For decades, the heartbeat of modern computing has been the electron. From the earliest vacuum tubes to the nanometer-scale transistors in today’s smartphones, we have relied on the movement of electricity to process information. But as we hit the physical limits of silicon and struggle with the massive energy demands of artificial intelligence, researchers are looking toward a faster, cooler alternative: light.
The goal is all-optical logic, a system where photons—rather than electrons—perform the calculations that power our software. While photonic devices promise speeds and energy efficiencies far beyond current electronics, they have historically faced a stubborn hurdle: the difficulty of manipulating light with light. To create a logical “switch,” you need a way to inform one beam of light to stop or start based on the presence of another, a process that usually requires bulky hardware or intense energy pulses that can damage the materials involved.
Now, an international research team led by the University of Ljubljana has developed a soft photonic switch that achieves this control using remarkably low light intensities. By utilizing “squishy” materials like liquid crystals and polymers, the team has demonstrated a method to steer light without altering the physical properties of the material, potentially opening the door to a new generation of flexible, energy-efficient computing.
The Nobel Inspiration: From Microscopy to Logic
The breakthrough began not in a computer lab, but during a physics conference in San Francisco. Igor Muševič, a professor of physics at the University of Ljubljana, was listening to a presentation by Stefan W. Hell regarding stimulated emission depletion (STED) microscopy.
Hell, who received the Nobel Prize in Chemistry in 2014, developed STED to bypass the diffraction limit of light, allowing scientists to see biological structures at a much smaller scale. The technique uses two lasers to create an incredibly precise, tiny beam. For Muševič, the realization was immediate: if the technique could employ one laser to “deplete” the emission of another to create a sharp image, it could be adapted to create a logical switch for computing.
This realization led to the creation of a device where the fate of a laser pulse—whether it exits the system or is extinguished—depends entirely on whether a second pulse is fired less than a nanosecond later. This binary “yes/no” capability is the fundamental requirement for any logic gate, the building blocks of all digital computation.
How the Soft Photonic Switch Works
Unlike traditional photonic chips made from rigid silicon, this device leverages soft matter. The core of the switch is a spherically shaped bead of liquid crystal, infused with a fluorescent dye. This bead is held in place by four cone-shaped polymer waveguides that act as “pipes” to guide light into and out of the crystal.
The process relies on a phenomenon known as whispering gallery mode resonance. When a laser pulse enters the liquid crystal, the photons don’t simply pass through. they reflect repeatedly off the inner surface of the sphere, circulating within the cavity. This keeps the light trapped long enough to excite the fluorescent dye molecules.
The “switching” happens when a second laser pulse, known as the STED beam, is introduced. If this second pulse—of a different color—enters the waveguide before the first pulse can exit, it triggers stimulated emission. This interaction causes the dye to release photons identical to the second pulse while simultaneously depleting the energy from the first. Essentially, the second beam “erases” the first, preventing it from being emitted. Because the outcome of the first pulse is controlled by the second, the team has successfully demonstrated light-by-light manipulation.
Efficiency and the Advantage of ‘Squishy’ Materials
One of the most significant aspects of this approach is its energy profile. Previous attempts to create photonic switches in soft matter typically required intense light fields to physically change the material’s index of refraction—essentially forcing the material to change its nature to bend the light. This process is energy-intensive and can be unstable.
The Ljubljana team’s method reduces the energy required by more than a factor of 100. Because the STED pulse circulates repeatedly within the liquid crystal bead, a single photon can deplete the energy of many dye molecules, creating a highly efficient amplification and depletion cycle.
Beyond energy, the move from hard silicon to soft polymers offers several manufacturing advantages. Silicon photonics requires high-temperature fabrication and rigid, precise etching. In contrast, soft matter devices can be manufactured at much lower temperatures, and the liquid crystal core can be inserted into the polymer structure in less than a second.
| Feature | Silicon Photonics | Soft Photonic Switch |
|---|---|---|
| Material | Rigid Crystalline Silicon | Liquid Crystals & Polymers |
| Manufacturing | High-temperature / Vacuum | Low-temperature / Rapid Insertion |
| Geometry | Fixed / Planar | Flexible / Variable Cavities |
| Energy Need | High for nonlinear effects | Low (STED-based depletion) |
The Road to Photonic Neural Networks
While the demonstration of a single switch is a milestone, the ultimate goal is to scale this into full photonic computing. Miha Ravnik, a theoretical physicist at the University of Ljubljana, notes that the ability to control when and in which direction light is generated allows for the creation of complex logical operations.
This has profound implications for the future of AI. Current neural networks rely on GPUs and TPUs that move massive amounts of data between memory and processors, generating immense heat and consuming vast amounts of electricity. A photonic neural network would theoretically process information at the speed of light with minimal thermal loss.
However, Ravnik is cautious about the timeline. He admits that this technology is not yet in a position to compete with current hardware implementations of neural networks. The challenge now lies in scaling these individual “beads” into integrated circuits that can handle billions of operations per second.
The next steps for the research involve experimenting with different cavity geometries and exploring how these soft switches can be integrated into larger, flexible arrays. As the team explores the “engineering space” provided by liquid crystals, the focus will shift toward maintaining signal integrity across multiple logic gates.
This report is for informational purposes and describes experimental physics research; it does not constitute technical advice for hardware implementation.
We want to hear from you. Do you think photonic computing will eventually replace silicon, or will it remain a niche tool for specialized AI tasks? Share your thoughts in the comments below.
