17 August 2026
Physicists at the University of Twente have improved the standard test for the quality of individual particles of light. By letting three photons interfere at the same time instead of two, they draw more information out of every measurement. Their experiment outperforms even a perfect, noise-free run of the old method. The work appeared in Physical Review Letters on 12 August.
Quantum technology based on light only works if the photons involved are as close to identical as possible. In practice they never quite are. Laboratories therefore measure in advance how well their photons match. Since 1987 that has been done with the Hong-Ou-Mandel experiment, in which two photons meet at a beam splitter. If the two are indistinguishable, they always leave the splitter together.
"Hong-Ou-Mandel is one of the first experiments you learn about in quantum optics," says Stefan van den Hoven, PhD candidate and first author of the study. "It is usually also the first benchmark that quantum researchers use for their experiments."
More information from the same measurement
Anyone building a photonic quantum computer has to check not one pair of photons but hundreds. The old test compares two photons at a time, so with ten photons there are already 45 pairs to work through. On top of that, a single measurement never gives an answer. A quantum experiment produces one outcome at a time, and only after thousands of repetitions does it become clear how often each outcome occurs. That distribution reveals how alike the photons are.
The Twente researchers placed a second beam splitter behind the first, so that a third photon can join in. That addition uses the very thing that gives photonic quantum technology its power: interference between photons. Each measurement therefore yields more information about how well the photons match. Roughly a fifth fewer repetitions are needed to reach the same precision.
To show that this is the best possible arrangement, the researchers turn to the Fisher information matrix, a measure from information science that captures how much a measurement tells you about an unknown quantity. For three photons their design is provably optimal. It can be extended to four or five photons, but whether it is still the best option at that point is not yet known.
A gamble you cannot lose
Most theoretical quantum advantages disappear in practice, because noise means particles can be lost along the way. With three photons, that risk is greater than with two. Here it works out differently. If the experiment does lose a photon, two remain and the setup automatically delivers the familiar two-photon measurement. The two tests sit side by side. Van den Hoven: "So it is a gamble you cannot lose."
For now the method matters most to quantum optics laboratories. In time it could serve as a calibration step, or as a built-in test module in a large-scale photonic quantum computer.
About the research
The article Quantum Advantage for Single-Photon State Characterization by Van den Hoven, Anguita, Marzban and Renema appeared in Physical Review Letters (137, 073604) on 12 August 2026. The research was carried out at the MESA+ Institute for Nanotechnology at the University of Twente. It was funded through PhotonDelta, a programme of the Dutch National Growth Fund, and by an NWO Vidi grant, "At the Quantum Edge" (VI.Vidi.223.075).
DOI: 10.1103/g28d-jzgj
