Short answer: Leakage is when a qubit leaves its two computational states for other energy levels, and in a 2023 Nature Physics paper the authors report removing it each cycle on a superconducting processor, with a tenfold drop in steady-state leakage of the data qubits. A 2023 Physical Review Letters paper reports an all-microwave leakage reduction unit with up to 99% efficacy in 220 ns. These are laboratory demonstrations on small codes, not a solved problem at scale.
Why leakage matters
The Nature Physics abstract describes leakage as a major challenge for quantum error correction. During an error-corrected algorithm, leakage builds over time and spreads through multi-qubit interactions. The authors say this produces correlated errors that degrade the exponential suppression of logical error with scale.
2023: leakage removed every cycle
In the Nature Physics paper, the authors demonstrate a distance-3 surface code and a distance-21 bit-flip code on a processor where leakage is removed from all qubits in each cycle. They report a tenfold reduction in the steady-state leakage population of the data qubits and an average leakage population below 1 x 10^-3 across the device. They say adding the removal process to a code circuit would prevent leakage from inducing correlated errors across cycles, and they conclude that leakage can be contained. That last statement is the authors' own assessment.
2023: an all-microwave leakage reduction unit
The Physical Review Letters paper realizes and extends an all-microwave leakage reduction unit for transmons in circuit QED, originally proposed by Battistel and coauthors. The authors report that it reduces leakage from the second and third excited transmon states with up to 99% efficacy in 220 ns, with minimal impact on the qubit subspace. As a first application to error correction, they show that several simultaneous units can reduce the error detection rate and hold leakage buildup within 1% in data and ancilla qubits over 50 cycles of a weight-2 stabilizer measurement.
A 2021 scheme
A 2021 PRX Quantum paper is titled "Hardware-Efficient Leakage-Reduction Scheme for Quantum Error Correction with Superconducting Transmon Qubits". Only its title was available when this post was written, so its results are not described here.
How the two results fit together
The two papers report different things. The Nature Physics work measures leakage population across a whole processor while running small codes, and the Physical Review Letters work characterizes one reduction unit and its effect on a short stabilizer measurement. Neither abstract compares the two methods directly, so it would be wrong to rank them. What they share is the aim stated in the first abstract: shorten the lifetime of leakage so it cannot spread and create correlated errors.
Limits
Both experiments use small codes or a short stabilizer measurement. Results are reported for the authors' own devices, and the abstracts do not claim performance on larger codes.
FAQ
What is leakage?
A qubit leaving its two computational states for other energy states, per the 2023 Nature Physics abstract.
Why is it a problem for error correction?
It builds up and spreads through multi-qubit interactions, causing correlated errors.
What did the Nature Physics paper show?
A distance-3 surface code and a distance-21 bit-flip code with leakage removed each cycle, and a tenfold reduction in steady-state data qubit leakage.
How fast was the microwave reduction unit?
Up to 99% efficacy in 220 ns, per the Physical Review Letters abstract.
Is leakage solved?
The authors of the Nature Physics paper say it can be contained. The experiments are small.
All field notes