Black Phosphorus/Ferroelectric P(VDF-TrFE) Field-Effect Transistors with High Mobility for Energy-Efficient Artificial Synapse in High-Accuracy Neuromorphic Computing.

The neuromorphic system is an attractive platform for next-generation computing with low power and fast speed to emulate knowledge-based learning. Here, we design ferroelectric-tuned synaptic transistors by integrating 2D black phosphorus (BP) with a flexible ferroelectric copolymer poly(vinylidene fluoride-trifluoroethylene) (P(VDF-TrFE)). Through nonvolatile ferroelectric polarization, the P(VDF-TrFE)/BP synaptic transistors show a high mobility value of 900 cm2 V-1 s-1 with a 103 on/off current ratio and can operate with low energy consumption down to the femtojoule level (∼40 fJ). Reliable and programmable synaptic behaviors have been demonstrated, including paired-pulse facilitation, long-term depression, and potentiation. The biological memory consolidation process is emulated through ferroelectric gate-sensitive neuromorphic behaviors. Inspiringly, the artificial neural network is simulated for handwritten digit recognition, achieving a high recognition accuracy of 93.6%. These findings highlight the prospects of 2D ferroelectric field-effect transistors as ideal building blocks for high-performance neuromorphic networks.

Memristors based on 2D MoSe2 nanosheets as artificial synapses and nociceptors for neuromorphic computing.

Neuromorphic computing inspired by the human brain is highly desirable in the artificial intelligence age. Thus, it is essential to comprehensively investigate the neuromorphic characteristics of artificial synapses and neurons which are the unit cells in an artificial neural network (ANN). Memristors are considered ideal candidates to serve as artificial synapses and neurons in the ANN. Herein, two-terminal memristors based on two-dimensional (2D) MoSe2 nanosheets are fabricated, demonstrating analog resistive switching (RS) behaviors. Unlike the digital RS behaviors with a sharp transition between the two resistance states, the analog RS provides a series of tunable resistance states, which is more suitable for the realization of synaptic plasticity. Thus, the fabricated memristors successfully implement the synaptic functions, such as paired-pulse facilitation, long-term potentiation and long-term depression. The analog memristors can be utilized to construct the ANN for image recognition, leading to a high recognition accuracy of 92%. In addition, the synaptic memristors can emulate the "learning-forgetting" experience of the human brain. Furthermore, to demonstrate the ability of single neuron learning in our devices, the memristors are studied as artificial nociceptors to recognize noxious stimuli. Our research provides comprehensive investigations on the neuromorphic characteristics of artificial synapses and nociceptors, suggesting promising prospects for applications in neuromorphic computing based on 2D MoSe2 nanosheets.

Controllable digital and analog resistive switching behavior of 2D layered WSe2 nanosheets for neuromorphic computing.

Memristors with controllable resistive switching (RS) behavior have been considered as promising candidates for synaptic devices in next-generation neuromorphic computing. In this work, two-terminal memristors with controllable digital and analog RS behavior are fabricated based on two-dimensional (2D) WSe2 nanosheets. Under a relatively high operating voltage of 4 V, the memristor demonstrates stable and reliable non-volatile bipolar digital RS with a high switching ratio of 6.3 × 104. On the other hand, under a relatively low operation voltage, the memristor exhibits analog RS with a series of tunable resistance states. The fabricated memristors can work as an artificial synapse with fundamental synaptic functions, such as long-term potentiation (LTP) and depression (LTD) as well as paired-pulse facilitation (PPF). More importantly, the memristor demonstrates high conductance modulation linearity with the calculated nonlinear parameter for conductance as -0.82 in the LTP process, which is beneficial to improving the accuracy of neuromorphic computing. Furthermore, the neuromorphic computing of file types and image recognition can be emulated based on a constructed three-layer artificial neural network (ANN) with a recognition accuracy that can reach up to 95.9% for small digits. In addition, memristors can be used to emulate the learning-forgetting experience of the human brain. Consequently, the memristor based on 2D WSe2 nanosheets not only exhibits controllable RS behavior but also simulates synaptic functions and is expected to be a potential candidate for future neuromorphic computing applications.

