The Signal
NIST Team Reads a Superconducting Qubit at 99.2% with Millimeter Waves
A NIST-led team measured a superconducting transmon with a 34.7 GHz resonator and reported 99.2% readout fidelity using 109 resonator photons. The experiment also found just 1.74% measurement efficiency, leaving substantial room to improve
Quantum.world Team · Written with AI · fact-checked ·
A superconducting transmon qubit reached 99.2% readout fidelity when researchers probed it with 109 photons in a resonator operating at 34.7 GHz. The result, published October 30 in Physical Review Letters, uses a readout frequency more than ten times the qubit’s 3.1 GHz transition frequency. The National Institute of Standards and Technology-led team says the separation suppresses drive-induced transitions that can disturb the state being measured.
A 34.7 GHz cavity beside a 3.1 GHz qubit
Conventional superconducting-qubit readout infers a qubit’s state from a shift in the resonator’s frequency. The measurement drive itself can also push the qubit into another state, undermining the result. The team’s approach keeps the transmon in its usual microwave range while moving the resonator into the millimeter-wave band, creating a large frequency gap between the two.
The experiment paired a transmon with the fundamental mode of a three-dimensional aluminum cavity. The device sat at 10 millikelvin in a dilution refrigerator; the cavity and qubit were coupled at 1.3 GHz. The authors write, “With such a large detuning between cavity and qubit, ωr/ωq>10, we are able to suppress drive induced unwanted resonant state transitions, enabling strong drives for qubit readout.” Akash V. Dixit, Zachary L. Parrott, Dennis Chunikhin, Trevyn F. Q. Larson, Bradley D. Hauer, and John D. Teufel
109 photons, 780 nanoseconds, 99.2%
For the readout test, the researchers prepared the qubit in either its ground or first excited state and collected the reflected signal for 780 nanoseconds. They repeated the preparation and measurement 50,000 times. At an average of 109 photons in the resonator, the team measured 99.2% fidelity for each of the two states.
That photon count refers to the readout resonator during measurement, not the separate drive-power test. In that test, the team found no measurable resonant transitions out of the qubit’s ground or first excited state at drive levels up to 100 photons. At around 1,000 drive photons, however, the transmon began to enter a quantum-to-classical transition in which strong driving spreads its population among many states.
High assignment fidelity, low measurement efficiency
The headline fidelity does not mean the measurement chain captures nearly all available information. The paper reports a measurement efficiency of 1.74%, with 28 photons of added system noise. The authors identify a millimeter-wave quantum-limited amplifier as a way to improve that efficiency.
The distinction matters: the experiment demonstrates accurate state assignment under its test conditions, while the low efficiency signals that much of the measurement signal is lost to noise in the readout chain. The researchers also point to qubit decay, readout duration, state preparation and reset as factors that could affect fidelity. They say further work should examine other possible forms of leakage, including effects from inelastic scattering of drive photons and quasiparticle generation.
A single-qubit demonstration, not a processor result
The work establishes a way to read one transmon using a much higher-frequency cavity; it does not demonstrate a multi-qubit processor or a complete readout architecture for one. The authors suggest that separating qubit and resonator frequencies could also reduce unwanted collisions with parametric processes used to connect qubits. Millimeter waves could also serve as an interface frequency for systems involving superconducting circuits, Rydberg atoms or optomechanics, but the paper presents those possibilities as directions for future work.
For now, the measured benchmark is specific: 99.2% fidelity at 109 resonator photons in a 780-nanosecond readout, alongside 1.74% measurement efficiency. Improving the efficiency of the millimeter-wave readout chain remains an open engineering task.