{"id":"cab3a652-dc51-4095-a4f0-76c82e9e8040","arxiv_id":"2412.15856","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"An anomalous low-field increase in the resonant frequency of ultra-thin TiN resonators is explained by a spin-flip two-level system model in which magnetic defects mix charge tunneling and spin flips.","lead":"Ultrathin TiN superconducting resonators show a puzzling increase in resonant frequency when a small in-plane magnetic field is applied, opposite to what standard superconductor theory predicts. The authors propose that magnetic defects create spin-flip two-level systems that mix electron tunneling with spin flips, quantitatively capturing the measured frequency shifts.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The θ-integral in Eq. (3) produces a T-dependent prefactor omitted from Eq. (4), predicting a growing sTLS signal with temperature, opposite to the observed thermal saturation.","rationale":"The experiment is reproducible, but the model's derivation of Eq. (4) from Eq. (3) has an internal inconsistency. The θ distribution used to justify θ≈0 also carries a temperature-dependent width that multiplies the dipole coupling. Omitting this factor changes the predicted temperature trend, undermining the main independent validation. The reader flagged the ad hoc θ≈0 assumption; my concern shows even a delta-function b leaves a T-dependent θ contribution. The authors would need to provide a full integration or an argument for why A itself scales as 1/T to restore the observed saturation. Without that, the quantitative claim fails. Verdict should move to reject, with possibility of conditional revision if the missing prefactor is corrected and fits remain.","tokens_in":890,"tokens_out":1543,"duration_ms":198143,"concrete_test":"Numerically evaluate the θ integral in Eq. (3) with Pθ as given, J=-1 meV, b0=17 μeV, B=200 mT, T=10 and 600 mK, retaining the factor (b sin(θ/2))^2. If the 600 mK integrated response exceeds the 10 mK one, Eq. (4) is missing a T-dependent prefactor; re-fit Fig. 3C/D and Fig. 4 with the corrected expression and check whether the anomaly can disappear at 600 mK.","verdict_should_be":"REJECT","load_bearing_attack":"The central quantitative claim rests on reducing Eq. (3) to Eq. (4). With Pθ as stated, J≈-1 meV, θ is peaked near zero. The electric-dipole factor (b sin(θ/2))^2 averages to about kBT/(2|J|), a factor proportional to T. This prefactor is absent in Eq. (4), where A is T-independent. From 10 mK to 600 mK this factor grows ~60x, while tanh(ε/2T) with t≈0.5 K drops only ~2.6x. The model would predict a much larger sTLS response at 600 mK than at 10 mK, opposite to the data. Therefore Eq. (4) is not the actual integral of Eq. (3) with the stated θ distribution, and the temperature dependence claimed as an independent check does not follow.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports an anomalous increase in resonant frequency at low in-plane magnetic fields in 7.9-nm TiN superconducting resonators, in contrast to the quadratic decrease expected from quasiparticle generation. It attributes the anomaly to a proposed 'spin-flip two-level system' (sTLS), in which an electron tunneling between two wells is coupled to magnetic defects whose inhomogeneous local fields mix charge tunneling and spin flips. A closed-form susceptibility, Eq. (4), is proposed and used together with a BCS quasiparticle term to fit frequency-field and frequency-temperature data, claiming quantitative reproduction with three parameters k, W, and t.","tokens_in":12284,"tokens_out":9528,"duration_ms":87851,"significance":"If correct, the sTLS mechanism would connect magnetic surface defects to TLS losses and would suggest magnetic-field engineering of decoherence in quantum devices. The experimental part has strengths: the anomaly is reported as reproducible across 12 resonators, hysteresis and vortex effects are discussed, and the high-temperature BCS baseline is consistent with known behavior. However, the central theoretical claim is compromised by the unshown reduction of Eq. (3) to Eq. (4) and by a temperature dependence that appears to contradict the stated θ-distribution. The model as presented is therefore not an independent