the visual system and tactile system enabled by synaptic devices will further accelerate the realization of artificial intelligence. This section highlights the recent progress in this area and summarizes the future directions and challenges.

4.1. Artificial sensorimotor systems

Different spatiotemporal action potentials produced by an external pulse will be processed and identified through biological neural systems and finally transmitted to the human brain with low power consumption. An imitation human neural network based on biological synapses has become the key technique in bio-inspired electronics, which can effectively promote the development of human-computer interaction and intelligent robots. As shown in Fig. 9a, Kwon et al.⁶⁵ in real time according to the mechanisms of the biological semi-circular canals by combining synaptic transistors with triboelectric rotation sensors. Rotational movement will trigger TENG rotation and generate voltage pulses to artificial synaptic transistors, which will be detected as different postsynaptic currents (Fig. 9b). This artificial nervous system is enhancing the prospects in neurorobotics and soft electronics across time and space. In living systems, the production of sensation and the domination of movement are dynamically linked and mutually reinforced. The real time communication between analog world information and robotic platforms could be realized through neuronal computation. Recently, processing and learning the target task with neuromorphic circuit is realized. The robot can navigate itself in a two-dimensional maze following a planned path after being trained with sensorimotor systems and organic neuromorphic circuits. However, the untrained robot is disoriented and cannot get out of the maze (Fig. 9c).⁶⁹ This demonstration paves the way for a sophisticated systems connection like rapid prototyping of intelligent and real-world systems. Moreover, Chen et al.¹²⁴ further combined a visual with a vestibule system to construct the artificial motion sensory system, in which TENG as a vestibule for sensing rotation and a synaptic transistor array as the retina are used. The human vision-vestibule integrated nervous system and artificial motion multisensory system are demonstrated in the left and right sight of Fig. 9d, respectively. The TENG and 1-TDA-CsPbBr₃ QDs in indacenodithiophene–benzothiadiazole (IDTBT) are used to simulate the rotation signal of the vestibule and receive the external optical signal, respectively. All these signals are eventually converted into electronic visual signals and electronic motion signals through synaptic transistors. These works provide a direction for development in multisensory integration, information processing, robot motion monitoring and artificial prosthetics.

4.2. The artificial visual system

The retina, as a light-sensitive layer, mainly transforms and memorizes light information and then adjusts brain activity. To date, many studies on optoelectronic synapses have been widely studied for their low crosstalk and high interference immunity, such as the imitating of colored and color-mixed pattern recognition of the vision system with a h-BN/WSe₂ heterostructure optic-neural synaptic device (Fig. 10a). This research demonstrates the stable conduction states (less than 1% variation), which accelerated the integration pace between neural networks and complex pattern recognition.⁶⁷ In addition, Zhang's group developed bionic optoelectronics based on OFETs with ultralow energy consumption of 0.07–34 fJ per spike by adjusting the source Schottky barrier and charge-carrier injection. Furthermore, flexible 8 × 8 OFET arrays for the artificial optic-neural network indicating excellent image recognition and contrast reinforcement abilities. The image recognition function was examined not only on a flat but also on a human eyeball mode with a bending radius of 11 mm (Fig. 10b). As the number of stimuli increases, the photocurrent weight also increases, and the image contrast enhancement function is realized, which demonstrates the potential for future applications such as to help blind people to regain their optesthesia.¹²⁵ In another example, Shen et al.¹²⁶ developed flexible artificial optoelectronic synapses (10 × 10 array, Fig. 10c) based on lead-free metal halide nanocrystals for neuromorphic computing and color recognition. The EPSC image mapping of the 10 × 10 array under 2, 20, and 40 s light duration proves enhanced visual perception function with a longer duration time (Fig. 10d). Moreover, the development of infrared artificial synaptic devices is conducive to night monitoring. Liu et al.⁷⁹ designed organic/inorganic hybrid heterojunction synaptic phototransistors for short-wave infrared night visual systems because the heterojunction structure can promote photogenerated exciton separation (Fig. 10e). As shown in Fig. 10f, the measured EPSC values and corresponding shape of “SIR” through a mask clearly appeared after 1100 nm SIR light even under a dark night, 200 and 500 cd m⁻² white back-ground light night. This provides the direction for the artificial night vision. Furthermore, Liu et al.¹²⁷ constructed hemispherical biomimetic eyes using ultra-flexible and ultrasensitive near-infrared organic phototransistors, which are more suitable for integration with bioelectronic skin. The imaging system of the ultra-flexible hemispherical bio-mimetic eyes is shown in Fig. 10g, compared with planar image arrays, the letter “H” pattern shows higher contrast compared with flat plan for a more consistent distance. Therefore, it can be used as higher-resolution image systems. These works especially the ultra-flexible NIR imaging system will expand the further application of NIR visualization in biomedicine, visual prosthetics and intelligent bioimaging systems.

