Abstract:
Memristive devices have attracted great attention in recent years for their application potential in memory and neuromorphic technologies. The information in these devices is stored in form of different resistance states, offering binary, multibit or even analog data storage. These characteristics combined with their non-volatility make memristive devices attractive candidates for embedded applications for the internet of things or as synapses in artificial neural networks.
This tutorial focuses on the most common class of memristive devices: redox-based resistive switching devices (ReRAMs). It consists of four parts: First, the basic switching behavior of these devices will be reviewed. Second, the current understanding of the physical processes will be explained. Based on the device physics, compact models will be derived and the open issues will be discussed. In particular, the impact of the device physics on computation-in-memory applications will be highlighted. Finally, the tutorial comprises an on-site training on different compact models, which will be made available for the attendees in MATLAB and Verilog-A.