Superconducting quantum circuits — SQUIDs (Superconducting QUantum Interference Devices) and superconducting qubits — exhibit an excess low-frequency magnetic flux noise with amplitude spectral density of order 10−6 Φ0/Hz10^{-6}\,\Phi_0/\sqrt{\mathrm{Hz}} at 1 Hz that is nearly universal across devices, materials, and geometries. Bialczak et al. (2007, on Josephson phase qubits) showed that the standard model of magnetic two-level-state (TLS) defects in the native surface oxides underpredicts the measured noise power by about four orders of magnitude, and posed the question of what surface or interface mechanism supplies the required areal density and magnetic coupling of fluctuators at millikelvin temperatures. Subsequent work has converged on surface-adsorbed paramagnetic spins as the source: on-chip electron-spin-resonance has directly detected a 2.2×1017 m−22.2\times10^{17}\,\mathrm{m}^{-2} surface spin density on alumina matching the flux-noise-inferred density, with candidate species including physisorbed atomic hydrogen and molecular oxygen. However, the chemical identity of the dominant fluctuators in working devices, their interaction and relaxation dynamics, and a quantitative theory reproducing the 1/fα1/f^\alpha spectrum (α≈0.6\alpha\approx0.6–11 over ten decades of frequency) and its weak temperature dependence remain open, as stated explicitly in primary works from 2023–2026; a complete account must in addition be consistent with the magnetic-field response documented after 2007 (Rower et al. 2023; Günzler et al. 2025). The problem is to give a complete, quantitatively verified microscopic identification and dynamical account of the flux-noise mechanism.

Background

Superconducting quantum circuits store quantum information in lithographed electrical circuits cooled to millikelvin temperatures. Two workhorses are the SQUID (Superconducting QUantum Interference Device — a superconducting loop interrupted by Josephson junctions, whose quantum interference makes it a sensitive flux-to-voltage magnetometer) and superconducting qubits such as the current-biased phase qubit and the flux qubit, whose transition frequencies depend on the magnetic flux Φ\Phi threading a superconducting loop. The relevant unit of flux is the flux quantum Φ0=h/2e≈2.07×10−15 Wb\Phi_0=h/2e\approx2.07\times10^{-15}\,\mathrm{Wb}. Because qubit transition frequencies shift with Φ\Phi, slow fluctuations of the ambient flux cause qubit dephasing.

In 1987, Wellstood, Urbina and Clarke measured, in dc SQUIDs below 1 K, an excess magnetic flux noise with power spectral density SΦ(f)S_\Phi(f) scaling as 1/fα1/f^\alpha with α≈0.6\alpha\approx0.6–11, an amplitude at 1 Hz corresponding to roughly 10−6 Φ0/Hz10^{-6}\,\Phi_0/\sqrt{\mathrm{Hz}} (a canonical value of about 2 μΦ0/Hz2\,\mu\Phi_0/\sqrt{\mathrm{Hz}}), and remarkably weak dependence on loop size, geometry, material, and temperature. They proposed spin flips of localized electrons on device surfaces, which would require areal spin densities of order 1017 m−210^{17}\,\mathrm{m}^{-2} — far above bulk defect densities. The issue lay largely dormant for two decades until qubit-era work revived it: Koch, DiVincenzo and Clarke (2007) formulated a phenomenological model of dilute, nearly freely relaxing surface spins whose diffusive dynamics generate low-frequency noise.

Bialczak et al. (2007) measured this noise inside a Josephson phase qubit using a new technique that reaches below 1 Hz: comparing frequency-fluctuation spectra taken at positive and negative current bias separates flux noise (which shifts the two biases oppositely) from junction critical-current noise (estimated at only the few-percent level), yielding flux noise of the same magnitude as in SQUIDs — its own measurement extrapolates to 4 μΦ0/Hz4\,\mu\Phi_0/\sqrt{\mathrm{Hz}} at 1 Hz with α=0.95\alpha=0.95. The paper then tested the standard model attributing the noise to magnetic two-level-system (TLS) defects — atoms or bonds tunneling between two stable configurations, the same defects invoked to explain dielectric loss — in the roughly 2 nm native surface oxides. Even granting TLS densities inferred from dielectric-loss and tunnel-junction data and maximal magnetic moments per defect, the model predicts SΦ(1 Hz)=1.1×10−3 (μΦ0)2/HzS_\Phi(1\,\mathrm{Hz})=1.1\times10^{-3}\,(\mu\Phi_0)^2/\mathrm{Hz}, about four orders of magnitude below the measurement; matching the data would require a defect-to-atomic-bond ratio about 10410^4 times larger than known bulk TLS densities. The paper therefore asked what surface or interface mechanism can supply the requisite areal density and magnetic coupling of fluctuators at millikelvin temperatures, concluding that any viable model "must arise from a surface defect mechanism that is very different than that described by the standard TLS model, as it must have a defect/atomic-bond ratio that is about 10410^4 times larger than for bulk TLS defects." As candidate directions it noted the surface electronic states suggested by Koch et al. — whose density had been estimated only at room temperature — and a proposed mechanism based on tunneling of conduction electrons into surface states (Faoro–Ioffe), along with a specific-heat consequence testable in amorphous structures thinner than about 11–10 μm10\,\mu\mathrm{m}.