prediction, and the quantitative statement is not supported in its current form.","major_comments":[{"comment":"The passage from the integral in Eq. (3) to the closed form in Eq. (4) is not derived. With the stated distribution Pθ for J < 0 and |J| >> kBT, the θ integral of the dipole factor (b sin(θ/2))² gives ⟨sin²(θ/2)⟩ ≈ kBT/(2|J|), a factor linear in T that is absent from the T-independent coefficient A in Eq. (4). Since ε ≈ sqrt((vB)² + t²) and the tanh factor saturates at low temperature, the resulting sTLS contribution would grow with temperature, whereas the data show the anomaly disappearing by 600 mK. As written, Eq. (4) is not the integral of Eq. (3), so the claimed quantitative reproduction and the temperature check do not follow. Please provide the full integration or revise the model so that the predicted temperature dependence matches the data, and redo the fits accordingly.","section":"§II, Eqs. (3)–(4)"},{"comment":"The replacement Pb = δ(b − b0) is asserted with only a reference to unshown numerical calculations ('we have verified through numerical calculations'), and the text concedes that Pb cannot be accurately defined. Because b enters both the sTLS energy and the electric-dipole matrix element, a broad Pb would change the field and temperature dependence of h(B,T). Without a derivation, a sensitivity analysis, or the actual numerical check, the three-parameter fit cannot discriminate the sTLS model from a generic empirical curve.","section":"§II, Pb definition"},{"comment":"The central claim of quantitative reproduction rests on fitting k, W, and t to the same 10-mK field sweep that defines the anomaly, and then refitting W and t for each temperature in Fig. 3D. There is no out-of-sample or parameter-free prediction; the statement that the temperature dependence is 'reproduced' is therefore overstated. A stronger test would be to fix W and t from one condition and predict the other curves, or to report residuals and parameter uncertainties for all 12 resonators.","section":"§III, Fig. 3C and 3D"}],"minor_comments":[{"comment":"The coefficient A is said to be 'positively related to nsTLS', but its exact relation to the microscopic parameters and to the integrals over Δ0 and θ is not given; please define it explicitly.","section":"Eq. (4)"},{"comment":"The numerical verification of the delta-function assumption for Pb is not shown; please include it in the Supplementary Material so that the assumption can be checked.","section":"§II, Pb paragraph"},{"comment":"The sTLS contribution in Fig. 3D is obtained by subtracting the BCS fit at 600 mK from data at all temperatures; please clarify why the modified formula's own quasiparticle contribution at each temperature is not used for this subtraction.","section":"§III, Fig. 3D"},{"comment":"Some references are cited indirectly, notably [42] for the θ-distribution and [46] for the sign of J; please cite primary sources for these assumptions so that the reader can verify them.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The experimental anomaly is interesting and potentially reproducible, but the theoretical derivation has a load-bearing gap: Eq. (4) does not appear to follow from Eq. (3) with the stated θ-distribution, and the predicted temperature dependence is opposite to the data. I would not reject outright because the empirical core may survive a corrected model, but the manuscript cannot be accepted without a complete derivation of Eq. (4) and a temperature behavior consistent with the data. The stress-test concern about the θ integral lands and is not a stylistic issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing you should know: the experiment is genuinely interesting. A reproducible low-field increase in resonant frequency in 12 ultra-thin TiN resonators, absent in perpendicular field, is not in the prior literature, and the authors have done the basic sanity checks (field direction, vortex pinning, hysteresis, reproducibility). The idea that spin-flip TLSs from magnetic defects couple to the resonator is physically plausible and connects to existing glass-model work. I'd want this on the record.