A new experiment uses superconducting qubits to prove that quantum mechanics violates a so-called local realism by allowing two objects to behave as one quantum system regardless of how large their separation is. The experiment isn’t the first to show that local realism isn’t the way the universe works — it’s not the first to do so with qubits.

But they are the first to separate the qubits by enough distance to ensure that light is not fast enough to travel between them while making measurements. It did this by cooling a 30-meter-long aluminum wire to a few millikelvins. Because qubits are so easy to control, the experiment provides a new precision for these types of measurements. And preparing the hardware may be essential for future quantum computing efforts.

## Getting real about realism

Albert Einstein was famously uncomfortable with some of the consequences of quantum entanglement. If quantum mechanics is correct, then a pair of entangled objects will behave as a single quantum system regardless of the distance between them. Changing the state of one must instantly change the state of the second, since the change seems to happen faster than the possibility of light traveling between the two objects. This is almost certainly wrong, Einstein argued.

Over the years, people have proposed different versions of so-called hidden variables—physical properties that are shared between objects, enabling entanglement-like behavior while keeping the information that dictates that behavior localized. Hidden variables maintain what is called a “local realism” but it turns out that they don’t actually describe our reality.

Physicist John Bell showed that all frameworks for local variables limit the degree to which the behavior of quantum objects can be related. But quantum mechanics expects the correlations to be even higher. By measuring the behavior of pairs of entangled particles, we can determine whether they violate Bell’s equations, and thus clearly prove that hidden variables do not explain their behavior.

Initial steps toward this demonstration were poor for hidden variables but allowed for loopholes – although Bell’s inequality was violated, it still could be that information travels between quantum objects at the speed of light. But over the past few decades, the loopholes have gradually closed and the Nobel Prizes have been handed out.

So why go back to experiments? Partly because qubits give us a great deal of control over the system, allowing us to quickly run a large number of experiments and investigate the behavior of this entanglement. And partly because it presents an interesting technical challenge. Superconducting qubits are controlled by microwave radiation, and their entanglement requires moving some very low-energy microwave photons between the two. And to do so without environmental noise spoiling everything is a serious challenge.

## Scary action at a distance of 30 metres

Violating Bell’s inequality is a relatively simple matter of repeatedly measuring entangled particles and showing that their states are correlated. If this correlation exceeds a critical value, then we know that hidden variables cannot explain this behaviour. And superconducting qubits, called translats, are made so that the measurement is trivial, accurate, and fast. So this part is simple.

Eliminating one of the major gaps in these measurements is where things get tricky. You need to show that correlation in measurements cannot be mediated by information traveling at the speed of light. Since measurements require so little time to occur, this means that you must separate the two qubits by enough distance to allow the measurement to be completed before the light travels between them. Based on how long the measurements take, the research team behind the new work, working at ETH Zürich, calculated that 30 meters would be sufficient.

While this is just down the hall in a different lab building, 30 meters is quite challenging because of the entanglement process, which involves the use of low-energy microwave photons, which can easily be lost in a sea of environmental noise. In practical terms, this means that anything attached to these photons must remain at the same millikelvin temperatures as the qubits themselves. Therefore, the 30-meter-long aluminum wire that serves as the waveguide for microwaves must be cooled to a fraction of a degree above absolute zero.

In practice, this meant giving the entire assembly built to keep the wire’s access cool to liquid helium cooling systems containing qubits at each end — and building a separate cooling system at the center point of the 30-meter tube. The system also needed flexible internal connections and external supports because the whole thing contracts exponentially as it cools.

However, it all worked out admirably. Because of the performance of qubits, researchers can perform more than a million individual experiments in just 20 minutes. The resulting correlations ended up being above the limit set by Bell’s equations by a staggering 22 standard deviations. In various words, the p-value of the result was less than 10^{-108}.

## Things to come?

The two main factors limiting system performance are errors in the qubits and the loss of photons used in their entanglement. The researchers think they can improve on both, potentially making qubits the most rigorous test for Bell inequalities. But the work may become more important because of how the qubits get entangled.

Everyone who works with superconducting qubits says that eventually we will need to combine thousands of them into a single quantum computer. Unfortunately, each of these qubits requires a lot of space on the chip, which means it’s difficult to make chips with more than a few hundred of them. So major players like Google and IBM are eventually planning to link multiple chips to a single computer (something startup Rigetti is already doing).

However, for tens of thousands of bits, we would definitely need so many chips that it would be hard to keep them all in one cooler bit. This means that we’ll eventually want to thread the chips into different cooling systems – exactly what’s shown here. So this is an important demonstration that we can, in fact, interconnect qubits across these types of systems.

Nature, 2023. DOI: 10.1038 / s41586-023-05885-0 (about DOIs).

“Infuriatingly humble analyst. Bacon maven. Proud food specialist. Certified reader. Avid writer. Zombie advocate. Incurable problem solver.”