New horizons in the quest to power future computation are being reached, according to researchers who propose a bright idea: using light to power computers instead of electrons.
Since some of the earliest modern computers that were developed beginning in the 1940s, electrons have remained the propellant force behind the technology. However, researchers at the University of Pennsylvania now say that with the 21st century demands of computation in a range of modern applications, addressing the limitations of electron-based hardware is becoming more crucial than ever.
One of the key issues with electron-driven computation involves the limitations of modern computer chips. As these charged particles make their way through these delicate electronic components, the heat they produce is a significant waste of valuable energy.
Add to this the further problems posed by such energy waste in applications that involve artificial intelligence (AI), which has seen an explosion of use in recent years, and the necessity for more energy-efficient computation becomes all-too-apparent.
A Light-Based Approach
Now, a research effort at the University of Pennsylvania led by physicist Bo Zhen in the School of Arts & Sciences is taking a very different approach: the idea that light particles might present a feasible alternative to electron-based computation.
Li He, the first author of a new paper that recently appeared in Physical Review Letters describing the team’s work, says that photons are an ideal alternative, mainly because they are able to “carry information quickly over long distances with minimal loss, dominating communications technology.”
Li, a former postdoctoral researcher in the Zhen Lab, says that one primary reason for this involves the fact that photons are virtually massless, and are neutral particles possessing no charge. However, that isn’t to say that the neutrality of photons doesn’t present a few issues of its own.
“That neutrality means they barely interact with their environment,” Li recently said, “making them bad at the sort of signal-switching logic that computers depend on.” As a result, while light is an optimal medium for the transmission of information—in addition to offering a very efficient alternative to electron-based systems—photonic applications are limited when it comes to the kinds of switching operations in computing at which electrons can excel.
A New Approach
A solution arises with the special case of what are known as exciton-polaritons, which are a variety of particles that form when photons are linked to electrons within a semiconducting material possessing remarkably thin properties of no greater width than a single atom.
Under such conditions, light can interact more effectively for signal switching applications, enabling the kinds of computational tasks that normally only electrons can reliably perform—a crucially important feature when it comes to AI systems, which are often major consumers of energy.
In fact, there are already several light-based technologies in use with AI systems that enable high-speed calculations, although for them to function, they must perform decision-based functions and other nonlinear activation tasks, where the light-based signals are converted back into electrical ones.
The result of this conversion is that the entire process becomes far slower, and overall energy use is increased—effectively canceling out the benefit of current photonic systems and contributing to wasted energy.
The Pennsylvania team says that exciton-polaritons can help to overcome this, as evidenced by their tests, which demonstrated entirely light-based switching using atomically thin materials, resulting in an energy expense of a minuscule 4 quadrillionths of a joule of energy.
Next, the team says, the technology must be scaled for practical use. However, if this can be achieved, photonic chip processing could potentially revolutionize a range of technologies and consumer-level devices, such as cameras, AI-systems, and eventually also future chip technologies that may become the driving force behind quantum computing systems.
The team’s research was supported by the U.S. Office of Naval Research and the Sloan Foundation, and was detailed in a new paper, “Strongly Nonlinear Nanocavity Exciton Polaritons in Gate-Tunable Monolayer Semiconductors,” which appeared in Physical Review Letters.
Micah Hanks is the Editor-in-Chief and Co-Founder of The Debrief. A longtime reporter on science, defense, and technology with a focus on space and astronomy, he can be reached at micah@thedebrief.org. Follow him on X @MicahHanks, and at micahhanks.com.