Light particles have long ignored one another – now researchers can make them talk

Quantum technology promises computers and sensors capable of solving problems on which even the world’s fastest supercomputers must give up. Nevertheless, a fundamental obstacle at the heart of the technology has persisted.

The tiny particles of light – photons – are ideal carriers of quantum information. They move quickly, lose almost no information along the way, and are only weakly affected by their surroundings. Precisely for that reason, making them work together is notoriously difficult.

“Photons are fantastic at transporting quantum information,” says Kasper Hede Nielsen, a PhD student at the Niels Bohr Institute, University of Copenhagen, Denmark. “But they hardly talk to anything at all – not even to each other.”

In other words, in an ordinary photonic circuit, one photon cannot change what the next one does. They can interfere, but they do not “push” each other in a way that produces a stable quantum operation that can be repeated reliably.

Without that possibility, photonic quantum technology has lacked the ability to perform genuine calculations – stable, perhaps, but poor at computing. If photons cannot influence one another, quantum circuits cannot be built in which information is actually processed rather than merely passed on.

“You can think of it as us having lacked a coupling between photons,” says Kasper Hede Nielsen – like gears in a machine that never quite mesh. They spin perfectly well on their own, but without that final engagement, no power is transferred.

Researchers have now found a way around this impasse.

Light is brilliant at carrying information – but poor at computing

For decades, photons have been among the most attractive building blocks in quantum technology. Light already plays a central role in quantum communication – and in many proposed architectures for quantum computers. Nevertheless, the very properties that make photons excellent carriers of quantum information are also their weaknesses.

When photons move through an optical circuit, only a limited set of operations usually takes place: they are split, mixed or sent onwards. They do not influence one another along the way. Such processes are known as linear operations – pure “split, mix and forward” transformations.

Here is the defining feature: in a purely linear circuit, the system itself does not change its behaviour depending on how many photons pass through it.

If you send in two photons, they may interfere and create correlations in the output. But the circuit behaves exactly as it would if only one photon were present. Nothing inside the system responds to the presence of the second photon.

This means that the kind of controllable, deterministic interaction in which one photon alters what the next one does simply does not occur.

And this is a serious limitation. A quantum computer must do more than route signals. It must carry out computational steps in which particles affect one another in a controlled and repeatable way.

In linear photonic circuits, this never truly happens. Two photons can pass through the same device without ever “noticing” each other – like cars travelling in parallel lanes that never intersect.

“If you only have linear operations, you can only get so far,” says Kasper Hede Nielsen. “At some point, you hit a hard limit on what purely linear photonic circuits can do.”

When photons cannot push each other

If photons cannot influence one another directly, researchers have for years had to make light behave as if an interaction had taken place – even though the photons never actually touch.

A common workaround has been to use quantum measurements as a kind of trick. Instead of enabling two photons to interact directly, they are sent through an optical set-up in which, under certain conditions, a measurement makes it appear as though an interaction has occurred.

But this is where the central difficulty lies.

“In quantum mechanics, a measurement never succeeds with 100% probability,” explains Kasper Hede Nielsen.

In practical terms, this means every computational step carries a genuine risk of failure. The measurement may produce the wrong outcome – or, in the worst case, destroy the photons entirely.

And when this happens, there is no way forward. The entire calculation must be restarted from scratch.

The problem: each step can fail – and you are back to square one

The more calculation steps you try to put together, the worse the problem becomes: each new step increases the risk that the entire process will break down and have to be repeated.

“It is like a game of Ludo, in which you are sent back to the start every time you fail,” says Kasper Hede Nielsen. “That is precisely the situation we are trying to move away from.”

The shortcuts have made a great deal possible in the laboratory. But they have also set a clear limit on what photonic quantum systems can become in practice.

The field has therefore long faced a dilemma: either large, advanced photonic circuits in which photons can interfere – but in which genuine, controllable interaction is still missing – or strong nonlinearity that could not be controlled and programmed.

Clever tricks, yes, but no real conversation between the photons.

It is precisely this gap that the new work seeks to close – the missing link that enables photons to truly influence one another.

The trick is time: the same chip, many photons

The key to the breakthrough is not more components – but a smarter way of using them. Instead of building a large and complex circuit, the researchers send the photons in as a train of short light pulses. The same physical elements are reused again and again – simply at different times.

“The clever part is that we use time as an extra dimension,” explains Kasper Hede Nielsen. “The photons pass through the same structure at different times, so that the same quantum dot can affect many photons – one after the other.”

And it is precisely the quantum dot that provides the nonlinearity: a small “active” piece of material that can react differently depending on whether one or two photons arrive.

