A narrow constriction in the chip's superconducting film — the same Josephson junction element already there — acts as a passive millikelvin thermometer, freeing chip debugging from microwave based qubit thermometry.
Quantum processors run at temperatures colder than deep space, but the engineers building them have had no clean way to ask how cold the chip actually is. A new arXiv preprint from a team working on superconducting qubits proposes a fix that is almost frictionless: take a Dayem bridge, a narrow thin-film Josephson junction already used in superconducting circuits, and let it read its own temperature.
The method sidesteps the dominant approach to on-chip thermometry, which uses a qubit itself as the thermometer. That approach requires dedicated control sequences, microwave calibration, and careful modeling because the qubit's response conflates temperature with non-thermal noise sources. The new technique measures the critical current of a Dayem bridge from a simple current–voltage curve and treats it as a direct local probe of chip temperature. No microwave pulses, no qubit-state readout, no per-chip calibration dance.
A Dayem bridge is a constriction in a thin superconducting film. When current flows through it, the bridge stays superconducting up to a well-defined critical current and becomes resistive beyond it. That critical current scales with temperature, falling steeply as the chip warms from its millikelvin base. The team calibrates the scaling curve against the mixing-chamber stage temperature and reads out chip temperature from the bridge's I-V response. The result is a thermometer that lives on the same substrate as the qubit but does not borrow any of the qubit's control infrastructure.
The motivation reads less like "new sensor" and more like "remove a debugging tax." Chip-level thermal gradients matter for superconducting quantum hardware because they shift qubit frequencies, drive excess quasiparticles, and modulate decoherence rates. Until now, a team trying to map those gradients on a multi-qubit chip either had to run a qubit-as-thermometer experiment on every device under test, attach an off-chip sensor near the sample, or infer temperature from dilution-fridge thermometry that mostly measures the cold finger, not the chip. The Dayem-bridge approach substitutes a passive, microwave-free element that any group already running transmon fabrication can drop in alongside qubits.
The work slots into a small but growing literature on local thermometry for superconducting hardware. Earlier preprints have proposed CMOS-integrated deep-cryogenic temperature sensors built for quantum computing applications, and a separate qubit-based thermometry protocol established the baseline method this work complements. The Dayem-bridge route is the first to make the thermometer an ordinary DC device on the same chip, which matters for fabrication flows: it does not require exotic lithography, and the readout fits inside standard I-V measurement setups already present in every dilution-fridge lab.
The obvious caveats apply. This is a single preprint from a single group, not yet peer-reviewed, and the reported millikelvin accuracy is the authors' measured result rather than an independently replicated benchmark. The Dayem bridge also does not solve every problem: it tells you the local temperature at the bridge, not at the qubit itself, and placing a bridge next to each qubit on a dense processor will require design work that the preprint does not yet show. The authors frame their result as a complementary local probe rather than a replacement for qubit-based thermometry.
For a field that has spent years treating temperature as something you measure at the fridge rather than the chip, the practical test is whether future processors ship with Dayem bridges as built-in thermometers. Independent replication on a different dilution fridge would be the obvious next milestone.