{"id":"1a541ebd-511d-4a42-a9c2-3593a7e7750d","arxiv_id":"2411.09071","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A nonlinear, steep-slope transduction scheme lowers the noise-equivalent power of a 140 GHz thermomechanical bolometer to about 30 pW/√Hz with all-electrical readout.","lead":"This paper shows that operating a tiny silicon-nitride trampoline detector at the steep edge of a nonlinear mechanical resonance lowers its noise-equivalent power to about 30 pW/√Hz for 140 GHz radiation. The trick is to use Duffing and Fano line-shape distortion to create a sharp transduction slope without changing the resonator's dissipation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 30 pW/√Hz NEP assumes dark Allan deviation equals the noise under illumination at the steep-slope frequency; a direct calibrated-power NEP measurement is needed to verify this.","rationale":"The paper's central claim is that nonlinear transduction along a steep slope reduces the NEP to about 30 pW/√Hz in an all-electrical sub-THz bolometer. The reader identified the weakest assumption as the unverified equivalence of dark and bright noise at the steep-slope operating frequency; my reading fully agrees. This is the single most load-bearing concern because the NEP is never measured directly with calibrated modulated power; it is inferred from dark Allan deviation divided by a responsivity that itself relies on the rigid-shift approximation. Any illumination-induced change in noise or any amplification of frequency noise by the slope directly changes the true NEP. The additional inconsistency between the reported 117x NEP reduction and the 5.6x slope asymmetry strengthens this concern: it suggests that either the noise varies with fD in ways not captured by the presented single representative Allan deviation, or the responsivity model is incomplete. Both possibilities undermine the quantitative headline. The paper otherwise reports a credible experimental platform with careful characterization of spectra, Bode response, and Allan deviation; the nonlinear transduction concept is plausible and supported by the data's general trends. Therefore, the proper verdict is CONDITIONAL: the mechanism is promising but the absolute NEP requires direct experimental validation before the claim can be accepted. Since the reader already set CONDITIONAL, my verdict recommendation is UNCHANGED.","tokens_in":11483,"tokens_out":8229,"duration_ms":90271,"concrete_test":"Operate the PyC device at fD=524.26 kHz and apply a calibrated square-wave-modulated 140 GHz power (amplitude ~0.1-1 µW, fM=10 Hz) while demodulating with the UHF lock-in. Record the output step amplitude and the output Allan deviation (or noise PSD) over the same bandwidth, then compute NEP_direct = P_in / SNR(√Hz). Repeat with the source OFF to obtain dark noise. If NEP_direct exceeds 30 pW/√Hz by more than about 2x, the dark-noise assumption fails. Additionally, measure the output voltage noise as a function of fD and incident power to test whether σ_AD scales with dV/df.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The absolute NEP claim depends on Eq. (3), which combines a dark Allan deviation σ_AD with a dynamic responsivity R built from a rigid frequency-shift model of the bright response. For the reported 30 pW/√Hz to hold, the output noise at fD=524.26 kHz under 140 GHz illumination must be identical to the dark noise at that same operating point, and the bright spectrum must be a rigid shift of the dark one (Eq. 1). Neither condition is demonstrated. If illumination heats the membrane and changes mechanical dissipation or the operating point on the nonlinear response, the noise could be larger; if the underlying frequency noise is merely transduced through the steep slope, the voltage noise would scale with dV/df and NEP would not improve at all. The paper further reports a 117x NEP reduction while the measured slope asymmetry in Fig. 3(b) is only 5.6x; under Eqs. (2)-(3), this inconsistency implies that σ_AD itself varies strongly with fD, which is exactly the unverified assumption that the dark noise is independent of slope and illumination. Thus the central quantitative claim is not yet established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes to improve the noise-equivalent power (NEP) of thermomechanical bolometers by operating the resonator on a steep slope of a nonlinear (Duffing) resonance, thereby increasing transduction responsivity without changing the dissipation rate. The authors characterize two Si3N4 trampoline devices with different absorbing layers (Cr/Au