{"id":"71453b9f-f89f-4a59-90e6-f67d941893ae","arxiv_id":"2506.20219","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An integrated optomechanical sensor with a suspended SiO2 membrane and high-Q Si3N4 microring reaches nano-Pascal-level noise-equivalent pressure, the best reported for integrated microcavity ultrasound detectors.","lead":"A chip-scale ultrasonic sensor that combines a suspended membrane with an embedded silicon nitride microring achieves record-low noise-equivalent pressures of 218 nPa/Hz^0.5 in air and 9.6 nPa/Hz^0.5 in water. The same sensor performs photoacoustic gas detection and underwater imaging, pointing toward compact, high-sensitivity ultrasound systems.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Water NEP of 9.6 nPa/√Hz is ~100-300× below the fluctuation-dissipation floor for the stated 450-µm-radius membrane (f0=52 kHz, Q=26); either the pressure calibration or the NEP extraction is off by that factor.","rationale":"The reader correctly identified absolute pressure calibration as the weakest point, but the problem is more acute: the claimed water NEP is not merely unverified, it is below the fluctuation-dissipation floor. I estimated the floor using the paper's quoted f0=52 kHz, linewidth 2 kHz, R_mem=450 µm, and the layer thicknesses in Methods. Even choosing A_eff equal to the full membrane area and m_eff equal to the total membrane mass—which minimizes the floor—gives p_th ≈ 2.1 µPa/√Hz; realistic mode-shape values give ≈2.9 µPa/√Hz. The claimed 9.6 nPa/√Hz is ~200-300× smaller. In air the floor is ~6 µPa/√Hz versus the claimed 218 nPa/√Hz, a ~30× inconsistency. Because the measured PSD peak is identified as thermomechanical, the NEP at resonance should equal the thermal floor if the pressure calibration is correct. The discrepancy forces a calibration error of orders of magnitude or an error in the NEP extraction. This invalidates the record-sensitivity claim and the quoted FOM. The demonstrations may still show relative sensitivity enhancement, but the central quantitative claim cannot stand as stated. I therefore recommend REJECT rather than CONDITIONAL: the issue is not a missing error bar but a physically impossible headline number. A two-hydrophone cross-calibration is the decisive check.","tokens_in":11522,"tokens_out":23885,"duration_ms":267806,"concrete_test":"Recompute the water NEP from the raw thermal-noise PSD and the calibrated response, and compare it with p_floor = sqrt(4 k_B T m_eff ω0/Q)/A_eff using the stated geometry (R=450 µm, f0=52 kHz, Q=26). Independently cross-calibrate the 52-kHz water pressure at the sensor location with a second hydrophone (e.g., a calibrated fiber-optic hydrophone); if the two calibrations agree to within 20%, the reported 9.6 nPa/√Hz will be ~100× below the thermal floor, confirming a calibration or extraction error rather than a record sensitivity.","verdict_should_be":"REJECT","load_bearing_attack":"At the mechanical resonance of a thermally equilibrated mode, the noise-equivalent pressure is bounded below by the fluctuation-dissipation theorem: p_th = sqrt(4 k_B T m_eff ω0/Q) / A_eff. Using the paper's own parameters — R_mem = 450 µm, f0 = 52 kHz, linewidth 2 kHz (Q=26), and a membrane mass estimated from the stated 4-µm BOX, 265-nm Si3N4, and 2-µm SiO2 cladding stack — even the most conservative choices (A_eff = full membrane area, m_eff = total mass) give p_th ≈ 2 µPa/√Hz; realistic mode-shape factors give ≈3 µPa/√Hz. The claimed 9.6 nPa/√Hz is therefore 200-300× below the thermal noise floor. In air (f0=289 kHz, Q≈29, R=450 µm) the floor is several µPa/√Hz, still more than 10× above the claimed 218 nPa/√Hz. Because the measured PSD peak is explicitly identified as thermomechanical noise (Fig. 3b), the reported NEP at resonance cannot be below this floor unless the absolute pressure calibration overestimates the acoustic field by ~10^2-10^3, or the NEP derivation is inconsistent. This is not merely a missing uncertainty budget; as written, the headline sensitivity is thermodynamically impossible.