MemrisTec
Tommaso Rizzi improves device models for new space applications to supply rural areas with internet by satellites based on radiation-hard electronics. These in-memory computing circuits consist of memristors (resistive RAM).
The chair of Computer Engineering is now lead by Dr. Marc Reichenbach, who substitutes Prof. Dr.-Ing. habil. Michael Hübner, the new Vice President for Research and Transfer at BTU.
Lautaro Petrauskas aims to protect harmless insects by replacing chemical pesticides that are not selective by large-area traps consisting of printable memcapacitive circuits with low-power consumption for pattern recognition in the MemrisTec project MemTrap.
As a measure to enhance the knowledge base, especially of junior scientists, and the interdisciplinary scientific discussion, MemrisTec established Paper Club. Here, enrolled scientists meet every two weeks in a virtual meeting and discuss on important papers and reviews. The MemrisTec Paper Club started today with an overview on emerging non-volatile memory technologies like PCM, STTRAM, RRAM, and FeFET.
Dr. Vasileios Ntinas aims to develop a hardware implementation of a memristor-based cellular nonlinear network (CNN) for ultra-fast image processing and in-sensor-computing in the MemrisTec project Mem²CNN.
Dr. Richard Schroedter aims to reduce the power consumption in computing with crossbar structures of memristive and memcapacitive devices in the MemrisTec project BioMCross.
Will adaptive “terminator” sensors be able to stop all those insect infestations that otherwise destroy entire harvests in the near future? Optoelectronics and AI experts at the TUD are at least on the right track: in the “MemTrap” (memory trap) project, they want to work together to build adaptive organic terminator chips that can distinguish “good” from “bad” insects and only catch the plagues. This was announced by Prof. Stefan Mannsfeld from the Centre for Advanced Electronics Dresden (cfaed). He is driving the project forward together with Prof. Frank Ellinger and Dr. Bahman K. Boroujeni from the Chair of Circuit Technology and Network Theory.
The researchers want to link neuromorphic adaptive circuits with shape recognition sensors. These chips will be set up in a similar way to fly traps. If an insect comes flying along, the intelligent trap recognises by its outline whether it is a dangerous Gracillariidae or an busy imme, for example – and then eliminates the “bad guys” and keeps the “good guys” flying. The latter is particularly important, as the number of insects that spread pollen and thus secure harvests decreases.
Such clever insect traps would have to be adaptive and efficient on the one hand, and very cheap and space-saving on the other. They should also only consume a few nanowatts of electrical energy so that they can establish themselves in orchards or agricultural fields. This is made possible by an innovation from the TUD: photonics experts from the cfaed and the “Dresden Integrated Center for Applied Physics and Photonic Materials” (IAPP) have developed novel organic light storage devices in recent years. These “PinMOS” memories are a combination of organic light-emitting diodes (OLEDs) and capacitors. Unlike classic computer chips, they can remember not only zeros and ones, but many different digits per cell. And this data can be read and written both by current and by light. In addition, circuits with such organic memory cells can also remember and then recognise patterns – for example, the different shapes of bees and moths.
This article was published in the Dresdner Universitätsjournal 07/2021 of 20 April 2021. The complete issue can be downloaded free of charge from the UJ’s online presence at https://tu-dresden.de/uj or here in pdf format.
Article by Heiko Weckbrodt
On 5 March 2021, the virtual kick-off workshop of the priority programme of the German Research Foundation (DFG) MemrisTec took place. Almost 70 scientists from Germany as well as guests from England and the United States came together to explore the many possible applications of memristors.
Memristors are nanoelectric components capable of storage and computation. They are so small that many of them can be placed very densely on a chip. In addition, memristors learn and process information. Because of these properties, they enable a completely new and much more powerful type of electronic circuit than we know today. They thus open up a multitude of applications that have not yet been sufficiently investigated. This is precisely the task of the DFG priority programme “MemrisTec: Memristive Devices for Intelligent Technical Systems”. Under the direction of Prof. Dr. Ronald Tetzlaff from the Chair of Fundamentals of Electrical Engineering, the interdisciplinary, international research team is striving to fathom the scientific fundamentals of the memristor in order to make their industrial application possible.
The cooperation within the framework of MemrisTec now got underway with the virtual workshop conducted in English. After words of welcome from the MemrisTec coordinators, Prof. Tetzlaff and Christian Bruchatz, and the representative of the DFG, Dr. Damian Dudek, the keynote speaker Prof. Dr. Themis Prodromakis from the University of Southampton presented the variety of applications of memristors. Thanks to numerous questions, his presentation “Emerging Applications of Memristor Technologies: From Bio-Interfaces To Radiation-Hard Electronics” formed the introduction to the discussion and the subsequent lectures. Subsequent presentations provided comprehensive insights into selected potential application areas of memristors. A special focus was on the highly efficient, robust and faster memory technologies that can be developed thanks to memristors and the use of memristors in neuromorphic electronic systems. With memristors, the challenges of the Internet of Things could be better mastered and biologically-inspired artificial brains (computers that can “think” and “learn”) could be developed. During the workshop, these possible applications were presented in detail.
The virtual format also offered the opportunity for dialogue and the development of ideas. In a digital World Café, participants were able to move between six parallel tables and jointly develop concepts for shaping interdisciplinary and cross-location cooperation over the next six years. These also include the promotion of young scientists, equal opportunities and strengthened international cooperation.
The workshop culminated with two more keynote lectures, which took place at the end of the day due to the time difference with the United States. Prof. R. Stanley Williams from Texas A&M University explained that it is impossible to measure isolated locally active memristors directly and that many previous measurements carried out worldwide would be subject to interference. Furthermore, a fundamental theory for the thermal behaviour of memristors is still lacking, since understanding the thermal circuit is as crucial to its description as that of the electrical circuit.
The evening ended with Prof. Dr. Qiangfei Xia from the University of Massachusetts at Amsherst, who presented, among other things, the expected resource savings in matrix multiplications in neural networks using memristors.
The event laid the foundation for further cooperation.
Article from Anna Fejdasz
The Deutsche Forschungsgemeinschaft (German Research Foundation) (DFG) establishes 14 new priority programs for 2020. 50 initiatives applied for the 6 year long research funding. Two of these are from program directors at the TU Dresden. Besides Thomas Heine, professor for theoretical chemistry, Ronald Tetzlaff, professor for Fundamentals of Electrical Engineering, convinced the DFG with his research.
Starting in 2020, researchers may write proposals for subprojects to be a part of these research projects. The goal of these interdisciplinary priority programs is foundational research of very current topics or in just developing new areas of research.
The 14 new programs will receive for the first three years in total about 85 million Euro. Additionally, they will receive a 22 % fixed rate for indirect costs resulting from the projects. The DFG is currently funding 97 priority programs.