4.3. The artificial tactile system

Effective tactile information processing and recognition by a synapse transistor system allows for the integration of environmental signals and human skin systems. The recognition and transfer of strain signals mainly originated from the following working mechanisms including resistive, piezoelectric, and triboelectric-capacitive coupling response, which will be discussed separately in chronological order. The development of milestone artificial tactile systems in the most recent five years are introduced in Fig. 11. In 2018, Bao et al.¹²⁸ constructed an artificial tactile system by connecting resistive pressure sensors, organic ring oscillators (convert resistance signal into voltage), and a synaptic transistor (Fig. 11a). The as-prepared system could detect the movement of an object, combine simultaneous pressure inputs, and distinguish braille characters. Furthermore, a monosynaptic reflex arc can be executed from postsynaptic currents induced by voltage stimulation converted from a ring oscillator. In the same year, Chen et al.¹²⁹ converts pressure stimuli into electric signals through the pressure sensor, which is coupled into a synaptic transistor to record information (Fig. 11b). In addition to transforming the pressure signal by means of a resistance sensor, the nanogenerator can convert the pressure signal into a discontinuous electrical signal. As shown in Fig. 11c, a stretchable neurologically integrated robot was created combining TENGs and synaptic transistors. The action potentials from TENGs are transmitted into artificial synapse counterparts to execute memory functions.⁷⁶ In addition, a self-powered synapse transistor enabled by touching TENG to imitate synaptic functions without any additional voltage to generate pre-synapse spikes are realized. Moreover, the “AND” and “OR” logic functions were achieved and mimicked, which shows the great potentials in the applications of e-skin and human-computer interaction (Fig. 11d).¹³⁰ Flexible neurological electronic skin is developed combining a ferro-electret FENG and artificial synapse transistor. The tactile signal of force is converted by the FENG sensor, and then applied to the synaptic transistor gate for different postsynaptic currents and next stage transform (Fig. 11e).¹³¹ Flexible strain organic synaptic systems controlled by capacitive coupling under touch are reported (Fig. 11f). By controlling the barium titanate nanocomposite composition in a ferroelectric poly(vinylidene fluoride-trifluoroethylene) dielectric to obtain memory functions.¹³²Fig. 11g and h are both strain synaptic systems based on resistive sensors, except for the same signal transition principle, both works use induced coupling of ions in the dielectric layer to achieve synaptic function, which provides promising potential in artificial intelligence.^68,133 An artificial somatosensory system was developed utilizing multiple tactile sensor arrays, synaptic transistors and artificial muscles. Multiple sensors and multi-gate synaptic transistors are good for spatio-temporal information-processing. In addition, an artificial muscle could be moved by an instant feedback function (Fig. 11i).¹³⁴ Xu et al. further integrated the stress sensing signals of different gestures with intelligent synaptic transistors (Fig. 11j), and trained the artificial intelligence networks with different drain currents corresponding to different gestures, finally recognizing gestures with high recognition accuracy.⁷⁸ This work is an important step toward wearable intelligent human–machine interfaces.

5. Conclusions and perspectives

In conclusion, the relationships between the specific structural design of semiconductor materials, devices and synaptic properties are reviewed. Furthermore, synaptic applications based on transistors in stretchable, biodegradable, multi-modal reconfigurable, robotic information processing, vision system and strain perception are systematically summarized. Based on this, it is found that the emerging synaptic devices and integrated systems would be a new breakthrough in brain-like chips and even the realization of robotic connection. More strictly, giving more complex functions to individual synaptic devices such as stretchability, degradability, reconfigurability, and developing high-density synaptic systems to further mimic human activity will further advance the connection between intelligent and biological synapses. For example, the introduction of fully degradable synaptic devices to simulate reflexes in animals, the combination of electret and electrochemical materials to achieve multimodal, reconfigurable devices based on single synaptic transistors, organic polymer-based stretchable synaptic electronic systems, and photosensitive vision systems and self-powered strain interconnection systems will be an important way to accelerate the development of multifunctional synaptic electrons. In addition, printing technology is not only convenient and low cost but it is also easier to achieve a high throughput output. Therefore, novel materials and device structures, multifunctional synaptic devices, diversity of synaptic properties and synaptic sensing interconnection systems combined with high-throughput printing processes are expected to achieve multi-mode, intelligent synaptic interconnection systems and brain-like chips (summarized in Fig. 12).

While the next wave of transistor-based synaptic electronic systems is on the horizon, challenges and opportunities still exist and need to be addressed.

5.1. The challenges

(1) There are relatively few studies related to multi-mode and reconfigurable synapse properties based on individual transistors, whose multi-stable current states need to be maintained by controlling the carrier in the channel.

(2) The selection of material systems and the integration of low temperature printing processes are urgently needed for e-skin synapse applications.

(3) Complex synaptic system integration and artificial neural network simulation procedures need to be deeply studied and connected.

(4) The multi-signal simulation principles of neural network interconnected systems need to be further explored.

Based on the proposed challenges, there are also many opportunities to further improve the development of intelligent synapses.

5.2. The opportunities

(1) The preparation of multimodal reconfigurable and low power consumption synaptic transistors can be effectively facilitated by controlling the carrier capture and induction processes.

(2) Preparation of fully printed stretchable synaptic transistors using material blending systems such as organic polymers and carbon nanotubes to increase viscosity suitable for printing processes.

(3) Development of high density integrated synaptic transistor systems driven by multiple sensing coupling interconnects.

(4) Implantable fully degradable synaptic transistor interconnect system possessing promising applications.

