{"id":"c092cdec-9a3b-4e03-bf89-c030d383ae50","arxiv_id":"2412.11280","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Fabrication optimizations yield Josephson parametric converters and amplifiers with internal quality factors exceeding 100,000 at the single-photon level, a record for tunable Josephson circuits.","lead":"This paper reports superconducting microwave converters and amplifiers with internal quality factors above 100,000 at single-photon power, a record for tunable Josephson devices. The result matters because low-loss parametric devices are central to quantum computing, microwave quantum networks, and ultra-low-noise amplification.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The JPA record Qint is not independently measured: the scattering fit is constrained by assuming JPA and JPC losses are identical, and the squeezing fit uses Qint as a free parameter; the written bound is also an impossible inequality.","rationale":"The reader's verdict is well-founded. I read the full manuscript and the central claim has two evidentiary pillars for the JPA: scattering and squeezing. Both have a weakness. The scattering pillar is explicitly built on the assumption of identical losses; the text itself notes that an accurate direct estimate is difficult and a coupling series would be needed. Because JPC and JPA have different circuits (JRM vs dc-SQUID, λ/2 vs λ/4, different coupling ports, extra pump line), 'identical conditions' is not a guarantee of identical Qint. This is a material gap, not a stylistic one. The squeezing pillar is presented as a confirmation, but it does not provide an independent measurement of Qint unless the fit is sensitive to it. With four free parameters, a profile-likelihood test is needed. I also note the impossible inequality in the constraint, which must be corrected for the scattering analysis to be reproducible. Despite these issues, the paper contains a detailed fabrication process and credible JPC measurements, so the appropriate disposition is the same conditional accept recommended by the reader.","tokens_in":17517,"tokens_out":5730,"duration_ms":52222,"concrete_test":"Refit the JPA squeezing and purity data with Qint fixed to 3×10^4, 5×10^4, 1×10^5, and 3×10^5, allowing the other parameters (χ(2), n'_J, δ, and the temperatures within the stated 10-50 mK range) to float, and report the minimized χ² or likelihood profile. If the goodness of fit changes by less than the noise level across this Qint range, the squeezing data cannot distinguish these loss levels and cannot confirm the >10^5 JPA claim; if χ² worsens sharply for Qint below 1×10^5, the confirmation is real.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section IV's JPA scattering analysis: after acknowledging that the JPA is 'strongly overcoupled' and 'an accurate estimation of Qint is difficult' because the resonance shape is dominated by external coupling, the authors do not follow the stated need for a coupling-strength series; instead they assume that JPC and JPA samples fabricated and measured under identical conditions have the same Qint, and use this assumption to constrain the JPA scattering fit. The resulting Qint = 1.1-1.5 × 10^5 is therefore not measured for the JPA but transferred from the JPC. The same section contains an impossible constraint, 1 × 10^-5 < Q_l^-1 - Q_ext^-1 < 5 × 10^-6, since 1 × 10^-5 > 5 × 10^-6; as written the constrained minimization has an empty feasible region. The later squeezing fit is called a confirmation, but there κint is one of four free parameters (χ(2), n'_J, δ free; Tatt, Tmxc only bounded to 10-50 mK), so agreement does not by itself fix Qint unless a profile-likelihood or identifiability check shows the fit is sensitive to it. Because JPC and JPA differ in circuit topology (JRM vs dc-SQUID, λ/2 vs λ/4, extra pump line), equal Qint is an assumption, not a consequence of identical processing.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports fabrication and cryogenic characterization of Josephson parametric converter (JPC) and Josephson parametric amplifier (JPA) devices, claiming internal quality factors Q_int exceeding 10^5 in the single-photon regime. The JPC is a λ/2 resonator coupled through a Josephson ring modulator; the JPA is a λ/4 resonator terminated by a dc-SQUID. The authors extract Q_int from reflection scattering data using circle fits and background corrections, and additionally analyze squeezing and purity of squeezed vacuum states generated by the JPA to support the extracted loss rates. The central claim is that both device types reach record-low internal loss, comparable to fixed-frequency resonators.","tokens_in":17857,"tokens_out":3644,"duration_ms":35130,"significance":"If substantiated, the claimed Q_int > 10^5 for tunable Josephson parametric devices would be a meaningful advance: low internal loss directly improves squeezed-state purity, amplifier noise performance, and the fidelity of microwave quantum operations. The paper provides a detailed fabrication protocol, including surface treatments and argon milling steps, which is valuable for reproducibility. The JPC analysis follows a standard circle-fit procedure to scattering data and is credible. However, the JPA-specific evidence is weaker: the scattering-based JPA Q_int relies on an explicit assumption