\n\nThe soft spot is the theory. The step from Eq. (3) to Eq. (4) is not shown, and when you try to do it with the distributions stated, it does not work. With Pθ as given and J ~ -1 meV, θ is peaked near zero, and the dipole factor (b sin θ/2)^2 averages to roughly kB T/(2|J|). That is a temperature-dependent prefactor that grows about 60x between 10 mK and 600 mK for t ~ 0.5 K, while the tanh in Eq. (4) falls only ~2.5x. So the sTLS susceptibility should grow with temperature, not saturate as the data show. Eq. (4) has no such prefactor. Unless the authors can exhibit the integration, the claimed quantitative temperature dependence does not follow.\n\nThere are also smaller issues: the b distribution is replaced by a delta function after admitting it cannot be defined, the fits use the same data that define the anomaly, and the extracted shifts in Fig. 2B are small enough that missing error bars matter. The consistency of the fitted BCS coefficient k with the 600 mK value is encouraging, and the extracted t ~ 300–600 mK is sensible for GHz-frequency TLSs.\n\nNet: this is a valuable experimental paper wrapped in a theory that is not yet established. The observation deserves a serious referee. I'd send it out, but with a request for the full derivation of Eq. (4), a numerical check of the θ integration, and error bars on the fractional shifts. If the derivation cannot be produced, the paper should be reframed as an experimental report with a speculative model.","headline":"A reproducible and striking experimental anomaly in ultra-thin TiN resonators, but the central quantitative model has a load-bearing derivation gap: the stated θ distribution gives a temperature-dependent prefactor absent from Eq. (4), so the claimed temperature check does not follow as written.","tokens_in":12875,"tokens_out":5457,"would_cite":true,"duration_ms":48381,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Spin-flip two-level systems explain the anomalous resonator frequency rise at low magnetic fields.","keywords":["spin-flip two-level systems","magnetic defects","superconducting resonators","TiN thin films","effective spin-orbit coupling","frequency shift","quantum device noise","two-level system saturation"],"falsifier":"Measure the frequency-field curve on an ultra-thin TiN resonator before and after a surface treatment that removes magnetic defects: if the low-field frequency rise survives the treatment, the sTLS attribution fails. Alternatively, compare the extracted spin-flip rate $b_0 \\approx 10\\,\\mu\\mathrm{eV}$ with electron-spin-resonance spectra; the amplitude of the anomaly should track the density of the specific spin species, and a material with negligible surface magnetism should show no rise at all.","tokens_in":11831,"feed_emoji":"🧲","tokens_out":10317,"duration_ms":79438,"temperature":0.7,"pith_summary":"Ultra-thin titanium nitride superconducting resonators show an unexpected increase in resonant frequency when a small in-plane magnetic field is applied, in contrast to the quadratic decrease predicted by BCS theory. The paper attributes this anomaly to spin-flip two-level systems (sTLSs), in which an electron tunneling between two sites also flips its spin because the local magnetic field from nearby magnetic defects points differently at the two sites. A model with an effective spin-orbit coupling yields a closed-form expression for the sTLS contribution to the resonator frequency shift, and the resulting formula quantitatively reproduces the measured frequency-field curves and their temperature dependence. If correct, this connects the long-known magnetic-defect noise in quantum devices to the TLS losses that limit superconducting circuits, and suggests magnetic fields could be used to engineer those losses.","feed_headline":"Spin-flip defects lift resonator frequency at low fields","feed_subtitle":"Ultrathin TiN resonators trace an anomalous frequency rise to magnetic surface defects that flip spins during tunneling.","key_machinery":"The load-bearing object is the spin-flip TLS Hamiltonian $H_{\\mathrm{sys}} = \\frac{1}{2}(m\\sigma_z + 2\\Delta_0 \\tau_x) + b(|L\\rangle\\langle L|\\sigma_x + |R\\rangle\\langle R|\\sigma_\\theta)$, where $b$ is the spin-flip rate induced by the perpendicular component of the local magnetic field at each well, $\\theta$ is the relative