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an integrated optomechanical ultrasound sensor consisting of a suspended SiO2 membrane with an embedded high-Q Si3N4 microring resonator. The authors claim record-low noise-equivalent pressures of 218 nPa/√Hz at 289 kHz in air and 9.6 nPa/√Hz at 52 kHz in water, and they demonstrate photoacoustic detection of acetylene at 2.9 ppm with 1 s integration and underwater ultrasound imaging with 1.89 mm resolution at drive pressures of 0.3 mPa. The sensor is fabricated with wafer-scale CMOS-compatible processes, packaged with fiber coupling, and characterized using calibrated ultrasound transducers.","tokens_in":11831,"tokens_out":9506,"duration_ms":109699,"significance":"If the reported sensitivities were correct, the work would be a substantial advance over existing integrated photonic ultrasound sensors, whose NEPs are typically in the mPa/√Hz to µPa/√Hz range. The device concept is interesting: a microring embedded in a suspended membrane exploits simultaneous optical and mechanical resonances, and the demonstrations of photoacoustic gas spectroscopy and underwater imaging are relevant to practical applications. The comparison of the measured acetylene spectrum with HITRAN is a useful external validation. However, the central quantitative claim of nano-Pascal-level NEP at mechanical resonance appears thermodynamically inconsistent with the stated device parameters, and the paper does not provide an uncertainty budget for the absolute pressure calibration. As presented, the headline sensitivity therefore cannot be accepted.","major_comments":[{"comment":"The reported NEP minima are below the thermal fluctuation-dissipation floor for the stated membrane. For a resonant mechanical mode with effective mass m_eff, pressure-coupling area A_eff, angular frequency ω0, and quality factor Q, the thermal-noise-equivalent pressure is p_th = sqrt(4 k_B T m_eff ω0 / Q) / A_eff. Using the manuscript's membrane radius of 450 µm, a thickness stack of about 6.27 µm (4 µm BOX + 265 nm Si3N4 + 2 µm SiO2 cladding), f0 = 52 kHz, and Q = 26 in water, even the most favorable choices of A_eff equal to the full membrane area and m_eff equal to the total mass give p_th ≈ 2 µPa/√Hz; realistic mode-shape factors increase this to about 3 µPa/√Hz. The claimed 9.6 nPa/√Hz is therefore roughly 200-300 times below this bound. Since the measured PSD peak in Fig. 3b is identified as thermomechanical noise, the NEP at resonance cannot be below p_th unless the absolute pressure used for calibration is overestimated by a comparable factor or the NEP derivation is inconsistent. This makes the headline water sensitivity physically impossible with the parameters given in the paper.","section":"Figs. 3b, 3e, 3f; Methods: Ultrasound sensitivity characterization"},{"comment":"The same thermodynamic bound applies in air. With f0 = 289 kHz, Q ≈ 29, and the same membrane geometry, p_th is several µPa/√Hz even under the most conservative assumptions for m_eff and A_eff, which is more than an order of magnitude above the claimed 218 nPa/√Hz. The air value is thus also not supported by the stated device parameters and thermomechanical noise identification.","section":"Figs. 3a, 3e; Methods: Ultrasound sensitivity characterization"},{"comment":"No uncertainty budget is provided for the absolute acoustic pressure calibration (needle hydrophone in water, scanning laser vibrometer in air), and the paper reports a 38% device-to-device sensitivity variation, indicating that the headline NEP values are single-device numbers without error bars. In light of the above discrepancy with the fluctuation-dissipation bound, the full calibration chain and the derivation of the NEP from the measured PSD and response spectra need to be re-examined and reported with uncertainties before any quantitative sensitivity claim can be evaluated.","section":"Methods: Ultrasound sensitivity characterization; Device characterization"}],"minor_comments":[{"comment":"The sentence 'we implement R_ring = 450 µm and R_mem = 235 µm' is inconsistent with the main text, which states membrane radius 450 µm and microring radius 235 µm; the stated optimum ratio of 0.52 indicates the main-text values are the intended ones.","section":"Methods: Device design"},{"comment":"The data and code are stated to be 'available upon publication'; a repository link or DOI should be provided, and the phrase 'upon publication' is too vague for a reproducibility standard.","section":"Data availability"},{"comment":"There are several typographical errors: 'inluding' in the Conclusion, 'derict' in the Fig. 4e caption, 'A 2silica' in the Extended Data Fig. 1d caption, and a missing closing parenthesis after 'Exail MX-LN-10' in the Methods.","section":"Throughout"},{"comment":"The benchmarking claims involving NEP×√A at the 10^-8 Pa mm/√Hz level inherit the same calibration and thermodynamic issues as the raw NEP values; a revised version should report this figure with uncertainty bars and a clear statement of which devices and frequencies are used.","section":"Fig. 5 and Conclusion"}],"recommendation":"reject","confidential_remarks":"The central claim of the paper—nano-Pascal-level NEP at mechanical resonance—appears to violate