Qubit coherence falls after radiation hits detector arrays

Introduction

Recent experiments at Lawrence Livermore National Laboratory have provided the first direct, real‑time evidence that cosmic radiation can degrade the performance of superconducting qubits. By placing a qubit between two microwave kinetic inductance detector (MKID) arrays, researchers observed a sharp, up‑to‑30.5% reduction in both energy‑relaxation time (T₁) and phase‑coherence time (T₂) immediately after a radiation event was registered on both detector layers. The findings challenge the common assumption that decoherence is purely random and uncorrelated, with important implications for error‑correction strategies and hardware shielding.

A novel three‑layer platform

The experimental stack consists of a superconducting qubit sandwiched between two MKID arrays, each capable of detecting quasiparticle bursts generated when high‑energy particles interact with the superconducting films. The MKIDs monitor resonance‑frequency shifts with a 12‑bit, ±350 MHz bandwidth, allowing continuous, high‑resolution tracking of kinetic‑inductance changes. When a penetrating particle—most often a muon—passes through the qubit region, both the top and bottom MKIDs register a coincident signal (a “dual” event). This geometry maximizes the likelihood of capturing events that truly traverse the qubit, distinguishing them from localized radiation that only triggers a single detector.

Correlating radiation with decoherence

Synchronizing MKID triggers with single‑shot measurements of T₁ and T₂ revealed a clear temporal link: qubit coherence drops sharply in the milliseconds following a dual detection. Statistical analysis showed that dual events deposit less energy per detector than single‑detector events, consistent with the passage of a high‑energy muon rather than a localized gamma or alpha interaction. The observed rate—roughly one correlated event per minute—implies a muon flux of about 0.015 events s⁻¹ cm⁻² at the experimental depth.

Implications for quantum error correction and hardware design

Most quantum error‑correction (QEC) codes assume errors are independent in space and time. The correlated nature of muon‑induced decoherence violates this assumption, potentially reducing the effectiveness of standard QEC thresholds. The LLNL platform offers a pathway to develop tailored mitigation techniques:

1. Real‑time error tagging – By flagging qubit operations that occur shortly after a dual MKID event, a processor can apply conditional correction or discard affected data.
2. Targeted shielding – Quantifying the muon flux and its impact enables engineers to design localized shielding (e.g., high‑Z materials) around critical qubit regions without over‑burdening the cryogenic system.
3. Algorithmic adaptation – Error‑correction protocols can be modified to account for bursty, correlated error patterns, improving logical error rates in larger processors.

Outlook

The MKID‑sandwich approach transforms radiation from a hidden, statistical background into a measurable, actionable signal. As quantum processors scale to thousands of qubits, the cumulative probability of radiation‑induced correlated errors will rise, making real‑time detection and mitigation increasingly vital. Future work will likely explore:

• Extending the platform to multi‑qubit arrays to map spatial error propagation.
• Integrating faster MKID readout electronics for sub‑microsecond latency.
• Testing various shielding configurations to quantify mitigation gains.

By turning a fundamental environmental challenge into a diagnostic tool, this research paves the way for more resilient quantum hardware and error‑correction schemes that can tolerate the inevitable presence of cosmic radiation.