that JPC and JPA have equal internal loss, and the written constraint bounding Q_l^{-1} - Q_ext^{-1} is mathematically impossible as stated. The squeezing fit, presented as confirmation, uses Q_int as a free parameter, so it does not independently establish the JPA loss rate without a sensitivity or identifiability analysis. The paper would be strengthened by correcting the constraint and providing such an analysis.","major_comments":[{"comment":"The constraint used in the constrained minimization is written as 1 × 10^{-5} < Q_l^{-1} − Q_ext^{-1} < 5 × 10^{-6}. Since Q_int^{-1} = Q_l^{-1} − Q_ext^{-1} and 1 × 10^{-5} > 5 × 10^{-6}, no value satisfies both inequalities; the feasible region is empty. The quoted JPA Q_int = 1.1 × 10^5–1.5 × 10^5 from scattering therefore cannot be a valid output of the procedure as described. Please correct the bound or re-analyze without it, and state clearly what constraint was actually enforced.","section":"§IV, JPA scattering analysis"},{"comment":"The JPA scattering Q_int is not independently measured. The text acknowledges that the JPA is strongly overcoupled and that an accurate Q_int estimate is difficult, then assumes that JPC and JPA samples fabricated and measured under identical conditions have the same Q_int. The JPC and JPA differ in nonlinear element (JRM vs dc-SQUID), resonator type (λ/2 vs λ/4), and the presence of a separate pump line, so equal internal loss is an untested assumption rather than a consequence of identical processing. A series of resonators with varying external coupling, or an independent loss measurement, is needed to support the JPA scattering value.","section":"§IV, paragraph beginning 'Our JPA is deliberately designed'"},{"comment":"The squeezing and purity fits are presented as confirmation of the JPA Q_int, but Q_int (κ_int) is one of four free fit parameters, together with χ^(2), n'_J, and δ, while T_att and T_mxc are only bounded to 10–50 mK. Agreement of a multi-parameter fit does not by itself establish Q_int unless the fit is shown to be sensitive to it. Please provide profile likelihoods, confidence intervals, or another identifiability check for κ_int. Without such an analysis, the stated 'excellent agreement' does not confirm the scattering-derived value.","section":"§IV, squeezing analysis, Eq. (8)–(10)"},{"comment":"The abstract states that Q_int values 'for both Josephson parametric converters and Josephson parametric amplifiers in the single-photon regime' are obtained 'by fitting the scattering data.' For the JPA, the scattering fit is constrained by the JPC-based assumption, and the only device-specific JPA estimate comes from the squeezing fit with free κ_int. The wording overstates the independence of the JPA scattering result and should be revised to match the actual analysis.","section":"Abstract and Conclusion"}],"minor_comments":[{"comment":"There is a typo in the sentence defining the thermal photon numbers: 'Tmsc' should presumably be 'Tmxc' to match the earlier notation.","section":"§IV, thermal bath paragraph"},{"comment":"The text says 'Using previously determined values of κ_ext and experimentally measured temperatures' but then states that T_att and T_mxc are constrained within 10–50 mK and fitted. Please clarify whether these temperatures are measured or treated as free parameters with bounded ranges.","section":"§IV, temperature treatment"},{"comment":"The caption states that red stars refer to both JPC and JPA devices, but the figure does not appear to distinguish them. Please add separate markers or labels so the reader can identify which point corresponds to which device.","section":"Fig. 5(b)"},{"comment":"For the JPC, the text states Q_ext < Q_int (slightly overcoupled) and later reports Q_ext ≈ 3.9–4.1 × 10^4 and Q_int ≈ 1.1–1.34 × 10^5; this is consistent, but it would be helpful to state the coupling efficiency Q_ext/Q_l explicitly for the JPA as well, since that quantity is used in Fig. 5(b).","section":"§IV, JPC coupling regime"}],"recommendation":"major_revision","confidential_remarks":"The fabrication details and the JPC Q_int measurement are solid and worth publishing, but the JPA record claim is currently not supported as written because of the impossible constraint, the unvalidated equal-Q_int assumption, and the free-parameter nature of the squeezing fit. I would not reject the paper if the authors can correct the constraint, re-analyze or reframe the JPA scattering result, and provide a sensitivity or profile-likelihood analysis for κ_int in the squeezing fit."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe JPC half of this paper is a real advance: internal Qint above 1e5 at single-photon power for a flux-tunable Josephson parametric converter, extracted with a standard circle-fit that looks credible. The JPA half is shakier: the record claim rests on an assumed equality of JPC and JPA losses and a squeezing fit where Qint is a free parameter. That does not sink the paper, but it needs fixing.\n\nWhat is genuinely good: the fabrication section is detailed and reproducible (surface treatment, Ar+ milling, Al bandaging), and the JPC data are systematic—Qint versus flux and photon number, with a sensible power dependence. The comparison plot against literature is useful, and the claimed JPC values sit well above prior tunable devices. I would trust the JPC measurement.