QNR.20.002 – Materials and Approaches for Green ICT

Route: The quantum / nano-revolution

Cluster question: 114 How can we connect 'things' (hardware) all the time and everywhere and boost processing speed while using less energy?

The explosive growth of digital data use and storage has led to an enormous rise in energy consumption of our information technology (IT), which already stands at 7% of the world electricity consumption. With new IT technologies, such as Artificial Intelligence and Internet-of-Things with its billions of sensor nodes, this energy consumption is exponentially increasing and soon becoming unsustainable and at odds with Europe’s goals for energy efficiency and reduction of greenhouse gasses. We propose revolutionary different approaches, using smart and nanostructured quantum materials as well as novel architectures to pave the way for IT beyond Moore. We will address important research questions associated with combining methods and materials from photonics, spintronics and neuromorphics. Spintronics exploits concepts like dissipationless spin currents, the intrinsic non-volatility of magnetism, and energy efficient switching of magnetic memory by light, whereas photonics offers huge signal bandwidth and ultralow transport loss, which has revolutionized data transport. Inspired by the energy-efficient architecture of the brain, neuromorphics explores adaptable materials that demonstrate synaptic and neuron like behaviour, to create a truly new paradigm: materials that learn. We will develop techniques and engineer nanomaterials down to the atomic scale, combining spintronic and photonic functionalities within integrated photonic devices, and will design and explore neuromorphic architectures based on manipulating topologically protected spin textures – aiming for ultralow power approaches inspired by the brain. The next frontier in photonics is information processing of light with light, at the spatial footprint of a transistor, and at energies per bit approaching attojoules – the energy of one photon. This requires efficient interconversion of light and matter excitations in atomic scale junctions and materials, light control through spin, chirality, and topological protection. Finally, we aim at merging local processing and storage, using adaptable physical interactions that can implement learning algorithms towards materials that learn.

Keywords

green ict, neuromorphic, photonics, spintronics

Other organisations

Amolf, Eindhoven University of Technology (TU/e), HBO, IBM, Lionix, Rijksuniversiteit Groningen (RUG), Technische Universiteit Delft (TUD), Thales, TNO, TU, UT

Submitter

Organisation Radboud Universiteit Nijmegen (RU)
Name Prof. dr. T.H.M. (Theo) Rasing
E-mail th.rasing@science.ru.nl
Website ru.nl/ssi