Silizium-Germanium-Chip bricht Bandbreitenrekord mit 500 Gbps

This function is critical in modern electronics, where systems need to handle massive amounts of data in real time.The team reports that the system can process more than 500 gigabits per second in a single channel using quadrature amplitude modulation.

In multi-channel setups, the data rate could exceed 100 terabits per second, a level relevant for long-distance communication networks.Faster data, lower energyThe new design uses silicon-germanium technology, which allows faster switching speeds while reducing energy consumption.

This combination is important for next-generation applications such as 5G and 6G networks, autonomous vehicles, and high-speed sensors.Silicon-based analog-to-digital converters already operate at extremely high speeds, but improving both bandwidth and sampling rate together has been a technical challenge.

The researchers focused on optimizing both

The researchers focused on optimizing both parameters to improve overall system performance.“Transceivers are ‘ambassadors’, so to speak, between analogue and digital.

They combine two functions: both sending digital data and receiving data from outside,” explained Maxim Weizel, a research associate involved in the project.Higher bandwidth allows more data to be transmitted in less time, which directly affects performance in servers, cloud systems, and data centers.

For example, network cards with higher bandwidth can significantly improve overall system efficiency.Pushing measurement limits furtherThe team also faced challenges in measuring performance at such high frequencies.

Even small errors can introduce phase

Even small errors can introduce phase noise or signal distortion, making accurate testing difficult.“We worked with extremely high frequencies, which in turn require extremely high precision,” said Weizel. “Even the smallest errors caused disruptive reflections or so-called phase noise.”To address this, the researchers relied on advanced simulations and high-performance computing resources to validate their design.

The chip’s performance was strong enough to push existing measurement systems to their limits.“Especially in the context of AI, high speed becomes a competitive advantage,” Weizel added, noting that large datasets and real-time communication demand faster processing.The development also highlights the growing role of advanced semiconductor materials in pushing computing limits.

Silicon-germanium combines the manufacturability of silicon with improved electronic performance, making it attractive for next-generation chips.

As demand for faster data processing rises, such hybrid materials could play a central role in scaling future communication and computing systems.The project involved collaboration between several institutions, including RWTH Aachen University, Karlsruhe Institute Technology, and DESY.The results are documented in the open-access book “Electronic-Photonic Integrated Systems Ultrafast Signal Processing.”