Sound Logic

In the world of computing, more is never enough. No matter how much compute power can be engineered into a system, we will always find a way to quickly exhaust it. Scientific simulations, machine learning, drug discovery, and whatever else tomorrow brings, will always demand more power than we have available to us.

This has led researchers to explore completely different types of computing architectures than those that are traditionally used. Electrons zipping through silicon have served us well, but to support quantum computing and beyond, we will need technologies that are faster, smaller, and more efficient than what we have today. One potential option is light-based chips. While they are still an emerging technology, they have been shown to be capable of dramatically improving performance while slashing energy consumption.

To date, these chips have only been used for certain applications, however. So to meet our general-purpose computing needs, further advances will be necessary. Researchers at Penn State University and the University of Science and Technology of China think that turning to sound waves might get us closer to that goal. But to be of much practical use, the sound must be able to switch on and off very rapidly, which is something that existing technologies struggle with.

That is where phononic circuits come into play. Instead of electrons or photons, these devices manipulate phonons (quanta of sound waves) to perform operations. While the idea of using sound to process information might seem odd at first, phonons offer unique advantages. In particular, they can be guided in extremely small structures, potentially enabling the development of compact chips that operate at high frequencies.

The team recently produced compact phononic circuits capable of reliably handling sound at 1.5 GHz. This was made possible through the use of microscopic waveguides arranged in special topological patterns. Much like highways for phonons, these pathways allow sound to move smoothly around corners and past defects, avoiding the scattering that typically disrupts acoustic devices. This topological robustness makes the circuits not only smaller, but also more reliable than earlier approaches.

To demonstrate their system, the researchers used a custom-built scanning optical vibrometer to visualize the motion of phonons as they traveled along the waveguides. They confirmed that the sound waves propagated cleanly through the edge channels without interference. They then put the circuits through a Mach-Zehnder interferometer test and showed that the pathways could be reconfigured on demand. This rapid switching ability is essential for real-world applications.

The work is still in its early stages, but by pushing phononic circuits into the gigahertz realm, the researchers have cleared a major hurdle. If further developed, sound-based chips could one day help us to handle larger computational workloads. But chances are that we will still need more power.