{"id":"415156fe-017c-4f3b-bd59-33ee68b2b8bd","arxiv_id":"2607.21329","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Fluxonium decoherence under quasiparticle injection is accurately described only when the asymmetric superconducting gaps across the junctions are included, reconciling previously contradictory quasiparticle-density bounds.","lead":"This experiment injects quasiparticles into a fluxonium qubit and measures how fast it gets excited and relaxes at different magnetic flux values. The data show that an often-ignored difference in superconducting energy gaps across a junction is essential to explaining the rates, and it may resolve earlier conflicting quasiparticle-density estimates.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"ΔΓ↓ flux dependence alone cannot distinguish δΔ+equal-density from δΔ=0+unequal-density; the choice rests on the unverified thermal Boltzmann T_qp from the transmon.","rationale":"The paper is a serious experimental study: controlled injection, background subtraction, independent δΔ estimation from both film thickness and the ΔΓ02 resonance, and a one-parameter fit that captures the main ΔΓ↓ feature. These are real strengths. The single most load-bearing weakness, however, is model degeneracy. The external-flux dependence of ΔΓ↓ alone cannot distinguish the paper's δΔ+equal-density interpretation from the earlier δΔ=0+unequal-density interpretation; the discrimination is provided almost entirely by the thermal Boltzmann distribution inferred from a transmon detailed-balance ratio. The reader's weakest-assumption statement already identifies the thermal/equal-density choice and the Appendix K2 athermal alternative; our attack sharpens it by framing it as an explicit model-selection problem. We do not see this as grounds to reject or unverdict the paper: the authors acknowledge the athermal possibility, and their physical arguments against it are plausible. But the concern is genuine and addressable, so the reader's CONDITIONAL verdict remains appropriate. We recommend no change to the verdict; the concrete test above would settle whether the concern actually lands.","tokens_in":34516,"tokens_out":8429,"duration_ms":101472,"concrete_test":"Reanalyze the raw Q1 population traces with a 4-level rate model that uses the simulated dispersive shifts of Fig. 13 to assign readout probabilities for states |3⟩–|6⟩, then perform a nested model comparison between (A) δΔ≠0 with x_array=x_small and Eq. (4), and (B) δΔ=0 with x_array free, fitting both to the leakage-corrected ΔΓ↓ and ΔΓ↑. If model B is not rejected at ΔAIC>4, the experiment has not isolated δΔ as essential. An alternative decisive check is to measure the transmon ΔΓ↑/ΔΓ↓ ratio at several transmon frequencies on the same chip: if the ratio is not described by a single k_BT_qp=1.2 GHz exponential, the thermal-distribution assumption underlying the model selection is falsified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that δΔ is essential for the HFQ suppression is not uniquely supported by the de-excitation data. The Q1 ΔΓ↓ data in Fig. 4(c) can be fit nearly equally well by (i) the paper's δΔ≠0, x_array=x_small model (one parameter) and (ii) the δΔ=0, x_array≠x_small simplified model of Appendix H (Table III gives x_small=2.1e-6, x_array=0.4e-6, with visually comparable fits in Fig. 9). The paper rejects model (ii) not by a direct statistical comparison of the fits to ΔΓ↓, but through the argument in Sec. IV A that δE_qp≪ℏω01 is violated because ΔΓ↑ is finite and because the co-fabricated transmon gives k_BT_qp/h=1.2 GHz via a detailed-balance ratio. That argument is the load-bearing step. It assumes the injected QPs are a single Boltzmann distribution with the same T_qp in both leads of both devices; Appendix K2 demonstrates that a narrow athermal distribution (δE_qp/h=0.20 GHz, δμ/δΔ=0.81) also reproduces the full ΔΓ↓/ΔΓ↑ dataset with δΔ fixed. The authors reject this alternative using the transmon-inferred width and relaxation-time estimates, but neither directly measures the QP distribution in the fluxonium array. The ΔΓ02 resonance near ω02/2π=1.72 GHz independently supports the presence of δΔ, but it does not establish that δΔ, rather than unequal QP densities, is what controls the HFQ peak-height comparison. The 'essential role of δΔ' therefore rests on an unverified thermal/equal-density prior.