In arXiv:2407.18778, Tuokkola et al. characterize a transmon qubit with energy relaxation and echo dephasing times surpassing those in the existing literature, utilizing a traveling-wave parametric amplifier (TWPA) from VTT. This TWPA is from an early prototype run, based on which AI-TWPA-C was created.
In relation to the TWPA specifically, the authors show that qubit coherence is not significantly affected when the strong microwave pump signal used to provide energy to the TWPA is on (see Sec. IV). Another interesting technical detail in the setup is how the pump signal gets to the TWPA. It is initially fed intentionally in the “wrong direction” through a circulator, after which it bounces back from a low-pass filter (see Fig. 3). Whether this is ultimately the best wiring configuration remains to be seen, but the qubit coherence results in the article are indisputably amazing, so one might be well advised to follow their example for the time being.
The median energy relaxation time (T1) of 502 µs for a qubit operating at a frequency of 2.890 GHz corresponds to an effective quality factor of 9 million. To put this in perspective, if you had a mechanical oscillators such as a pendulum with this quality factor and you would give it a push, it would swing back and forth a million times before losing half of its energy! This energy relaxation time is state of the art, but even more impressive is the qubit coherence time. After applying an echo pulse sequence that reduces the impact of low-frequency parameter fluctuations, Tuokkola et al. measure a median dephasing time of 541 µs and a maximum of 1.057 ms. As far as the authors know, those are the longest coherence times for transmon-type qubits reported in the literature.
It is very exciting to see whether these record-breaking results can be reproduced in larger scale devices. Even if that would not turn out to be the case in the immediate future, this is an important result that demonstrates there is no deep reason preventing transmon-type qubits from reaching millisecond coherence time. This shows that the largest superconducting quantum processing units (QPUs) out there today, which are generally based on transmons, can become far more powerful without drastic changes to the type of qubit used.
Note: Cover image of this post is Fig. 1 from arXiv:2407.18778 by Tuokkola et al. (CC-BY 4.0)
