Research On Switching Mechanism And Physical Model Of Resistive Switching Memory

With the rapid development and popularization of new information technologies such as mobile intelligent terminals, cloud computing, big data etc., the demands for nonvolatile memory (NVM) are growing dramatically in recent years. Flash memory dominates the NVM market and is the fastest-growing segment in today’s solid-state memories. However, during the advancement of semiconductor fabrication technology, traditional Flash memory has faced formidable challenges in scaling-down, high storage reliability and so on. Emerging memory technologies may bring enormous opportunities for the development of information technologies in the future. Among various emerging concepts, resistive random access memory (RRAM) stands out in the past a few years for its superior scalability, low programming current, fast switching speed and good retention characteristics. It is selected by ITRS as one of the highest priority emerging technologies worth developing for future NVM application. However, the ambiguous resistive switching (RS) mechanism strongly impedes RRAM’s innovation, and high density 3-dimensional (3D) integration of RRAM is also a burning issue up to date.To address above scientific issues of RRAM, this dissertation has focused on detailed investigations of revealing the underiying physical nature of RS phenomena, modeling and simulation on both thermal and electrical behaviors during RS process of 3D RRAM crossbar array, respectively. Main works and innovative results of the thesis include:1) A microcosmic physical mechanism for forming behavior in oxygen vacancy based RRAM (Valence Change Memory, VCM) is proposed, in which the dependence of forming time on voltage is attributed to the combined effects of both generation and migration effects of oxygen vacancy. When migration rate of oxygen vacancy is relatively low, forming time decreases exponentially with the increase of pulse voltage amplitude, and the dependence will deviate from exponential relationship when migration rate is relatively high.2) Key microcosmic processes such as generation, reduction, adsorption and transport of metal ion in electrochemical metallization cell (ECM) are modeled with Monte Carlo method. Related physical effects of device performance are quantified, such as device current, potential distribution, morphology evolution of conductive path inside the device and so on. This model enables the investigations of material/fabrication/measurement etc. on RRAM device parameters, and visualize the nanoscale morphology evolutions of conductive filament (CF). Based on the stochastic model, dynamic growth process of CF in ECM devices are simulated, the simulated I-V characteristics and CF morphology evolution agree well with the experimental result. Different migration rate of metal ion in the electrolyte layer may result in opposite growth directions and different morphologies of CF.3) Theoretical investigation on Joule heating effect during reset process in a single RRAM device is performed. It is found that CF rupture occurs at transient thermal state and its location is not the hottest point, this conclusion is different with previous point views. In addition, the position with highest electric field value changes with time dynamically. The above conclusions established the foundation for the further research of thermal effects in 3D RRAM crossbar array integration.4) Thermal effects in 3D RRAM crossbar array are investigated through physical modeling and numerical computation. It is revealed that the time system consumes to reach steady thermal state increases with the increase of array stack layer, and individual device models based on thermal steady state are not applicable in 3D arrays. Thermal crosstalk phenomena could deteriorate device retention performance and even lead to data storage state failure (from LRS to HRS) of the disturbed RRAM cell. Moreover, the endurance performance of 1D1R storage element would be degraded under thermal effect at small feature size. Feasible methods to alleviate these thermal effects while further advancing the scaling potential are provided and verified by numerical simulation.The research achievements in this work would not only help to elucidate the physical mechanisms of RS phenomena, but also provide valuable guidelines for optimizing device performance and boosting the development of 3D integration technologies of RRAM devices.