the fluctuation-dissipation bound for the stated device geometry. This is not a mere missing-error-bar issue; it indicates a fundamental inconsistency in either the absolute pressure calibration or the NEP derivation. The paper would require a major re-analysis and likely a fundamentally revised conclusion, so I recommend rejection. The device concept and the demonstration of photoacoustic spectroscopy with HITRAN comparison are valuable, and a corrected manuscript with realistic sensitivity numbers could still be a useful contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the device work is real and interesting, but the reported NEP cannot be correct as written. The stress-test note is right: for the membrane parameters given (R = 450 μm, f0 = 52 kHz, Q = 26), the thermal fluctuation-dissipation floor for pressure sensing at resonance is about 2 μPa/√Hz, not 9.6 nPa/√Hz—a 200× discrepancy. In air the floor is roughly 5 μPa/√Hz versus the claimed 218 nPa/√Hz. Unless the absolute pressure calibration is off by a factor of hundreds, or the NEP is computed by mixing off-resonance noise with on-resonance response, the result is thermodynamically impossible.\n\nWhat the paper does well: the device itself is well engineered—wafer-scale, CMOS-compatible, packaged with fiber pigtails, optical Q above 10^6 in water. The photoacoustic spectroscopy demonstration is credible: the measured C2H2 spectrum matches HITRAN within 5%, and the air and water tests show clear mechanical resonance enhancement. The suspended vs non-suspended comparison is also useful.\n\nThe critical problem is the NEP calculation. The paper identifies the noise peak at resonance as thermomechanical. If that is true, the NEP at resonance cannot go below the FDT floor. The absence of error bars and an uncertainty budget is secondary; the primary issue is that the number itself violates a fundamental bound. I would want to see the full derivation of NEP and the raw calibration data for acoustic pressure before taking the record claim seriously. The 38% device-to-device variation and reliance on single-device headline values make this more urgent.\n\nThis paper is for a specialist in optomechanical sensing and integrated photonics. It is not a harmless overclaim; it affects the central message. Still, the underlying engineering and demonstrations are worth refereeing because a corrected calibration could make the NEP higher while the device remains competitive.\n\nSend it to review, but ask for a rigorous response to the FDT bound. If the authors cannot explain the factor of 200, the headline should be revised and the claims recast. As is, I would not cite the NEP value.","headline":"The device engineering is credible, but the nano-Pascal NEP claim is thermodynamically impossible as stated.","tokens_in":12364,"tokens_out":4260,"would_cite":false,"duration_ms":43475,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"An integrated optomechanical ultrasound sensor reports noise-equivalent pressures of 9.6 nanopascals per root hertz in water and 218 nanopascals per root hertz in air.","keywords":["optomechanical sensing","ultrasound detection","microring resonator","noise-equivalent pressure","photoacoustic spectroscopy","underwater imaging","suspended membrane","silicon nitride"],"falsifier":"An independent acoustic-pressure calibration at the membrane—for example, interferometric measurement of the membrane displacement combined with the known mechanical susceptibility, or a second calibrated reference hydrophone—would settle the claim: if the independently inferred pressure differs from the needle-hydrophone (water) or laser-vibrometer (air) calibration used in the paper by more than the expected uncertainty, the reported NEP values shift by exactly that factor.","tokens_in":11358,"feed_emoji":"🔊","tokens_out":9390,"duration_ms":94561,"temperature":0.7,"pith_summary":"This paper claims that an integrated optomechanical ultrasound sensor can detect sound with noise-equivalent pressures of $218\\ \\mathrm{nPa}/\\sqrt{\\mathrm{Hz}}$ in air at 289 kHz and $9.6\\ \\mathrm{nPa}/\\sqrt{\\mathrm{Hz}}$ in water at 52 kHz. The device is a suspended SiO$_2$ membrane with an embedded high-$Q$ Si$_3$N$_4$ microring, and it uses the membrane's mechanical flapping mode to amplify ultrasound-induced displacement while the ring's optical resonance converts that displacement into a laser-intensity modulation. The paper demonstrates two consequences of this sensitivity: photoacoustic detection of acetylene down to 2.9 ppm with 1 s integration, and underwater transmission imaging with 1.89 mm resolution at drive pressures of 0.3 