Much has since been established. Sendelbach et al. (2008) found SQUID magnetization behavior indicating spin-glass-like freezing of surface spins (a spin glass is a disordered spin system with frustrated interactions whose magnetization relaxes extremely slowly). Choi, Lee, Louie and Clarke (2009) proposed localized metal-induced gap states (MIGS — electronic states induced inside the insulator's band gap at a metal–insulator interface) as spin carriers. Density-functional-theory studies identified specific paramagnetic adsorbates as candidates: molecular oxygen (a spin triplet) on alumina with moment ≈1.8 μB\approx1.8\,\mu_B and spin-reorientation barriers of order 10 mK (H. Wang et al. 2015), and physisorbed or embedded hydrogen with moments 0.80.8–0.9 μB0.9\,\mu_B and similarly small barriers (Z. Wang et al. 2018). Experimentally, Kumar et al. (2016) showed that adsorbed O2O_2 dominates the residual magnetism of superconducting thin films and that ion-milling or improved vacuum suppresses both this magnetism and the flux noise. Most directly, de Graaf et al. (2017), using an on-chip electron-spin-resonance (ESR — spectroscopy detecting the microwave-frequency resonance of spin magnetic moments) technique, detected dilute surface spins on Al2O3\mathrm{Al_2O_3} with total density n=2.2×1017 m−2n=2.2\times10^{17}\,\mathrm{m}^{-2}, matching the density inferred from SQUID flux noise, and assigned two of three ESR lines to physisorbed atomic hydrogen (a by-product of water dissociation) and the third tentatively to O2−O_2^- captured at surface defect sites.

What is still missing is a quantitative, closed account. It is not settled which species dominates in working devices (hydrogen, oxygen, dangling bonds, interface gap states, bulk magnetic impurities — the last argued to matter for aluminum in the Pippard nonlocal-electrodynamics regime by Yapa et al. 2019 — or several at once, potentially material-dependent), and no identified spin ensemble with independently determined interaction parameters has been shown to generate the observed 1/fα1/f^\alpha spectrum over ten decades of frequency with the measured magnitude, weak temperature dependence below ∼100\sim100 mK, and the observed magnetic-field behavior (a low-frequency crossover toward Lorentzian spectra and field-dependent noise evolution found by Rower et al. 2023; suppression of MHz-range flux noise in fields up to 1 T found by Günzler et al. 2025). Monte Carlo studies (Mickelsen, Wu and Yu 2024) find that only glassy spin ensembles yield α≈1\alpha\approx1 at low temperature, and parameter-free simulations of disordered O2O_2 on Al2O3\mathrm{Al_2O_3} (Ray et al. 2026) reproduce field-dependent trends but not yet a complete quantitative account. Reviews (Paladino et al. 2014) and primary papers through 2025–2026 continue to describe the microscopic origin as an open question.

