Best Oral Presentation Award für Johannes Hellwig (PGI-7)
Auf der Memrisys-Konferenz in Turin wurde Johannes Hellwig (PGI-7) für seinen Vortrag Resolving the Physical Origin of LRS Relaxation in Valence Change Memory mit dem Best Oral Presentation Award ausgezeichnet. Johannes Hellwig ist Doktorand am PGI 7 und forscht unter anderem im Rahmen des Projektes „Memristive time-difference decoder“ im DFG-Schwerpunktprogramm 2262 „Memristec“ sowie im Strukturwandelprojekt Neurotec II. Seine primären Forschungsziele, die weitere Vertiefung der Valence Change Memory (VCM) Zellen, die Entwicklung von flüchtigen VCM-Zellen für neuromorphe Anwendungen sowie die Verbindung von elementarer Device-Physik und neuromorpher Anwendung, tragen zu beiden Projekten bei.
Resolving the Physical Origin of LRS Relaxation in Valence Change Memory
Johannes Hellwig*1, Carsten Funck1, Hugh Greatorex2,3, Elisabetta Chicca2,3, Regina Dittmann1
1 Peter-Grünberg-Institute (PGI 7), Forschungszentrum Jülich, Germany
2 Bio-Inspired Circuits and Systems (BICS) Lab, Zernike Institute for Advanced Materials, University of
Groningen, Netherlands
3 Groningen Cognitive Systems and Materials Center (CogniGron), University of Groningen, Netherlands
*e-mail: j.hellwig@fz-juelich.de
In Neuromorphic computing systems, e.g. event-based sensing circuits [1], volatile memristive devices may be utilized to determine the time correlation between inputs. Here, the time constant of the device determines in which time domain the system can operate. So far, in common CMOS-based event-based sensing circuits a capacitor limits the time domain to milliseconds. Showing decays in the second regime, a volatile memristive device may extend the field of operation. Promising candidates are valance change mechanism (VCM) devices, where time constants can be engineered [2]. The decay is not yet well understood, and a physical decay model has not yet been reported. However, incorporating these devices into a circuit, a model is crucial.
Here, we present an evaluation of the VCM-decay process of the low-resistive state (LRS) with respect to its time constants using a crystalline Pt/SrTiO3/Nb:SrTiO3 device as model-system. This model system is well understood, and its material parameters are well known [3], [4]. The decay in shows at least two time constants, a sharp drop in current in the millisecond-range moving into a long-term relaxation up to 50 s. By applying the Schottky-Read-Hall model and concepts of the Gibb’s free energy to our measurements, we demonstrate that the observed short and long time constants can best be attributed to ionic migration near the Pt/SrTiO3 and to the oxygen exchange at this interface. We propose an ionic model, which aims to clarify confusion around the relaxation mechanism, and may be used as a tool for developing guidelines and design rules for future volatile memristive technology. We demonstrate experimentally the possible application of SrTiO3-based devices in event-based sensing for different time domains.
References
[1] M. B. Milde et al. Neural Computation, 30, 2384-2417 (2018).
[2] Xiong et al. Advanced Functional Materials, 29, 1807316/1- (2019).
[3] D. Cooper et al. Advanced Materials, 29, 1700212 (2017).
[4] C. Baeumer et al. Nanoscale, 8, 13967-13975 (2016).