Each photon is assigned its own place in the sequence in the form of small time slots, known as time bins. These can be thought of as seats on a train: the photons are separated but travel through the same system. First, the time bins are sent through a programmable network of optical nodes that can mix and delay the light with great precision.

Crucially, the set-up is inherently stable and can run for long periods without constant fine-tuning.

“We are already familiar with the linear elements,” says Kasper Hede Nielsen. “The difference here is that we can control them extremely precisely – connect them directly to a nonlinearity we can control.”

In this way, time becomes not just something that passes but something that is actively used to build a larger quantum system without making it physically larger. It is this time-based architecture that enables advanced photonic circuits to be connected directly with genuine photon–photon interactions.

The meeting point: one quantum dot, many photons

The decisive moment in the experiment takes place in one specific location: where the photons encounter a single quantum dot – the element that gives the circuit the nonlinearity that linear photonic circuits lack.

After the first linear section, the light pulses are sent into a small nanophotonic chip, where they are forced to pass very close to an embedded quantum dot. The chip is designed so that the interaction between light and matter becomes extremely strong. In practice, this means that around nine out of 10 photons actually interact with the quantum dot.

That figure is crucial. If the photons simply rush past without “hitting” the dot, no real conversation can arise between them.

When the photons interact with the quantum dot, something happens that linear circuits can never produce: the first photon slightly alters the system. When the next photon arrives, it therefore encounters a system that is no longer exactly the same.

“The quantum dot acts as an active element,” explains Kasper Hede Nielsen. “It reacts differently depending on whether one or two photons arrive – so that one photon can actually be felt by the other.”

Instead of using the quantum dot as a light source – its most common role – the researchers turn the process on its head. Here, it is used as a meeting place where incoming photons can influence one another.

“We are actually using the same system that normally emits photons,” says Kasper Hede Nielsen. “Just in reverse.”

The photons pass through the same physical structure with small time delays. In this way, the same quantum dot can be reused again and again – as an active hub that causes many light particles to remain correlated rather than simply pass one another by.

This is where the light particles truly begin to interact.

Two buttons to turn: how strong – and when

The crucial point is not just that the photons can influence one another – but that the researchers can control both when and how strongly this happens.

The strength of the nonlinear interaction can be programmed in several ways. The researchers can adjust the quantum dot’s “tone” with an electric field – like tuning an instrument – so that it matches the colour of the photons precisely while the experiment is running.

In addition, they can control how short or long the light pulses are, determining how long the photons remain at the quantum dot and therefore how strongly they influence one another.

“This gives us two independent knobs to turn,” says Kasper Hede Nielsen. “We can adjust the interaction between the photons without disturbing the linear parts.”

This is crucial because strong nonlinearity in previous experiments often made the rest of the circuit unstable. Here, the researchers can turn the conversation between the photons up or down without losing control of how the light otherwise moves through the system.

After scattering from the quantum dot, the photons are recorded individually. The researchers can then directly observe whether they emerge independently or whether they remain correlated. This is the most tangible test of whether the light particles have actually influenced one another.

In short: the interaction is not only strong but controllable.

The proof in the data: photons fall into patterns

When the quantum dot is brought into resonance with the photons, something new happens: the light particles cease to behave as independent entities and begin to bunch together. Instead of appearing randomly, the photons begin to appear in patterns. They follow one another in time – as if one photon influences when the next one passes through.

The researchers can see this directly in the measurements as temporal coincidences: maps showing when two photons are registered relative to one another. When the circuit is linear, the maps are flat and random. When the nonlinear interaction is switched on, clear structures emerge.

“Without nonlinearity, the photons are statistically independent,” explains Kasper Hede Nielsen. “But as soon as we tune the quantum dot correctly, significant correlations arise.”

This is the most direct experimental evidence in the work: the photons respond to one another.

How far can you turn it up before it tips over?

When the photons begin to influence one another more strongly, a natural question arises: does the system still hold together – or does it fall apart?

The researchers observe that there is a practical balance. When the interaction is increased, some photons are more likely to be sent in the wrong direction. But this is not a fundamental limitation of the method – it is a technical issue that can be improved.

At the same time, the measurements show that the system produces a significant effect: a clear and measurable “twist” in the quantum signal that directly reflects how strongly the photons influence one another. The effect is large enough to be relevant for real quantum operations – not only in principle but in practice.

More importantly, the results closely follow the theory. The effect is not a random experimental artefact but something the researchers can predict and control.