Problem Statement

Give a complete, quantitatively verified microscopic account of the excess low-frequency magnetic flux noise in superconducting quantum circuits — the mechanism whose surface-oxide TLS interpretation Bialczak et al. (2007) showed to fail by four orders of magnitude. A complete answer must deliver both of the following linked parts of this single objective: (1) Identification: the chemical/physical identity and location of the fluctuating magnetic moments (surface adsorbates such as hydrogen or oxygen, dangling bonds, interface gap states, bulk magnetic impurities, or a coexisting, possibly material-dependent combination), their areal or volume density, and their magnetic moments, consistent with direct spectroscopy and magnetometry on the relevant surfaces (Al2O3\mathrm{Al_2O_3} and other native oxides, superconductor–air and metal–insulator interfaces, device edges) and with the existing discriminating experimental record — deliberate surface dosing or removal, surface treatments and encapsulation, isotope or material substitution, geometry variation, and direct magnetic spectroscopy, with the manipulated spin density tracking the measured noise. (2) Dynamics and quantitative spectrum: from the identified species and interaction parameters (dipolar coupling, exchange where relevant, anisotropy barriers, hyperfine-mediated relaxation, disorder) determined independently of the noise data being reproduced, derive the flux-noise spectral density SΦ(f,T)S_\Phi(f,T) reproducing the source-supported experimental benchmarks: amplitude at 1 Hz of order (2–4) μΦ0/Hz(2\text{–}4)\,\mu\Phi_0/\sqrt{\mathrm{Hz}} (the canonical SQUID value of about 2 μΦ0/Hz2\,\mu\Phi_0/\sqrt{\mathrm{Hz}} and Bialczak et al.'s phase-qubit measurement of 4 μΦ0/Hz4\,\mu\Phi_0/\sqrt{\mathrm{Hz}} at α=0.95\alpha=0.95); 1/fα1/f^\alpha form with α≈0.6\alpha\approx0.6–11 from below 1 Hz to the MHz regime; weak temperature dependence below roughly 100 mK; and the observed near-universality across loop geometry, inductance, and materials. The account must also be consistent with the post-2007 observational record documented in the Background — the measured dependence on applied magnetic field and the observed low-frequency crossover toward Lorentzian spectra (Rower et al. 2023; Günzler et al. 2025) — although quantitative reproduction of that later dataset is not itself a closure condition. An answer delivering only one part — a spin species identified without quantitative spectral accounting, or a phenomenological or post-hoc-fitted spectrum without species identification — does not close the problem.