Introduction of defects in hexagonal boron nitride for vacancy-based 2D memristors.

Two-dimensional (2D) materials have become potential resistive switching (RS) layers to prepare emerging non-volatile memristors. The atomically thin thickness and the highly controllable defect density contribute to the construction of ultimately scaled memory cells with stable switching behaviors. Although the conductive bridge random-access memory based on 2D hexagonal boron nitride has been widely studied, the realization of RS completely relying on vacancies in 2D materials has performance superiority. Here, we synthesize carbon-doped h-BN (C-h-BN) with a certain number of defects by controlling the weight percentage of carbon powder in the source. These defects can form a vacancy-based conductive filament under an applied electric field. The memristor displays bipolar non-volatile memory with a low SET voltage of 0.85 V and shows a long retention time of up to 104 s at 120 °C. The response times of the SET and RESET process are less than 80 ns and 240 ns, respectively. The current mapping by conductive atomic force microscopy demonstrates the electric-field-induced current tunneling from defective sites of the C-h-BN flake, revealing the defect-based RS in the C-h-BN memristor. Moreover, C-h-BN with excellent flexibility can be applied to wearable devices, maintaining stable RS performance in a variety of bending environments and after multiple bending cycles. The vacancy-based 2D memristor provides a new strategy for developing ultra-scaled memory units with high controllability.

Tunable Resistive Switching in 2D MXene Ti3C2 Nanosheets for Non-Volatile Memory and Neuromorphic Computing.

An artificial synapse is essential for neuromorphic computing which has been expected to overcome the bottleneck of the traditional von-Neumann system. Memristors can work as an artificial synapse owing to their tunable non-volatile resistance states which offer the capabilities of information storage, processing, and computing. In this work, memristors based on two-dimensional (2D) MXene Ti3C2 nanosheets sandwiched by Pt electrodes are investigated in terms of resistive switching (RS) characteristics, synaptic functions, and neuromorphic computing. Digital and analog RS behaviors are found to coexist depending on the magnitude of operation voltage. Digital RS behaviors with two resistance states possessing a large switching ratio exceeding 103 can be achieved under a high operation voltage. Analog RS behaviors with a series of resistance states exhibiting a gradual change can be observed at a relatively low operation voltage. Furthermore, artificial synapses can be implemented based on the memristors with the basic synaptic functions, such as long-term plasticity of long-term potentiation and depression and short-term plasticity of the paired-pulse facilitation and depression. Moreover, the "learning-forgetting" experience is successfully emulated based on the artificial synapses. Also, more importantly, the artificial synapses can construct an artificial neural network to implement image recognition. The coexistence of digital and analog RS behaviors in the 2D Ti3C2 nanosheets suggests the potential applications in non-volatile memory and neuromorphic computing, which is expected to facilitate simplifying the manufacturing complexity for complex neutral systems where analog and digital switching is essential for information storage and processing.

Low-Power Memristor Based on Two-Dimensional Materials.

The memristor is an excellent candidate for nonvolatile memory and neuromorphic computing. Recently, two-dimensional (2D) materials have been developed for use in memristors with high-performance resistive switching characteristics, such as high on/off ratios, low SET/RESET voltages, good retention and endurance, fast switching speed, and low power and energy consumption. Low-power memristors are highly desired for recent fast-speed and energy-efficient artificial neuromorphic networks. This Perspective focuses on the recent progress of low-power memristors based on 2D materials, providing a condensed overview of relevant developments in memristive performance, physical mechanism, material modification, and device assembly as well as potential applications. The detailed research status of memristors has been reviewed based on different 2D materials from insulating hexagonal boron nitride, semiconducting transition metal dichalcogenides, to some newly developed 2D materials. Furthermore, a brief summary introducing the perspectives and challenges is included, with the aim of providing an insightful guide for this research field.

Memristor Based on Inorganic and Organic Two-Dimensional Materials: Mechanisms, Performance, and Synaptic Applications.

