Resistive random access memory (RRAM) is considered as one of the most promising candidates as future embedded nonvolatile memory. Due to the explosive expansion of new markets such as internet of things (IoT), artificial intelligence (AI), smart car and virtual reality (VR) in the next decade, high density embedded memory for storage and calculation of big data in electronic terminal is necessary. The 3D Vertical structure and crossbar array (CBA) provides a cost-effective approach for high density integration. Unfortunately, when ReRAM devices are used in the 3D architecture, sneak-path currents flow through the unselected neighboring cells in the CBA, leading to a read disturb error and a reduction in device density.
To solve this problems, other group has introduced rectification diodes HfOx and a chalcogenide-based selector device in each cell and ReRAM with self-reacting characteristics of bi-layer structures using a switching layer (RS layer) and a barrier layer. We have studied selector devices that are ruptured spontaneously through unstable filament by Ag-doped into HfOx, ZrNx and SiOxNy-based materials. In self-rectifying ReRAM, we have researching rectification characteristics using bandgap and work function difference of each layer.