{"id":"be850d21-15cb-4f66-9839-c1672d8cc588","arxiv_id":"2502.04911","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A graphene Josephson parametric amplifier operated in its nonlinear regime works as a fast, broadband bolometer, with sideband response enhanced by about 100 times and a reported NEP of about 500 aW/√Hz.","lead":"The authors built a graphene Josephson parametric amplifier and used it as a bolometer, finding that operating it in the nonlinear (Kerr) regime boosts the detected sideband signal by about 100 times. This could lead to faster, broader-band microwave and terahertz sensors, with a reported noise-equivalent power of about 500 aW/√Hz.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"NEP of 500 aW/√Hz rests on a COMSOL-derived 35% heat-loss fraction whose parameters are calibrated against the same experimental dataset; without an independent constraint, the headline sensitivity is not robust.","rationale":"The reader's verdict correctly identifies the weakest point. The paper's central claim—that the Kerr nonlinearity produces a strong (≈20 dB) sideband enhancement and a lower detection threshold—is directly supported by the experimental sideband data in Fig. 3f and the SNR map in Fig. 3e; this part of the argument does not depend on absolute power calibration because the nonlinear and linear curves are measured under identical readout conditions. However, the paper also claims a specific NEP of ~500 aW/√Hz, and this number depends on converting the applied heater power into the heat actually absorbed at the Josephson junction. The conversion in Eq. S2 uses a 35% phonon-loss fraction that comes solely from a COMSOL model whose material parameters are fitted to the same experimental temperature data that the model is supposed to explain. No independent measurement constrains η_e-p: the etched-heater control in Fig. 4 is qualitative, and the literature values for the thermal parameters listed in Table 1 appear mis-scaled relative to known material properties, undermining confidence in the simulation's absolute heat-flux ratio. Because the NEP is inversely proportional to the heat reaching the junction, an incorrect η_e-p directly rescales the headline sensitivity. This does not invalidate the qualitative nonlinear-enhancement claim, but it means the quantitative NEP is not yet established at the level implied by the abstract. This justifies the CONDITIONAL verdict and suggests additional calibration experiments or at minimum a sensitivity analysis of the thermal model.","tokens_in":17441,"tokens_out":21603,"duration_ms":268090,"concrete_test":"Run a parameter-sweep study with the COMSOL model used in Supplementary Sec. XI: vary Σ_e-p and graphene k independently over literature-plausible ranges (e.g., Σ_e-p = 0.1–10 W/K^4 m^2, k = 50–500 W/mK), refit to the experimental temperature-lookup table, and record the resulting η_e-p values. If the spread in η_e-p exceeds approximately 0.15–0.55, then the factor (1-η_e-p) in Eq. S2 ranges from 0.45 to 0.85, making the extracted NEP uncertain by up to a factor of ~2; this would require presenting the 500 aW/√Hz figure as model-dependent rather than a measured value.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equation S2 corrects the applied heater power by (1-|Γ|^2)(1-η_e-p), with η_e-p ≈ 35% obtained from a COMSOL simulation. The simulation's thermal parameters (Σ_e-p = 0.6×10^6 W/K^4 m^2, graphene k = 100 W/mK) are fitted so that the simulated electron temperature matches the experimental temperature-versus-heater-current lookup table (Fig. S12), which was itself inferred from the resonance shift. This makes η_e-p a fitting-dependent output rather than an independently measured quantity. The control experiment (Fig. 4) demonstrates qualitatively that phonons can carry heat to the junction, but it does not provide a quantitative fraction. The tabulated literature values for graphene thermal properties in Table 1 also appear anomalous (e.g., thermal conductivity 3×10^-4 W/mK), further weakening confidence in the model's absolute heat-flux output. Because the measured NEP is directly proportional to the (1-η_e-p) factor, an uncertainty in η_e-p of ±20% translates to a comparable relative uncertainty in the NEP, while the claimed order-of-magnitude improvement over the linear regime is unaffected because the same correction applies to both bias conditions. Thus the central qualitative claim is secure, but the headline 500 aW/√Hz value is not.