A memristor is a two-terminal device with nonvolatile resistive switching (RS) behaviors. Recently, memristors have been highly desirable for both fundamental research and technological applications because of their great potential in the development of high-density memory technology and neuromorphic computing. Benefiting from the unique two-dimensional (2D) layered structure and outstanding properties, 2D materials have proven to be good candidates for use in gate-tunable, highly reliable, heterojunction-compatible, and low-power memristive devices. More intriguing, stable and reliable nonvolatile RS behaviors can be achieved in multi- and even monolayer 2D materials, which seems unlikely to be achieved in traditional oxides with thicknesses less than a few nanometers because of the leakage currents. Moreover, such two-terminal devices show a series of synaptic functionalities, suggesting applications in simulating a biological synapse in the neural network. In this review article, we summarize the recent progress in memristors based on inorganic and organic 2D materials, from the material synthesis, device structure and fabrication, and physical mechanism to some versatile memristors based on diverse 2D materials with good RS properties and memristor-based synaptic applications. The development prospects and challenges at the current stage are then highlighted, which is expected to inspire further advancements and new insights into the fields of information storage and neuromorphic computing.

A synaptic memristor based on two-dimensional layered WSe2 nanosheets with short- and long-term plasticity.

Neural synapses with diverse synaptic functions of short- and long-term plasticity are highly desired for developing complex neuromorphic systems. A memristor with its two terminals serving as pre- and post-neurons, respectively, can emulate two neuronal-based synaptic functions. In this work, multilayer two-dimensional (2D) layered WSe2 nanosheets are synthesized by a salt-assisted chemical vapor deposition (CVD) method. Two-terminal memristors with a planar structure are fabricated based on the CVD-grown triangular WSe2 nanosheets. The fabricated devices exhibit typical bipolar nonvolatile resistive switching behaviors with a high current ON/OFF ratio of up to 6 × 103 and good retention and endurance properties, suggesting good stability and reliability of the WSe2-based memristors. Furthermore, the developed memristors demonstrate synaptic functions of short- and long-term plasticity (STP and LTP), as well as a transition from STP to LTP by applying consecutive pulse voltages. Moreover, the WSe2-based memristors exhibits biological synaptic functions of long-term potentiation and depression, and paired-pulse facilitation. Thus, our 2D WSe2 nanosheet based memristors not only exhibit stable and reliable nonvolatile resistive switching behaviors, but also show potential applications in mimicking biological synapses.

An Artificial Electrical-Chemical Mixed Synapse Based on Ion-Gated MoS2 Nanosheets for Real-Time Facilitation Index Tuning.

Tunability of facilitation in short-term memory (STM) provides great potential in bioinspired computing. Recently, several doping strategies were proposed to modify the intrinsic features of materials, resulting in the optimization of the facilitation index (FI). However, real-time scale tuning, which is implemented on the same synaptic device, has not yet been demonstrated. Inspired by the chemical-electrical mixed synapse structure in the brain, we propose a three-terminal artificial synapse based on an ion-gated MoS2 memristor. The gate terminal serves as a nonvolatile ionic pump via chemical intercalation, which effectively affects both the conductance baseline and the hysteresis degree of the STM effect of the memristor. We further modeled the postsynaptic current (PSC) behavior and used it for reservoir computing. Simulation results show that, due to the real-time tuning ability, the built reservoir can be programmed for specific handwritten recognition tasks with the pruning of neurons from 784 to 50. The developed artificial mixed synapse is promising for a downsampling module in neural network design.

Reversible Transformation between Bipolar Memory Switching and Bidirectional Threshold Switching in 2D Layered K-Birnessite Nanosheets.

Birnessite-related manganese dioxides (MnO2) have recently been studied owing to their diverse low-dimensional layered structures and potential applications in energy devices. The birnessite MnO2 possesses a layered structure with edge-shared MnO6 octahedra layer stacked with interlayer of cations. The unique layered structure may provide some distinct electrical properties for the 2D layered nanosheets. In this work, layered K-birnessite MnO2 samples are synthesized by a hydrothermal method. The resistive switching (RS) devices based on single K-birnessite MnO2 nanosheets are fabricated by transferring the nanosheets onto SiO2/Si substrates through a facile and feasible method of mechanical exfoliation. The device exhibits nonvolatile memory switching (MS) behaviors with high current ON/OFF ratio of ∼2 × 105. And more importantly, reversible transformation between the nonvolatile MS and volatile threshold switching (TS) can be achieved in the single layered nanosheet through tuning the magnitude of compliance current (Icc). To be more specific, a relatively high Icc (1 mA) can trigger the nonvolatile MS behaviors, while a relatively low Icc (≤100 µA) can generate volatile TS characteristics. This work not only demonstrates the memristor based on single birnessite-related MnO2 nanosheet, but also offers an insight into understanding the complex resistive switching types and relevant physical mechanisms of the 2D layered oxide nanosheets.