and pyrolitic carbon) at 140 GHz, measure dark and bright spectra, extract a dynamic responsivity from the spectral slope, combine it with dark Allan deviation noise to compute NEP as a function of drive frequency and modulation frequency, and report a best NEP of about 30 pW/Hz^(1/2) for the PyC device. They also discuss operation in the multistable regime for threshold detection.","tokens_in":11699,"tokens_out":9226,"duration_ms":81128,"significance":"The idea of using nonlinearity to steepen the transduction slope is physically motivated and could offer a practical alternative to high-Q engineering; the all-electrical readout is an advance over earlier optical-readout thermomechanical bolometers. The manuscript includes careful spectral characterization, a comparison of two absorber materials, and a clear discussion of dynamic-range limitations. However, the central quantitative claims—the slope-induced NEP reduction and the absolute 30 pW/Hz^(1/2) value—rest on assumptions about the noise floor that are not directly verified, and there is an internal inconsistency between the reported NEP reduction and the measured slope asymmetry. If these points are resolved, the work would constitute a meaningful contribution to room-temperature sub-THz detection.","major_comments":[{"comment":"The reported NEP reduction factor of 117 for the PyC device is inconsistent with the measured slope asymmetry of about 5.6 in Fig. 3(b). According to Eq. (3), for a noise level sigma_AD independent of fD, the NEP ratio between the maximum and minimum slope operating points should equal the inverse of the slope ratio, i.e., about 5.6, not 117. The observed factor of 117 implies that sigma_AD at the steep-slope frequency is roughly 20 times smaller than at the shallow-slope frequency, but the manuscript neither reports sigma_AD as a function of fD nor offers any mechanism for such a strong variation. This point must be addressed before the NEP reduction can be attributed to the engineered slope.","section":"Section III, Eq. (3) and Fig. 3"},{"comment":"The NEP calculation combines a dark Allan deviation sigma_AD (in volts) with a responsivity R that is proportional to dV/df. If the dominant noise at the operating point is frequency noise rather than additive voltage noise, then sigma_AD itself scales with dV/df and the NEP becomes independent of the slope. The paper does not identify which noise source dominates or demonstrate that sigma_AD is independent of fD; the observed NEP reduction therefore does not by itself prove that transduction along the slope improves the signal-to-noise ratio. A direct measurement of NEP under calibrated modulated illumination at fD, or an explicit characterization of sigma_AD(fD), is needed.","section":"Section III, Eq. (3)"},{"comment":"The absolute NEP of about 30 pW/Hz^(1/2) is inferred from dark Allan deviation measurements divided by a dynamic responsivity derived from a rigid-shift model of the bright spectrum. The assumptions that the noise under 140 GHz illumination at fD is identical to the dark noise, and that the bright spectrum is a simple frequency shift of the dark one, are not experimentally verified. If illumination changes the mechanical dissipation or the operating point on the nonlinear response, the true NEP could be significantly higher. An end-to-end measurement with a calibrated modulated power source would strengthen the claim.","section":"Section III and Fig. 2"},{"comment":"No error bars or uncertainty analysis are provided for the NEP values. Since the absolute NEP depends on several calibrations (Golay cell power, COC window transmission of about 50%, assumed absorbing area of 60 um x 85 um), the reported 30 pW/Hz^(1/2) should be accompanied by an uncertainty estimate.","section":"Section III, Fig. 3(e)"}],"minor_comments":[{"comment":"The phrase 'the PyC devicenegative derivative peak' contains a missing space; it should read 'the PyC device negative derivative peak'.","section":"Section III, p. 5"},{"comment":"The axis labels and tick marks for fM are difficult to read; please increase the font and ensure the axis range is clear.","section":"Figure 3(e)"},{"comment":"The statement that 'All NEP curves remain relatively constant as a function of fM' is surprising given the 20 Hz cut-off in the Bode response shown in Fig. 2(b); a brief explanation of why the NEP does not degrade at modulation frequencies above the thermal cut-off would improve clarity.","section":"Section III, Fig. 3(e) vs Fig. 2(b)"},{"comment":"The assumption that both devices have the same vibrational amplitude at