Make photons interact on chip

Photonic quantum hardware has a weird problem. Light is excellent for carrying information, but photons usually pass through each other like ghosts. That is great for communications. It is terrible if you want a quantum processor, because useful quantum logic and many simulation tasks need particles to interact. The new result is that a team at the University of Copenhagen showed a programmable on-chip circuit where single photons can be made to influence one another directly. ### Why is that hard? Ordinary photonic chips are mostly linear. You can split light, delay it, interfere it, and recombine it with very high precision, but one photon does not naturally shove another photon around. That missing nonlinearity has been one of the central bottlenecks in photonic quantum technology, because without it you usually fall back on measurement-based tricks that only work probabilistically. ### So what changed here? The Copenhagen group built a multimode nonlinear photonic circuit that can be programmed at the single-photon level. They combined a programmable linear interferometer with a deterministic nonlinear element, so the same platform can set both ordinary phase operations and genuinely nonlinear ones instead of hardwiring one special-purpose device. The paper appeared in *Nature Communications* on December 11, 2025. ### What is the nonlinear element? It is a quantum dot embedded in a nanophotonic waveguide. Basically, the quantum dot acts like an artificial atom sitting in the path of the light. A lone photon interacts strongly with that emitter, while a different photon arriving in the right conditions can see a changed optical response. That is the mechanism that turns “photons as noninteracting messengers” into “photons as particles that can mediate logic.” ### Why does “programmable” matter so much? Because a lot of nonlinear photonics has been one-off. You build a device for one operation, test the physics, and move on. Here, the point is reconfigurability. The team says the circuit can be reprogrammed to realize different linear and nonlinear settings with high precision, which is much closer to how a useful processor or simulator has to behave in practice. ### What did they actually do with it? They used the platform to run a proof-of-concept quantum simulation of anharmonic molecular dynamics. That sounds niche, but it is a good stress test because anharmonic systems are exactly where simple, noninteracting models stop being enough. If your hardware can only do linear optics, those effects are awkward to represent. A controllable on-chip nonlinearity makes that kind of simulation much more natural. ### Is this replacing superconducting or trapped-ion systems? Not directly. It is more like photonics is picking up a missing capability. Superconducting circuits and trapped ions already have strong interactions, but they come with their own scaling and operating constraints. Photonic systems bring speed, low noise, and easy networking. The catch is that direct interactions have been the weak spot in a way that fits integrated chips. ### What is still missing? A lot of engineering. One programmable nonlinear element is not the same thing as a large fault-tolerant photonic quantum computer. The field still needs better fabrication yield, lower loss, tighter control of emitters, and ways to scale many interacting components on one platform. But this result matters because it shows the core ingredient is no longer just theoretical hand-waving — it can be built, tuned, and used on chip. ### Bottom line The advance is not “photons finally became matter.” It is simpler than that. Researchers found a practical way to bolt a controllable interaction onto a programmable photonic chip. That gives light-based quantum hardware a shot at doing more than elegant interference experiments — it can start tackling the messy, interaction-heavy problems that make quantum simulation interesting in the first place.