mPa. A sympathetic reading is that the work establishes chip-scale, photonic ultrasound sensors as a sensitivity class above conventional piezoelectric transducers, with a route to wafer-scale fabrication. If correct, it narrows the gap between bench-top optical microcavity sensors and field-deployable integrated devices.","feed_headline":"Chip sensor hears sound down to 9.6 nanopascals per root hertz","feed_subtitle":"A suspended membrane and microring also enable ppm-level gas detection and millimeter-resolved underwater imaging.","key_machinery":"The load-bearing mechanism is simultaneous optical and mechanical resonance in a suspended-membrane–embedded microring. The mechanical flapping mode of the circular SiO$_2$ membrane amplifies the displacement response to incident ultrasound, and the high-$Q$ optical resonance of the embedded Si$_3$N$_4$ ring transduces that amplified displacement into a measurable intensity change. The two resonances work together because the narrow optical linewidth gives a steep transduction slope, while the mechanical linewidth (about 10 kHz in air and 2 kHz in water) provides resonant gain at 289 kHz and 52 kHz, respectively. The sensor is packaged with mode-conversion fibers, so it operates without free-space alignment. The design rule that carries the optimization is the radius ratio $R_\\mathrm{ring}/R_\\mathrm{mem}=0.52$, which maximises the radial strain on the ring for the $\\nu_{(0,0)}$ mode.","core_discovery":"The central claim is that operating a microring resonator on a suspended membrane at the coincidence of the optical resonance and the fundamental mechanical flapping mode $\\nu_{(0,0)}$ yields nano-Pascal-level noise-equivalent pressures. The membrane, 450 µm in radius and clamped at its periphery, is maximally displaced by incident ultrasound at the flapping mode; the embedded Si$_3$N$_4$ ring, radius 235 µm, is stretched by that displacement, shifting its optical resonance. With the laser locked to the blue-detuned slope of a resonance with intrinsic $Q\\approx1.35\\times10^6$, the shift appears as an intensity modulation. The paper reports NEP minima of $218\\ \\mathrm{nPa}/\\sqrt{\\mathrm{Hz}}$ at 289 kHz in air and $9.6\\ \\mathrm{nPa}/\\sqrt{\\mathrm{Hz}}$ at 52 kHz in water, and supports the claim with photoacoustic spectroscopy of C$_2$H$_2$ (minimum detectable concentration 2.9 ppm, 1 s integration) and underwater imaging of an 'F'-shaped groove with 1.89 mm resolution at 0.3 mPa drive pressure. The paper presents the sensitivity as a record for microcavity-based ultrasonic sensors.","pith_inferences":["Beyond the paper, the same suspended-membrane mechanical amplifier could be repurposed for other force or displacement sensors (magnetometers, electric-field sensors) by changing the transduction layer, because the flapping-mode gain is generic.","Beyond the paper, the demonstrated 0.3 mPa imaging pressure suggests the sensor could operate as a passive underwater listener at much lower ambient acoustic pressures; a direct test would be recording natural underwater sound at frequencies near 52 kHz.","Beyond the paper, the reported 38% sensitivity variation across nine devices implies that the single-device record NEP is not yet a population-level guarantee; a straightforward extension is to identify the fabrication parameter (e.g., membrane stress or ring width) that controls the spread.","Beyond the paper, the lack of an uncertainty budget for the pressure calibration means a metrological cross-calibration of the two reference instruments (needle hydrophone and laser vibrometer) could make the absolute NEP values reproducible across labs."],"forward_implications":["The demonstrated sensitivity places integrated photonic ultrasound sensors in the same NEP range as fiber-taper-coupled microdisk optomechanical sensors, without free-space alignment or taper fragility.","The photoacoustic demonstration implies that chip-scale sensors can detect trace gases at ppm concentrations with a 1 s integration time, which is relevant for breath analysis and environmental monitoring.","The underwater imaging result implies that acoustic imaging at drive pressures of 0.3 mPa is possible, three orders of magnitude below the pressure needed by a commercial hydrophone in the same experiment.","The figure of merit NEP $\\times \\sqrt{A}$ near $10^{-8}\\ \\mathrm{Pa\\,mm}/\\sqrt{\\mathrm{Hz}}$ suggests the sensitivity is not a consequence of a large sensing area, and therefore similar sensitivity could be expected in scaled-down or arrayed devices."],"supporting_citations":[{"why":"Reports a chip-scale optomechanical ultrasound sensor with µPa/√Hz-level