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a graphene Josephson parametric amplifier (JPA) operated as a bolometer, with an on-chip heater on the same graphene flake. A modulated heater current at frequency fh produces sidebands at fs ± 2fh in the reflected microwave signal. The authors compare sideband response in linear and nonlinear biasing, observe roughly 20 dB higher sideband power and a lower onset heater power in the nonlinear regime, and extract a best NEP of about 500 aW/√Hz after correcting the applied heater power for impedance mismatch and an estimated ~35% phonon-mediated heat loss. They also measure a thermal time constant of 4.26 μs and an intrinsic JPA response time of 70 ns, and support the phonon heat-loss channel with a control device in which the graphene heater is etched to remove the galvanic path.","tokens_in":17752,"tokens_out":4852,"duration_ms":50096,"significance":"The central qualitative claim—that Kerr nonlinearity enhances the bolometric sideband response—is directly visible in the data and is not circular: the linear-versus-nonlinear comparison is an independent measurement. The time-constant measurements are clean, and the paper is accompanied by extensive supplementary characterization, including a control experiment for phonon-mediated heating. If the absolute NEP value is robust, this would be a useful demonstration of a JPA-based bolometer with fast readout and gate tunability. However, the headline NEP depends on a COMSOL heat-loss correction whose parameters are calibrated against the same experimental dataset used to infer heating, so the numerical NEP is not yet established to the precision implied by the abstract.","major_comments":[{"comment":"The corrected heater power in Eq. S2 contains the factor (1 − ηe−p) with ηe−p ≈ 35% taken from the COMSOL simulation in Sec. XI. That simulation is calibrated by matching the simulated electron temperature to the experimental temperature-versus-heater-current lookup table derived from the same resonance-shift data that defines the bolometer response. The control experiment in Fig. 4 demonstrates phonon-mediated heating qualitatively but does not provide an independent quantitative constraint on ηe−p. Since the extracted NEP in Eq. S1 is directly proportional to P′_heater, an uncertainty of, say, ±20% in ηe−p maps directly onto a comparable uncertainty in the quoted 500 aW/√Hz. The authors should either provide an independent thermal calibration, a conservative upper bound, or a propagation of the full parameter uncertainty into the NEP.","section":"Supplementary Sec. V, Eq. S2; Supplementary Sec. XI"},{"comment":"The thermal parameters used in the COMSOL model are internally inconsistent and are load-bearing for the heat-loss fraction. Table 1 lists a graphene literature thermal conductivity of 3×10⁻⁴ W/mK and an hBN simulation conductivity of 10⁻¹⁰ W/mK, both far from accepted values, and the simulation heat capacities differ from literature values by orders of magnitude. Table 2 lists a graphene thickness of 1 μm, which is not physical for a monolayer flake unless it is an effective modeling thickness. These entries suggest unit or notation errors, and they undermine confidence in the absolute heat flux that produces ηe−p ≈ 0.35. The authors should correct the table and quantify how ηe−p varies over a plausible range of the fitted parameters.","section":"Supplementary Sec. XI, Tables 1 and 2"},{"comment":"The quoted bolometer thermal time constant of 4.26 μs is measured with the device biased in the linear regime, while the high-sensitivity bolometric operation is in the nonlinear regime. Figure 5d shows that the JPA response slows considerably as the drive power moves from linear to nonlinear and dissipative regimes. The manuscript should state the operating point for the 4.26 μs time constant explicitly and discuss whether the faster linear-regime time constant is representative of the nonlinear-biased bolometer used for the NEP measurement.","section":"Main text, 'Time constants' paragraph; Fig. 5"}],"minor_comments":[{"comment":"The horizontal axis is labeled in dBm and also written as P′_heater; please clarify whether the plotted quantity is the corrected heater power from Eq. S2 or the raw applied power, and state the units of the vertical NEP axis consistently.","section":"Supplementary Fig. S5"},{"comment":"The parameter tr in the current-phase relation is not defined in the main text; please define the averaged transparency and