Executive Summary : | The microelectronic industry is increasingly seeking devices that can enable multi-bit storage and have high packing density. Artificial intelligence (AI) and machine learning (ML) algorithms have addressed the limitations of the von-Neumann architecture, which transfers data between memory and processor. RRAM devices, with their gradual reset process, can efficiently implement AI/ML algorithms through in-memory computing. These devices also have the potential to store multiple bits and have extremely high packing density due to their 3D crossbar architecture. However, there is no consensus on the dominant electron transport mechanism for different materials and the impact of ambient temperature on RRAM devices. This project proposes a comprehensive simulation environment that accounts for all electron transport mechanisms in the Master equation formulation, reducing computational burden and allowing exploration of design space with different oxide and electrode materials. The ion and ion-vacancy dynamics will be simulated using kinetic Monte Carlo to capture device-to-device and cycle-to-cycle variability. The comprehensive simulation framework will be employed to understand the working and dominant transport in RRAM devices across a wide temperature range and different materials. The insights gained from this study will provide helpful guidelines for optimizing devices from multi-bit storage, in-memory computing, efficient implementation of AI/ML algorithms, and variability perspectives. |