\n\nSoft spots, in order of severity. First, the JPA scattering analysis: after correctly noting that a strongly overcoupled resonator hides Qint, the authors do not measure a coupling series; they assert equal JPC/JPA loss because both are fabricated and measured under identical conditions. But the circuits differ topologically (JRM vs dc-SQUID, half-wave vs quarter-wave, extra pump line), so equal loss is an assumption, not a consequence. The bound they write, 1e-5 < Q_l^-1 - Q_ext^-1 < 5e-6, is empty as written because the lower limit exceeds the upper. That looks like a typo, but it must be corrected. Second, the squeezing fit: Qint is one of several free parameters (alongside chi^(2), n'_J, delta), with bath temperatures only bounded. The good agreement (S=11.75 dB, purity ~99%) is consistent with low loss, but it is not an independent confirmation. A profile-likelihood or sensitivity check showing the fit actually constrains Qint would materially strengthen the JPA claim.\n\nNet: this is a solid fabrication paper with one very good measurement (JPC) and one overreaching claim (JPA). It deserves peer review, not desk rejection. I would ask for a revision that fixes the inequality, reports uncertainties on Qint for both devices, and either adds a coupling-series measurement for a JPA or reframes the JPA Qint as consistent with 1e5 rather than independently measured. The fabrication community will want to read this regardless.\n\nRecommendation: send to review, conditional on the above.","headline":"The JPC internal quality factor is a credible record; the JPA claim needs an independent measurement or a softer claim.","tokens_in":18441,"tokens_out":2056,"would_cite":true,"duration_ms":18643,"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":"The authors claim a refined surface-treatment and fabrication recipe yields Josephson parametric devices with internal quality factors above 10^5 at the single-photon level, confirmed by squeezed-state purity.","keywords":["Josephson parametric amplifier","Josephson parametric converter","internal quality factor","single-photon regime","microwave squeezing","superconducting fabrication","surface treatment"],"falsifier":"Fabricate a second JPA with a much smaller coupling capacitance so that Qext is comparable to Qint, then fit its background-corrected reflection circle without imposing the converter-derived bound; a directly fitted Qint below 100,000 would contradict the paper's central claim.","tokens_in":17326,"feed_emoji":"⚛️","tokens_out":6974,"duration_ms":62617,"temperature":0.7,"pith_summary":"The paper reports a fabrication recipe for two kinds of flux-tunable Josephson microwave devices, a parametric converter and a parametric amplifier, that reaches internal quality factors above $10^{5}$ when probed with roughly one photon in the resonator. Internal loss is the main barrier to clean operation of these devices, because it injects noise and degrades squeezed microwave states. The authors extract the quality factor from calibrated scattering measurements and independently confirm it by fitting the measured purity of squeezed vacuum states generated by the amplifier. If the claim is correct, tunable Josephson circuits no longer have to be much lossier than fixed-frequency superconducting resonators.","feed_headline":"Tunable Josephson devices reach internal Q factors above 100,000","feed_subtitle":"A refined surface treatment cuts loss in parametric amplifiers and converters, lifting squeezed-state purity near 99 percent.","key_machinery":"The devices are a Josephson parametric converter (two half-wavelength coplanar-waveguide resonators coupled through an inductively shunted Josephson ring modulator) and a Josephson parametric amplifier (a quarter-wavelength resonator shorted to ground through a dc-SQUID). The argument is carried by two independent probes of loss. One is a calibrated scattering-parameter analysis in which cable delay, background, impedance mismatch, and Fano distortion are removed so that the corrected complex resonance circle can be fit to extract the total and external loss rates. The other is a two-bath input-output model of the amplifier whose quadrature variances depend on pump power through the internal loss rate; fitting measured squeezing and purity gives $Q_\\mathrm{int}$ without using the scattering constraint. On the fabrication side, piranha and buffered-oxide etches plus argon-ion milling clean the substrate-metal, metal-air, and metal-metal interfaces that otherwise host lossy defects.","core_discovery":"The central claim is that a combination of wet-chemical surface treatments, e-beam-defined Al/AlOx/Al Manhattan junctions, and optimized Ar+ milling for the Nb-Al contact produces Josephson parametric devices whose internal quality factors exceed 100,000 at the single-photon level. For the converter, fits to background-corrected reflection data give $Q_\\mathrm{int}$ between $1.1\\times10^5$ and $1.34\\times10^5$. For the strongly overcoupled amplifier, the same scattering analysis is constrained by assuming the converter loss rate applies and yields $1.1\\times10^5$ to $1.5\\times10^5$; a separate multi-parameter fit to the measured squeezing and purity returns $Q_\\mathrm{int}=1.26\\times10^5$, consistent with the