specify the regime in which the cubic expansion in φ is valid.","section":"Main text, Eq. (1)"},{"comment":"The claim of an 'order of magnitude improvement in sensitivity' is supported by sideband power data in Fig. 3f, but no linear-regime NEP curve is shown. Since the same (1 − ηe−p) correction applies to both regimes the ordering is robust, but a direct linear-regime NEP extraction would make the sensitivity comparison quantitative rather than inferred from sideband response.","section":"Main text, abstract and conclusion"},{"comment":"The control experiment in Fig. 4 would be more convincing with an overlay of the resonance-frequency shift before and after etching, including the heater-current range, so that the quantitative fraction of phonon-mediated transmission can be visually assessed.","section":"Fig. 4c and Supplementary Sec. XI"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the manuscript is a strong experimental demonstration, and the qualitative nonlinear-enhancement claim is well supported. The main risk is the absolute NEP value, which rests on a COMSOL correction calibrated to the same experiment. I would ask the authors to substantially strengthen the treatment of this correction—ideally with an independent measurement or a careful propagation of parameter uncertainties—before publication. The novelty and fit with the journal are appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear X,\n\nThe core claim is credible: this is the first JPA used as a bolometer, and the sideband data in Fig. 3f show that biasing in the nonlinear Kerr regime raises sideband power by ~20 dB and moves the onset down by ~10 dB of heater power. That 100x enhancement is the real result. The time-constant measurements are clean, and the etched-graphene control gives reasonable qualitative evidence that substrate phonons carry a significant share of the heat.\n\nThe soft spots all sit around the headline NEP. The 500 aW/√Hz value is extracted through Eq. S2, which corrects the applied heater power by (1-|Γ|²)(1-η_e-p) with η_e-p ≈ 35% taken from a COMSOL model. That model's thermal parameters, e.g. graphene k = 100 W/mK and Σ_e-p = 0.6×10^6 W/K⁴ m², are tuned so the simulated electron temperature matches the experimental temperature lookup table from the same device. So η_e-p is a fitting output, not an independent measurement. The Table 1 entries also look anomalous — graphene thermal conductivity listed as 3×10^-4 W/mK and hBN simulation k as 10^-10 W/mK are not physical — which further weakens confidence in the absolute heat-flux budget. An uncertainty of ±20% in η_e-p translates directly into the same relative uncertainty in NEP. In addition, the \"order-of-magnitude improvement over the linear regime\" is inferred from the sideband onset shift, not from a measured linear-regime NEP; a direct linear NEP curve would make that comparison airtight.\n\nThe central qualitative claim, nonlinear biasing enhances bolometric responsivity, is well supported and is not circular. The NEP figure, however, should be treated as model-dependent until the heat-loss fraction is independently constrained.\n\nThis paper is for the graphene bolometry and quantum-sensing community. It deserves a serious referee. I would recommend conditional acceptance after the authors either measure η_e-p independently or present NEP with a conservative range, and fix the simulation parameters.\n\n— [signed]","headline":"The nonlinear-enhancement effect is real and directly visible in the sidebands, but the 500 aW/√Hz NEP rests on a COMSOL heat-loss correction calibrated to the same experiment, so treat it as model-dependent.","tokens_in":18316,"tokens_out":3324,"would_cite":true,"duration_ms":30529,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A graphene Josephson parametric amplifier, biased in its nonlinear Kerr regime, operates as a bolometer with a noise-equivalent power of about 500 aW/√Hz.","keywords":["graphene Josephson junction","Josephson parametric amplifier","bolometer","Kerr nonlinearity","noise equivalent power","sideband readout","microwave detection","hot-electron thermalization"],"falsifier":"Measure the bolometer's NEP with an independent, absolute absorbed-power calibration, for example by feeding a known microwave power through a calibrated attenuator directly into the junction and comparing the resulting resonance shift with the heater calibration. If the true delivered power differs from (1 − |Γ|²)(1 − ηe−p)Pheater, the 500 