More generally, future research should consider the potential effects of muti-mode, reconfigurable, flexible, printable, fully degradable, stretchable and system-integrated capabilities more carefully. Under this context, it believes that multidisciplinary cross-collaborative research will enable the successful fabrication and application of numerous synaptic electronic systems in the future.

Author contributions

Xin Wang: investigation, artical analysis, writing original draft. Yixin Ran: investigation, writing part of the original draft. Xiaoqian Li: revision of the manuscript. Xinsu Qin: investigation. Wanlong Lu: correction. Yuanwei Zhu: editing. Guanghao Lu: editing, funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (Grant No. 51873172, 51907148 and 52273026), and the Key Scientific and Technological Innovation Team Project of Shaanxi Province (2021GXLH-Z-055). G. L. thanks the support of the Cyrus Tang Foundation, and the Key Scientific and Technological Innovation Team Project of Shanxi Province (2020TD-002).

References

1. G. Indiveri and S.-C. Liu, Proc. IEEE, 2015, 103, 1379–1397 CAS .
2. P. Feng, W. Xu, Y. Yang, X. Wan, Y. Shi, Q. Wan, J. Zhao and Z. Cui, Adv. Funct. Mater., 2017, 27, 1604447 CrossRef .
3. R. W. Picard, E. Vyzas and J. Healey, IEEE Trans. Pattern Anal. Mach. Intell., 2001, 23, 1175–1191 CrossRef .
4. S. Furber, J. Neural Eng., 2016, 13, 051001 CrossRef PubMed .
5. R. A. Sporea, K. M. Niang, A. J. Flewitt and S. R. P. Silva, Adv. Mater., 2019, 31, 1902551 CrossRef PubMed .
6. B. M. Yu, Nature, 2016, 532, 449–450 CrossRef CAS PubMed .
7. Q. Guo, J. Koo, Z. Xie, R. Avila, X. Yu, X. Ning, H. Zhang, X. Liang, S. B. Kim, Y. Yan, M. R. MacEwan, H. M. Lee, A. Song, Z. Di, Y. Huang, Y. Mei and J. A. Rogers, Adv. Funct. Mater., 2019, 29, 1905451 CrossRef CAS .
8. F. Molina-Lopez, T. Z. Gao, U. Kraft, C. Zhu, T. Öhlund, R. Pfattner, V. R. Feig, Y. Kim, S. Wang, Y. Yun and Z. Bao, Nat. Commun., 2019, 10, 2676 CrossRef CAS PubMed .
9. H. Shim, F. Ershad, S. Patel, Y. Zhang, B. Wang, Z. Chen, T. J. Marks, A. Facchetti and C. Yu, Nat. Electron., 2022, 5, 670–671 Search PubMed .
10. C. Qian, Y. Choi, Y. J. Choi, S. Kim, Y. Y. Choi, D. G. Roe, M. S. Kang, J. Sun and J. H. Cho, Adv. Mater., 2020, 32, 2002653 CrossRef CAS PubMed .
11. L. Liu, Y. Ni, J. Liu, Y. Wang, C. Jiang and W. Xu, Adv. Funct. Mater., 2023, 33, 2210119 CrossRef CAS .
12. J. Gong, H. Wei, J. Liu, L. Sun, Z. Xu, H. Huang and W. Xu, Matter, 2022, 5, 1578–1589 CrossRef .
13. Nat. Electron., 2020, 3, 347 DOI:10.1038/s41928-020-0457-1.
14. Y. Hu, M. Dai, W. Feng, X. Zhang, F. Gao, S. Zhang, B. Tan, J. Zhang, Y. Shuai, Y. Fu and P. Hu, Adv. Mater., 2021, 33, 2104960 CrossRef CAS PubMed .
15. J. Wang, J. Wang, J. Zhang, W. Huang, X. Wang and M. Zhang, Adv. Electron. Mater., 2022, 8, 2100922 CrossRef CAS .
16. R. Martins, A. Nathan, R. Barros, L. Pereira, P. Barquinha, N. Correia, R. Costa, A. Ahnood, I. Ferreira and E. Fortunato, Adv. Mater., 2011, 23, 4491–4496 CrossRef CAS PubMed .
17. J. Wang, X. Guo, Z. Yu, Z. Ma, Y. Liu, Z. Lin, M. Chan, Y. Zhu, X. Wang and Y. Chai, Adv. Funct. Mater., 2020, 30, 2003859 CrossRef CAS .
