UMD-NIST Researcher Helps Develop Novel ‘Quantum Refrigerator’

To solve a math problem on an old-fashioned chalkboard, you want the board clear of half-erased markings so you have space to work. Quantum computers have a similar need for a clean workspace, and a team including scientists at the National Institute of Standards and Technology (NIST) and the University of Maryland have found an innovative and effective way to create and maintain it.

The research, a collaboration with physicists at Sweden’s Chalmers University of Technology, could address one of the main issues confronting quantum computer designers: how to keep bits in a superconducting quantum processor free of errors and ready to perform calculations. These “qubits” are notoriously sensitive to heat and radiation, which can spoil their calculations just as stray chalk marks might make a 1 look like a 7.

Erasing qubits after a calculation involves cooling them to a fraction of a degree above absolute zero and then keeping them there. The team’s method is not only more effective than other state-of-the-art methods for erasing the qubit chalkboard because of the lower temperatures it achieves, but it also works in a novel way—by powering the eraser with the heat that flows between two parts of this “quantum refrigerator.” And the approach could even prove itself more broadly useful.

“It could address one of the problems in quantum computer design, and it also shows that we can siphon heat from one part of the computer’s refrigerator and convert the heat into work,” said Nicole Yunger Halpern, a physicist at NIST and fellow in UMD’s Joint Center for Quantum Information and Computer Science. “It could introduce technological capabilities we haven’t even thought of yet.”

The team’s proof-of-principle demonstration of the method was published this week in the journal Nature Physics.

Although quantum computers are far from reaching maturity, they remain the object of intense research because they offer the potential to perform certain tasks that conventional computers cannot do easily, including simulating complex molecular structures that are important in drug design. These projected capabilities derive from a difference between qubits and the bits in a conventional computer: While a conventional bit can exist in two states, 1 or 0, a qubit can have both values simultaneously, nominally allowing a quantum computer to sift through vast numbers of potential solutions at once.

A promising way to make qubits is to build them from superconducting circuits, which is the process the team used in its study. The advantages of superconducting qubits include tunability, allowing researchers to change the properties of the qubits as desired. However, qubits—even those that superconduct—can develop errors very quickly, which can ruin calculations.

Erasing a superconducting qubit means resetting it to its lowest energy state, which has proved tricky. An effective way to reset the qubit would be to make it as cold as possible, down in the tens of millikelvins (mK), or thousandths of a degree above absolute zero. The best reset methods previously brought qubits to a range of 40–49 mK. While those numbers might sound good, they aren’t good enough, said co-author and quantum physicist Aamir Ali of Chalmers University of Technology, where the team’s experimental work was conducted, supervised by principal investigator Simone Gasparinetti.

“In a quantum computer, initial errors can compound as the calculation proceeds,” Ali said. “The more you can get rid of them at the outset, the more effort you will save later.”

The team’s method can cool the qubit to 22 mK, reducing the likelihood of initial errors causing trouble down the line.

“If you didn’t cool the qubit to that low a temperature, you wouldn’t be able to erase the board as thoroughly,” explained Yunger Halpern, who is also a senior investigator in UMD’s NSF Quantum Leap Challenge Institute for Robust Quantum Simulation.

The team has achieved these performance numbers using a “quantum refrigeration” technique that has never been harnessed in a practical machine before. A refrigerator cools objects by using some sort of energy to draw heat away from the fridge’s interior. In a conventional kitchen fridge, the energy source is electricity, but the quantum refrigerator would use heat from elsewhere in the computer to do the job.

The team’s fridge uses two other quantum bits as its components. One qubit, which would be connected to a warmer part of the computer, would serve as the energy supply. The second quantum bit would serve as a heat sink into which the computational qubit’s undesired extra heat could flow. In an actual quantum computer, if the computational qubit—the chalkboard—got too warm, the fridge’s first qubit would pump heat from the computational qubit into the heat sink, which would carry the heat away, returning the computational qubit to nearly its “ground state” of absolute zero and erasing the board.

The process works autonomously, requiring minimal external control or additional resources to maintain the computational qubit’s ability to calculate.

“It’s hard to manage errors in quantum computers right now,” Ali said. “Beginning closer to the ground state will compound into fewer errors you’d need to correct down the line, reducing errors before they occur.”

—This article was adapted from a story by Chad Boutin of NIST Communications