bifurcation, used to choose comparable driving voltages, is not justified; a sentence explaining why different absorber layers do not affect the bifurcation amplitude would be helpful.","section":"Section II.A"}],"recommendation":"major_revision","confidential_remarks":"The main concerns are the unexplained discrepancy between the reported 117x NEP reduction and the 5.6x slope asymmetry for the PyC device, and the unverified noise model underlying Eq. (3). These issues are load-bearing for the paper's central quantitative claim. The topic is within the journal's scope, and the idea is promising, but the authors should be asked to provide direct NEP measurements or a clear resolution of the noise-floor question before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The useful thing here is the combination, not the physics: using the steep slope of a Duffing-distorted Fano resonance to improve transduction in a thermomechanical bolometer with all-electrical readout. That precise combination is new, and the paper does it properly at the level of characterization—spectra, Bode response, Allan deviation, and an NEP spectrogram. The PyC absorber versus Au absorber comparison is also well conceived, and the citation pattern is appropriate, covering both the TMB literature and the nonlinear-sensing literature. Credit where due: this is a competent group making a plausible and interesting claim.\n\nThe soft spots are real but mostly addressable. The headline 30 pW/√Hz is not directly measured. NEP comes from dark Allan deviation divided by dynamic responsivity, and nothing verifies that the noise floor is unchanged when the 140 GHz source is on and the device sits at the steep-slope operating point. If illumination shifts the operating point or changes dissipation, the true NEP could be higher. More fundamentally, if the dominant output noise is frequency noise rather than additive voltage noise, the steep slope amplifies signal and noise together, so the NEP gain would be much smaller. The paper never distinguishes these cases.\n\nThe quantitative inconsistency is the thing that would make me pause as a referee: the text states a 117x NEP reduction for the PyC device while the measured slope asymmetry in Fig. 3(b) is 5.6x. Under their own Eq. (3), that forces the dark noise to vary by roughly 21x between the two operating points. That cannot be called \"good agreement\". Either the factor is misstated or the noise is strongly fD-dependent—which is exactly the assumption that needs testing. The Au device's 37.5x matches its slope ratio, so this may be PyC-specific, but it needs a clear explanation.\n\nMinor but worth noting: no error bars anywhere, and the absolute responsivity calibration rests on beam-shape assumptions and a ~50% window transmission.\n\nWho gets value from this: people working on room-temperature THz bolometers or nonlinear MEMS sensing. I would send it to peer review—the idea is worth referee time—but with the expectation of major revision. The 30 pW/√Hz number is a conditional claim until the authors add a direct calibrated-power NEP measurement at the steep-slope frequency with the source on, and reconcile the 117x figure.","headline":"A credible but unverified NEP claim: steep-slope transduction is genuinely new for thermomechanical bolometers, but the headline 30 pW/√Hz is inferred from dark noise and the reported 117x reduction sits uneasily with the 5.6x slope asymmetry.","tokens_in":12304,"tokens_out":4322,"would_cite":true,"duration_ms":46062,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["07.57.Kp"],"model":"deepseek-v4-flash","headline":"By engineering steeper resonance slopes through Duffing nonlinearity instead of raising the Q-factor, this paper demonstrates a room-temperature sub-THz thermomechanical bolometer with about 30 pW/√Hz NEP under all-electrical readout.","keywords":["thermomechanical bolometer","sub-THz detection","Duffing nonlinearity","noise equivalent power","trampoline resonator","Fano resonance","pyrolitic carbon absorber","all-electrical readout"],"falsifier":"Operate the PyC device at fD = 524.26 kHz with a 90 mV drive, illuminate with a calibrated amplitude-modulated 140 GHz source, and measure the signal-to-noise ratio end-to-end; if the resulting NEP exceeds 30 pW/√Hz, the dark-noise assumption is violated and the claimed value is optimistic.","tokens_in":1804,"feed_emoji":"📡","tokens_out":2712,"duration_ms":115974,"temperature":0.7,"pith_summary":"This paper proposes an alternative to high-Q resonators for sensitive transduction: instead of making the resonance line