18. X. Wang, M. Zhu, X. Li, Z. Qin, G. Lu, J. Zhao and Z. Zhang, Adv. Mater., 2022, 34, 2204066 CrossRef CAS PubMed .
19. M. M. Waldrop, Nat. News, 2016, 530, 144 CrossRef CAS PubMed .
20. V. Balasubramanian, Proc. Natl. Acad. Sci. U. S. A., 2021, 118, e2107022118 CrossRef CAS PubMed .
21. X. Wang, W. Lu, P. Wei, Z. Qin, N. Qiao, X. Qin, M. Zhang, Y. Zhu, L. Bu and G. Lu, ACS Appl. Mater. Interfaces, 2022, 14, 48948–48959 CrossRef CAS PubMed .
22. Y. Shen, N. C. Harris, S. Skirlo, M. Prabhu, T. Baehr-Jones, M. Hochberg, X. Sun, S. Zhao, H. Larochelle, D. Englund and M. Soljačić, Nat. Photonics, 2017, 11, 441–446 CrossRef CAS .
23. I. Cong, S. Choi and M. D. Lukin, Nat. Phys., 2019, 15, 1273–1278 Search PubMed .
24. T. Zhao, Y. Han, L. Qin, H. Guan, L. Xing, X. Li, X. Xue, G. Li and Y. Zhan, Nano Energy, 2021, 85, 106006 Search PubMed .
25. X. Xiao, Y. Fang, X. Xiao, J. Xu and J. Chen, ACS Nano, 2021, 15, 18633–18646 CrossRef CAS PubMed .
26. J.-Q. Yang, R. Wang, Y. Ren, J.-Y. Mao, Z.-P. Wang, Y. Zhou and S.-T. Han, Adv. Mater., 2020, 32, 2003610 CrossRef PubMed .
27. Z. Cheng, C. Rios, W. H. P. Pernice, C. D. Wright and H. Bhaskaran, Sci. Adv., 2017, 3, e1700160 CrossRef PubMed .
28. X. Ji, B. D. Paulsen, G. K. K. Chik, R. Wu, Y. Yin, P. K. L. Chan and J. Rivnay, Nat. Commun., 2021, 12, 2480 CrossRef CAS PubMed .
29. K. Liang, R. Wang, B. Huo, H. Ren, D. Li, Y. Wang, Y. Tang, Y. Chen, C. Song, F. Li, B. Ji, H. Wang and B. Zhu, ACS Nano, 2022, 16, 8651–8661 CrossRef CAS PubMed .
30. S. Kim, B. Choi, M. Lim, J. Yoon, J. Lee, H.-D. Kim and S.-J. Choi, ACS Nano, 2017, 11, 2814–2822 CrossRef CAS PubMed .
31. M. Hu, J. Yu, Y. Chen, S. Wang, B. Dong, H. Wang, Y. He, Y. Ma, F. Zhuge and T. Zhai, Mater. Horiz., 2022, 9, 2335–2344 RSC .
32. K. Wang, S. Dai, Y. Zhao, Y. Wang, C. Liu and J. Huang, Small, 2019, 15, 1900010 CrossRef PubMed .
33. J. Choi, J. S. Han, K. Hong, S. Y. Kim and H. W. Jang, Adv. Mater., 2018, 30, 1704002 CrossRef PubMed .
34. C. Giotis, A. Serb, V. Manouras, S. Stathopoulos and T. Prodromakis, Sci. Adv., 2022, 8, eabn7920 CrossRef CAS PubMed .
35. H. Wang, Q. Zhao, Z. Ni, Q. Li, H. Liu, Y. Yang, L. Wang, Y. Ran, Y. Guo, W. Hu and Y. Liu, Adv. Mater., 2018, 30, 1803961 CrossRef PubMed .
36. B. Yang, Y. Wang, Z. Hua, J. Zhang, L. Li, D. Hao, P. Guo, L. Xiong and J. Huang, Chem. Commun., 2021, 57, 8300–8303 RSC .
37. D. Yang, H. Yang, X. Guo, H. Zhang, C. Jiao, W. Xiao, P. Guo, Q. Wang and D. He, Adv. Funct. Mater., 2020, 30, 2004514 CrossRef CAS .
38. N. He, F. Ye, J. Liu, T. Sun, X. Wang, W. Hou, W. Shao, X. Wan, Y. Tong, F. Xu and Y. Sheng, Adv. Electron. Mater., 2023, 9, 2201038 CrossRef CAS .
39. H. Yu, H. Wei, J. Gong, H. Han, M. Ma, Y. Wang and W. Xu, Small, 2021, 17, 2000041 CrossRef CAS PubMed .
40. X. Li, Y. Liu, J. Zhang, F. Wu, M. Hu and H. Yang, Adv. Intelligent Syst., 2022, 4, 2200015 CrossRef .
41. S. Dai, Y. Zhao, Y. Wang, J. Zhang, L. Fang, S. Jin, Y. Shao and J. Huang, Adv. Funct. Mater., 2019, 29, 1903700 CrossRef CAS .
42. C. Eckel, J. Lenz, A. Melianas, A. Salleo and R. T. Weitz, Nano Lett., 2022, 22, 973–978 CrossRef CAS PubMed .
43. T. Chen, X. Wang, D. Hao, S. Dai, Q. Ou, J. Zhang and J. Huang, Adv. Opt. Mater., 2021, 9, 2002030 CrossRef CAS .
