Back
Technology

Penn Researchers Develop Excitonic Device for All-Optical Switching

View source

An all-optical switch, operating on a sliver of energy, could bypass a major bottleneck in photonic computing.

A team of researchers at the University of Pennsylvania has created a quasiparticle that combines the speed of light with the strong interactions required for computing.

The Breakthrough

The device utilizes an exciton-polariton—a hybrid particle formed by coupling photons with electrons in an atomically thin semiconductor. This hybrid nature allows light to interact strongly enough to enable all-optical signal switching.

Key Performance Metrics

  • Operates at approximately 4 quadrillionths of a joule.
  • This represents significantly less energy than required for traditional electronic switching.

The work was published in Physical Review Letters and led by Bo Zhen, Jin K. Lee Presidential Associate Professor in the Department of Physics and Astronomy.

The Problem with Current Computing

Electronic computing is hitting a physical wall. As chips become more complex, they face increasing limitations from heat generation and electrical resistance.

Photonic computing offers a solution—using photons instead of electrons for fast data transmission with minimal loss. However, photons interact with each other too weakly to perform the logic operations essential for computing.

How Exciton-Polaritons Solve This

By coupling light with matter inside an atomically thin semiconductor, researchers created a quasiparticle that retains the speed of light while gaining the strong interaction properties of matter. This makes all-optical switching possible.

Implications for AI and Computing

Photonic AI chips currently face a critical inefficiency. They must convert light signals into electronic signals for nonlinear activation steps—a conversion that reduces both speed and efficiency.

All-optical switching using exciton-polaritons could eliminate this conversion entirely. Potential applications include:

  • Improved performance in AI systems by removing the electronic bottleneck
  • Direct processing of light signals from cameras without conversion
  • Quantum computing capabilities integrated directly onto photonic chips

Funding Support

This research was supported by the US Office of Naval Research and the Sloan Foundation.