A New 1P1R Image Sensor with In-Memory Computing Properties Based on Silicon Nitride Devices
Nikolaos Vasileiadis, Vasileios Ntinas, Iosif-Angelos Fyrigos, Rafailia-Eleni Karamani, V. Ioannou-Sougleridis, P. Normand, Ioannis G. Karafyllidis, Georgios Ch. Sirakoulis, Panagiotis Dimitrakis
Abstract
Research progress in edge computing hardware, capable of demanding in-the-field processing tasks with simultaneous memory and low power properties, is leading the way towards a revolution in IoT hardware technology. Resistive random access memories (RRAM) are promising candidates for replacing current non-volatile memories and realize storage class memories, but also due to their memristive nature they are the perfect candidates for in-memory computing architectures. In this context, a CMOS compatible silicon nitride (SiN) device with memristive properties is presented accompanied by a data-fitted model extracted through analysis of measured resistance switching dynamics. Additionally, a new phototransistor-based image sensor architecture with integrated SiN memristor (1P1R) was presented. The in-memory computing capabilities of the 1P1R device were evaluated through SPICE-level circuit simulation with the previous presented device model. Finally, the fabrication aspects of the sensor are discussed.