Chair of Nanoelectronics

Memristive Devices and neuromorphic circuits

Andreas Bourani - Nur in meinem Kopf

Abstract 

Memristische Bauteile und neurale Systeme

Memristive devices are two terminal variable resistors exhibiting a (resistive) memory effect. Such devices offer the exciting opportunity to develop a new circuit architecture which might overcome the current restrictions of digital computers when it comes to cognitive tasks. The aim of our research is to spotlight benefits of memristive devices in artificial neuronal circuits with the purposes: 

  • Explore new physical concepts for devices to emulate neuronal functionality 
  • Memristive device fabrication 
  • Adaptation of biological concepts for non-volatile memories and circuit engineering with memristive devices In general our research interest belongs to subject of bio-inspired computing. 

Based on fundamental ingredients of data processing in synapses and neurons, such as spike signals (action potentials) and the axon-hillock threshold logic, the static and dynamic features of neuromorphic circuits comprising memresistive devices is try to be explored and memristive devices are developed. 

State of the Art 

An emerging task in the field of neuromorphic engineering is to mimic neural path ways via elegant technological approaches to close the gap between biological and digital computing.  While in digital computing the von Neumann architecture induces a strict separation of serial digital processing and storage, data processing and storage are inseparably linked in the human brain. In nature, memory and learning alter the function and structure of neurons and their interconnection strength. Nowadays the concepts of biological neural networks are dominated by the Neuroinformatic and Silicon based analogue VLSI circuits. Nonetheless, recently an important feature of neuronal signal processing, the spike time dependent plasticity (STDP), was demonstrated in memristive devices. Moreover the principle of associative learning (Palvov´s dog) and anticipation have been recently realised electronically. 

In order to achieve non-biological systems acting as basic building blocks of brain new physical concepts such as ferroelectrics, phase change in higher chalcogenides, and valence-change in transition metal oxides has been recently employed in the field of neuromorphic engineering. Furthermore, a memristive operation mode of floating gate transistors (MemFlash) has been recently presented, which can be considered as a potential substitute for any memristive device and is basically compatible with current Si-fabrication technology. All this different physical device concepts are united by the common memristive behaviour. In particular, they are electrical resistance switches that can retain a state of internal resistance based on the history of applied voltage and current. 

Publications

  • Martin Ziegler, Karlheinz Ochs, Mirko Hansen, and Hermann Kohlstedt, Appl. Phys. A,  2013, 10.1007/s00339-013-7615-5.
  • M. Ziegler, R. Soni, T. Patelczyk, M. Ignatov, T. Bartsch, P. Meuffels, and H. Kohlstedt, Adv. Funct. Mater., 2012, 22, 2744.
  • M. Ziegler, M. Oberländer, D. Schroeder, W. H. Krautschneider, and H. Kohlstedt, "Memristive Operation Mode of Floating Gate Transistors: A two-terminal MemFlash-Cell", Appl Phys. Lett. 101, 263504 (2012).
  • M. Ziegler, H. Kohlstedt: Mimic synaptic behavior with a single floating gate transistor: A MemFlash synapse, J. Appl. Phys. Volume:114 Issue:19 (2013).

 

Review Paper

  • Doo Seok Jeong, Inho Kim, Martin Ziegler, and Hermann Kohlstedt, RSC Advances 2013, 3, 3169.