References

  1. Radoslaw C. Bialczak, R. McDermott, M. Ansmann, M. Hofheinz, N. Katz, Erik Lucero, Matthew Neeley, A. D. O'Connell, H. Wang, A. N. Cleland, John M. Martinis (2007). "1/f Flux Noise in Josephson Phase Qubits". Physical Review Letters 99, 187006. DOI: 10.1103/PhysRevLett.99.187006; arXiv:0708.2125. https://arxiv.org/abs/0708.2125
  2. Frederick C. Wellstood, Cristian Urbina, John Clarke (1987). "Low-frequency noise in dc superconducting quantum interference devices below 1 K". Applied Physics Letters 50, 772. DOI: 10.1063/1.98041. https://doi.org/10.1063/1.98041
  3. Roger H. Koch, David P. DiVincenzo, John Clarke (2007). "Model for 1/f Flux Noise in SQUIDs and Qubits". Physical Review Letters 98, 267003. DOI: 10.1103/PhysRevLett.98.267003. https://doi.org/10.1103/PhysRevLett.98.267003
  4. S. Sendelbach, D. Hover, A. Kittel, M. Mück, John M. Martinis, R. McDermott (2008). "Magnetism in SQUIDs at Millikelvin Temperatures". Physical Review Letters 100, 227006. DOI: 10.1103/PhysRevLett.100.227006; arXiv:0802.1518. https://arxiv.org/abs/0802.1518
  5. SangKook Choi, Dung-Hai Lee, Steven G. Louie, John Clarke (2009). "Localization of Metal-Induced Gap States at the Metal-Insulator Interface: Origin of Flux Noise in SQUIDs and Superconducting Qubits". Physical Review Letters 103, 197001. DOI: 10.1103/PhysRevLett.103.197001. https://doi.org/10.1103/PhysRevLett.103.197001
  6. Hui Wang, Chuntai Shi, Jun Hu, Sungho Han, Clare C. Yu, R. Q. Wu (2015). "Candidate Source of Flux Noise in SQUIDs: Adsorbed Oxygen Molecules". Physical Review Letters 115, 077002. DOI: 10.1103/PhysRevLett.115.077002; arXiv:1504.08075. https://arxiv.org/abs/1504.08075
  7. P. Kumar, S. Sendelbach, M. A. Beck, J. W. Freeland, Zhe Wang, Hui Wang, Clare C. Yu, R. Q. Wu, D. P. Pappas, R. McDermott (2016). "Origin and Reduction of 1/f Magnetic Flux Noise in Superconducting Devices". Physical Review Applied 6, 041001. DOI: 10.1103/PhysRevApplied.6.041001; arXiv:1604.00877. https://arxiv.org/abs/1604.00877
  8. S. E. de Graaf, A. A. Adamyan, T. Lindström, D. Erts, S. E. Kubatkin, A. Ya. Tzalenchuk, A. V. Danilov (2017). "Direct Identification of Dilute Surface Spins on Al2O3: Origin of Flux Noise in Quantum Circuits". Physical Review Letters 118, 057703. DOI: 10.1103/PhysRevLett.118.057703; arXiv:1609.04562. https://arxiv.org/abs/1609.04562
  9. Zhe Wang, Hui Wang, Clare C. Yu, R. Q. Wu (2018). "Hydrogen as a source of flux noise in SQUIDs". Physical Review B 98, 020403(R). DOI: 10.1103/PhysRevB.98.020403; arXiv:1709.06208. https://arxiv.org/abs/1709.06208
  10. Pramodh Senarath Yapa, Tyler Makaro, Rogério de Sousa (2019). "Impact of nonlocal electrodynamics on the flux noise and inductance of superconducting wires". Physical Review Applied 11, 024041. DOI: 10.1103/PhysRevApplied.11.024041; arXiv:1806.09705. https://arxiv.org/abs/1806.09705
  11. Amrit De (2019). "1/f Flux Noise in low-Tc SQUIDs due to Superparamagnetic Phase Transitions in Defect Clusters". Physical Review B 99, 024305. DOI: 10.1103/PhysRevB.99.024305; arXiv:1710.09114. https://arxiv.org/abs/1710.09114
  12. E. Paladino, Y. M. Galperin, G. Falci, B. L. Altshuler (2014). "1/f noise: Implications for solid-state quantum information". Reviews of Modern Physics 86, 361–418. DOI: 10.1103/RevModPhys.86.361; arXiv:1304.7925. https://arxiv.org/abs/1304.7925
  13. José Alberto Nava Aquino, Rogério de Sousa (2022). "Flux noise in disordered spin systems". arXiv:2207.10033. https://arxiv.org/abs/2207.10033
  14. David A. Rower, Lamia Ateshian, Lauren H. Li, Max Hays, Dolev Bluvstein, Leon Ding, Bharath Kannan, Aziza Almanakly, Jochen Braumüller, David K. Kim, Alexander Melville, Bethany M. Niedzielski, Mollie E. Schwartz, Jonilyn L. Yoder, Terry P. Orlando, Joel I-Jan Wang, Simon Gustavsson, Jeffrey A. Grover, Kyle Serniak, Riccardo Comin, William D. Oliver (2023). "Evolution of 1/f Flux Noise in Superconducting Qubits with Weak Magnetic Fields". Physical Review Letters 130, 220602. DOI: 10.1103/PhysRevLett.130.220602; arXiv:2301.07804. https://arxiv.org/abs/2301.07804
  15. M. Herbst, A. Fleischmann, D. Hengstler, D. Mazibrada, L. Münch, A. Reifenberger, C. Ständer, C. Enss (2023). "Measuring magnetic 1/f noise in superconducting microstructures and the fluctuation-dissipation theorem". Superconductor Science and Technology 36, 113001. DOI: 10.1088/1361-6668/acf166; arXiv:2303.15257. https://arxiv.org/abs/2303.15257
  16. D. L. Mickelsen, Ruqian Wu, Clare C. Yu (2024). "Monte Carlo spin simulations of magnetic noise: The search for pivoting". Physical Review B 109, 144510. DOI: 10.1103/PhysRevB.109.144510; arXiv:2403.09078. https://arxiv.org/abs/2403.09078
  17. S. Günzler, J. Beck, D. Rieger, N. Gosling, N. Zapata, M. Field, S. Geisert, A. Bacher, J. K. Hohmann, M. Spiecker, W. Wernsdorfer, I. M. Pop (2025). "Spin environment of a superconducting qubit in high magnetic fields". Nature Communications 16, 9564. DOI: 10.1038/s41467-025-65528-y; arXiv:2501.03661. https://arxiv.org/abs/2501.03661
  18. Keith G. Ray, Yaniv Rosen, Jonathan L Dubois, Vincenzo Lordi (2026). "Field-Dependent Qubit Flux Noise Simulated from Materials-Specific Disordered Exchange Interactions Between Paramagnetic Adsorbates". arXiv:2602.09471. https://arxiv.org/abs/2602.09471