sensitivity; the present record claim must be benchmarked against it, and its noise model is adopted.","marker":"[42]"},{"why":"Demonstrates air-coupled megahertz ultrasound detection with on-chip microcavities and supplies the pre-calibrated ultrasound transducer used in the sensitivity measurements.","marker":"[43]"},{"why":"Reports micropascal-sensitivity ultrasound sensors based on optical microcavities; provides a key comparison point for the NEP versus sensing-area figure of merit.","marker":"[44]"},{"why":"Introduces a silicon photonic ring resonator on an acoustical membrane; the concept the present work extends by adding membrane suspension and mechanical-resonance operation.","marker":"[39]"},{"why":"Presents a sensitive, broadband optomechanical ultrasound sensor in silicon photonics; the integrated platform whose sensitivity is claimed to be improved by the suspended membrane.","marker":"[40]"},{"why":"Reports an ultrasensitive plano-concave optical microresonator for ultrasound; a benchmark for high-sensitivity optical readout of acoustic waves.","marker":"[26]"}],"fun_headline_variants":["Ultrasound chip hits 9.6 nPa/√Hz record in water","Chip sensor detects 9.6 nPa/√Hz underwater","Record sensitivity: 9.6 nPa/√Hz ultrasound on a chip","9.6 nPa/√Hz in water: chip ultrasound record","Nano-Pascal ultrasonic sensor on a photonic chip"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"All reported nano-Pascal NEP values depend on the absolute calibration of the ultrasound pressure that reaches the membrane, and a constant error in that calibration would rescale every NEP value and the derived figure of merit by the same factor.","fun_headline_variants_meta":{"raw":{"variants":["Ultrasound chip hits 9.6 nPa/√Hz record in water","Chip sensor detects 9.6 nPa/√Hz underwater","Record sensitivity: 9.6 nPa/√Hz ultrasound on a chip","9.6 nPa/√Hz in water: chip ultrasound record","Nano-Pascal ultrasonic sensor on a photonic chip"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001574,"raw_usage":{"total_tokens":6312,"prompt_tokens":1005,"completion_tokens":5307,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":621,"completion_tokens_details":{"reasoning_tokens":5209}},"tokens_in":621,"tokens_out":5307,"duration_ms":43163,"temperature":1.0,"reasoning_tokens":5209,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:53:02.752979+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"An independent acoustic-pressure calibration at the membrane—for example, interferometric measurement of the membrane displacement combined with the known mechanical susceptibility, or a second calibrated reference hydrophone—would settle the claim: if the independently inferred pressure differs from the needle-hydrophone (water) or laser-vibrometer (air) calibration used in the paper by more than the expected uncertainty, the reported NEP values shift by exactly that factor.","supporting_citations":[{"cited_title":"& Bowen, W","cited_arxiv_id":null,"evidence_quote":"Reports a chip-scale optomechanical ultrasound sensor with µPa/√Hz-level sensitivity; the present record claim must be benchmarked against it, and its noise model is adopted."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates air-coupled megahertz ultrasound detection with on-chip microcavities and supplies the pre-calibrated ultrasound transducer used in the sensitivity measurements."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports micropascal-sensitivity ultrasound sensors based on optical microcavities; provides a key comparison point for the NEP versus sensing-area figure of merit."},{"cited_title":"M.et al.A sensitive optical micro-machined ultrasound sensor (omus) based on a silicon photonic ring resonator on an acoustical membrane.Sci","cited_arxiv_id":null,"evidence_quote":"Introduces a silicon photonic ring resonator on an acoustical membrane; the concept the present work extends by adding membrane suspension and mechanical-resonance operation."},{"cited_title":"J.et al.Sensitive, small, broadband and scalable optomechanical ultrasound sensor in silicon pho- tonics.Nat","cited_arxiv_id":null,"evidence_quote":"Presents a sensitive, broadband optomechanical ultrasound sensor in silicon photonics; the integrated platform whose sensitivity is claimed to be improved by the suspended membrane."},{"cited_title":"A.et al.Ultrasensitive plano-concave optical microresonators for ultrasound sensing.Nat","cited_arxiv_id":null,"evidence_quote":"Reports an ultrasensitive plano-concave optical microresonator for ultrasound; a benchmark for high-sensitivity optical readout of acoustic waves."}],"review_version":1}