Mitigation of leakage-induced two-qubit-gate errors for quantum error correction with superconducting transmon qubits

In recent years, quantum error correction using superconducting circuits has seen significant experimental progress [1,2]. Nevertheless, continued reduction of physical error rates is essential for useful quantum error correction. Qubit leakage out of the computational subspace is an error that can persist for many error correction cycles and induces the spread of correlated errors to neighbouring qubits via two-qubit-gates. To mitigate this error channel, leakage reduction units are usually employed once per cycle. These units bring the leaked qubit back to the computational subspace, thereby reducing the leakage lifetime to a single error correction cycle [3,4]. However, leaked qubits can still induce correlated errors on neighbouring qubits in the cycle in which the leakage event occurs. Here, we characterize leakage-induced two-qubit-gate errors in small scale quantum error correction experiments using flux-tuneable transmons. Furthermore, we demonstrate a two-qubit-gate that prevents leakage-induced errors, complementing the use of leakage reduction units.

[1] Krinner, S., Lacroix, N. et. al., Nature 605, pages 669–674 (2022)

[2] Google Quantum AI and Collaborators, Nature 638, pages 920–926 (2025)

[3] K. C. Miao et al., Nat. Phys. 19, 1780–1786 (2023)

[4] N. Lacroix et al., Phys. Rev. Lett. 134, 120601 (2025)