New materials for more powerful optical data transmission

Enormously growing amounts of data pose problems for information and communication technology. In particular, digitally training so-called large language models for AI applications is a mammoth computational task. The bottleneck here is the communication between thousands of processors in huge parallel computers. Optical transceivers play a central role here: They convert electrical information into optical signals, which can then be transmitted efficiently and quickly via a glass fiber or optical fiber. Until now, silicon components have generally been used for signal conversion in transceivers. However, this approach is increasingly reaching its limits because pure silicon components are too slow for the increasingly large amounts of data. In addition, there is high energy consumption of the existing transceivers, which leads to high CO₂-Supporting the output of AI models.

Better data transfer with lower energy consumption

The ATHENS project investigated new material systems and components for converting electronic signals into optical signals. “Our goal is to make transceivers not only more powerful, but also more efficient in order to enable higher data transmission rates with the same or even lower energy consumption than before,” explains Professor Christian Koos from the Institute of Photonics and Quantum Electronics and the Institute of Microstructure Technology KIT. “The funding from the ERC Synergy Grant now allows us to implement ATHENS comprehensively – from the selection of suitable materials and the simulation of organic molecules to a functioning transmission system in the laboratory.”

Combination of silicon with other materials

The four-person project team, which, in addition to Koos, includes Professor Stefan Bräse from the Institute of Organic Chemistry and the Institute of Biological and Chemical Systems at KIT, Professor Carsten Ronning from the Friedrich Schiller University Jena and Professor Tobias Kippenberg from the École Polytechnique Fédérale de Lausanne , follows a hybrid approach: the researchers combine silicon with other materials. “Silicon components are inexpensive and available in large quantities, but their optical properties are limited.” “We combine silicon with other material systems to compensate for this deficit and continue to exploit the advantages of silicon,” says Koos. On the one hand The team is testing the use of organic materials, including carbon-based compounds. “We can first simulate these molecules on the computer before we then produce the substances with the desired properties synthetically in the laboratory and then print them on silicon wafers,” explains Bräse. A second method is the combination of silicon photonic chips with other chips. Additional material platforms play a role here, such as so-called crystal-on-insulator platforms, with which a thin, single-crystalline layer of a material is transferred to an insulating carrier substrate and further processed there into optical components.

Funding strengthens KIT’s position in photonic technologies

“The rapid developments in artificial intelligence are a major challenge for information technology. “The aim here is to quickly find sustainable solutions,” says Professor Oliver Kraft, Vice President Research at KIT. “I am pleased that the European Research Council is funding ATHENS, a project at the interface between materials science and information technology. This research strengthens KIT’s outstanding position in photonic technologies, which is becoming even more important with the Karlsruhe Center for Optics and Photonics, or KCOP for short, which is currently being created.”

Also useful for quantum technology and medical technology

Hybrid material systems for transceivers could find application not only in AI models, but also in the field of quantum technology and medical technology, for example in sensors for wearable devices or in optical lab-on-chip applications for blood value analysis.

ERC Synergy Grant

The European Research Council supports promising research teams with Synergy Grants. The funded projects should only be possible through collaboration between the named scientists and should lead to discoveries at the interfaces between established disciplines and to significant progress at the frontiers of knowledge. A total of 548 applications were received for the 2024 call for proposals. The ERC selects 57 projects for a Synergy Grant. Germany is represented in 34 of the selected projects and is therefore by far the most strongly represented. The prestigious ERC Synergy Grant has now gone to KIT researchers for the fourth time.