Phase coexistence induced strong piezoelectricity in K0.5Na0.5NbO3-based lead-free ceramics.

For perovskite ceramics, the ferroelectric phase boundary plays an important role in improving the piezoelectricity of the materials. In this work, (1 - x)(K0.5Na0.5)NbO3 - x[NaSbO3 + Bi0.5(Na0.8K0.2)0.5(Zr0.5Hf0.5)O3] lead-free ceramics with R-O-T ferroelectric phase coexistence were developed and the relationship between the phase structure and piezoelectricity was investigated in detail. As x increases, the transition temperature of the rhombohedral and orthorhombic phases (TR-O) increases, while the orthorhombic-tetragonal phase transition temperature (TO-T) decreases. When 0.04 ≤ x ≤ 0.05, three ferroelectric phases (R-O-T) coexist near room temperature in the ceramics. Due to the highly consistent orientation of the ferroelectric dipole and the free energy flattening and a low energy barrier induced by the coexistence of three ferroelectric phases (R-O-T), excellent piezoelectric performances of d33 = 452 pC N-1, kp = 63% and εr = 4414 are achieved at x = 0.04. Our study suggests that compared with two ferroelectric phase boundaries (R-O and O-T), the coexistence of the three ferroelectric phases (R-O-T) can effectively enhance the piezoelectric properties of (K0.5Na0.5)NbO3-based ceramics.

Ferroelectric and Piezoelectric Effects on the Optical Process in Advanced Materials and Devices.

Piezoelectric and ferroelectric materials have shown great potential for control of the optical process in emerging materials. There are three ways for them to impact on the optical process in various materials. They can act as external perturbations, such as ferroelectric gating and piezoelectric strain, to tune the optical properties of the materials and devices. Second, ferroelectricity and piezoelectricity as innate attributes may exist in some optoelectronic materials, which can couple with other functional features (e.g., semiconductor transport, photoexcitation, and photovoltaics) in the materials giving rise to unprecedented device characteristics. The last way is artificially introducing optical functionalities into ferroelectric and piezoelectric materials and devices, which provides an opportunity for investigating the intriguing interplay between the parameters (e.g., electric field, temperature, and strain) and the introduced optical properties. Here, the tuning strategies, fundamental mechanisms, and recent progress in ferroelectric and piezoelectric effects modulating the optical properties of a wide spectrum of materials, including lanthanide-doped phosphors, quantum dots, 2D materials, wurtzite-type semiconductors, and hybrid perovskites, are presented. Finally, the future outlook and challenges of this exciting field are suggested.

Lanthanide Yb/Er co-doped semiconductor layered WSe2 nanosheets with near-infrared luminescence at telecommunication wavelengths.

Atomically thin layers of transition metal dichalcogenides (TMDs) have recently drawn great attention. However, doping strategies and controlled synthesis for wafer-scale TMDs are still in their early stages, greatly hindering the construction of devices and further basic studies. In this work, we develop the fast deposition of wafer-scale layered lanthanide ion Yb/Er co-doped WSe2 using pulsed laser deposition. WSe2 nanosheets were chosen as the host, while Yb3+ and Er3+ ions served as the sensitizer and activator, respectively. The obtained Yb/Er co-doped WSe2 layers exhibit good uniformity and high crystallinity with highly textured features. Under the excitation of a diode laser at 980 nm, down-conversion emission is observed at around 1540 nm, assigned to the emission transition between the 4I13/2 and 4I15/2 states of Er3+. Considering the significance of 1540 nm luminescence in the application of photonic technologies, this observation in the WSe2:Yb/Er nanosheets down to the monolayer provides a new opportunity for developing photonic devices at the 2D limit. Our work not only offers a general method to prepare wafer-scale lanthanide doped TMDs, but also to widely modulate the luminescence of atomically layered TMDs by introducing lanthanide ions.