“We see very good agreement between theory and experiment,” says Kasper Hede Nielsen. “This gives us confidence that we actually have control over the process.”

In short: the interaction between the photons can be made stronger – without the system losing its stability.

Why direct interaction outperforms measurement-based schemes

The crucial difference is simple: direct interactions succeed more often than the measurement-based approach.

In measurement-based photonic systems, each computational step depends on a measurement producing the “right” outcome. If it does not, the entire process must be repeated. The more steps you chain together, the greater the risk of being sent back to the beginning.

However, this reliance on chance disappears when photons can influence one another directly through quantum entanglement. Far more computational steps succeed on the first attempt, and the system can be scaled up without the explosive growth in resources that measurement-based methods require.

“Even in our current set-up, we perform better than the best measurement-based schemes,” says Kasper Hede Nielsen. “And in principle, fully deterministic operation can be achieved with known improvements.”

Test in practice: the water molecule’s “neat” and “crooked” vibrations

The difference becomes clear when the researchers use the circuit to simulate vibrations in a water molecule.

With purely linear operations, regular, “neat” oscillations can be mimicked – like a perfect spring. But real molecules act differently. They vibrate unevenly and irregularly, with small deviations that are crucial to their chemistry.

These irregularities require photons to genuinely influence one another. Without direct interaction between the light particles, they disappear – and with them, the correct physics.

“If you only use linear operations, you simply get the wrong physics,” says Kasper Hede Nielsen. “The important nonlinear effects are missing.”

When light goes from messenger to building block

With the new platform, light is changing its role in quantum technology. Photons are no longer just neutral messengers that carry quantum information – they are becoming active building blocks that perform computational steps themselves.

“This is not just an improvement on existing photonic circuits,” says Kasper Hede Nielsen. “It is a new functionality that fundamentally changes what you can do with light in quantum mechanical systems.”

The consequences are concrete. Today, many quantum operations require enormous additional resources: more photons, repeated attempts and complex control schemes, because one must rely on fortunate measurement outcomes. When photons can instead influence one another directly, a large part of that overhead disappears.

“If you create nonlinearity through measurements, the operation may only succeed half the time – or even less frequently,” says Kasper Hede Nielsen. “Direct interactions mean that far more steps succeed the first time.”

In short: when light not only carries information but also acts on it, quantum technology becomes simpler, more efficient – and far more realistic to scale.

Where it matters most: chemistry and molecules

One area in which the new platform makes the greatest difference is in simulating molecules and chemical processes – problems that are among the most difficult for classical computers to solve.

Molecules rarely act “neatly”. In simple models, atoms oscillate like perfect springs, but in real molecules the motions are skewed, coupled and full of small deviations. It is precisely these non-neat vibrations that govern chemical reactions, energy transfer and function.

The linear parts of a photonic circuit can mimic the regular motions. But as soon as the molecule is described more realistically, the methods fail: the crucial irregularities are missing.

Here, direct photon–photon interaction becomes essential.

As a demonstration, the researchers use the circuit to simulate vibrations in a water molecule. This sounds simple, but water is a classic example in which small deviations from the “neat” model are crucial. When the nonlinear part is switched on, the simulation follows the full dynamics. Without it, you obtain the wrong physics.

“If you only use linear operations, you miss the important nonlinear effects,” says Kasper Hede Nielsen. “Then you are describing a different – and simpler – system than the one nature actually gives us.”

The implications extend far beyond water. Many chemical reactions, biological processes and material properties are governed by these nonlinear motions. By making them accessible in a controllable photonic quantum system, the work opens up simulations that were previously out of reach.

A first step towards scalable quantum platforms

In the longer term, the researchers point to applications in precision measurements, quantum-inspired neural networks and other areas in which nonlinear quantum effects can provide access to new states and improved performance.

In addition, Kasper Hede Nielsen points to a more fundamental challenge in the field:

“Every time someone says they have achieved quantum advantage, the problem is that if this is really true, we cannot prove it,” he says. “It takes a quantum computer to show that they are right.”

That is precisely why flexible quantum platforms are important.

“When you can programme both linear and nonlinear operations, you gain an extremely powerful tool,” says Kasper Hede Nielsen. “Predicting all the applications is difficult – but that is often the case when something fundamentally new becomes possible.”

There are still technical challenges, including those related to loss and stability, but there are no fundamental barriers.

“We see this as a first, but crucial, step,” says Kasper Hede Nielsen. “It shows that direct, programmable nonlinearity in photonic circuits is not only possible – but practically useful.” This brings an entire class of quantum systems, which until now have been theoretically attractive but practically unusable, closer to reality.