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports measurements of quasiparticle (QP) induced excitation and de-excitation rates in a fluxonium qubit under controlled on-chip QP injection. The authors isolate QP-induced rates by measuring quasi-instantaneous transition rates with and without injection, and map their dependence on external flux. The central claim is that a superconducting gap difference δΔ between the two leads of each Josephson junction, estimated independently from film thickness and from a resonant enhancement of the |0⟩→|2⟩ transition, plays an essential role in QP-induced transitions: near the half-integer flux quantum (HFQ), where the qubit frequency is small compared with δΔ, the gap difference suppresses the de-excitation rate. After including δΔ and assuming equal QP densities in the small junction and the array, a one-parameter fit reproduces the external-flux dependence of the measured de-excitation rate ΔΓ↓, yielding x_qp = 3.6×10^-6, consistent with an independent transmon estimate. The measured excitation rate ΔΓ↑ is only qualitatively reproduced; the discrepancy is attributed to leakage to higher qubit states that are misidentified during readout. An alternative athermal QP distribution model (Appendix K2) is shown to also reproduce the data, but the authors argue it is inconsistent with the transmon-inferred QP temperature and with relaxation-time expectations.","tokens_in":35007,"tokens_out":4404,"duration_ms":48571,"significance":"If the central claim holds, the work resolves a long-standing puzzle in fluxonium research: the apparent order-of-magnitude disparity between QP density bounds in the small junction and the junction array could be an artifact of neglecting gap asymmetry. It also demonstrates that gap engineering can suppress QP-induced decoherence in fluxonium, extending similar results in transmon qubits. The experimental methodology — controlled injection with quasi-instantaneous rate extraction and background subtraction — is a valuable contribution, and the independent determination of δΔ from the |0⟩→|2⟩ resonance is a particularly clean result. The paper is thorough, with extensive appendices, and the theoretical framework is carefully connected to prior work. However, the uniqueness of the δΔ interpretation is not fully established, because the same ΔΓ↓ data can be fit by a δΔ=0 model with unequal densities, and the alternative athermal model reproduces both ΔΓ↓ and ΔΓ↑. The central claim is therefore plausible and well supported but not uniquely determined by the presented data.","major_comments":[{"comment":"The rejection of the δΔ=0, x_array≠x_small simplified model is not based on a quantitative statistical comparison of the fits to ΔΓ↓. The simplified model (Appendix H, Fig. 9) and the δΔ≠0 model (Fig. 4c) appear visually similar, and Table III reports different fitted densities but no fit-quality metrics (e.g., chi-square, AIC, or residuals). Since the central claim is that δΔ is 'necessary' to model the flux dependence, the paper should provide a direct comparison of goodness-of-fit for the two models on the ΔΓ↓ data. Without this, the claim that δΔ, rather than unequal densities, is essential is underdetermined by the de-excitation data alone.","section":"Sec. IV C and Appendix H, Table III, Fig. 9"},{"comment":"The load-bearing step that rules out δΔ=0 is the argument that δE_qp≪ℏω01 is violated, based on the finite ΔΓ↑ and the transmon detailed-balance estimate k_B T_qp/h = 1.2 GHz. However, Appendix K2 presents an alternative narrow athermal distribution (δE_qp/h = 0.20 GHz, δμ/δΔ = 0.81) that reproduces both ΔΓ↓ and ΔΓ↑ with δΔ fixed and equal densities. The authors reject this alternative using the transmon-inferred width and relaxation-time estimates, but neither directly measures the QP energy distribution in the fluxonium array. To make the 'essential role of δΔ' claim conclusive, the paper should either provide a direct experimental constraint on the fluxonium QP distribution or weaken the wording from 'essential/necessary' to 'consistent with' and acknowledge that the data do not uniquely distinguish the thermal equal-density model with δΔ≠0 from the athermal model with δΔ≠0.","section":"Sec. IV A and Appendix K2"},{"comment":"The quantitative discrepancy in ΔΓ↑ is attributed to leakage to higher qubit states that are misidentified as |1⟩. This explanation is plausible, and the three-level analysis in Appendix J shows that including |2⟩ does not resolve the inversion. However, no quantitative model of the apparent ΔΓ↑ under leakage is presented; the paper states that such a description 'would require additional parameters... beyond