Short answer: Fixed-frequency transmon qubits can be tuned after fabrication by heating their Josephson junctions with a laser, which raises the junction resistance and so lowers the qubit frequency. A 2019 APS March Meeting abstract reported raising junction resistance by up to 15%. A 2021 paper in npj Quantum Information reported a nearly tenfold improvement in frequency precision, and a 2021 Science Advances paper reported a tuning precision of 18.5 MHz across hundreds of qubits. These are fabrication results, not claims of a general solution to scaling.
The frequency crowding problem
The npj Quantum Information abstract explains that as superconducting circuits grow, frequency crowding becomes a formidable problem in fixed-frequency architectures. The reason is limited control over the fabrication of aluminum junctions, as the 2019 abstract puts it, which makes accurate targeting of qubit transition frequency difficult.
2019: the first method
The APS March Meeting abstract describes a selective increase in junction resistance by localized thermal annealing with a focused diode laser at room temperature. By tuning irradiation time and laser power, the authors controllably raise junction resistance by up to 15%. They compare transmon transition frequencies and coherence before and after annealing. This is a meeting abstract and not a peer-reviewed paper.
2021: precision and yield modelling
In the npj Quantum Information paper, the authors adjust qubit frequencies systematically after fabrication and show a nearly tenfold improvement in the precision of setting them. To judge scalability, they identify the types of frequency collisions that impair a transmon and cross-resonance gate architecture, and use statistical modelling to compute the probability of avoiding all of them as a function of precision. They find that without post-fabrication tuning, the probability of finding a workable lattice quickly approaches zero, while with the demonstrated precision it is possible to find collision-free lattices with favorable yield. They state the techniques are currently employed in available quantum systems, which is their claim and was not checked here.
2021: statistics over hundreds of qubits
The Science Advances abstract says that fixed-frequency transmons are attractive for long coherence and noise immunity, but need precise relative frequencies. The authors use laser annealing to tune transmons into desired frequency patterns. Statistics over hundreds of annealed qubits give an empirical tuning precision of 18.5 MHz with no measurable impact on coherence. They report gate error statistics on a tuned 65-qubit processor, with a median two-qubit gate fidelity of 98.7%. Baseline tuning statistics give a frequency-equivalent resistance precision of 4.7 MHz, which they say is sufficient for high-yield scaling beyond the 10 cubed qubit level. They anticipate that selective laser annealing will play a central role in scaling fixed-frequency architectures, which is their expectation.
Limits
The precision figures come from different datasets and definitions, so the 2019, nearly tenfold and 18.5 MHz results should not be read as one trend line. The 98.7% gate fidelity describes one processor in the authors' report. Frequency collisions are one scaling constraint among many.
FAQ
How does laser annealing tune a qubit?
It raises the junction resistance locally, which shifts the transition frequency, per the 2019 abstract.
How precise is it?
18.5 MHz empirical precision over hundreds of qubits, per the 2021 Science Advances authors.
Does it hurt coherence?
The authors report no measurable impact on coherence.
Why does frequency matter?
Frequency collisions can make a lattice of fixed-frequency qubits unworkable, the authors say.
Does it solve scaling?
No. It addresses one problem, according to the authors' own framing.
All field notes