cfaed Cluster of Excellence 2012-2019
Information processing is today still dominated by complementary metal oxide semiconductor (CMOS) technology. Since the 1960s, the advancements in electronics due to scaling of integration densities along Moore’s law have established a general expectation for very short innovation cycles with ever-new application possibilities. Indeed, the huge advancement of electronics over the past decades has been the driving force for innovation in various application fields and has significantly shaped the world we live in today. As CMOS technology is reaching atomic boundaries, it increasingly fails to deal with the negative impacts on device behavior due to shrinkage (short channel effects, leakage, etc.), and Moore’s law is projected to end. Thus, our cluster was motivated by the insight that long-term innovation in electronics cannot be based solely on higher planar integration densities. Instead, we aimed at new ways to address current and future challenges of electronic information processing systems as there are:
- Physical Size: Size does matter: small volumes are required for practicability and ease of use in many applications (especially in mobile devices). As the planar scaling ends, alternatives for decreasing size per functionality were investigated.
- Speed: For decades, transistor transit frequency has scaled with the inverse of the squared gate length until recently reaching saturation. Thus, new ways to increase circuit speeds were to be explored. This is especially important for high bandwidth communications.
- Energy Efficiency: For ecological and technical reasons, energy consumption and heat dissi-pation in silicon chips is highly problematic. Many large computing applications (still) have sig-nificant environmental impact, often unknown to the users. Hence, new materials and system architectures were researched to increase the energy efficiency of electronic processing while still meeting future application demands.
- New Functionality: To enter novel application domains, new integrated functions are required. New types of sensors (e.g., for biomolecules) could enable completely new applications and new processing paradigms could enable information processing not yet attainable.
- Self-Assembly/-Organization: From the materials and devices perspective (bottom-up fabri-cation) all the way to the software perspective (e.g., Self-* / Self-X), self-assembly (design time) and self-organization (run time) mechanisms were considered to handle the complex tuning of components to a desirable state without centralized control. This should enable cost effective and parallel synthesis from materials all the way to highly adaptive run time (mostly software) systems.
- Adaptivity: Still today, many solutions consist of flexible software running on fixed hardware. To optimize systems flexibly for efficient execution of applications, a new degree of adaptivity was investigated involving both hard- and software.
- Resilience: As materials and devices are driven to their fundamental limits of operation, transi-ent errors start occurring. Therefore, system designs solving the task of enabling reliable pro-cessing with unreliable components were investigated.
- Cost: Ground-breaking research seldom takes the cost of commercial production into account. However, the analysis of fundamental barriers for cost competitive fabrication possibilities pro-vided important insight into the future potential of new technologies.
In 2012, there was still room for improvement in CMOS technology and industry roadmaps reached out to 2022. The first decade of the new millennium had seen significant advances in new materials, which led to many promising discoveries. While materials research was continued, some discoveries had reached a point that warranted exploring device fabrication, circuits, and information processing systems for potential applications. As CMOS scaling was pro-jected to end soon after 2020, we assumed that industry would stop being preoccupied with ad-vancing CMOS and would eagerly look out for new ideas. Given these two developments, we be-lieved that university-based research had a unique opportunity to integrate discoveries on new materials and technological innovations with the potential for advancing electronic information pro-cessing beyond 2020. It was the vision of the Center for Advancing Electronics Dresden (cfaed) that future CMOS technology would be complemented with new technologies (augmented CMOS), resulting in heterogeneous architectures to form highly efficient information processing systems.