narrower, make its sides steeper by combining Fano interference with Duffing nonlinearity while keeping dissipation constant. The test case is a silicon-nitride trampoline bolometer for 140 GHz radiation, read out electrically. Operating the detector at the frequency of maximum slope of the distorted resonance reduces the noise equivalent power to about 30 pW/√Hz, roughly three times better than the authors' earlier 100 pW/√Hz device that used optical readout. The same nonlinearity can be pushed past a bifurcation to make a threshold detector, at the cost of a narrow dynamic range.","feed_headline":"Vibrational nonlinearity cuts bolometer noise to 30 pW/√Hz","feed_subtitle":"Reading out at the steepest slope of a Duffing-distorted resonance beats the authors' earlier optical-readout record.","key_machinery":"The central object is the transduction operating point: the detector is read out at a fixed drive/demodulation frequency fD, and the signal is the first derivative of the resonance amplitude, through ΔVLI ~ Ad(fD) - Ad(fD+δf) ~ (dAd/df) δf. The slope is amplified by two coexisting line-shaping effects: Fano interference, which makes the bare resonance asymmetric, and the Duffing (cubic) mechanical nonlinearity, which hardens the resonance and steepens one edge at higher drive amplitudes. The noise-equivalent power is NEP(fM,fD) = σAD√(2τ)/R, so the figure of merit is the ratio of the measured Allan-deviation noise floor to the dynamic responsivity at the chosen frequency.","core_discovery":"The authors demonstrate experimentally that transduction sensitivity is governed by the local slope dAd/df of the mechanical resonance, and that this slope can be engineered by driving the resonator into the Duffing nonlinear regime, where the already asymmetric Fano lineshape develops a much steeper edge. By choosing the transduction frequency fD at the steepest point, they obtain a noise-equivalent power of about 30 pW/√Hz for a pyrolitic-carbon-coated device under 140 GHz illumination at room temperature with all-electrical inductive readout, compared with about 100 pW/√Hz for their earlier optical-readout device. The improvement tracks the derivative enhancement (factors of roughly 117 and 37.5 for the two devices), and the absorber choice matters: the pyrolitic-carbon layer's ~40% sub-THz absorbance gives about a seven-fold better static responsivity than a thin Cr/Au layer. Past the bifurcation point, the derivative becomes delta-like, which precludes intensity detection but suggests a threshold-sensor operating mode.","pith_inferences":["Because the steep-slope effect is a property of the resonance lineshape rather than of bolometry, the operating-point strategy should transfer to other nonlinear resonant sensors, such as mass or gas detectors, whenever a Duffing nonlinearity is available.","A direct test of the absolute sensitivity claim would be to measure the noise floor with the 140 GHz source on, at the same transduction frequency and drive amplitude, and compare it with the dark Allan-deviation floor; if the bright floor is higher, the true NEP exceeds 30 pW/√Hz.","The inferred roughly 10% absorbance of the granular Cr/Au film suggests that engineering grain morphology in ultra-thin metal coatings is a practical lever for impedance matching, potentially bringing metal absorbers closer to the 188 Ω sheet-resistance ideal.","At drive amplitudes past the bifurcation, the readout becomes a binary jump rather than an analog slope, which could be developed into a click-style threshold detector for individual THz pulses if the thermal time constant is short enough."],"forward_implications":["At its best operating point, the PyC device reaches about 30 pW/√Hz at 140 GHz with all-electrical readout, roughly three times better than the earlier 100 pW/√Hz optical-readout version.","Within the same device, moving the transduction frequency from a flat region to the steepest slope reduces the NEP by factors of roughly 117 (PyC) and 37.5 (Au), with no change to dissipation or fabrication.","The steep-slope regime has a derivative linewidth of about 50 Hz, limiting linear detection to signals below roughly 100 nW; lower drive amplitudes restore a broader dynamic range at higher NEP.","Increasing the drive past the bifurcation point produces a delta-like derivative, which is unsuited to intensity measurement but opens a threshold-detection mode for events such as light pulses or mass loading.","All-electrical readout at a 20 Hz operating speed makes the device competitive with commercial sub-THz detectors that achieve NEP around 10 