44. G. Liu, Q. Li, W. Shi, Y. Liu, K. Liu, X. Yang, M. Shao, A. Guo, X. Huang, F. Zhang, Z. Zhao, Y. Guo and Y. Liu, Adv. Funct. Mater., 2022, 32, 2200959 CrossRef CAS .
45. Y.-S. Zhang, B.-J. Chen, X. Deng, Z. Guan, B.-B. Chen, Y. Chen, N. Zhong, P.-H. Xiang and C.-G. Duan, J. Mater. Chem. C, 2022, 10, 11654–11663 RSC .
46. L. Danial, E. Pikhay, E. Herbelin, N. Wainstein, V. Gupta, N. Wald, Y. Roizin, R. Daniel and S. Kvatinsky, Nat. Electron., 2019, 2, 596–605 CrossRef CAS .
47. Y. Zhu, Y. Fan, S. Li, P. Wei, D. Li, B. Liu, D. Cui, Z. Zhang, G. Li, Y. Nie and G. Lu, Mater. Horiz., 2020, 7, 1861–1871 RSC .
48. Z.-D. Luo, X. Xia, M.-M. Yang, N. R. Wilson, A. Gruverman and M. Alexe, ACS Nano, 2020, 14, 746–754 CrossRef CAS PubMed .
49. Y. Yang, R. C. da Costa, M. J. Fuchter and A. J. Campbell, Nat. Photonics, 2013, 7, 634–638 CrossRef CAS .
50. M. E. Beck, A. Shylendra, V. K. Sangwan, S. Guo, W. A. Gaviria Rojas, H. Yoo, H. Bergeron, K. Su, A. R. Trivedi and M. C. Hersam, Nat. Commun., 2020, 11, 1565 CrossRef CAS PubMed .
51. Y. Lee, J. Y. Oh, W. Xu, O. Kim, T. R. Kim, J. Kang, Y. Kim, D. Son, J. B.-H. Tok, M. J. Park, Z. Bao and T.-W. Lee, Sci. Adv., 2018, 4, eaat7387 CrossRef CAS PubMed .
52. S. Zhang, E. Hubis, G. Tomasello, G. Soliveri, P. Kumar and F. Cicoira, Chem. Mater., 2017, 29, 3126–3132 CrossRef CAS .
53. Y. Dai, H. Hu, M. Wang, J. Xu and S. Wang, Nat. Electron., 2021, 4, 17–29 CrossRef CAS .
54. Z. Bao and X. Chen, Adv. Mater., 2016, 28, 4177–4179 CrossRef CAS PubMed .
55. S. Zhang and W. Xu, J. Mater. Chem. C, 2020, 8, 11138–11144 RSC .
56. L. Li, K. Wang, H. Fan, X. Zhu, J. Mu, H. Yu, Q. Zhang, Y. Li, C. Hou and H. Wang, Mater. Horiz., 2021, 8, 1711–1721 RSC .
57. Y. Jiang and B. Tian, Nat. Rev. Mater., 2018, 3, 473–490 CrossRef PubMed .
58. M. Cai, X. Tong, H. Zhao, X. Li, Y. You, R. Wang, L. Xia, N. Zhou, L. Wang and Z. M. Wang, Small, 2022, 18, 2204495 CrossRef CAS PubMed .
59. Y. Yang, X. Zhao, C. Zhang, Y. Tong, J. Hu, H. Zhang, M. Yang, X. Ye, S. Wang, Z. Sun, Q. Tang and Y. Liu, Adv. Funct. Mater., 2020, 30, 2006271 CrossRef CAS .
60. B. Yang, Y. Lu, D. Jiang, Z. Li, Y. Zeng, S. Zhang, Y. Ye, Z. Liu, Q. Ou, Y. Wang, S. Dai, Y. Yi and J. Huang, Adv. Mater., 2020, 32, 2001227 CrossRef CAS PubMed .
61. S. Dai, X. Wu, D. Liu, Y. Chu, K. Wang, B. Yang and J. Huang, ACS Appl. Mater. Interfaces, 2018, 10, 21472–21480 CrossRef CAS PubMed .
62. J. Li, W. Fu, Y. Lei, L. Li, W. Zhu and J. Zhang, ACS Appl. Mater. Interfaces, 2022, 14, 8587–8597 CrossRef CAS PubMed .
63. H. Lian, Q. Liao, B. Yang, Y. Zhai, S.-T. Han and Y. Zhou, J. Mater. Chem. C, 2021, 9, 640–648 RSC .
64. X. Deng, S.-Q. Wang, Y.-X. Liu, N. Zhong, Y.-H. He, H. Peng, P.-H. Xiang and C.-G. Duan, Adv. Funct. Mater., 2021, 31, 2101099 CrossRef CAS .
65. D. Choi, M.-K. Song, T. Sung, S. Jang and J.-Y. Kwon, Nano Energy, 2020, 74, 104912 CrossRef CAS .
66. H. Wei, Y. Ni, L. Sun, H. Yu, J. Gong, Y. Du, M. Ma, H. Han and W. Xu, Nano Energy, 2021, 81, 105648 CrossRef CAS .
67. S. Seo, S.-H. Jo, S. Kim, J. Shim, S. Oh, J.-H. Kim, K. Heo, J.-W. Choi, C. Choi, S. Oh, D. Kuzum, H.-S. P. Wong and J.-H. Park, Nat. Commun., 2018, 9, 5106 CrossRef PubMed .