angle between the local fields in the two wells, $\\Delta_0$ is the usual tunneling amplitude, and $m = g\\mu_B B$ is the Zeeman energy. This Hamiltonian converts the field-dependent spin-flip rate into a field-dependent energy splitting $\\varepsilon = \\sqrt{m^2 + 4b^2}$ for $\\theta \\approx 0$, and into an electric-dipole matrix element $p_{\\mathrm{eff}} = p \\sin\\Gamma$ with $\\sin\\Gamma \\approx (b \\sin(\\theta/2))/(\\Delta_0 - \\varepsilon^2/4\\Delta_0)$. Substituting the assumed distributions $P_{\\Delta_0} \\sim n_{\\mathrm{sTLS}}/\\Delta_0$, $P_{\\cos\\phi}$ uniform, $P_b = \\delta(b-b_0)$, and $P_\\theta$ from a negative-$J$ angular distribution into the susceptibility integral yields the closed form of Eq. (4), which then enters the observed frequency shift through Eq. (5). The key step is that the spin-flip rate $b$, not the tunnel splitting $\\Delta_0$, sets the low-energy scale of the defect.","core_discovery":"The paper's central discovery is that the anomalous low-field rise in frequency is not a quasiparticle effect, but the signature of a new class of two-level defects. In each sTLS, an electron tunnels between two wells while the perpendicular component of the local field from magnetic defects, oriented differently at the two wells, induces a spin flip; the inhomogeneous field thereby creates an effective spin-orbit coupling that mixes orbital and spin states. As a result, the sTLS energy takes the field-dependent form $\\varepsilon = \\sqrt{(g\\mu_B B)^2 + 4b^2}$ rather than the field-independent tunnel splitting of a conventional TLS, and its electric-dipole coupling is reduced by the spin-orbit mixing angle $\\Gamma$. Integrating the susceptibility of such systems over parameter distributions gives the closed form of Eq. (4), and adding the BCS quasiparticle term, $\\Delta f_r/f_r = kB^2 + h(B,T)$, fits the measured curves at 10 mK and at higher temperatures with only three parameters. The extracted zero-field sTLS energies lie between 300 and 600 mK and the implied magnetic defect density is around $10^{17}\\,\\mathrm{m}^{-2}$, placing these defects squarely in the surface layer of the TiN film.","pith_inferences":["The delta-function replacement for $P_b$ is a strong simplification; a direct way to test it would be to measure the frequency shift at fields well beyond the fitted range and check that the functional form of $h(B,T)$ still holds.","The model predicts a specific dependence on the direction of the in-plane field relative to the local disorder axes; rotating the field in the plane while keeping its magnitude fixed could reveal an anisotropic component of the anomaly that the current data do not address.","Because the sTLS susceptibility is nonresonant (the photon energy $hf_r$ is far from $\\varepsilon$), the same mechanism should also produce a small shift in the resonator's internal quality factor; measuring $Q_i$ as a function of $B$ and $T$ would provide an independent check.","The sTLS picture could be extended to other disordered superconductors that show surface magnetism; the model predicts that their sTLS spectra, and hence their optimal operating fields, depend on the specific defect species."],"forward_implications":["The sTLS zero-field energies (300–600 mK, i.e., 6–12 GHz) fall directly in the operating band of superconducting qubits and microwave resonators, so these defects contribute to loss and noise at device-relevant frequencies.","Because sTLSs couple to the electric field through the reduced dipole $p \\sin\\Gamma$, they saturate at higher drive power than conventional TLSs, which can explain the stronger loss suppression with magnetic field observed at intermediate power.","Polarizing defect spins with a magnetic field reduces the inhomogeneity of the local field and thereby weakens the effective spin-orbit coupling, suggesting a path to magnetically decouple sTLSs from the device.","An optimal magnetic field should exist that minimizes decoherence, balancing sTLS suppression against quasiparticle generation and other field-induced losses.","The estimated defect density of about $10^{17}\\,\\mathrm{m}^{-2}$ supports the view that surface magnetic defects, not bulk TLSs, are responsible for the anomaly, which is