pW/√Hz, while remaining above the thermal-fluctuation fundamental limit."],"supporting_citations":[{"why":"Reports the 100 pW/√Hz NEP of the earlier optical-readout 140 GHz thermomechanical bolometer that this work improves on with electrical readout.","marker":"[32]"},{"why":"Supplies the measured roughly 40% sub-THz absorbance of the pyrolitic-carbon film used on the best-performing device.","marker":"[49]"},{"why":"Provides the methodology for defining illuminating power on the absorber area and a near-fundamental-limit optical-readout comparison.","marker":"[50]"},{"why":"Models the thermal relaxation, Fano cross-talk, and multiplexed array behavior used to interpret the device spectra and the 20 Hz thermal cutoff.","marker":"[36]"},{"why":"Documents the Duffing nonlinearity and multistability that create the steep slopes and bifurcation regime exploited in this work.","marker":"[39]"},{"why":"Establishes the 188 Ω sheet-resistance condition for near-perfect absorption in thin metal films, used to interpret the Cr/Au absorber response.","marker":"[46]"},{"why":"Provides the basis for magnetomotive and inductive all-electrical readout that replaces the optical probing of earlier devices.","marker":"[38]"},{"why":"Defines the thermal-fluctuation fundamental NEP limit used to gauge how far the measured device stands from the ideal.","marker":"[52]"},{"why":"Provides the commercial sub-THz detector NEP benchmark of about 10 pW/√Hz against which the all-electrical result is compared.","marker":"[53]"}],"fun_headline_variants":["Nonlinear resonance slope cuts bolometer noise to 30 pW/√Hz","Steep Duffing slope enables 30 pW/√Hz NEP in sub-THz bolometer","Nonlinearity, not high Q, sharpens bolometer readout to 30 pW/√Hz","Duffing nonlinearity improves sub-THz bolometer NEP to 30 pW/√Hz"],"cache_read_input_tokens":14464,"weakest_assumption_plain":"The sensitivity claim assumes the detector's noise floor measured in the dark is unchanged when light shines on it at the steepest-response frequency, so that sensitivity can be computed as dark noise divided by response strength rather than measured directly.","fun_headline_variants_meta":{"raw":{"variants":["Nonlinear resonance slope cuts bolometer noise to 30 pW/√Hz","Steep Duffing slope enables 30 pW/√Hz NEP in sub-THz bolometer","Nonlinearity, not high Q, sharpens bolometer readout to 30 pW/√Hz","Duffing nonlinearity improves sub-THz bolometer NEP to 30 pW/√Hz"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000833,"raw_usage":{"total_tokens":3602,"prompt_tokens":879,"completion_tokens":2723,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":495,"completion_tokens_details":{"reasoning_tokens":2620}},"tokens_in":495,"tokens_out":2723,"duration_ms":19033,"temperature":1.0,"reasoning_tokens":2620,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:06:05.151100+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Operate the PyC device at fD = 524.26 kHz with a 90 mV drive, illuminate with a calibrated amplitude-modulated 140 GHz source, and measure the signal-to-noise ratio end-to-end; if the resulting NEP exceeds 30 pW/√Hz, the dark-noise assumption is violated and the claimed value is optimistic.","supporting_citations":[{"cited_title":"Jorudas, H","cited_arxiv_id":null,"evidence_quote":"Supplies the measured roughly 40% sub-THz absorbance of the pyrolitic-carbon film used on the best-performing device."},{"cited_title":"Martini, K","cited_arxiv_id":null,"evidence_quote":"Provides the methodology for defining illuminating power on the absorber area and a near-fundamental-limit optical-readout comparison."},{"cited_title":"Gregorat, M","cited_arxiv_id":null,"evidence_quote":"Models the thermal relaxation, Fano cross-talk, and multiplexed array behavior used to interpret the device spectra and the 20 Hz thermal cutoff."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents the Duffing nonlinearity and multistability that create the steep slopes and bifurcation regime exploited in this work."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the 188 Ω sheet-resistance condition for near-perfect absorption in thin metal films, used to interpret the Cr/Au absorber response."},{"cited_title":"Rogalski, Infrared Detectors, 2nd ed","cited_arxiv_id":null,"evidence_quote":"Defines the thermal-fluctuation fundamental NEP limit used to gauge how far the measured device stands from the ideal."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the commercial sub-THz detector NEP benchmark of about 10 pW/√Hz against which the all-electrical result is compared."}],"review_version":1}