Computing

A superfluid helium qubit could make 100 times fewer errors, on paper

Key takeaways

  • University of Surrey researchers published the first reported design for a qubit based on superfluid helium-3.
  • Their calculations predict error rates around 100 times lower than conventional superconducting qubits.
  • No prototype exists yet, and building one is the team's next step.

Around 100 times fewer errors. That is what calculations predict for a new superfluid qubit design from the University of Surrey, built on superfluid helium-3 rather than the superconducting circuits most quantum computers use today. The study, published in npj Quantum Information, is the first reported design for a superfluid-based qubit, according to the university's release on Nanowerk.

The important word there is design, because nobody has built one yet. Every number in the story is a prediction from the maths, not a measurement from a chip.

What is a superfluid helium qubit?

A qubit is the basic unit of quantum information. Unlike an ordinary bit, which is either 0 or 1, a qubit can hold a blend of both states at once, and quantum algorithms exploit that blend. The catch is that the blend is fragile, and any disturbance from the surroundings can destroy it. Engineers call that damage an error.

Helium-3 is a rare, light form of helium. Cooled to extremely low temperatures, it becomes a superfluid, a liquid that flows without friction. The Surrey proposal uses that fluid as the working material of the qubit instead of a circuit made of superconducting metal.

Superfluid helium-3 only appears at a few thousandths of a degree above absolute zero, colder than deep space. That sounds like a drawback, but superconducting quantum computers already live in dilution refrigerators that run close to those conditions, so the cooling is familiar territory for the industry.

How does the SHOQ device work?

The proposed device is called the Superfluid Helium Oscillator Quantum (SHOQ) device. It holds quantum information in quantised oscillations of the superfluid inside a microfluidic chip. "Quantised" means the oscillations can only take certain fixed energy levels, which is what lets them act as the 0 and 1 of a qubit.

The appeal is that the helium carries no electric charge. Superconducting qubits are very sensitive to electromagnetic noise and stray charges, which scramble the information they hold and make error rates hard to control as machines grow. A charge-neutral fluid should simply ignore some of that noise.

That is the mechanism behind the headline figure. The team's calculations predict error rates around 100 times lower than conventional superconducting qubits, because a major source of disturbance has little to grab onto.

How does it compare with superconducting qubits?

The table below sets out the differences the study describes. The error figure for the superfluid design is a prediction only.

FeatureSuperconducting qubitProposed SHOQ qubit
Working materialSuperconducting circuitSuperfluid helium-3
Carries electric chargeYes, so sensitive to stray chargesNo, charge-neutral fluid
Sensitivity to electromagnetic noiseHighLower, according to the design
Predicted error rateBaselineAround 100 times lower
StatusBuilt and widely usedDesign only, no prototype yet
Likely roleGeneral-purpose qubitsQuantum memory, linked to existing hardware

Why do fewer quantum errors matter?

Errors are the bottleneck for useful quantum computing. Today's machines spend most of their physical qubits on error correction, which is why the gap between the two kinds of qubit matters so much; we explain it in logical vs physical qubits. A qubit that starts out 100 times cleaner would need far less of that overhead.

Error correction is also moving quickly on the software side. Our report on continuous recalibration of quantum processors shows one way teams are fighting noise in existing hardware, and the story of how quantum computers got more reliable covers recent progress with the circuits already in labs. A cleaner qubit would attack the same problem from the hardware end.

Why does electric charge cause errors?

A superconducting qubit is an electrical circuit, so anything that wobbles an electric field nearby can nudge its state. Stray charges trapped in materials, and background electromagnetic radiation, both do exactly that. Each nudge can flip a qubit or blur the phase relationship that quantum algorithms depend on.

Because helium-3 atoms are electrically neutral, those sources of trouble couple to the superfluid far less. The design does not remove every source of error, since heat and vibration still exist, but it takes a major one off the list, according to the Surrey calculations.

Is this meant to replace current quantum computers?

The team is not pitching it as a replacement. Lead author Dr Priya Sharma says "the maths tells us that it should work", and the paper shows how the device could connect to existing superconducting hardware.

One suggested use is quantum memory, holding information while other qubits do the calculating. A memory is judged mainly on how long it keeps data intact, so a qubit that ignores charge noise suits the job well. In that role the superfluid device would sit alongside the superconducting chips that dominate the field now, rather than competing with them head on.

Surrey worked with Professor Jens Koch at Northwestern University, one of the researchers behind the transmon, which has become the most widely used superconducting qubit design. That matters because the proposal comes partly from people who know the incumbent technology well.

What is a transmon, and why does it matter here?

The transmon is a superconducting qubit design that reduced sensitivity to charge noise compared with earlier circuits, and it is now the standard choice in many labs. Having one of its co-developers on the Surrey paper gives the comparison credibility, because the benchmark being beaten on paper is the one its own designer helped to build.

It also suggests why the team frames the device as a partner to current chips. A memory that connects to transmon-based processors could be adopted step by step, without asking the industry to abandon the hardware it has already invested in.

What happens next?

The next milestone is a prototype. The team is now working toward one, backed by a commercialisation fellowship awarded to Sharma, and the very low temperatures the device needs have already been reached in superfluid helium-3 experiments.

Building the device is a separate problem from calculating that it should work. Real chips pick up imperfections that a design on paper does not include, and only a working prototype can show whether the predicted error rate survives contact with the lab.

Until a real chip hits that predicted error rate, 100 times is a target rather than a result. Readers following the wider push for practical machines can also see how governments are treating the field in our piece on the White House quantum computer order.

More from Future Technology HQ