Researchers at the National Institute of Standards and Technology (NIST) have developed an optical switch that is capable of rerouting light between computer chips within 20 billionths of a second. The new device is faster than any similar devices, and it could be integrated into low-cost silicon chips due to its low voltages. When it redirects light, the chip suffers very low signal loss.
The new chip will have big implications for computing, and it will help develop a computer that processes information using light rather than electricity. There are several advantages to using photons in order to transport data including faster travel and energy efficiency. With the use of electricity, computer components are heated up which wastes energy, and it limits computer performance.
The newly developed switch uses nanometer-scale gold and silicon optical, electrical and mechanical components. These are all densely packed, and they send light into and out of a channel. This affects its speed and direction of travel.
The device was described by the NIST-led international team in Science.
According to co-author Christian Haffner of NIST, ETH Zurich and the University of Maryland, the switch has a lot of potential applications. It could be used in driverless vehicles to redirect light beams that scan a roadway in order to measure the distance to other vehicles and pedestrians. The switch could also be used within neural networks, utilizing more powerful light-based circuits rather than electricity-based ones.
One of the major benefits of the new switch is that it uses very little energy to redirect light signals, which could be extremely important in quantum computing. A quantum computer has a fragile relationship between pairs of subatomic particles, which processes data. Because of their fragile nature, a computer needs to operate at extremely low temperatures and low power so that particle pairs are not disturbed. Since the newly developed switch requires a lot less energy, it could prove to be an important aspect of quantum computing.
Challenging Long Held Beliefs
According to Haffner, along with his colleagues Vladimir Akysuk and Henri Lezec of NIST, the new findings contradict many long-held beliefs within the scientific community. Many researchers believe that these types of switches would not be practical due to their bulky size, and they would operate at high voltages that cause slow performance.
The setup includes a tube-shaped channel called a waveguide, and a light beam travels inside of it. There is an off-ramp where some of the light exits into a cavity that is a few nanometers away.
The switch also utilizes a thin gold membrane that is suspended a few tens of nanometers above a silicon disk, which has the cavity etched into it. When the light travels around, some of it leaks out and hits the membrane. This activity induces groups of electrons that are on the membrane’s surface to oscillate. The oscillations are called plasmons, and they are a mix between a light wave and an electron wave. The oscillating electrons have a shorter wavelength that allows researchers to manipulate the plasmons over nanoscale distances. All of this helps the optical switch remain extremely compact.
If the researchers change the gap between the silicon disk and the gold member by a few nanometers, the phase of the hybrid light wave is delayed or advanced. When the phase of the wave recombines with light traveling in the tube-shaped channel, the two beams cause the light to either be obstructed or continue in its original direction. This allows the light to be transferred to any other computer chips at will.
The team’s next steps involve shortening the distance between the silicon disk and the gold membrane in order to make the device smaller. This would help further reduce signal loss, making the switch even more useful to different industries.
- The Black Box Problem in LLMs: Challenges and Emerging Solutions
- Alex Ratner, CEO & Co-Founder of Snorkel AI – Interview Series
- Circleboom Review: The Best AI-Powered Social Media Tool?
- Stable Video Diffusion: Latent Video Diffusion Models to Large Datasets
- Donny White, CEO & Co-Founder of Satisfi Labs – Interview Series