Study On The Resistive Switching And The Application Of Artificial Synapse Of HfO₂/biFeO₃ And Two-dimensional TiS₃ Memristors

With the development of science and technology,higher requirements for information processing and information storage have been presented now.In this era,it has become increasingly difficult for the traditional computing system based on the von Neumann system to meet the demand due to the separation of the data storage unit and the data processing unit in terms of data processing speed and energy consumption.In contrast,the human brain is seen as a powerful information processing center.The neuromorphic computing system inspired by the brain is expected to solve the bottleneck of the traditional von Neumann system and achieve low energy consumption and high efficiency for computing and storing data.In the brain,the transmission and processing of information rely on the coordination of nerves and synapses.Simulating the behavior of the human brain,studying its mechanism and the realization of electronic devices with synaptic function are the key steps to realize the neuromorphic computing system.Traditional methods,which use multiple electronic components to build circuits to simulate synaptic behaviors and functions will lead to complex structures and high energy consumption of the circuit.Because of its low power consumption,fast response speed,good stability,easy integration and compatibility,traditional CMOS,its similar structure with synapses and reproducible conductance regulation,the memristor has become a strong competitor for artificial electronic synapses and is used in the field of neuromorphic computing.The paper first studies the resistive switching characteristics of the HfO₂/BFO double-layer memristor.The ultra-thin bismuth ferrite film is inserted into the HfO₂-based memristor to prepare the Pt/BFO/HfO₂/Ti N memristor.The multi-level resistance state performance and conductance gradual changing performance of the memristor device under DC voltage scanning are obtained.And the realization of the conductance modulation performance of the memristor in the pulse mode shows that the device has the function to realize the simulation of the gradual adjustment process of the synaptic weight in the synaptic system.In addition,we also realize the simulation of the spiking time dependent plasticity(STDP)of the learning rule in the synapse by the Pt/BFO/HfO₂/Ti N memristive device.Finally,a 3×3 memristor array is constructed to complete the learning and memory functions.Subsequently,the two-dimensional material TiS₃ nanoribbons is successfully prepared and characterized.In response to the demand of memristors for the realization of multifunctional devices,Al/TiS₃/ITO photoelectric memristive devices are prepared based on TiS₃ nanoribbons.Based on the characteristics of TiS₃ nanoribbons,the paper further studies the resistive performance of the device.Al/TiS₃/ITO photoelectric memristor has a simple operation process,stable bipolar resistance characteristics,excellent on-off ratio,retention performance and needn't set compliance current,.In addition,under the stimulation of light signals of different wavelengths,the device shows different I-V curves with different high-resistance current values.Further by fitting the mechanism of the device and combining the energy band diagram and the resistance change model diagram analysis,the resistance switching mechanism of the device is obtained.Base on the resistive switching characteristics and photoelectric properties of photoelectric memristors of TiS₃ nanoribbons,the paper further focuses on its synaptic bionic function.Firstly,the conductance modulation of the photoelectric memristor device in the pulse mode is presented,which means successful realization of simulation to adjust the synaptic weight in the synaptic system.In addition,the Al/TiS₃/ITO photoelectric memristor has completed the realization of the spiking time dependent plasticity(STDP)and photonic enhanced and electrical inhibition in the synapse.Finally,by applying appropriate light stimulation signals and electrical stimulation signals to the device,the photoelectric device realizes the simulation of the typical associative learning behavior of Pavlov's conditioned reflex.In summary,the paper presents the memristor based on the HfO₂/BiFeO₃ double-layer memristor and the photoelectric memristor based on the two-dimensional material TiS₃nanoribbons through the method of structural design and material optimization,studies their resistive switching performance and realizes the function of synapse.This is of great significance to the application of memristors in the field of non-volatile storage and neuromorphic fields,and provides a feasible idea for the realization of subsequent multi-functional devices and the completion of a perceptual learning system.