68. C. Jiang, J. Liu, L. Yang, J. Gong, H. Wei and W. Xu, Adv. Sci., 2022, 9, 2106124 CrossRef CAS PubMed .
69. I. Krauhausen, D. A. Koutsouras, A. Melianas, S. T. Keene, K. Lieberth, H. Ledanseur, R. Sheelamanthula, A. Giovannitti, F. Torricelli, I. Mcculloch, P. W. M. Blom, A. Salleo, Y. van de Burgt and P. Gkoupidenis, Sci. Adv., 2021, 7, eabl5068 CrossRef CAS PubMed .
70. X. Zhao, S. Wang, Y. Ni, Y. Tong, Q. Tang and Y. Liu, Adv. Sci., 2021, 8, 2004050 CrossRef CAS PubMed .
71. J. Kim, Y. Kim, O. Kwon, T. Kim, S. Oh, S. Jin, W. Park, J.-D. Kwon, S.-W. Hong, C.-S. Lee, H.-Y. Ryu, S. Hong, J. Kim, T.-Y. Heo and B. Cho, Adv. Electron. Mater., 2020, 6, 1901072 CrossRef CAS .
72. C. Wang, W.-Y. Lee, D. Kong, R. Pfattner, G. Schweicher, R. Nakajima, C. Lu, J. Mei, T. H. Lee, H.-C. Wu, J. Lopez, Y. Diao, X. Gu, S. Himmelberger, W. Niu, J. R. Matthews, M. He, A. Salleo, Y. Nishi and Z. Bao, Sci. Rep., 2015, 5, 17849 CrossRef CAS PubMed .
73. W. Huang, P. Hang, Y. Wang, K. Wang, S. Han, Z. Chen, W. Peng, Y. Zhu, M. Xu, Y. Zhang, Y. Fang, X. Yu, D. Yang and X. Pi, Nano Energy, 2020, 73, 104790 CrossRef CAS .
74. S. J. Kim, J.-S. Jeong, H. W. Jang, H. Yi, H. Yang, H. Ju and J. A. Lim, Adv. Mater., 2021, 33, 2100475 CrossRef CAS PubMed .
75. Y. Choi, S. Oh, C. Qian, J.-H. Park and J. H. Cho, Nat. Commun., 2020, 11, 4595 CrossRef CAS PubMed .
76. H. Shim, K. Sim, F. Ershad, P. Yang, A. Thukral, Z. Rao, H.-J. Kim, Y. Liu, X. Wang, G. Gu, L. Gao, X. Wang, Y. Chai and C. Yu, Sci. Adv., 2019, 5, eaax4961 CrossRef CAS PubMed .
77. H. Wang, Y. Chen, Z. Ni and P. Samorì, Adv. Mater., 2022, 34, 2205945 CrossRef CAS PubMed .
78. L. Liu, W. Xu, Y. Ni, Z. Xu, B. Cui, J. Liu, H. Wei and W. Xu, ACS Nano, 2022, 16, 2282–2291 CrossRef CAS PubMed .
79. X. Huang, Y. Liu, G. Liu, K. Liu, X. Wei, M. Zhu, W. Wen, Z. Zhao, Y. Guo and Y. Liu, Adv. Funct. Mater., 2023, 33, 2208836 CrossRef CAS .
80. J. Tang, C. He, J. Tang, K. Yue, Q. Zhang, Y. Liu, Q. Wang, S. Wang, N. Li, C. Shen, Y. Zhao, J. Liu, J. Yuan, Z. Wei, J. Li, K. Watanabe, T. Taniguchi, D. Shang, S. Wang, W. Yang, R. Yang, D. Shi and G. Zhang, Adv. Funct. Mater., 2021, 31, 2011083 CrossRef CAS .
81. M. Xu, X. Mai, J. Lin, W. Zhang, Y. Li, Y. He, H. Tong, X. Hou, P. Zhou and X. Miao, Adv. Funct. Mater., 2020, 30, 2003419 CrossRef CAS .
82. Q. Zhang, T. Jin, X. Ye, D. Geng, W. Chen and W. Hu, Adv. Funct. Mater., 2021, 31, 2106151 CrossRef CAS .
83. S. G. J. Mathijssen, M.-J. Spijkman, A.-M. Andringa, P. A. van Hal, I. McCulloch, M. Kemerink, R. A. J. Janssen and D. M. de Leeuw, Adv. Mater., 2010, 22, 5105–5109 CrossRef CAS PubMed .
84. J. Zhang, T. Sun, S. Zeng, D. Hao, B. Yang, S. Dai, D. Liu, L. Xiong, C. Zhao and J. Huang, Nano Energy, 2022, 95, 106987 CrossRef CAS .
85. J. Yu, Y. Wang, S. Qin, G. Gao, C. Xu, Z. Lin Wang and Q. Sun, Mater. Today, 2022, 60, 158–182 CrossRef CAS .
86. J. Shi, S. D. Ha, Y. Zhou, F. Schoofs and S. Ramanathan, Nat. Commun., 2013, 4, 2676 CrossRef PubMed .
87. J.-K. Qin, F. Zhou, J. Wang, J. Chen, C. Wang, X. Guo, S. Zhao, Y. Pei, L. Zhen, P. D. Ye, S. P. Lau, Y. Zhu, C.-Y. Xu and Y. Chai, ACS Nano, 2020, 14, 10018–10026 CrossRef CAS PubMed .
