Short answer: A fluxonium qubit is a superconducting qubit made from a loop interrupted by a chain of more than 100 Josephson junctions. A 2019 paper in Physical Review X by Nguyen and colleagues reported fluxonium qubits with T2 coherence times above 100 microseconds, and above 400 microseconds in one device, largely limited by energy relaxation. These are laboratory results from 2019, and a later paper studies what limits relaxation in fluxonium.
What a fluxonium qubit is
The Physical Review X summary describes fluxonium as a relatively unexplored superconducting artificial atom with properties desirable for engineering fast, controllable interactions between qubits. It says a notable feature is that a fluxonium consists of a superconducting loop interrupted by over 100 Josephson junctions, which it describes as strips of insulating material a few nanometres thick sandwiched between superconducting layers.
A perspective in Applied Physics Letters places this in context. It says flux qubits are among the first qubits ever demonstrated, and have advantages when compared to capacitively shunted charge qubits, which are now commonly used for building prototypes of quantum processors. It says flux qubits are intrinsically nonlinear systems, remaining so even with low charging energies, which is important for suppressing large charge noise in solids. It adds that, despite these advantages, applications of flux qubits in multi-qubit devices are still limited.
The 2019 coherence results
The Physical Review X paper, "High-Coherence Fluxonium Qubit", reports superconducting fluxonium qubits with coherence times largely limited by energy relaxation, reproducibly satisfying T2 above 100 microseconds, and above 400 microseconds in one device. The summary describes the work as reaching 0.5 ms and calls it record long coherence. That is the paper's own description and is specific to its date and to superconducting qubits of the time.
The authors also say that, given the state-of-the-art values of the surface loss tangent and the 1/f flux-noise amplitude, the coherence time can be further improved beyond 1 ms. That is a projection, not a measurement.
A claim the paper pushes back on
The abstract says the results violate a common viewpoint that the number of Josephson junctions in a superconducting circuit, over 100 here, must be minimized for best qubit coherence. It outlines how fluxonium's combination of long coherence time and large anharmonicity could benefit both gate-based and adiabatic quantum computing. This is the authors' argument.
What limits relaxation
A paper in Physical Review Applied studies a fluxonium qubit with in situ tunability of its Josephson energy and characterizes its energy relaxation at different flux biases and Josephson energies. It reports that the relaxation rate over a range of qubit energies can be quantitatively explained by a combination of dielectric loss and 1/f flux noise, with a crossover point. In the dielectric-loss regime, the loss is consistent with electric dipole interaction with two-level-system defects. It also reports that raising the Josephson energy, which lowers the qubit frequency at the flux insensitive spot, gave increasingly weaker coupling to those defects, which it calls desirable for high-fidelity operations. The authors say their work establishes a generic noise model for designing high-coherence fluxonium and future noise-resilient superconducting qubits.
What this does not tell you
This article summarises abstracts and summaries of three papers. It does not compare fluxonium with other qubit types on current hardware, and it does not claim that fluxonium processors are in use. Coherence time alone does not determine gate quality.
FAQ
What is fluxonium?
A superconducting artificial atom: a loop interrupted by over 100 Josephson junctions, per the Physical Review X summary.
How long did the 2019 qubits stay coherent?
T2 above 100 microseconds reproducibly, and above 400 microseconds in one device, according to the paper's abstract.
What limited the coherence?
Largely energy relaxation. A later paper says relaxation is explained by dielectric loss and 1/f flux noise.
Are many junctions bad for coherence?
The 2019 paper says its results violate the common viewpoint that the junction count must be minimized.
Is fluxonium used in commercial processors?
These sources do not say so. They describe laboratory devices.
All field notes