aW/√Hz value changes by the same factor. A direct measurement of ηe−p by varying the heater-to-junction distance or changing the substrate would also test the correction.","tokens_in":17230,"feed_emoji":"📡","tokens_out":4932,"duration_ms":44545,"temperature":0.7,"pith_summary":"This paper reports the first use of a Josephson parametric amplifier (JPA) as a bolometer. The central claim is that operating the JPA in the nonlinear, Kerr-dominated regime makes the device far more sensitive to heat: modulated heating produces sidebands roughly 100 times larger than in the linear regime, improving the noise-equivalent power by an order of magnitude to about 500 aW/√Hz. The device is built on a graphene Josephson junction with an on-chip heater, and it combines a thermal time constant of 4.26 µs with a broad, gate-tunable microwave band. If true, this points to a way of turning the intrinsic nonlinearity of Josephson devices into improved quantum sensors without sacrificing speed.","feed_headline":"Kerr nonlinearity makes graphene bolometer 100x more sensitive","feed_subtitle":"Nonlinear biasing boosts sidebands ~100x and cuts noise-equivalent power to 500 aW/√Hz.","key_machinery":"The central object is the Kerr nonlinear inductance of the graphene Josephson junction, approximated from the current-phase relation Is(φ) ≈ αφ + βφ³, which gives a junction inductance LJ(φ) ≈ L0 + L2φ². This nonlinear inductance makes the reflected phase highly sensitive to temperature-induced changes in the junction, and it is what converts a modulated heater signal into sidebands at fs ± 2fh. The sideband power and the phase sensitivity in the nonlinear regime are the readout mechanism, while the on-chip graphene heater and the COMSOL thermal model with electron-phonon coupling supply the calibrated input power.","core_discovery":"The paper claims that the Kerr nonlinearity of a graphene JPA, arising from the cubic term in the junction current-phase relation, is not merely tolerable but useful for bolometry. When the junction is biased with a pump tone in the nonlinear regime, small temperature changes from an injected heater signal produce large shifts in the reflected phase, and a modulated heater generates sidebands at fs ± 2fh whose power is about 20 dB higher than in the linear regime. Using this sideband response, the authors extract a noise-equivalent power of about 500 aW/√Hz at 1 MHz heater modulation, an order-of-magnitude improvement over linear biasing, and they measure thermal time constants τon = 4.45 µs and τoff = 4.26 µs, with an intrinsic JPA response of 70 ns. A control experiment in which the graphene heater is electrically severed shows that phonon-mediated heat transfer through the substrate also contributes to the response, and finite-element simulations estimate that about 35% of the injected heat is lost to the phonon bath.","pith_inferences":["If the Kerr enhancement transfers to other Josephson devices, nonlinear biasing could become a standard sensitivity boost for bolometers and calorimeters beyond graphene-based ones.","The frequency-multiplexed sideband readout may allow large arrays of JPA bolometers on a single feed line, since each resonator is gate-tunable to a distinct frequency.","The roughly 35% phonon-loss correction implies that reducing substrate coupling, for example by suspending the graphene or using lower-gap superconducting contacts, could push the NEP well below 500 aW/√Hz.","The SNR hotspots observed in the nonlinear phase diagram suggest that the operating point could be further optimized, possibly approaching the bistable edge where phase sensitivity is largest."],"forward_implications":["A single device achieves NEP ≈ 500 aW/√Hz with a thermal time constant of 4.26 µs and an intrinsic JPA time constant of 70 ns.","Biasing in the nonlinear regime raises sideband signals by about 20 dB, roughly 100 times, and improves sensitivity by an order of magnitude over linear biasing.","The gate-tunable resonance, spanning roughly 4 to 5.7 GHz, and an on-flake heater operating from DC to 100 MHz give the bolometer broad bandwidth and a fast, direct readout.","Because the nonlinear response is the sensing mechanism, the device can operate as a detector without relying on the slow switching histograms used in earlier Josephson bolometers.","The architecture could extend to direct irradiation in the THz and near-infrared ranges by using the graphene flake as a hot-electron bus, with further sensitivity gains from lower-gap