Litcius/Paper detail

CMOS-compatible self-aligned 3D memristive elements for reservoir computing systems

Aleksandra Koroleva, Dmitry S. Kuzmichev, Maxim G. Kozodaev, Ivan V. Zabrosaev, Е. В. Коростылев, Andrey M. Markeev

2023Applied Physics Letters23 citationsDOI

Abstract

Neuromorphic capabilities of a self-aligned complementary metal-oxide-semiconductor compatible W/WOx/HfO2/Ru cell in a 3D vertical memristive structure were investigated. We show that the device exhibits nonfilamentary forming-free multilevel resistive switching with gradual resistance change. In addition, the poor retention of a low resistance state allows integration of these structures in architectures that require short-term memory characteristics such as reservoir computing systems. The ability of the device to rely on the temporal sequence of the stream was tested with the digit recognition task. Since a WOx layer was obtained by thermal oxidization and HfO2 and Ru layers were grown by atomic layer deposition methods, the device is suitable for high-density systems with high connectivity within a neural network.

Topics & Concepts

Neuromorphic engineeringMaterials scienceAtomic layer depositionReservoir computingNon-volatile memoryLayer (electronics)CMOSOptoelectronicsResistive touchscreenComputer scienceResistive random-access memoryNanotechnologyDeposition (geology)High resistanceArtificial neural networkVoltageElectrical engineeringEngineeringArtificial intelligenceRecurrent neural networkPaleontologyBiologyAgronomyComputer visionSedimentAdvanced Memory and Neural ComputingNeural dynamics and brain functionNeural Networks and Reservoir Computing