Successful half-time milestone for the NEUROTEC project reached after one year
The successful acceptance of the central milestone by the project officer from VDI/VDE/IT on November 19, 2020 marks the first successful stage on the way to the electronics of the future for the BMBF-funded project NEUROTEC. The alliance of Forschungszentrum Jülich and RWTH Aachen University envisions future computers for artificial intelligence (AI) made in Germany and the EU and aims to contribute to Europe's technological sovereignty.
This will involve the use of new types of hardware that are more closely modelled on the brain. This will require several new concepts for making computer architectures function more efficiently when it comes to AI tasks, such as pattern recognition or autonomous driving. A whole series of new technological steps are needed, ranging from the model-based design of circuits to the production of components and their characterization.
The computer chips are to be neuro-inspired or even neuromorphic, i.e. they are to imitate the functionality of the brain in hardware. The long road towards this goal leads deep into material and component development and is inspired by basic research on brain function at FZ Jülich. Coating systems, measurement technology and software from regional manufacturers are used for the technology. For the milestone, new systems from local high-tech suppliers have been set up and are soon ready for operation. The structural change in the Rhine coal mining area is thus being specifically promoted.
A specialty of the alliance of Forschungszentrum Jülich and Aachen RWTH University are very small electronic components that can be used for neuromorphic computing and that are integrable with conventional microelectronics. In particular, the name of the devices is Memristor (memory + resistor). They store information in their resistance value even after the current has been switched off. The resistance can be adjusted to store intermediate values, i.e. not only digital ones and zeros. Placed between two conductors, the memristive components determine how much current flows. They can then be used to build structures that multiply and add up values and to simulate synapses, which weigh the connections between neurons differently, similar to the brain. The calculation is done directly in memory. Designed as a neuronal network, future computers learn in hardware just like the brain does: associatively, plastically and randomly distributed. As the simulations in the project show, this neuromorphic AI electronics will become very energy efficient.
The Chair of Electronic Devices (ELD) conducts research with memristors based on two-dimensional (2D) materials, an innovative material system for the realization of ultra-thin devices composed of single (or only a few) atomic layers. ELD investigates corresponding manufacturing processes and is responsible for the characterization of memristive devices with the aim of describing their behavior and operation principle. First results were published Open Access in a recent paper titled "Nonvolatile Resistive Switching in Nanocrystalline Molybdenum Disulfide with Ion-Based Plasticity"1 in the journal Advanced Electronic Materials. In addition to the investigation of individual components, integration concepts for 2D materials are also developed by ELD within the framework of NEUROTEC.
In NEUROTEC I, initial concepts are now being tested, while a new proposal, NEUROTEC II aims for demonstrators in laboratory samples. NEUROTEC I and the planned NEUROTEC II are ideal bridge projects to develop the excellent basic research in the Aachen/Jülich area into a sustainable structural change. Both projects are strategically embedded in a series of current research activities that pursue the common goal of establishing the greater Aachen/Jülich region as a world-leading research and development location and, in the long term, also as a production location for neuromorphic technologies. To this end, the results achieved in NEUROTEC strengthen also the NeuroSys Clusters4Future Initiative, which brings together interdisciplinary techno-socio-economic players from RWTH Aachen University, Forschungszentrum Jülich, the NRW State Institute AMO GmbH and regional technology companies as a regional innovation network. The NeuroSys initiative is coordinated by Prof. Max Lemme and was successfully evaluated for funding of the so-called conception phase in the context of the BMBF's call for proposals for future clusters. It is currently evaluated in the final round of the selection process with 15 competitors.
Through all these activities, regional companies are strengthened in their innovation power. This relevance for structural change was recently confirmed by the award of the 3rd star for NEUOROTEC II by the Future Agency Rheinisches Revier in the immediate programPLUS. This paves the way for the application for five years of further funding by the BMBF under the Coal Region Investment Act.
Funding Code: 16ES1133K, 16ES1134