the scope of the present work.' Since the finite ΔΓ↑ is part of the evidence against δE_qp≪ℏω01, and since the leakage hypothesis is not directly verified, this is an acknowledged limitation of the excitation-rate analysis. The paper should state more prominently that the leakage scenario is a hypothesis, not a tested correction, and that the central conclusions rest on the ΔΓ↓ data.","section":"Sec. IV D and Appendix K1"}],"minor_comments":[{"comment":"The caption says the red dotted curve is the prediction for ΔΓ↑, but the curve is computed without leakage corrections; consider adding 'in the two-level model' or 'neglecting leakage to higher states' for clarity.","section":"Fig. 4(c) caption"},{"comment":"The third column reports x_qp from the δΔ≠0 model, but for Q1 the main-text one-parameter fit gives 3.6×10^-6 while Table III gives 3.1×10^-6 from a two-parameter fit with T_qp free. The difference should be noted in the table caption or text to avoid confusion.","section":"Table III"},{"comment":"The transmon injection response that yields k_B T_qp/h = 1.2 GHz is stated in the main text; please specify explicitly that this measurement uses the same injection parameters (V_inj = 5.0 Δ_Al/e, t_inj = 4 μs) as the data in Fig. 4, or note the difference if not.","section":"Sec. IV A"},{"comment":"The text equates δE_qp/h with the Boltzmann temperature T_qp of the athermal model, but the distribution also has a chemical potential offset. Clarify how δE_qp is defined for this model, especially because the comparison to the transmon width δE_qp/h ≈ 1.2 GHz assumes a common definition.","section":"Appendix K2, Eq. (K1)"},{"comment":"The sentence 'This similarity deserves a comment' is informal; consider rewording to 'This equality is notable' or similar.","section":"Sec. V"}],"recommendation":"major_revision","confidential_remarks":"The paper is well-executed and the experimental methodology is strong, but the central claim about the 'essential role' of δΔ is not uniquely established by the data as presented. The major comments ask for a quantitative model comparison and a tempering of the claim, which should be feasible within the scope of a revision. The alternative athermal model in Appendix K2 is a serious competing explanation that the current arguments do not fully exclude. If the authors can provide a statistical comparison of the two fits to ΔΓ↓ and/or a direct constraint on the fluxonium QP distribution, the paper would be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: this is a well-executed experiment with a plausible central claim—gap asymmetry δΔ, not unequal quasiparticle densities, explains why the HFQ peak in ΔΓ↓ is not three to four times the IFQ peak. The controlled on-chip injection plus quasi-instantaneous rate extraction is genuinely new for fluxonium, and the measurement of a resonant ΔΓ02 peak at ω02/2π≈1.72 GHz is a nice independent handle on δΔ. I came away believing the main conclusion is probably right.\n\nWhat the paper does well: it isolates QP-induced rates by subtracting background and fitting short-time dynamics rather than doing a standard T1; it separately extracts T_qp from a co-fabricated transmon; it shows the simplified equal-gap model (Appendix H) gives x_array << x_small and then demonstrates that accounting for δΔ removes that apparent imbalance. The four-qubit comparison in the appendices gives the analysis some breadth. The treatment of the apparent rate inversion is honest: they do not claim to resolve it quantitatively, and they flag leakage as the likely cause while testing an athermal alternative.\n\nSoft spots, in proportion. The stress-test note is right that the flux dependence of ΔΓ↓ alone cannot distinguish δΔ + equal densities from δΔ = 0 + unequal densities. It is also right that the load-bearing thermal-Boltzmann T_qp comes from the transmon, not from a direct measurement of the fluxonium QP distribution. But the paper does not merely assert this—the finite ΔΓ↑ and the independent ΔΓ02 resonance provide some support, and the athermal model in Appendix K2 is rejected on physical timescale grounds. That rejection is reasonable, not airtight. A skeptic can still demand a direct measurement of the QP energy distribution or a leakage-resolved readout. The x_array/x_small = 1.2 result is presented without uncertainty, which is a smaller but real gap. And no data or code are released, so the extraction procedure cannot be independently checked. All of these are addressable; none looks fatal to the δΔ explanation.