| Traditional computing chips are based on the Von Neumann architecture,which cycles through the basic sequence of fetching instructions from memory,decoding instructions,executing instructions,and storing results.Although current computers have made rapid progress in improving computing speed and processing complexity,the most advanced supercomputers still cannot compete with the human brain in terms of adaptability,fault tolerance,size and energy efficiency.As a new type of storage device,resistive random access memory(RRAM)has attracted much attention due to its simple structure,low power consumption and high integration density.With the increases demand of chip integration density,RRAM devices need to have better performance to improve their practicality,such as achieving multi-level storage or multi-functional integration.This article mainly focus on nickel oxide(NiO)thin film resistive memory and conducts research on its non-volatile storage,threshold switching,quantum conductance phenomenon and neural synapse biomimetic simulation.The main research contents are as follows:1.A Pt/NiO/FTO resistive memory was prepared using the sol-gel method and its switching characteristics,threshold switching characteristics,quantum conductance phenomena and negative differential resistance characteristics were tested.Nickel oxide resistive memory is a non-volatile resistance switch.Due to the ease of obtaining raw materials and stable device performance,it has recently attracted widespread interest.FTO has good transparency and good application prospects in the next generation of electronic products.By preparing nickel oxide devices on FTO using the sol-gel method,we found that under low limiting currents,Pt/NiO/FTO devices exhibit good resistance switching characteristics with a switch ratio of about 103.When the limiting current is 2 m A,the SET process threshold switching phenomenon is stable,and the RESET process exhibits quantum conductance phenomena.The principle of this phenomenon was preliminarily explained by diffusion effects and contact potential barriers,which is inspiring for the application of devices in integrated circuits and the search for similar functional devices.Secondly,an explanation was given for the quantum conductance behavior of the device when a reverse scan voltage is applied.This provides an important basis for optimizing multi-level memory storage ultra-high-density circuit design and neural morphological computing applications.The principle of this phenomenon was preliminarily explained by diffusion effects,which is inspiring for the application of devices in integrated circuits and the search for similar functional devices.We used a core-shell model to explain the process of current-controlled negative differential resistance characteristics.The current-voltage characteristics generated by this model can be determined by feedback generated by current distribution and local Joule heating.2.An Ag/NiO/n+-Si resistive memory was prepared using the sol-gel method and its resistance switching characteristics,the effect of limiting current on the device,and the memory-forgetting-memory process of synapses were tested using pulse signals.Currently,most chips are integrated on silicon wafers,and the conductive filament mechanism of silver ion migration has been attempted in resistive memories of different materials.This mechanism has advantages such as fast migration rate and good cycling characteristics,which are more conducive to achieving fast devices.Therefore,this paper uses Ag as the upper electrode to prepare Ag/NiO/n+-Si devices by the sol-gel method.The device exhibits low dispersion and high stability in both high-resistance and low-resistance states during 200 complete scan cycles.At a measurement voltage of 0.4 V,it maintains for more than 104 s.The switch ratio is always around 105.Adjusting the limiting current will change the resistance of the device in the low-resistance state.The size of Vreset is also affected by the limiting current.A continuous adjustable conductivity phenomenon was demonstrated in Ag/NiO/n+-Si resistive random access memory(RRAM).Using a continuous square wave with a pulse width of 10 ms and an amplitude of+1.7 V applied to the device can gradually increase its conductivity,while applying a square wave of-1.7 V can linearly decrease its conductivity.By regarding the conductivity of the device as the weight of synapses,we simulated the memory-forgetting-rememory process of synapses and showed the utillzation potentiality of artificial synapses. |