88. S.-K. Lee, Y. W. Cho, J.-S. Lee, Y.-R. Jung, S.-H. Oh, J.-Y. Sun, S. Kim and Y.-C. Joo, Adv. Sci., 2021, 8, 2001544 CrossRef CAS PubMed .
89. C. Zhang, S. Wang, X. Zhao, Y. Yang, Y. Tong, M. Zhang, Q. Tang and Y. Liu, Adv. Funct. Mater., 2021, 31, 2007894 CrossRef CAS .
90. S.-H. Kim and W.-J. Cho, Polymers, 2022, 14, 1372 CrossRef CAS PubMed .
91. H.-S. Kim, H. Park and W.-J. Cho, Nanomaterials, 2022, 12, 2596 CrossRef CAS PubMed .
92. R. Wang, P. Chen, D. Hao, J. Zhang, Q. Shi, D. Liu, L. Li, L. Xiong, J. Zhou and J. Huang, ACS Appl. Mater. Interfaces, 2021, 13, 43144–43154 CrossRef CAS PubMed .
93. H. R. Lee, D. Lee and J. H. Oh, Adv. Mater., 2021, 33, 2100119 CrossRef CAS PubMed .
94. M. Li, Q. Shu, X. Qing, J. Wu, Q. Xiao, K. Jia, X. Wang and D. Wang, J. Mater. Chem. C, 2023, 11, 5208–5216 RSC .
95. D. Kireev, S. Liu, H. Jin, T. Patrick Xiao, C. H. Bennett, D. Akinwande and J. A. C. Incorvia, Nat. Commun., 2022, 13, 4386 CrossRef CAS PubMed .
96. R. Liu, L. Q. Zhu, W. Wang, X. Hui, Z. P. Liu and Q. Wan, J. Mater. Chem. C, 2016, 4, 7744–7750 RSC .
97. Y. Yang, X. Zhao, S. Wang, C. Zhang, H. Sun, F. Xu, Y. Tong, Q. Tang and Y. Liu, J. Mater. Chem. C, 2020, 8, 16542–16550 RSC .
98. Y. Ke, R. Yu, S. Lan, L. He, Y. Yan, H. Yang, L. Shan, H. Chen and T. Guo, J. Mater. Chem. C, 2021, 9, 4854–4861 RSC .
99. S. Wang, C. Chen, Z. Yu, Y. He, X. Chen, Q. Wan, Y. Shi, D. W. Zhang, H. Zhou, X. Wang and P. Zhou, Adv. Mater., 2019, 31, 1806227 CrossRef PubMed .
100. S. Iqbal, L. T. Duy, H. Kang, R. Singh, M. Kumar, J. Park and H. Seo, Adv. Funct. Mater., 2021, 31, 2102567 CrossRef CAS .
101. M.-K. Kim and J.-S. Lee, Adv. Mater., 2020, 32, 1907826 CrossRef CAS PubMed .
102. P. C. Harikesh, C.-Y. Yang, D. Tu, J. Y. Gerasimov, A. M. Dar, A. Armada-Moreira, M. Massetti, R. Kroon, D. Bliman, R. Olsson, E. Stavrinidou, M. Berggren and S. Fabiano, Nat. Commun., 2022, 13, 901 CrossRef CAS PubMed .
103. G. Cao, P. Meng, J. Chen, H. Liu, R. Bian, C. Zhu, F. Liu and Z. Liu, Adv. Funct. Mater., 2021, 31, 2005443 CrossRef CAS .
104. J. Xu, S. Wang, G.-J. N. Wang, C. Zhu, S. Luo, L. Jin, X. Gu, S. Chen, V. R. Feig, J. W. F. To, S. Rondeau-Gagné, J. Park, B. C. Schroeder, C. Lu, J. Y. Oh, Y. Wang, Y.-H. Kim, H. Yan, R. Sinclair, D. Zhou, G. Xue, B. Murmann, C. Linder, W. Cai, J. B.-H. Tok, J. W. Chung and Z. Bao, Science, 2017, 355, 59–64 CrossRef CAS PubMed .
105. D. H. Park, H. W. Park, J. W. Chung, K. Nam, S. Choi, Y. S. Chung, H. Hwang, B. Kim and D. H. Kim, Adv. Funct. Mater., 2019, 29, 1808909 CrossRef .
106. S. Wang, J. Xu, W. Wang, G.-J. N. Wang, R. Rastak, F. Molina-Lopez, J. W. Chung, S. Niu, V. R. Feig, J. Lopez, T. Lei, S.-K. Kwon, Y. Kim, A. M. Foudeh, A. Ehrlich, A. Gasperini, Y. Yun, B. Murmann, J. B.-H. Tok and Z. Bao, Nature, 2018, 555, 83–88 CrossRef CAS PubMed .
107. Q. Yang, Z.-D. Luo, D. Zhang, M. Zhang, X. Gan, J. Seidel, Y. Liu, Y. Hao and G. Han, Adv. Funct. Mater., 2022, 32, 202207290 CrossRef CAS .
108. S. M. Kwon, S. W. Cho, M. Kim, J. S. Heo, Y.-H. Kim and S. K. Park, Adv. Mater., 2019, 31, 1906433 CrossRef CAS PubMed .
109. K. Lu, X. Li, Q. Sun, X. Pang, J. Chen, T. Minari, X. Liu and Y. Song, Mater. Horiz., 2021, 8, 447–470 RSC .