\n\nWho this is for: people working on quasiparticle loss in fluxonium and on gap engineering generally. It deserves a serious referee—I would send it out. My own verdict would be conditional: accept after the authors provide error bars on the density ratio and either quantify leakage with a higher-state readout or soften the claim on the excitation-rate comparison.","headline":"Careful, novel fluxonium injection experiment in which gap asymmetry plausibly resolves the apparent QP-density imbalance, though the thermal-QP assumption and one unquantified ratio keep it from being fully airtight.","tokens_in":35469,"tokens_out":2114,"would_cite":true,"duration_ms":25441,"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":"Quasiparticle-induced transitions in a fluxonium qubit are governed by the superconducting gap difference between the junction leads, not by an imbalance in quasiparticle densities.","keywords":["fluxonium qubit","quasiparticle-induced decoherence","superconducting gap asymmetry","quasiparticle tunneling","quasiparticle injection","Josephson junction array","qubit energy relaxation","transmon"],"falsifier":"Measure the quasiparticle-induced transition rates with a readout that resolves states |3⟩ and higher: if the excess excitation rate near the half-integer flux quantum persists after separating leakage states, the leakage explanation fails and the thermal-Boltzmann-plus-gap-difference model would need revision; conversely, if the |0⟩→|1⟩ excitation rate drops to the predicted value, the model is confirmed. A more direct test of the δΔ mechanism would be to vary the lead-thickness asymmetry on co-fabricated devices and check that the half-integer-flux peak height tracks δΔ − ℏω01.","tokens_in":34433,"feed_emoji":"❄️","tokens_out":4679,"duration_ms":46860,"temperature":0.7,"pith_summary":"Fluxonium qubits are hard to protect from quasiparticles because, unlike transmons, they are insensitive to offset charge, so charge-parity detection is impossible. This paper measures quasiparticle-induced excitation and de-excitation rates under controlled on-chip injection and finds that the de-excitation rate as a function of magnetic flux has two comparable peaks, at the integer and half-integer flux quantum. The standard assumptions—equal superconducting gaps in the two junction leads and a narrow quasiparticle energy distribution—cannot explain this flux dependence. The authors' central claim is that the gap difference δΔ between the leads suppresses quasiparticle tunneling near the half-integer flux point, where the qubit transition energy is small; once δΔ is included, a one-parameter fit with equal quasiparticle densities in the small junction and the array reproduces the full flux dependence and yields x_qp = 3.6×10^-6. If correct, this resolves a long-standing apparent disparity between quasiparticle-density bounds for the small junction versus the junction array and suggests that gap engineering can suppress quasiparticle decoherence in fluxonium.","feed_headline":"Gap mismatch explains fluxonium quasiparticle rates","feed_subtitle":"Equal densities in the small junction and array reproduce all transition-rate data, once gap asymmetry is included.","key_machinery":"The central object is the quasiparticle tunneling structure factor S±(ω) for a Josephson junction with unequal superconducting gaps ΔL and ΔH in its two leads. It enters the transition-rate formula together with flux-dependent matrix elements of sin(φ/2), cos(φ/2), and φ/2, which weight the small-junction and array contributions. The key mechanism is the combination of the gap difference δΔ = ΔH − ΔL with the energy width of the quasiparticle distribution: for de-excitation near the half-integer flux quantum, the condition k_BT_qp ≲ δΔ − ℏω01 leaves too few quasiparticle states available, suppressing the rate. The paper separately measures δΔ from a resonant peak in the |0⟩→|2⟩ rate when ℏω0","core_discovery":"The central finding is that the superconducting gap difference δΔ between the two leads of a Josephson junction, which arises naturally when the aluminum films have different thicknesses, is essential for modeling quasiparticle-induced transitions in fluxonium. When the qubit transition energy ℏω01 is small compared with δΔ—which happens near the half-integer flux quantum—the gap difference strongly suppresses the quasiparticle-induced transition rate, making the half-integer-flux peak comparable to the integer-flux peak rather than several times larger. With