Forschungszentrum Jülich scientists publish a guide for the design of memristor hardware

They are several times faster than flash memory and require much less energy: memristive memory cells could revolutionize the energy efficiency of neuromorphic computers. In these computers, modeled on the functioning of the human brain, the memristive cells function like artificial synapses. Many groups around the world are working on the use of corresponding neuromorphic circuits, but often with a lack of understanding of how they work and with faulty models. The Jülich researchers have now summarized the physical principles and models in a comprehensive review article in the renowned journal advances in physics.

Some tasks – such as pattern and language recognition – are performed very efficiently by a human brain, requiring only about one ten thousandth the energy of a conventional computer, says “von Neumann”. One of the reasons lies in the structural differences: in a von Neumann architecture, there is a clear separation between memory and processor, which requires constant movement of large amounts of data. It takes time and energy – the so-called von Neumann bottleneck. In the brain, the calculation operation is carried out directly in the memory of the data and the biological synapses ensure the tasks of memory and processor at the same time.

At Jülich, scientists have been working for more than 15 years on special data storage devices and components that may have properties similar to synapses in the human brain. So-called memristive memory devices, also called memristors, are considered to be extremely fast, energy efficient and can be very well miniaturized down to the order of a nanometer. The operation of memristive cells is based on a very specific effect: their electrical resistance is not constant, but can be modified and reset by applying an external voltage, theoretically continuously. The change in resistance is controlled by the movement of oxygen ions. If these come out of the semiconductive metal oxide layer, the material becomes more conductive and the electrical resistance drops. This resistance change can be used to store information.

The processes that can occur in cells are very complex and vary depending on the hardware system. Three researchers from the Jülich Institute Peter Grünberg — Prof. Regina Dittmann, Dr. Stephan Menzel and Prof. Rainer Waser — therefore compiled their research results in a detailed review article, “Nanoionic memristive phenomena in metal oxides: the valence change mechanism.” They explain in detail the various physical and chemical effects in memristors and shed light on the influence of these effects on the switching properties of memristive cells and their reliability.

“If you look at current research activities in the field of neuromorphic memristor circuits, they are often based on empirical approaches to material optimization,” said Rainer Waser, director of the Peter Grünberg Institute. “Our goal with our review article is to give researchers something to work with to enable insight-based optimization of materials.” The team of authors worked on the approximately 200-page article for ten years and naturally had to keep incorporating advances in knowledge.

“The analogous functioning of memristive cells necessary for their use as artificial synapses is not the normal case. Usually there are sudden jumps in resistance, generated by the mutual amplification of ion motion and Joule heat,” explains Regina Dittmann of the Peter Grünberg Institute. . “In our review article, we provide researchers with the necessary understanding of how to alter cell dynamics to enable an analog mode of operation.”

“You see over and over again groups simulating their memristor circuits with models that don’t account for the high dynamics of the cells at all. Those circuits will never work.” said Stephan Menzel, who leads modeling activities at the Peter Grünberg Institute and has developed powerful compact models that are now in the public domain ( “In our test article we provide the extremely useful basics for the correct use of our compact models.”

Neuromorphic Computing Roadmap

The “Roadmap for Neuromorphic Computing and Engineering”, published in May 2022, shows how neuromorphic computing can help reduce the enormous energy consumption of computing worldwide. In this document, researchers from the Peter Grünberg Institute (PGI-7), together with leading experts in the field, have compiled the various technological possibilities, computational approaches, learning algorithms and areas of application.

According to the study, applications in the field of artificial intelligence, such as pattern recognition or speech recognition, are likely to benefit in particular ways from the use of neuromorphic hardware. Indeed, they are based, much more than conventional numerical calculation operations, on the movement of large quantities of data. Memristive cells make it possible to process these gigantic sets of data directly in memory without transporting them back and forth between the processor and the memory. This could reduce the energy efficiency of artificial neural networks by several orders of magnitude.

Memristive cells can also be interconnected to form high-density arrays that allow neural networks to learn locally. This so-called edge computing thus moves the calculations from the data center to the factory, the vehicle or the homes of people in need of care. Thus, monitoring and controlling processes or initiating rescue measures can be done without sending data via a cloud. “This achieves two things at the same time: you save energy and, at the same time, personal data and security-relevant data remain on site,” says Professor Dittmann, who played a key role in the creation of the roadmap as an editor.

Comments are closed.