Unlocking scalable entanglement will enable next-generation quantum computing
For years, photonic quantum computing has been stuck at two modes. A team at the University of Central Florida has now demonstrated five.
In a paper published in Science in March 2026, researchers Andrea Blanco-Redondo, Javad Zakeri, and Armando Perez-Leija showed that entangled photons could be generated across five topological modes in a silicon photonic chip — a meaningful jump from the two-mode ceiling that has constrained the field. The result comes from a collaboration between UCF's Center for Research and Education in Optics and Lasers (CREOL) and Saint Louis University, funded by the National Science Foundation under its ExpandQISE program (award No. 2328993).
The work is an entangled photon source, not a quantum computer. That distinction matters. Photonic quantum systems need bright, stable sources of entangled photons to feed into computation circuits — and the quality of that source determines what you can do downstream. This paper is about that upstream component, and on those terms, the results are worth examining closely.
The team used topological photonics — a design approach where light is forced to travel along edges of a patterned structure, protected from the disorder that typically degrades quantum states in real fabricated devices. They built three array designs with four, five, and six waveguides per unit cell, producing three, four, and five entangled topological modes respectively. The chip generates photon pairs via degenerate four-wave mixing in silicon waveguides (nonlinear parameter gamma = 120 W^-1 m^-1), pumped at 1550nm with signal and idler outputs around 1545nm and 1555nm.
The practical headline is fabrication tolerance. State-of-the-art electron-beam lithography produces waveguide variations of roughly ±5nm between runs. The team fabricated four copies of each design and measured them against each other. The entanglement held — the Schmidt number stayed stable across devices, even with those inevitable nanoscale differences between chips. That's the part that matters for scaling: a fabrication process that can produce consistent quantum performance despite real-world manufacturing tolerances.
There are caveats worth naming. Fidelity drops slightly with higher mode count — the five-mode states are less robust than the three-mode ones, a known trade-off in high-dimensional entanglement. The paper measures coincidence counts accumulated over 300 seconds per data point, which is a characterization timescale, not a runtime for a practical device. The team demonstrated the effect in a lab setting; integration into a full quantum computation stack is a separate engineering challenge.
The group previously published related work in Nature Materials in 2025, establishing the platform for controlling topological properties in silicon photonics. This Science paper is the sequel that closes the loop on entanglement generation.
Photonic quantum computing has attracted serious investment from PsiQuantum, Xanadu, and a range of academic groups precisely because integrated photonics offers a path to room-temperature operation — unlike superconducting or trapped-ion qubits. But the field has had to grapple with the fact that entangling many modes in a chip while keeping them stable and fabrication-tolerant has been hard. This result doesn't solve that problem. It takes a specific piece of it — high-dimensional topological entanglement in silicon photonics — and shows the fab tolerance holds at the five-mode level. Whether that translates into practical computation circuits is a different question. But for the people building those circuits, knowing your photon source will survive normal manufacturing variations is not nothing.
Sources: