Short answer: Cosmic rays and other ionizing radiation can create bursts of quasiparticles in superconducting chips that cause errors across many qubits at once, which is a problem because error-correcting codes usually assume errors are mostly independent. Three papers in Nature journals measure this directly. One found that cosmic rays caused correlated errors at a rate of about one per 592 seconds in its 10-qubit chip, accounting for 17.1 percent of such events. These are experimental findings that affect how error correction must be designed.
Why correlated errors matter
A 2021 Nature Physics paper says scalable quantum computing with error correction rests on the premise that physical errors remain small and sufficiently uncorrelated as devices grow, so logical error rates can be suppressed exponentially. It says impacts from cosmic rays and latent radioactivity violate those assumptions: an impinging particle can ionize the substrate and induce a burst of quasiparticles that destroys qubit coherence throughout the device. It says the effect of high-energy radiation on large-scale algorithms and error correction was an open question.
A chip-wide failure
Using space- and time-resolved measurements on a large-scale quantum processor, the 2021 authors identify bursts of quasiparticles produced by high-energy rays. They track the events from a localized initial impact as they spread, simultaneously and severely limiting the energy coherence of all qubits and causing chip-wide failure. They say their results show the damaging effect of these bursts and the necessity of mitigation to enable scaling. A 2025 paper notes that large-scale correlated errors were demonstrated in Google's Sycamore processor by monitoring simultaneous energy relaxations of multiple qubits.
Measuring the cosmic-ray share
A 2025 paper in Nature Communications directly measured the cosmic-ray contribution to spatiotemporally correlated qubit errors. The authors synchronously monitored cosmic-ray detectors and the energy-relaxation dynamics of 10 transmon qubits on a 5 by 5 by 0.35 mm silicon chip. They report that cosmic rays caused correlated errors at a rate of one per 592 seconds, with an uncertainty of about plus 48 and minus 41 seconds, accounting for 17.1 plus or minus 1.3 percent of all such events. The qubits responded to essentially all of the cosmic rays and secondary particles incident on the chip, consistent with the independently measured flux. They also found that the landscape of the superconducting gap near the Josephson junctions dramatically affects how the qubits respond. Because shielding cosmic rays is practically difficult, the authors say radiation hardening, such as superconducting gap engineering, is important for robust error correction.
Separating muons from gamma rays
Another 2025 Nature Communications paper used charge-parity jumps and bit flips to monitor quasiparticle bursts, and two muon detectors inside the dilution refrigerator. The authors say they directly observed bursts leading to correlated errors induced solely by muons, and separated the contributions of muons and gamma rays. They also investigated the dynamics of a burst and the impact of quasiparticle trapping. They suggest the method might also be applied to detect cosmic-ray particles, low-mass dark matter and far-infrared photons.
What this does not tell you
These papers measure specific chips and setups, and the 17.1 percent figure applies to that device and its set of correlated events. They do not say how any particular error-correction scheme performs under these bursts, or how much mitigation would be needed at scale.
FAQ
What is a quasiparticle burst?
A sudden excess of quasiparticles in a superconducting circuit after an ionizing particle hits the substrate, which can disrupt qubit coherence.
Why do correlated errors matter for error correction?
Conventional codes assume errors are small and sufficiently uncorrelated as devices scale.
How common are cosmic-ray errors?
In one 10-qubit chip, cosmic rays caused correlated errors about once per 592 seconds, per the 2025 paper.
Can shielding fix it?
The authors of that paper note practical difficulties with shielding and point to radiation hardening, such as gap engineering.
Do muons cause it?
A 2025 paper reports bursts induced solely by muons and separates their contribution from gamma rays.
All field notes