Conductive Mechanism And Bionic Characteristics Of Analog Memristors Based On Metal Oxides

As the fourth passive device,the nonlinear resistance of memristor,which can remember current,will become the most promising memory element beyond Moore's law.Analog memristor,which is an important classification of memristors,is recognized as being extremely similar to the plastic response of the human brain synapse under the stimulation of bioelectrical signals.The analog memristor is one of the most commonly known devices in the electronic components that are the closest to the synapse in the nervous system.In this paper,the analog memristors based on the metal oxide film were studied.The research contents include preparation process,performance characteristic,mathematical and physical model and its application in the bionics.The preparation process and performance characteristics of the memristor are studied by means of the combination of experiment and theory.Based on the advantages of the sputtering method,all the films in the experiment process are prepared by a magnetron sputtering or DC sputtering method.Metal oxide comprises four materials including ZnO,CuO,Al2O3 and TiO2 as object,and the metal/oxide/metal(MIM)is a basic structure.In order to study the effect of the heterostructure and the electrode on the memristive performance of the film,the individual thin films and the electrode materials were combined into various structures,and the memristive characteristics of these structures were studied The I-V characteristic of memristor is measured by digital source meter.The structure,appearance and composition of the films are analyzed and characterized by means of XRD,SEM,TEM and EDSThe results show that all the metal oxides of the MIM structure prepared by the sputtering method can exhibit the characteristics of the analog memristor,and the current and the voltage relation of the MIM structure exhibit a hysteresis curve shape.However,the experimental process also shows that whether the films have resistance memory is constrained by the preparation process and structure.For the specific metal oxide material,the ratio of oxygen and argon,the pressure of the gas,the duration of the sputtering(the thickness of the film),and the type of the bottom electrode in the sputtering process,have a decisive influence on the formation of the memristive performance.It is also found in the study that the metal oxide of the multi-layer heterostructure tends to obtain the memristive performance more easily than the single structure,for example,the contents of the fourth chapter show that the Au/CuO/Cu/Pt has better resistance memory and anti-fatigue property than the Au/CuO/Pt structure.Based on the above conclusions,the results show that the electrical properties of metal oxide thin film memristors can be effectively regulated by structural design and technological conditions.On the basis of the experiment,the paper establishes an effective model for simulating the memristor.The model equation is based on the fermi gold law,and the transport process of charge in the MIM structure is described in terms of the number of defects,the coupling strength between the defects and the probability of the carrier jump.In this work,metal oxide and metal electrode interface are used to model the equations.After solving the equations by Matlab,the voltage-current relationship curve of the simulator can be well simulated.The results of the test and simulation show that the memristor device prepared in this work has the learning ability and the neurosynaptic simulation ability similar to the conditional reflex with the continuous and memory-based conductance change of the memristor.A single analog memristor can simulate the conditioned reflex of the babbitt's dog.A plurality of analog memristors can be used to simulate and complete the high-efficiency image recognition function.The experiment and the simulation also prove that the prepared memristor has the relevant plasticity of the pulse time possessed by the nerve synapse,short-term plasticity,long-term plasticity and learning-forget-restudy and other biological characteristics.