When a beam of light passes through a cloud of atoms, photons (particles of light) sometimes appear to spend a negative amount of time there, with light seeming to exit the cloud before it even enters. Now, physicists have confirmed this quantum quirk by asking the atoms themselves.
“This doesn’t mean that we’re on the verge of building a time machine or anything like that,” study co-author Howard Wiseman, a theoretical quantum physicist at Griffith University in Australia, told Live Science. “It can all be understood with standard physics, but it’s yet one more weird property of quantum physics that people hadn’t suspected.”
Photons that pass through an atomic cloud can be temporarily absorbed. They vanish as particles of light and reappear as atomic excitations — a kind of stored energy — before being reemitted. Some photons, called transmitted photons, make it through in roughly the same direction they entered Others scatter off in random directions.
You may like
Experiments dating back to 1993 had already hinted that transmitted photons tend to arrive at a detector before the center of their own pulse even enters the cloud. That implies a negative transit time.
But there was a problem with this setup: Photons at the front of a pulse may be more likely to make it through than photons at the back. If you look only at the ones that are transmitted, of course, they look early. But this left a door open for a simpler explanation.
“People were convincing themselves that this is not actually as crazy as it sounds,” Wiseman told Live Science.
Confirming the crazy
In a new paper published April 13 in the journal Physical Review Letters, physicists tried a different approach. Rather than watching when a photon arrived at a detector, they monitored whether the atoms were in an excited state while the photon was passing through.
When a photon is absorbed by an atom, it is stored as energy, causing the atom to enter what physicists call an excited state. The atom remains in this excited state until it reemits the photon. Therefore, measuring the duration of the atom’s excited state reveals how long the photon was absorbed by the atom.
The team measured this using a second beam of light, which picked up a tiny phase shift depending on the atoms’ excitation levels. The light beam acted as a live readout of what the atoms were experiencing from moment to moment.
This atomic readout confirmed the quantum craziness of the earlier experiments.
What to read next
“You get the same answer if you ask the atoms, ‘How long was the photon staying with you?’” Wiseman said. “They will also tell you an answer, which is a negative time.”
A million-test milestone
Getting that answer wasn’t easy, because measuring quantum systems disturbs them. In this case, it potentially prevents the photon from being absorbed at all. So the team used “weak measurements,” which are gentle but extremely noisy. Any single run of the experiment was swamped by noise — random fluctuations that made it impossible to tell signal from static in any individual measurement. Only after averaging roughly 1 million runs did a clear signal emerge. Across roughly seven sets of experimental parameters, the total data collection ran to approximately 70 hours.
“Even in this really simple thing — a photon interacting with atoms — people were already doing calculations on that almost 100 years ago,” Wiseman said. “Just the fact that it can still show surprises after all this time is interesting.”
The team’s next target is the photons that don’t make it through the cloud. Theory predicts that those scattered photons carry extra positive excitation time. That is enough to balance the negative time of the transmitted ones, keeping the overall average for the beam of light at zero or above. That prediction has never been tested.
Angulo, D., Thompson, K., Nixon, V., Jiao, A., Wiseman, H. M., & Steinberg, A. M. (2026). Experimental observation of negative weak values for the time atoms spend in the excited state as a photon is transmitted. Physical Review Letters, 136(15), 153601. https://doi.org/10.1103/gjfq-k9dv