Research focuses on the development and optimisation of memristors, novel components with variable resistances that can be adjusted and stored using electrical pulses. These memristors are embedded in integrated circuits to enable neuromorphic hardware systems with simultaneous computing and storage.

New 2D materials are being explored as potential active materials for memristive components due to their atomic thickness and miniaturization potential. Their chemical and structural properties, especially within the devices, can significantly influence device behavior under electrical bias conditions due to microscopic mechanisms. Project A2 thus aims to take the metrology of memristive materials to a new level, enabling the precise and local characterization of material properties and compositions.

Project B1 aims to realize neuromorphic hardware using integrated photonic circuits in which information is transmitted using light. Optical transmission systems enable extremely high data rates and a substantial reduction in latency during signal transmission. The neuromorphic hardware system will be demonstrated in a silicon photonics chiplet combined with high-frequency electronics. Additionally, the use of vertically emitting lasers (VCSELs) for co-packaged optics in AI centers will be investigated.

Project B2 focuses on a component that is essential for extremely fast and energy-efficient neuromorphic photonic computations: electro-optical modulators. To this end, we aim to research a novel graphene-based modulator whose dimensions can be minimized by maximizing light-matter interaction. The new modulators will be implemented in a so-called photonic tensor core for vector-matrix multiplication and used for initial demonstration experiments. Black Semiconductor, as an associated partner, is supporting the high-frequency characterization of the modulators and is interested in further exploiting these components.

The long-term vision of the NeuroSys cluster is the future technological independence of Germany and Europe in the field of hardware for artificial intelligence applications. Our goal is to lay the scientific, technological, and socio-economic foundations to attract the large-scale investments required to establish a production line for AI hardware in the Aachen region. With our goals, we pursue the vision of technology development committed to the European community of values. Project C, "Algorithm-Hardware Co-Design," plays an integrative role between the more technology-oriented projects A and B and the application- and transfer-oriented projects D and E.

Project D aims to efficiently implement machine learning applications on future neuromorphic hardware and ensure their optimal use from a user perspective. The focus is on persistent network architectures developed specifically for neuromorphic hardware, algorithmic approaches to complexity reduction, and practice-oriented software tools. Practical solutions will be demonstrated using a variety of data formats, for example, in medical systems for improving diagnostics and therapy or in speech technology applications. Technology transfer is ensured through close collaborations with companies that translate research results into practice and simultaneously incorporate application requirements into research to promote sustainable innovations.

Project E completes the seven interlocking projects from science, business, and society. The accompanying innovation processes and business model developments are analyzed to ensure long-term economic success and social and moral benefits. This type of strategy lays the foundation for subsequent transfer into business practice.