Time-dependent transport characteristics of graphene tuned by ferroelectric polarization and interface charge trapping.

Graphene-based field effect transistors (FETs) were fabricated by employing ferroelectric Pb(Mg1/3Nb2/3)O3-PbTiO3 (PMN-PT) as a gate insulator. The co-existing effects of ferroelectric gating and interface charge trapping on the transport properties of graphene were investigated with respect to the FET structure. The sheet resistance (Rs) of graphene shows a slight decay under a small applied voltage, which is much less than the coercive voltage of the ferroelectric PMN-PT, suggesting non-negligible charge trapping effects. Moreover, when the applied voltage is increased up to a value larger than the coercive voltage, Rs exhibits three states: an initial rapid change, followed by a slow nearly exponential evolution, and finally a saturated state either during the applied voltage is retained or after it is released. In particular, a high-resistance state is finally reached due to the ferroelectric gating, implying that ferroelectric effects dominate this process. The underlying physical mechanism was fully investigated to effectively address the observed evolution of time-dependent Rs. Such a finding provides us an opportunity to understand the co-existing effects of ferroelectric gating and charge trapping and tune the transport properties of graphene through the interface effects.

Observation of Room-Temperature Magnetoresistance in Monolayer MoS2 by Ferromagnetic Gating.

Room-temperature magnetoresistance (MR) effect is observed in heterostructures of wafer-scale MoS2 layers and ferromagnetic dielectric CoFe2O4 (CFO) thin films. Through the ferromagnetic gating, an MR ratio of -12.7% is experimentally achieved in monolayer MoS2 under 90 kOe magnetic field at room temperature (RT). The observed MR ratio is much higher than that in previously reported nonmagnetic metal coupled with ferromagnetic insulator, which generally exhibited MR ratio of less than 1%. The enhanced MR is attributed to the spin accumulation at the heterostructure interface and spin injection to the MoS2 layers by the strong spin-orbit coupling effect. The injected spin can contribute to the spin current and give rise to the MR by changing the resistance of MoS2 layers. Furthermore, the MR effect decreases as the thickness of MoS2 increases, and the MR ratio becomes negligible in MoS2 with thickness more than 10 layers. Besides, it is interesting to find a magnetic field direction dependent spin Hall magnetoresistance that stems from a combination of the spin Hall and the inverse spin Hall effects. Our research provides an insight into exploring RT MR in monolayer materials, which should be helpful for developing ultrathin magnetic storage devices in the atomically thin limit.

Wafer-Scale Synthesis of High-Quality Semiconducting Two-Dimensional Layered InSe with Broadband Photoresponse.

Large-scale synthesis of two-dimensional (2D) materials is one of the significant issues for fabricating layered materials into practical devices. As one of the typical III-VI semiconductors, InSe has attracted much attention due to its outstanding electrical transport property, attractive quantum physics characteristics, and dramatic photoresponse when it is reduced to atomic scale. However, scalable synthesis of single phase 2D InSe has not yet been achieved so far, greatly hindering further fundamental studies and device applications. Here, we demonstrate the direct growth of wafer-scale layered InSe nanosheets by pulsed laser deposition (PLD). The obtained InSe layers exhibit good uniformity, high crystallinity with macro texture feature, and stoichiometric growth by in situ precise control. The characterization of optical properties indicates that PLD grown InSe nanosheets have a wide range tunable band gap (1.26-2.20 eV) among the large-scale 2D crystals. The device demonstration of field-effect transistor shows the n-type channel feature with high mobility of 10 cm2 V-1 s-1. Upon illumination, InSe-based phototransistors show a broad photoresponse to the wavelengths from ultraviolet to near-infrared. The maximum photoresponsivity attains 27 A/W, plus a response time of 0.5 s for the rise and 1.7 s for the decay, demonstrating the strong and fast photodetection ability. Our findings suggest that the PLD grown InSe would be a promising choice for future device applications in the 2D limit.

Layer-dependent nonlinear optical properties and stability of non-centrosymmetric modification in few-layer GaSe sheets.