110. H. Ling, D. A. Koutsouras, S. Kazemzadeh, Y. van de Burgt, F. Yan and P. Gkoupidenis, Appl. Phys. Rev., 2020, 7, 011307 CAS .
111. K. Liu, Y. Bian, J. Kuang, X. Huang, Y. Li, W. Shi, Z. Zhu, G. Liu, M. Qin, Z. Zhao, X. Li, Y. Guo and Y. Liu, Adv. Mater., 2022, 34, 2107304 CrossRef CAS PubMed .
112. Y. Wang, D. Liu, Y. Zhang, L. Fan, Q. Ren, S. Ma and M. Zhang, ACS Nano, 2022, 16, 8283–8293 CrossRef CAS PubMed .
113. X. Wang, Y. Yan, E. Li, Y. Liu, D. Lai, Z. Lin, Y. Liu, H. Chen and T. Guo, Nano Energy, 2020, 75, 104952 CrossRef CAS .
114. X. Wang, E. Li, Y. Liu, S. Lan, H. Yang, Y. Yan, L. Shan, Z. Lin, H. Chen and T. Guo, Nano Energy, 2021, 90, 106497 CrossRef CAS .
115. Q. Ou, B. Yang, J. Zhang, D. Liu, T. Chen, X. Wang, D. Hao, Y. Lu and J. Huang, Small, 2021, 17, 2007241 CrossRef CAS PubMed .
116. Y. Li, C. Zhang, X. Zhao, Y. Tong, Q. Tang and Y. Liu, ACS Appl. Electron. Mater., 2022, 4, 316–325 CrossRef CAS .
117. Y. Yao, X. Huang, S. Peng, D. Zhang, J. Shi, G. Yu, Q. Liu and Z. Jin, Adv. Electron. Mater., 2019, 5, 1800887 CrossRef .
118. K. Toprasertpong, M. Takenaka and S. Takagi, Appl. Phys. A: Mater. Sci. Process., 2022, 128, 1114 CrossRef CAS .
119. W. Jie and J. Hao, Nanoscale, 2017, 10, 328–335 RSC .
120. C. Zhang, K. Wang, F. Zhao, R. Pan, J. Zhang, H. Yu and J. Li, Nano Energy, 2020, 78, 105324 CrossRef CAS .
121. A. A. Pil’nik, A. A. Chernov and D. R. Islamov, Sci. Rep., 2020, 10, 15759 CrossRef PubMed .
122. P. Wei, X. Wang, X. Li, S. Han, N. Qiao, P. Zhang, Y. Deng, W. Zhang, L. Bu and G. Lu, Adv. Funct. Mater., 2021, 31, 2103369 CrossRef CAS .
123. Z. Wang, X. Liu, X. Zhou, Y. Yuan, K. Zhou, D. Zhang, H. Luo and J. Sun, Adv. Mater., 2022, 34, 2200032 CrossRef CAS PubMed .
124. Z. Chen, R. Yu, X. Yu, E. Li, C. Wang, Y. Liu, T. Guo and H. Chen, ACS Nano, 2022, 16, 19155–19164 CrossRef CAS PubMed .
125. J. Shi, J. Jie, W. Deng, G. Luo, X. Fang, Y. Xiao, Y. Zhang, X. Zhang and X. Zhang, Adv. Mater., 2022, 34, 2200380 CrossRef CAS PubMed .
126. Y. Li, J. Wang, Q. Yang and G. Shen, Adv. Sci., 2022, 9, 2202123 CrossRef CAS PubMed .
127. H. Yu, X. Zhao, M. Tan, B. Wang, M. Zhang, X. Wang, S. Guo, Y. Tong, Q. Tang and Y. Liu, Adv. Funct. Mater., 2022, 32, 2206765 CrossRef CAS .
128. Y. Kim, A. Chortos, W. Xu, Y. Liu, J. Y. Oh, D. Son, J. Kang, A. M. Foudeh, C. Zhu, Y. Lee, S. Niu, J. Liu, R. Pfattner, Z. Bao and T.-W. Lee, Science, 2018, 360, 998–1003 CrossRef CAS PubMed .
129. C. Wan, G. Chen, Y. Fu, M. Wang, N. Matsuhisa, S. Pan, L. Pan, H. Yang, Q. Wan, L. Zhu and X. Chen, Adv. Mater., 2018, 30, 1801291 CrossRef PubMed .
130. Y. Liu, J. Zhong, E. Li, H. Yang, X. Wang, D. Lai, H. Chen and T. Guo, Nano Energy, 2019, 60, 377–384 CrossRef CAS .
131. H. Wan, Y. Cao, L.-W. Lo, J. Zhao, N. Sepulveda and C. Wang, ACS Nano, 2020, 14, 10402–10412 CrossRef CAS PubMed .
132. Y. R. Lee, T. Q. Trung, B.-U. Hwang and N.-E. Lee, Nat. Commun., 2020, 11, 2753 CrossRef CAS PubMed .
133. R. Yu, Y. Yan, E. Li, X. Wu, X. Zhang, J. Chen, Y. Hu, H. Chen and T. Guo, Mater. Horiz., 2021, 8, 2797–2807 RSC .
134. F. Sun, Q. Lu, M. Hao, Y. Wu, Y. Li, L. Liu, L. Li, Y. Wang and T. Zhang, npj Flex Electron, 2022, 6, 1–8 CrossRef .

This journal is © The Royal Society of Chemistry 2023