A preprint reports a sharp contrast in a cryogenic test of a SQUID-based Josephson parametric amplifier, or JPA: its input-referred added noise was 0.06 photons in phase-sensitive mode, compared with 0.71 photons in phase-preserving mode. The photon figures express the extra noise attributed to the amplifier as though it were present at the input, making the two operating modes directly comparable.
For system noise, phase-sensitive operation reached a minimum noise temperature of 94 ± 12 millikelvin, below the reported 288-millikelvin standard quantum limit. Phase-preserving operation produced 351 ± 53 millikelvin at 6 gigahertz and was described as near-quantum-limited. System noise temperature is the paper’s temperature-based measure of the noise seen by the setup, while added noise refers to the contribution assigned to the amplifier itself.
The contrast depends on phase
In phase-sensitive mode, the setup is aligned with one chosen signal component, called the amplified quadrature. The study compares that regime with phase-preserving operation in the same JPA and uses a TWPA cascade to probe noise when JPA gain is low.
Phase control was central to that comparison. The generated signals had relative phase stability of 3 milliradians over five hours, and phase-locking the local oscillator to the amplified quadrature was associated with a 6-decibel increase in measured gain.
A cryogenic two-stage test
The hardware was a SQUID-based JPA fabricated with VTT SWAPS technology and characterized in a three-wave-mixing, reflection-based setup at millikelvin temperatures. The cascade included a TWPA as an output pre-amplifier for measurements at low JPA gain.
Following initial calibration with both amplifiers off, the TWPA was biased in phase-preserving mode and provided approximately 20 decibels of pre-amplification. Its added noise was approximately 700 millikelvin, and its system noise was approximately 900 millikelvin.
Noise across the chain was estimated with the Friis formula, which combines the contributions of successive amplifier stages, and calibrated with a standard Y-factor method that heats the mixing chamber. The reported calibration uncertainty was estimated at ±1 decibel. Added noise was calculated from spectrum-analyzer measurements of gain and signal-to-noise improvement when the amplifier was turned on.
The quantum comparison
In phase-preserving operation, added noise increased with gain and saturated at 200 ± 54 millikelvin, about 56 millikelvin above the reported 144-millikelvin quantum reference. In phase-sensitive operation, added noise saturated at 17 ± 12 millikelvin. Expressed as input-referred photons, the reported values were 0.71 and 0.06, respectively.
The phase-sensitive system-noise contribution was also partitioned: approximately 89 millikelvin came from quantum fluctuations and intrinsic JPA noise, while approximately 5 millikelvin came from the subsequent amplifier chain. That breakdown indicates that the later stages made a smaller contribution to the reported phase-sensitive total under the tested conditions.
A proposed measurement route, with caveats
The authors present the JPA–TWPA cascade as a route to direct probing of a quantum-device state without subtracting measurement noise. The experiment described here, however, is a cryogenic characterization of the amplifier chain itself in a reflection-based setup. It does not report the direct-probing outcome; that remains a proposed use of the cascade.
The caution is important because the Y-factor calibration is described as approximate and may be affected by thermal gradients, cable losses and system matching. The report also does not give the number of independently fabricated devices, repeated runs or replicate measurements, so reproducibility across hardware and runs remains unresolved.
Phase-sensitive results are likewise tied to phase control and the selected amplified quadrature, as reflected in the reported phase stability and gain change. Open questions include whether the metrics reproduce across independently fabricated devices and repeated runs and whether the cascade can directly probe a quantum-device state without background-noise subtraction.
The manuscript is a preprint, identified as arXiv:2608.25645v1 and dated 26 August 2026.
Paper data and sources
Original title: Phase-sensitive cascade quantum amplifier with nearly noiseless operation
Authors: Ilari Lilja, Ekaterina Mukhanova, Michael Perelshtein et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text