δΔ/h = 1.72 GHz independently determined from a resonantly enhanced |0⟩→|2⟩ transition, and an effective quasiparticle temperature k_B","pith_inferences":["The paper's own Appendix K2 shows that an athermal quasiparticle distribution with different chemical potentials per lead can also reproduce the same data; a clean test would be to measure the quasiparticle distribution width directly, for example by probing transitions out of higher fluxonium states after injection.","The resonant |0⟩→|2⟩ enhancement when ℏω02 = δΔ offers a new spectroscopic probe of junction gap asymmetry that could be applied to other fluxonium devices without destructive fabrication measurements.","Because δΔ depends on the thickness of the aluminum films, fabricating junctions with a controlled lead-thickness asymmetry could be used as a deliberate engineering knob to suppress quasiparticle-induced transitions at the half-integer flux point; the paper demonstrates the effect but does not perform such a systematic test.","Re-analyzing historical fluxonium T1 data with δΔ ≠ 0 and a finite quasiparticle energy width might change inferred quasiparticle-density bounds for devices where the two junction leads have different thicknesses."],"forward_implications":["Previous bounds on quasiparticle densities in fluxonia that assumed δΔ = 0 overestimated the array's sensitivity near the half-integer flux quantum; accounting for δΔ makes the inferred bound less tight, so the reported x_array ≪ x_small disparity may be an artifact of the model.","Gap engineering, already demonstrated in transmons, should also suppress quasiparticle-induced decoherence in fluxonium qubits.","A single quasiparticle density shared by the small junction and the junction array, x_qp = 3.6×10^-6, reproduces the full flux dependence of the measured de-excitation rate, and this value is consistent with the independently inferred transmon density of 3.0×10^-6.","The measured quasiparticle-induced excitation rate exceeds the model prediction near the half-integer flux quantum; the paper attributes this to leakage into higher qubit states misidentified as |1⟩ during readout, which motivates readout schemes that resolve those states.","The fluxonium results underscore that all relevant energy scales—ℏω01, δΔ, and δE_qp—must be retained when analyzing quasiparticle-induced relaxation in low-frequency qubits."],"fun_headline_variants":["Gap asymmetry settles fluxonium quasiparticle puzzle","Fluxonium rates: gap mismatch is the missing piece","Half-flux peak traceable to junction gap difference","Gap asymmetry tames quasiparticle noise in fluxonium","Fluxonium quasiparticle rates need gap asymmetry"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The central fit assumes the injected quasiparticles follow a thermal Boltzmann energy distribution with a single effective temperature for both junction leads (Eq. 4), and that the quasiparticle density is the same in the small junction and the array; if the distribution is genuinely athermal, the same data can be reproduced by a different model, as the paper's own Appendix K2 demonstrates.","fun_headline_variants_meta":{"raw":{"variants":["Gap asymmetry settles fluxonium quasiparticle puzzle","Fluxonium rates: gap mismatch is the missing piece","Half-flux peak traceable to junction gap difference","Gap asymmetry tames quasiparticle noise in fluxonium","Fluxonium quasiparticle rates need gap asymmetry"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000203,"raw_usage":{"total_tokens":1178,"prompt_tokens":656,"completion_tokens":522,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":400,"completion_tokens_details":{"reasoning_tokens":443}},"tokens_in":400,"tokens_out":522,"duration_ms":6073,"temperature":1.0,"reasoning_tokens":443,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T07:45:33.358466+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the quasiparticle-induced transition rates with a readout that resolves states |3⟩ and higher: if the excess excitation rate near the half-integer flux quantum persists after separating leakage states, the leakage explanation fails and the thermal-Boltzmann-plus-gap-difference model would need revision; conversely, if the |0⟩→|1⟩ excitation rate drops to the predicted value, the model is confirmed. A more direct test of the δΔ mechanism would be to vary the lead-thickness asymmetry on co-fabricated devices and check that the half-integer-flux peak height tracks δΔ − ℏω01.","supporting_citations":[],"review_version":1}