Superconducting qubits lose quantum information because they interact with real-world degrees of freedom. Coherence is usually discussed in terms of T1 energy relaxation and T2 dephasing, and the same physical processes often degrade both. Identifying and mitigating these processes has been central to advances in the field and is supported by experimental and theoretical work from groups with deep expertise in materials, microwave engineering, and circuit quantum electrodynamics.
Dominant microscopic loss mechanisms
One of the clearest, repeatedly observed limits is dielectric loss produced by two-level systems in amorphous oxides and interfaces. John M. Martinis University of California, Santa Barbara and Google and collaborators traced energy relaxation in Josephson qubits to microscopic tunneling defects in surface oxides and substrate interfaces. These defects absorb microwave energy and produce temperature- and power-dependent loss that shortens T1 and contributes to dephasing. Materials research thus matters: oxide chemistry, substrate choice, and device geometry change the participation of lossy surfaces and directly alter coherence.
Another pervasive mechanism is coupling to electromagnetic modes and the surrounding circuitry. The Purcell effect—enhanced spontaneous emission into a resonant environment—was formalized in circuit QED theory and applied to superconducting qubits by Antonio Blais Université de Sherbrooke and colleagues. Qubits deliberately coupled to readout resonators or unintended spurious modes can radiate energy away, lowering T1 unless the environment is filtered or engineered.
Magnetic, charge, and quasiparticle noise
Magnetic noise from fluctuating surface spins produces low-frequency flux noise that dephases flux-sensitive qubits. John Clarke University of California, Berkeley documented how magnetic fluctuations in device surfaces and adsorbed contaminants give rise to 1/f-type noise that limits long-term phase stability. Charge noise historically limited Cooper-pair-box qubits, and the transmon architecture introduced by Jens Koch Yale University reduced that sensitivity by design, illustrating how circuit choices trade different noise channels.
Nonequilibrium quasiparticles—unpaired electrons that break Cooper pairs—produce both relaxation and stochastic jumps in qubit frequency. Robert J. Schoelkopf Yale University and Michel H. Devoret Yale University and their groups have studied quasiparticle-induced errors and advocated mitigation strategies such as quasiparticle traps, improved shielding, and gap engineering. Pair-breaking radiation from the environment or cosmic events can create bursts of quasiparticles and cause correlated errors across devices, a concern for scaling.
Materials processing, device geometry, and the lab environment all shape the relative importance of these mechanisms. Surface treatments, vacuum packaging, careful filtering, and three-dimensional cavity designs implemented by multiple groups have demonstrably extended coherence by reducing surface participation and unwanted coupling. Such improvements are empirical and cumulative, often requiring iteration between fabrication facilities, cryogenic microwave engineering, and targeted diagnostics.
The consequences are practical: limited coherence increases the overhead for quantum error correction and constrains algorithm depth. Addressing limits requires cross-disciplinary work spanning condensed-matter physics, materials science, and systems engineering. Progress depends not only on individual device physics but also on manufacturing quality, regional access to cleanroom infrastructure, and collaboration among institutions with complementary expertise. In practice, sustained gains come from combining careful materials selection, circuit design that minimizes sensitivity to dominant noise channels, and environmental controls that suppress radiative and quasiparticle sources.