Regensburg 2016 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
DS: Fachverband Dünne Schichten
DS 3: Phase Change / Resistive Switching
DS 3.4: Talk
Monday, March 7, 2016, 10:15–10:30, H8
Memristive Devices for Neuromorphic Systems — •Mirko Hansen1, Martin Ziegler1, Finn Zahari1, Lukas Kolberg1, Sven Dirkmann2, Thomas Mussenbrock2, and Hermann Kohlstedt1 — 1AG Nanoelektronik, Christian-Albrechts-Universität zu Kiel, Germany — 2Lehrstuhl für Theoretische Elektrotechnik, Ruhr-Universität Bochum, Germany
The intensified development of memristive devices for memory applications led to several other possible fields of operation. Among them, their use for neuromorphic systems is one of the most promising applications. While several neuron-based learning concepts have been successfully demonstrated using single devices, the realization of neuromorphic systems using many memristive devices remains challenging.
We present TiOx and NbxOy-based memristive devices [1] with different device properties. The devices are fabricated on 4 inch wafers using a four mask lithography process with etching and deposition steps to yield a large number of high quality devices. A wafer-scale electronic characterization of these devices allows obtaining device parameters from a large number of devices. These parameters are employed for a network simulation which is able to recognize patterns [2]. Based on simulations, essential device requirements and device parameters for neuromorphic systems will be discussed.
Financial support by the German Research Foundation through
FOR 2093 is gratefully acknowledged.
[1] M. Hansen et al., Scientific Reports, vol. 5, p. 13753 (2015)
[2] F. Zahari et al., AIMS Materials Science 2: 203-216 (2015)