scattering estimate. The authors state that these are the highest internal quality factors reported so far for such tunable Josephson devices.","pith_inferences":["The paper does not show it, but a direct test of the amplifier claim would be to fabricate a series of JPAs with different coupling capacitors and fit each reflection curve without imposing the converter-derived bound; agreement would make the JPA $Q_\\mathrm{int}$ fully standalone.","An unstated consequence is that the same surface-treatment and bandaging steps might improve flux-tunable qubits, whose coherence times currently lag fixed-frequency qubits, though junction operating energies and impedances differ.","The two-bath input-output fit treats internal loss as a single lumped channel; splitting $Q_\\mathrm{int}$ into substrate, interface, and quasiparticle contributions would show which part of the $10^5$ ceiling remains to be attacked.","The purity-versus-pump-power curve used to confirm $Q_\\mathrm{int}$ depends on the added pump-noise model, so an independent check using a weakly coupled JPA would remove that coupling as well."],"forward_implications":["Tunable Josephson circuits can be operated at internal loss rates comparable to fixed-frequency superconducting resonators, removing a long-standing gap.","Squeezed microwave states from such amplifiers should reach higher purity at a fixed squeezing level, since purity degrades roughly as the internal loss rate.","Josephson parametric converters with $Q_\\mathrm{int}$ above $10^5$ become practical low-noise frequency converters and quantum memories at the single-photon level.","The fabrication recipe, especially the surface treatment and argon milling steps, can be carried over to other flux-tunable superconducting circuits to raise their coherence times.","The maximum squeezing level achievable, which scales inversely with internal loss, increases correspondingly once $Q_\\mathrm{int}$ exceeds $10^5$."],"supporting_citations":[{"why":"supplies the circle-fit and background-correction method used to extract the loaded and external quality factors from reflection data.","marker":"[41]"},{"why":"provides the quantum input-output model and reference-state tomography used for the squeezing and purity fit that independently determines Qint.","marker":"[13]"},{"why":"gives the Josephson ring modulator design and the flux-tunable frequency relation used to characterize the converter.","marker":"[27]"},{"why":"gives the dc-SQUID-terminated amplifier model and distributed-element frequency expression used for the JPA.","marker":"[26]"},{"why":"motivates the materials-loss approach and the constrained estimate of Qint when external coupling dominates the resonance shape.","marker":"[17]"},{"why":"lists previously reported Qint values for tunable Josephson devices against which the authors compare and stake the record claim.","marker":"[46–59]"}],"fun_headline_variants":["Josephson devices hit record Q beyond 100,000 at single-photon level","Low-loss Josephson circuits break 100,000 Q barrier","Surface-treated Josephson devices reach Q above 100k","Squeezed states verify record-low loss in Josephson circuits","Tunable parametric devices achieve internal Q over 100k"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The amplifier's internal quality factor is assumed to equal the converter's because both are fabricated and measured under identical conditions; the amplifier's own reflection data, being strongly overcoupled, cannot determine Qint on its own.","fun_headline_variants_meta":{"raw":{"variants":["Josephson devices hit record Q beyond 100,000 at single-photon level","Low-loss Josephson circuits break 100,000 Q barrier","Surface-treated Josephson devices reach Q above 100k","Squeezed states verify record-low loss in Josephson circuits","Tunable parametric devices achieve internal Q over 100k"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000243,"raw_usage":{"total_tokens":1504,"prompt_tokens":894,"completion_tokens":610,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":510,"completion_tokens_details":{"reasoning_tokens":521}},"tokens_in":510,"tokens_out":610,"duration_ms":5686,"temperature":1.0,"reasoning_tokens":521,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T15:05:35.469508+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fabricate a second JPA with a much smaller coupling capacitance so that Qext is comparable to Qint, then fit its background-corrected reflection circle without imposing the converter-derived bound; a directly fitted Qint below 100,000 would contradict the paper's central claim.","supporting_citations":[{"cited_title":"Sandberg, C","cited_arxiv_id":null,"evidence_quote":"supplies the circle-fit and background-correction method used to extract the loaded and external quality factors from reflection data."},{"cited_title":"Di Candia, F","cited_arxiv_id":null,"evidence_quote":"provides the quantum input-output model and reference-state tomography used for the squeezing and purity fit that independently determines Qint."},{"cited_title":"Wang, Candidate Source of Flux Noise in SQUIDs: Adsorbed Oxygen Molecules, Phys","cited_arxiv_id":null,"evidence_quote":"gives the Josephson ring modulator design and the flux-tunable frequency relation used to characterize the converter."}],"review_version":1}