A Cryogenic Muon Tagging System Based on Kinetic Inductance Detectors for Superconducting Quantum Processors

TL;DR

Ionizing radiation, particularly atmospheric muons, threatens superconducting qubits by generating quasiparticles that degrade coherence. The authors implement a cryogenic muon-tagging system using Kinetic Inductance Detectors in a vertical detector arrangement operated at $20$ mK, achieving about $90\%$ tagging efficiency with negligible dead time. Geant4-based simulations and a measured prototype show excellent agreement on muon-tagging rates, with a muon-induced coincidence rate of $R_{T-B}^{\mu\mu} \approx (192 \pm 9) \times 10^{-3}$ s$^{-1}$ versus a predicted $(195 \pm 12) \times 10^{-3}$ s$^{-1}$, and an accidental γ-induced background that does not compromise live time. The results validate the feasibility of real-time muon veto or correction for above-ground superconducting processors and lay the groundwork for integrating such tagging with multi-qubit chips to mitigate muon-induced correlated errors.

Abstract

Ionizing radiation has emerged as a potential limiting factor for superconducting quantum processors, inducing quasiparticle bursts and correlated errors that challenge fault-tolerant operation. Atmospheric muons are particularly problematic due to their high energy and penetration power, making passive shielding ineffective. Therefore, monitoring the real-time muon flux is crucial to guide the development of alternative error-correction or protection strategies. We present the design, simulation, and first operation of a cryogenic muon-tagging system based on Kinetic Inductance Detectors (KIDs) for integration with superconducting quantum processors. The system consists of two KIDs arranged in a vertical stack and operated at ~20 mK. Monte Carlo simulations based on Geant4 guided the prototype design and provided reference expectations for muon-tagging efficiency and accidental coincidences due to ambient $γ$-rays. We measured a muon-induced coincidence rate among the top and bottom detectors of (192 $\pm$ 9) $\times$ 10$^{-3}$ events/s, in excellent agreement with the Monte Carlo prediction. The prototype achieves a muon-tagging efficiency of about 90% with negligible dead time. These results demonstrate the feasibility of operating a muon-tagging system at millikelvin temperatures and open the path toward its integration with multi-qubit chips to veto or correct muon-induced errors in real time.

Figures (6)

• Figure 1: Microscope image of a kinetic inductance detector (KID). The device consists of a meandered superconducting inductor ($\sim$6 cm in total length and 62.5 µm in width) connected to a two-finger interdigitated capacitor.
• Figure 2: (a) Picture of the three KID-based detectors: the top (A) and bottom (C) detectors form the muon-tagging system, while the central detector (B) serves as a proxy for the qubit chip in the prototype. (b) Vertical stack of the detectors mounted in the copper holder, showing the top–center–bottom configuration with a 4.5 mm inter-layer spacing. (c) The assembled detector stack installed at the mixing-chamber stage of the dilution refrigerator at La Sapienza University.
• Figure 3: Simulated energy deposition of atmospheric muons in the central chip (blue) and events identified by the muon-tagging system (orange). The ratio of tagged to total events (4589/5073) corresponds to a muon-tagging efficiency of about 90%.
• Figure 4: Distribution of the Pulse Peak Delay between the bottom and top detectors. The shaded area ($|\Delta t| \leq 170$ µs) represents the 340 µs-wide coincidence window used to select true coincident events, while the flat regions at larger delays correspond to random coincidences (sidebands) employed to estimate the accidental background.
• Figure 5: Scatter plot of pulse amplitudes in the top and bottom detectors before (blue) and after (orange) applying the selection cuts described in the text. Before the cuts the dataset is populated by noise and uncorrelated events, whereas after the cuts a cluster of true coincidences stands out more clearly.