Gallium selenide, an important second-order nonlinear semiconductor, has received much scientific interest. However, the nonlinear properties in its two-dimensional (2D) form are still unknown. A strong second harmonic generation (SHG) in bilayer and multilayer GaSe sheets is reported. This is also the first observation of SHG on 2D GaSe thin layers. The SHG of multilayer GaSe above five layers shows a quadratic dependence on the thickness; while that of a sheet thinner than five layers shows a cubic dependence. The discrepancy between the two SHG responses is attributed to the weakened stability of non-centrosymmetric GaSe in the atomically thin flakes where a layer-layer stacking order tends to favor centrosymmetric modification. Importantly, two-photon excited fluorescence has also been observed in the GaSe sheets. Our free-energy calculations based on first-principles methods support the observed nonlinear optical phenomena of the atomically thin layers.

Mass transport mechanism of cu species at the metal/dielectric interfaces with a graphene barrier.

The interface between the metal and dielectric is an indispensable part in various electronic devices. The migration of metallic species into the dielectric can adversely affect the reliability of the insulating dielectric and can also form a functional solid-state electrolyte device. In this work, we insert graphene between Cu and SiO2 as a barrier layer and investigate the mass transport mechanism of Cu species through the graphene barrier using density functional theory calculations, second-ion mass spectroscopy (SIMS), capacitance-voltage measurement, and cyclic voltammetry. Our theoretical calculations suggest that the major migration path for Cu species to penetrate through the multiple-layered graphene is the overlapped defects larger than 0.25 nm2. The depth-profile SIMS characterizations indicate that the "critical" thickness of the graphene barrier for completely blocking the Cu migration is 5 times smaller than that of the conventional TaN barrier. Capacitance-voltage and cyclic voltammetry measurement reveal that the electrochemical reactions at the Cu/SiO2 interface become a rate-limiting factor during the bias-temperature stressing process with the use of a graphene barrier. These studies provide a distinct roadmap for designing controllable mass transport in solid-state electrolyte devices with the use of a graphene barrier.

Graphene-based hybrid structures combined with functional materials of ferroelectrics and semiconductors.

Fundamental studies and applications of 2-dimensional (2D) graphene may be deepened and broadened via combining graphene sheets with various functional materials, which have been extended from the traditional insulator of SiO2 to a versatile range of dielectrics, semiconductors and metals, as well as organic compounds. Among them, ferroelectric materials have received much attention due to their unique ferroelectric polarization. As a result, many attractive characteristics can be shown in graphene/ferroelectric hybrid systems. On the other hand, graphene can be integrated with conventional semiconductors and some newly-discovered 2D layered materials to form distinct Schottky junctions, yielding fascinating behaviours and exhibiting the potential for various applications in future functional devices. This review article is an attempt to illustrate the most recent progress in the fabrication, operation principle, characterization, and promising applications of graphene-based hybrid structures combined with various functional materials, ranging from ferroelectrics to semiconductors. We focus on mechanically exfoliated and chemical-vapor-deposited graphene sheets integrated in numerous advanced devices. Some typical hybrid structures have been highlighted, aiming at potential applications in non-volatile memories, transparent flexible electrodes, solar cells, photodetectors, and so on.

Exceptional tunability of band energy in a compressively strained trilayer MoS2 sheet.

Tuning band energies of semiconductors through strain engineering can significantly enhance their electronic, photonic, and spintronic performances. Although low-dimensional nanostructures are relatively flexible, the reported tunability of the band gap is within 100 meV per 1% strain. It is also challenging to control strains in atomically thin semiconductors precisely and monitor the optical and phonon properties simultaneously. Here, we developed an electromechanical device that can apply biaxial compressive strain to trilayer MoS2 supported by a piezoelectric substrate and covered by a transparent graphene electrode. Photoluminescence and Raman characterizations show that the direct band gap can be blue-shifted for ~300 meV per 1% strain. First-principles investigations confirm the blue-shift of the direct band gap and reveal a higher tunability of the indirect band gap than the direct one. The exceptionally high strain tunability of the electronic structure in MoS2 promising a wide range of applications in functional nanodevices and the developed methodology should be generally applicable for two-dimensional semiconductors.