The Choice Of Materials And Investegation On The Machanism In The Memristive System Based On Oxides

Synapses are specialized junctions through which neurons signal to each other and to non-neuronal cells such as those in muscles or glands, and synaptic plasticity is important neurochemical foundation of learning and memory. Therefore the function simulation of the synapse is one of the most important components of artificial neural networks. The primary characteristic of memristive system is that it changes the resistance states, depending on the entire dynamical history of the charge flowing through the system, so that it "memorizes" the dynamical history of the system. It is the most likely candidate for the electronic components to achieve synaptic function in neural networks since its work mode resembles that of human brain. The artificial neural network can be expected to be built on the memristive technology, and then to acquire bionic brain functional hardware.The HP lab reported that they built the first working prototypes of memristor in 2008. The conception of memristor was proposed earlier by Leon Chua reasoned from symmetry arguments for the sake of logic completeness of circuit theory in 1971. Memristive system has attracted more attention and interest from researchers and engineers since the physical memristor was reported. The memristive behavior has been found to be ubiquitous in many materials and nanoscale system, and is related to a variety of physical mechanisms. For the complex of the materials and the mechanisms of the memristive system related, there is still lack of a general theory to describe and predict the memristive system. Nevertheless, we can still confirm that: first, memristive systems have some asymmetry in device structure, and the electrical response of such structures has rectification current due to the asymmetry; second, a fundamental characteristic of memristive system is that it changes the resistance states, depending on the entire dynamical history of the charge flowing through the system, and works as if it is "memorizing". This characteristic has been used as a criterion for the memristive system; third, the variation of the state variables manifested by the resistance modulation in memristive system depends on the migration of carriers with the changes in external conditions. Therefore, the migration rate of carriers and its changes to different conditions are essential to determine the state variables of the system.With the analysis of the types and the key factors of memristive system, we fabricated prototypical "digital" and "analog" memristive devices respectively, and investigated the origin of the memristive behaviors, the changes of the state variables with the applied voltage, the migration rate of carriers, and the device integrating method by analysis the electrical behavior and other experimental investigation:(1) Demonstrated the role of the electrochemical reaction of the active electrode in the memristive behaviors.Based on the investigation ofⅠ-Ⅴcharacteristics under the electrical field of Ti/Al2O3/Pt "digital" memristive prototypical device, its micro-structure and the composition, we found that:the resistive switching of the device can be realized without the strong-field forming process by the electrochemical reaction of the active electrode Ti under the electrical field;The redox of the active electrode Ti causes the drift of oxygen ions in the oxide layer, while bringing a large amount of oxygen vacancies to the oxide layer at the same time. The accumulation and dissipation of oxygen vacancies at the interface as well as the redox reaction at the electrode Ti interface layer modulates the interface contact potential barrier, and consequently modulates the resistance state of the device.Furthermore, we investigated the failure mechanism of Ti/Al2O3/Pt "digital" memristor and proposed an approach by which the device can be recovered.(2) Established the "analog" memristive system based on ferroelectric/metal heterojunction structure.We confirmed that the Pt/LiNbO3/Pt structure prototypical device has all the basic characteristics of memristive system, and its resistance state exhibits a damping decay with the external periodic voltage modulations. The tunneling mechanisms were also displayed in the transport behavior in each period.By analysis, it has been found out that the resistance state modulation of the system reflects the changes in the potential barrier. The thickness of the barrier decreased with the increasing modulated period. The modulation of the interface barrier originates from the accumulation and dissipation of oxygen vacancies at the bottom interface under the electric field.We have observed the resistance relaxation phenomenon, which is closely related to the migration and migration rate of ions (oxygen vacancies) in the material. This relaxation phenomenon is similar with the re-absorption of the neurotransmitters and the decrease of the synaptic weights. The attenuation of the synaptic weights is related to the "forget" behavior in the brain memory process.(3) The migration rate of oxygen vacancies were obtained based on the analysis of the transient current of the Pt/LiNbO3/Pt structure at different temperatures and electric fields.(4) 1×n integrated arrays of memoristor unit with the crossbar structure were fabricated combined with the lithography, "lift off", the thermal evaporation and the pulsed laser deposition techniques. The resistive switching characteristic of the memrstor arrays has also been testified in this work.