consistent with earlier surface-treatment experiments."],"supporting_citations":[{"why":"Supplies the standard formalism that converts the TLS electrical susceptibility into the resonator fractional frequency shift, which the sTLS model adapts.","marker":"[34]"},{"why":"Provides the conventional two-level-system tunneling model of glasses, the baseline that the spin-flip extension modifies.","marker":"[35, 36]"},{"why":"Shows that ultra-thin high-kinetic-inductance nanowire resonators maintain performance in in-plane magnetic fields, enabling the measurements.","marker":"[18]"},{"why":"Demonstrates that removing interface magnetic defects changes TLS-related loss and charge noise, motivating the connection between spins and TLSs.","marker":"[9]"},{"why":"Gives the classical treatment of the spin exchange interaction as a local magnetic field acting on the electron, used to define the spin-flip rate.","marker":"[41]"},{"why":"Supplies the angular distribution $P_\\theta$ of the relative local-field angle, used to justify the $\\theta\\approx 0$ limit for negative defect-defect coupling $J$.","marker":"[42]"}],"fun_headline_variants":["Spin-flip defects explain resonator's odd low-field frequency rise","Magnetic surface defects flip spins to boost resonator frequency","Spin-flip TLSs: key to anomalous resonator frequency shift","Ultrathin resonators expose spin-flip two-level systems","Effective spin-orbit coupling flips resonator frequency behavior"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calculation replaces the unknown distribution of spin-flip rates by a single constant $b_0$, treating $P_b = \\delta(b-b_0)$, and assumes the relative local-field angle $\\theta$ is near zero because the defect-defect coupling $J$ is negative, so that $\\varepsilon = \\sqrt{m^2 + 4b^2}$; if the real distributions of $b$ or $\\theta$ are broad, the predicted field and temperature dependence would differ from the fitted form.","fun_headline_variants_meta":{"raw":{"variants":["Spin-flip defects explain resonator's odd low-field frequency rise","Magnetic surface defects flip spins to boost resonator frequency","Spin-flip TLSs: key to anomalous resonator frequency shift","Ultrathin resonators expose spin-flip two-level systems","Effective spin-orbit coupling flips resonator frequency behavior"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000221,"raw_usage":{"total_tokens":1455,"prompt_tokens":952,"completion_tokens":503,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":568,"completion_tokens_details":{"reasoning_tokens":420}},"tokens_in":568,"tokens_out":503,"duration_ms":4731,"temperature":1.0,"reasoning_tokens":420,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T11:01:53.287578+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the frequency-field curve on an ultra-thin TiN resonator before and after a surface treatment that removes magnetic defects: if the low-field frequency rise survives the treatment, the sTLS attribution fails. Alternatively, compare the extracted spin-flip rate $b_0 \\approx 10\\,\\mu\\mathrm{eV}$ with electron-spin-resonance spectra; the amplitude of the anomaly should track the density of the specific spin species, and a material with negligible surface magnetism should show no rise at all.","supporting_citations":[{"cited_title":"Gao, The Physics of Superconducting Microwave Res- onators, Ph.D","cited_arxiv_id":null,"evidence_quote":"Supplies the standard formalism that converts the TLS electrical susceptibility into the resonator fractional frequency shift, which the sTLS model adapts."},{"cited_title":"Samkharadze, A","cited_arxiv_id":null,"evidence_quote":"Shows that ultra-thin high-kinetic-inductance nanowire resonators maintain performance in in-plane magnetic fields, enabling the measurements."},{"cited_title":"Jayaraman, A","cited_arxiv_id":null,"evidence_quote":"Demonstrates that removing interface magnetic defects changes TLS-related loss and charge noise, motivating the connection between spins and TLSs."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the angular distribution $P_\\theta$ of the relative local-field angle, used to justify the $\\theta\\approx 0$ limit for negative defect-defect coupling $J$."}],"review_version":1}