A Traveling-Wave Parametric Amplifier and Converter

A Traveling-Wave Parametric Amplifier and Converter
Notice: This research summary and analysis were automatically generated using AI technology. For absolute accuracy, please refer to the [Original Paper Viewer] below or the Original ArXiv Source.

High-fidelity qubit measurement is a critical element of all quantum computing architectures. In superconducting systems, qubits are typically measured by probing a readout resonator with a weak microwave tone that must be amplified before reaching the room temperature electronics. Superconducting parametric amplifiers have been widely adopted as the first amplifier in the chain, primarily because of their low noise performance, approaching the quantum limit. However, they require isolators and circulators to route signals up the measurement chain and to protect qubits from amplified noise. While these commercial components are wideband and simple to use, their intrinsic loss, size, and magnetic shielding requirements impact overall measurement efficiency and scalability. Here we report a parametric amplifier that achieves both broadband forward amplification and backward isolation in a single, compact, non-magnetic circuit that could be integrated on chip with superconducting qubits. The approach relies on a nonlinear transmission line that supports traveling-wave parametric amplification of forward propagating signals, and isolation via frequency conversion of backward propagating signals. This traveling-wave parametric amplifier and converter has the potential to reduce the readout hardware overhead when scaling up the size of superconducting quantum computers.


💡 Research Summary

High‑fidelity readout of superconducting qubits requires a weak microwave probe tone reflected from a readout resonator to be amplified by many orders of magnitude before reaching room‑temperature electronics. Today this is achieved with a low‑noise traveling‑wave parametric amplifier (TWPA) as the first stage, preceded by bulky, magnetic, and lossy circulators/isolators that enforce directionality and protect the qubit from amplified noise. These components limit scalability because of their size, insertion loss, and the need for magnetic shielding.

In this work the authors present a single, compact, non‑magnetic circuit that simultaneously provides forward gain and backward isolation: the Traveling‑Wave Parametric Amplifier and Converter (TWPA‑C). The device is a current‑biased nonlinear transmission line (NLTL) composed of 2650 unit cells, each containing a single‑junction Nb/Al‑AlOx/Nb Josephson inductor (I_c≈5 µA) and a low‑loss amorphous‑silicon parallel‑plate capacitor (tan δ≈4×10⁻⁴). The line has a characteristic impedance of ≈50 Ω and a phase velocity ≈2 % of the speed of light.

Two strong pumps are injected in opposite directions. A forward pump at ω_a≈14 GHz (≈2 × ω_s) amplifies a co‑propagating signal at ω_s≈7 GHz and its idler at ω_i=ω_a−ω_s via three‑wave mixing. A backward pump at ω_c≈3.15 GHz mixes any signal traveling opposite to the forward direction, converting it either down‑converted to ω_d=ω_s−ω_c (≈3 GHz) or up‑converted to ω_u=ω_s+ω_c (≈12 GHz). The converted frequencies fall inside engineered stop‑bands created by periodic loading (six‑cell periodic capacitors and tank resonators inserted every six cells). Because signals in these stop‑bands cannot propagate, the backward‑propagating tone is effectively removed, providing isolation without any ferrite devices.

Phase‑matching for both processes is achieved by tailoring the dispersion of the NLTL: the periodic capacitors and the interleaved tank circuits produce strong frequency‑dependent phase velocity near the stop‑bands, allowing the forward pump, signal, and idler to satisfy the momentum‑conservation condition in the 6–8 GHz band, and allowing the backward pump, signal, and down‑converted tone to satisfy a similar condition near 3 GHz.

A coupled‑mode theory (CME) model that includes three‑wave mixing predicts up to ~20 dB forward gain and ~10 dB backward isolation over a several‑GHz bandwidth. Experimental measurements at 12 mK show ~7 dB forward gain across an octave (≈6–12 GHz) and at least 5 dB isolation over ≈800 MHz, with peaks of −20 dB isolation at specific frequencies. The discrepancy with the simple CME is traced to additional four‑wave mixing terms, fabrication‑induced parameter spread, and impedance mismatches that generate pump depletion and high‑gain peaks near stop‑band edges. Time‑domain WRspice simulations that incorporate the full Josephson nonlinearity and all mixing pathways reproduce the measured scattering parameters quantitatively, confirming the physical picture.

Noise performance was characterized using a calibrated shot‑noise tunnel junction (SNTJ) as a calibrated source at the input of the measurement chain. As the TWPA‑C gain increases, the system‑added noise N_sys drops from the HEMT‑limited value (~10 quanta) down to an average of 5.2 quanta between 5.5 and 8.5 GHz. The contribution of the TWPA‑C itself (including its preceding microwave components) is 1.7 quanta, indicating near‑quantum‑limited operation comparable to conventional TWPAs. The 1 dB compression point is measured at ≈−90 dBm, confirming sufficient dynamic range for typical qubit readout powers.

Overall, the TWPA‑C demonstrates that forward amplification and backward isolation can be realized in a single, on‑chip, non‑magnetic device, eliminating the need for external circulators and isolators. This reduces insertion loss, magnetic shielding requirements, and physical footprint, directly addressing scalability challenges for large‑scale superconducting quantum processors. Remaining challenges include increasing the forward gain beyond 7 dB, widening the isolation bandwidth, and further suppressing unwanted four‑wave mixing; these will be the focus of future work.


Comments & Academic Discussion

Loading comments...

Leave a Comment