| The coming of the Big Data Era brings an urgent demand for high-performance computing,however,the enhancement of computing ability to cope with the complex real-time data（including unstructured sensory data such as images,video and sound）is severely impeded by the Von Neumann bottleneck（the data transfer rate between CPU and memory is limited）.Under such a background,researchers are eager to build new computer architecture.The human brain is not only good at handling unstructured information,but also has efficient and intelligent parallel architecture combining storage and computing.It is the tendency of computers’development and mankind’s dream to build brian-inspired computer.Synapse as an essential part of the human brain nervous system,realizing synaptic simulation is the critical step to build massively artificial neuromorphic system.The unique electrical transmission characteristics of the memristors are similar to biological synapses,therefore,they have great potential as electronic synapse devices applied in neuromorphic computing,In this thesis,a series of research about ferroelectric memristor were carried out based on perovskite ferroelectrics--BaTiO3.The effects of temperature and oxygen pressure on the Ba Ti O3 films grown by pulse laser deposition were investigated to find the optimal growth window,the quality of BaTiO3 films were mainly evaluated by lattice structure,surface morphology,ferroelectricity,etc.And then we explored the resistive switching performance of the devices when using perovskite-type nickelates（SmNiO3）instead of traditional perovskite oxide electrodes（SrRuO3）as the bottom electrode.The experimental results suggested that the devices have better performance when using SmNiO3 as the bottom electrode,including big ON/OFF ratio,excellent endurance and retention properties and multi-resistance properties.Based on the analysis of our results,the resistive switching behavior in the devices can be explained qualitatively by the polarization modulation of potentital barriers of p-n junction formed at the BaTiO3/SmNiO3 interface,the carrier concentration of p-type semiconducting SmNiO3 could be changed by the polarization switching.Finally,this thesis successfully simulated biological synapse performance using ferroelectric memristor based on SmNiO3 as bottom electrode structure,including the nonlinear transmission property,synaptic weight saturation characteristic,and senior learning rule in the brain--Spike-time-dependent plasticity（STDP）,in addition,the results of fatigue and power consumption tests shown that the devices have good endurance and the potential to realize ultra-low power consumption. |
