RESUMO
van der Waals heterostructures realized by stacking different two-dimensional materials offer the possibility to design new devices with atomic-level precision. The realization of memristive synapses based on full two-dimensional materials provides an avenue for developing neuromorphic computing systems with excellent mechanical flexibility in future. Herein, we first develop the flexible full two-dimensional memristive synapse of graphene/WSe2-xOy/graphene exhibiting stable volatile resistive switching. The oxidation of WSe2 through O2 plasma treatment plays a key role in improving the memristive performance of the devices. And the switching behavior is associated with the migration of oxygen vacancies at the graphene/WSe2-xOy interface. Versatile synaptic functions have been realized, including short-term plasticity, long-term plasticity, the transition from short-term plasticity to long-term plasticity and high-pass temporal filtering, with ultralow energy consumption of â¼16.1 pJ per spike. The flexible full two-dimensional devices fabricated on flexible polyimide substrate show excellent mechanical flexibility, including a good endurance against repeatable bending of 1000 times and a bending strain of 2.5%.
RESUMO
Fabrication of highly crystalline oxide films onto silicon wafers has long been a critical obstacle for integrating multi-functional oxides into silicon-based technology. Herein, Pt/Ti is used as a buffer layer for the integration of highly oriented crystalline LaBaCo2O5+δ (LBCO) thin films onto silicon via pulsed laser deposition. LBCO films are highly (00l) oriented with smooth and sharp LBCO/Pt interfaces. The highly oriented LBCO films exhibit a high magnetic transition temperature (TC) and large coercive field (HC) with superparamagnetism over those deposited on single crystal substrates. What is more, the metallic-like behavior with enhanced magnetoresistance is also observed. The opportunity of using a Pt/Ti buffer layer as the growth template opens an alternative route for integrating functional transition metal oxides with tunable magnetic properties into Si-based technology.
RESUMO
Monolayer of 2D transition metal dichalcogenides, with a thickness of less than 1 nm, paves a feasible path to the development of ultrathin memristive synapses, to fulfill the requirements for constructing large-scale high density 3D stacking neuromorphic chips. Herein, memristive devices based on monolayer n-MoS2 on p-Si substrate with a large self-rectification ratio, exhibiting photonic potentiation and electric habituation, are successfully fabricated. Versatile synaptic neuromorphic functions, such as potentiation/habituation, short-term/long-term plasticity, and paired-pulse facilitation, are successfully mimicked based on the inherent persistent photoconductivity performance and the volatile resistive switching behavior. These findings demonstrate the potential applications of ultrathin transition metal dichalcogenides for memristive synapses. These memristive synapses with the combination of photonic and electric neuromorphic functions have prospective in the applications of synthetic retinas and optoelectronic interfaces for integrated photonic circuits based on mixed-mode electro-optical operation.
RESUMO
Phase engineering of two-dimensional transition metal dichalcogenides has received increasing attention in recent years due to its atomically thin nature and polymorphism. Here, we first realize an electric-field-induced controllable phase transition between semiconducting 2H and metallic 1T' phases in MoTe2 memristive devices. The device performs stable bipolar resistive switching with a cycling endurance of over 105, an excellent retention characteristic of over 105 s at an elevated temperature of 85 °C and an ultrafast switching of â¼5 ns for SET and â¼10 ns for RESET. More importantly, the device works in different atmospheres including air, vacuum and oxygen, and even works with no degradation after being placed in air for one year, indicating excellent surrounding and time stability. In situ Raman analysis reveals that the stable resistive switching originates from a controllable phase transition between 2H and 1T' phases. Density functional theory calculations reveal that the Te vacancy facilitates the phase transition in MoTe2 through decreasing the barrier between 2H and 1T' phases, and serving as nucleation sites due to the elimination of repulsive forces. This electric-field-induced controllable phase transition in MoTe2 devices offers new opportunities for developing reliable and ultrafast phase transition devices based on atomically thin membranes.
RESUMO
The development of novel synaptic device architectures with a high order of synaptic plasticity can provide a breakthrough toward neuromorphic computing. Herein, through the thermal oxidation of two-dimensional (2D) WSe2, unique memristive synapses based on the lateral heterostructure of 2D WSe2 and WO3, with multi-gate modulation characteristics, are firstly demonstrated. An intermediate transition layer in the heterostructure is observed through transmission electron microscopy. Raman spectroscopy and detailed electrical measurements provide insights into the mechanism of memristive behavior, revealing that the protons injected into/removed from the intermediate transition layer account for the memristive behavior. This novel memristive synapse can be used to emulate two neuron-based synaptic functions, like post-synaptic current, short-term plasticity and long-term plasticity, with remarkable linearity, symmetry, and an ultralow energy consumption of â¼2.7 pJ per spike. More importantly, the synaptic plasticity between the drain and source electrodes can be effectively modulated by the gate voltage and visible light in a four-terminal configuration. Such multi-gate tuning of the synaptic plasticity cannot be accomplished by any previously reported multi-gate synaptic devices that only mimic two neuron-based synapses. This new synaptic architecture with electrical and optical modulation enables a realistic emulation of biological synapses whose synaptic plasticity can be additionally regulated by the surrounding astrocytes, greatly improving the recognition accuracy and processing capacity of artificial neuristors, and paving a new way for highly efficient neuromorphic computation devices.
Assuntos
Materiais Biomiméticos/química , Modelos Biológicos , Óxidos/química , Compostos de Selênio/química , Tungstênio/química , Plasticidade Neuronal , Neurônios/fisiologiaRESUMO
Neural networks based on memristive devices have achieved great progress recently. However, memristive synapses with nonlinearity and asymmetry seriously limit the classification accuracy. Moreover, insufficient number of training samples in many cases also have negative effect on the classification accuracy of neural networks due to overfitting. In this work, dropout neuronal units are developed based on stochastic volatile memristive devices of Ag/Ta2O5:Ag/Pt. The memristive neural network using the dropout neuronal units effectively solves the problem of overfitting and mitigates the negative effects of the nonideality of memristive synapses, eventually achieves a classification accuracy comparable to the theoretical limit. The stochastic and volatile switching performances of the Ag/Ta2O5:Ag/Pt device are attributed to the stochastical rupture of the Ag filament under high electrical stress in the Ta2O5 layer, according to the TEM observation and the kinetic Monte Carlo simulation.
RESUMO
Artificial neurons with functions such as leaky integrate-and-fire (LIF) and spike output are essential for brain-inspired computation with high efficiency. However, previously implemented artificial neurons, e.g., Hodgkin-Huxley (HH) neurons, integrate-and-fire (IF) neurons, and LIF neurons, only achieve partial functionality of a biological neuron. In this work, quasi-HH neurons with leaky integrate-and-fire functions are physically demonstrated with a volatile memristive device, W/WO3 /poly(3,4-ethylenedioxythiophene): polystyrene sulfonate/Pt. The resistive switching behavior of the device can be attributed to the migration of protons, unlike the migration of oxygen ions normally involved in oxide-based memristors. With multifunctions similar to their biological counterparts, quasi-HH neurons are advantageous over the reported HH and LIF neurons, demonstrating their potential for neuromorphic computing applications.