Editorial for the JECR special issue on resistive switching: Oxide materials, mechanisms, and devices

Abstract

Emulation of neural networks by redox-based Resistive Random Access Memories (ReRAMs) with components such as thin films of ceramic materials are considered by the technological roadmap (ITRS) as a promising concept for the next generation non-volatile memory storage and as an important key towards computation with neuromorphic algorithms. ReRAMs are regarded as conceptually new building units in modern nanoelectronics, finding application not only as a memory, but also as selectors, for logic operations and neuromorphic computing circuits beyond the von Neumann concept, being capable of bio-inspired cognitive functions, such as machine learning and pattern recognition. The information is saved in ReRAMs as particular resistances of the devices adjustable by voltage stimuli, where in the most simple case the high resistive state represents the Boolean 0 and the low resistive state – the Boolean 1. The devices show outstanding potential for scaling down to the atomic level, integration, low-power consumption, sub-nanosecond operation time range and digital and/or analog volatile and/or non-volatile information storage. In these devices, the switching relies on redox reactions and mixed ionic-electronic transport at the nanoscale, in cells/devices of the type MEM (metal-electrolyte-metal) or MIM (metal-insulator-metal) where oxides and higher chalcogenides are typically used as ion conducting materials. ReRAM devices are operated at extremely harsh conditions characterized by high electric fields (up to ~ 108 V/m) and extremely high current densities in the range above MA/cm2 and show perspective for future memory and neuromorphic computing devices.