superconductors such as aluminum."],"supporting_citations":[{"why":"Provides the prior graphene Josephson junction microwave bolometer whose switching-readout limitation this device addresses.","marker":"[10]"},{"why":"Supplies a cQED-threshold bolometer benchmark that this work compares against for sensitivity and speed.","marker":"[11]"},{"why":"Establishes the graphene Josephson junction as a quantum-limited microwave amplifier, the platform used here as a bolometer.","marker":"[23]"},{"why":"Demonstrates a gate-tunable graphene Josephson parametric amplifier, giving the architectural basis for this device.","marker":"[24]"},{"why":"Represents the fastest graphene bolometer with Johnson-noise readout, used as a comparison for NEP and thermal relaxation time.","marker":"[9]"},{"why":"Provides the best microwave single-photon sensitivity benchmark, against which the NEP improvement is placed.","marker":"[7]"},{"why":"Informs the thermal-transport and electron-phonon coupling corrections used to estimate the heat reaching the junction.","marker":"[27]"},{"why":"Supplies the hot-electron bolometer concept in graphene and the desirable thermal properties at charge neutrality.","marker":"[8]"}],"fun_headline_variants":["Kerr nonlinearity makes graphene bolometer 100x more sensitive","Nonlinear graphene JPA as bolometer: 100x sensitivity boost","Graphene bolometer gets 100x sensitivity from Kerr effect","Kerr effect amplifies graphene bolometer response 100x","Graphene bolometer hits 500 aW/√Hz via Kerr nonlinearity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The reported 500 aW/√Hz noise-equivalent power assumes that only the fraction (1 − |Γ|²)(1 − ηe−p) of the heater power reaches the junction, with ηe−p ≈ 35% taken from a COMSOL simulation calibrated against the same experimental temperature data.","fun_headline_variants_meta":{"raw":{"variants":["Kerr nonlinearity makes graphene bolometer 100x more sensitive","Nonlinear graphene JPA as bolometer: 100x sensitivity boost","Graphene bolometer gets 100x sensitivity from Kerr effect","Kerr effect amplifies graphene bolometer response 100x","Graphene bolometer hits 500 aW/√Hz via Kerr nonlinearity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000648,"raw_usage":{"total_tokens":3027,"prompt_tokens":1048,"completion_tokens":1979,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":664,"completion_tokens_details":{"reasoning_tokens":1884}},"tokens_in":664,"tokens_out":1979,"duration_ms":15641,"temperature":1.0,"reasoning_tokens":1884,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T21:00:15.347916+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the bolometer's NEP with an independent, absolute absorbed-power calibration, for example by feeding a known microwave power through a calibrated attenuator directly into the junction and comparing the resulting resonance shift with the heater calibration. If the true delivered power differs from (1 − |Γ|²)(1 − ηe−p)Pheater, the 500 aW/√Hz value changes by the same factor. A direct measurement of ηe−p by varying the heater-to-junction distance or changing the substrate would also test the correction.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the prior graphene Josephson junction microwave bolometer whose switching-readout limitation this device addresses."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies a cQED-threshold bolometer benchmark that this work compares against for sensitivity and speed."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the graphene Josephson junction as a quantum-limited microwave amplifier, the platform used here as a bolometer."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates a gate-tunable graphene Josephson parametric amplifier, giving the architectural basis for this device."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Represents the fastest graphene bolometer with Johnson-noise readout, used as a comparison for NEP and thermal relaxation time."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the best microwave single-photon sensitivity benchmark, against which the NEP improvement is placed."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Informs the thermal-transport and electron-phonon coupling corrections used to estimate the heat reaching the junction."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the hot-electron bolometer concept in graphene and the desirable thermal properties at charge neutrality."}],"review_version":1}