{"id":"51a48892-eefd-43a0-ba43-300c2b6bd6e8","arxiv_id":"2501.03161","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A nanomechanical membrane with a broadband platinum absorber reaches a specific detectivity of 3.8 x 10^9 cm sqrt(Hz)/W, within a factor of three of the thermal fluctuation limit for a perfect 50% absorber.","lead":"This paper builds a small vibrating silicon nitride drum coated with a few nanometers of platinum, which absorbs infrared light and changes the drum's vibration frequency, creating a room-temperature infrared detector. The best detector is within a factor of three of the theoretical sensitivity limit for a perfect 50% absorber, a step toward cheap and broadband thermal sensing.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Broadband absorptance claim rests on measurement of only the mid-IR fiber-transmitted window; outside that window the 3 nm Pt FSIM absorber's spectral behavior is unverified, so the near-IR-to-far-IR and factor-of-three claims are conditional.","rationale":"The reader's verdict already rests on this gap, and I agree it is the single most load-bearing soft spot. The D* and NEP numbers are computed from measurements made with a fiber-coupled IR source, so they are trustworthy for that mid-IR window; the responsivity and thermal time-constant data, including comparison to FEM/analytical models across 18 membranes, provide independent support. However, the abstract and conclusion elevate the result by claiming near-IR-to-far-IR operation and a broadband average absorptance of 47%—claims that would be invalid if the 3 nm Pt film's absorption falls outside the measured window. The paper's own evidence shows the absorber is not at the flat 50% FSIM condition (5 nm thickness) and therefore the spectral extrapolation is the weakest link. A second, less central issue is that the measured frequency noise is dominated by additive phase noise rather than thermal fluctuation noise, so 'near the fundamental limit' should be read as a figure-of-merit comparison, not a statement about the dominant noise source; this does not change the numerical claim. No internal contradiction is apparent. The condition is to provide full-range absorptance data; if that data confirms α≈0.47 across the claimed band, the claims stand.","tokens_in":9853,"tokens_out":13240,"duration_ms":119781,"concrete_test":"Re-analyze the recorded FTIR absorptance spectra without applying the fiber-transmission mask: compute the average α over the full measured FTIR range and, if possible, measure α/reflectance of the same 3 nm Pt-on-SiN membranes from 0.8–2.5 μm (UV-Vis-NIR with integrating sphere) and 25–1000 μm (FTIR far-IR beamsplitter or THz-TDS). If the full-range average falls materially below 0.47 (e.g., <0.35), or if α drops sharply outside the fiber window, the 'near-IR to far-IR' and 'broadband average 47%' statements, and the factor-of-three comparison to the fundamental limit over that extended range, should be restricted to the measured mid-IR window.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative result—D* = 3.8e9 cm√Hz/W, less than 3× below the 50%-absorptance thermal-fluctuation limit—is derived from a responsivity measurement using an IR source delivered through an optical fiber. The absorptance used in that derivation (α = 0.47) is the average over only the spectral interval transmitted by that fiber (Supplementary Information, 'Calculation of Platinum Absorptance'); absorptance in the grey non-transmitted intervals is explicitly excluded. But the paper's headline contribution is an 'extended spectral range from near-IR to far-IR' and a 'broadband spectral absorptance on average of 47%.' The fabricated absorber is stated to have Pt thickness ~3 nm, not the 5 nm needed for flat 50% FSIM behavior, and the shown FTIR spectra (Fig. 2c, Supp. Fig. 1b) exhibit noticeable spectral dependence. No measurement supports α below ~2 μm or above ~25 μm, so the near-IR and far-IR/THz ends of the broadband claim—and the corresponding statement that D* is only ~3× from the fundamental limit across that extended range—are extrapolations, not observations. This is not internal inconsistency, but it is a load-bearing gap in the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a nanomechanical silicon nitride membrane resonator with a platinum free-space impedance-matched (FSIM) absorber for uncooled thermal infrared detection. A circular clearance in the Pt absorber lets the readout laser hit the bare SiN, suppressing photothermal back-action; measurements in vacuum and an artificial thermal bath characterize thermal time constants, power responsivity, and frequency stability for 18 membranes of three sizes. For the best 1 mm resonator in the (2,2) mode, the authors report NEP = 27 pW/√Hz and D* = 3.8×10^9 cm√Hz/W, stating this is less than a factor of three below the fundamental thermal-fluctuation limit for a 50%-absorptance detector at room temperature, with an absorber claimed to be broadband from near-IR to far-IR. The results are compared with analytical models and FEM simulations, using material-parameter uncertainty bands.","tokens_in":10125,"tokens_out":4195,"duration_ms":39167,"significance":"If substantiated, the central claim is significant: a room-temperature, broadband, uncooled IR detector operating within a factor of three of the fundamental thermal-fluctuation limit would be a notable advance for nanomechanical sensing and could compete with state-of-the-art pyroelectric and optomechanical detectors. The design contribution of a clearance in the absorber to avoid photothermal back-action is elegant, is supported by a direct comparison of Allan deviations with and without the clearance, and is a transferable idea. The paper also offers a fairly complete characterization against analytical and FEM models with uncertainty bands, and uses machine-checkable formulas for the figures of merit. The main limitation is that the broadband and near-limit claims rest on an absorptance value measured only over the mid-IR window of the fiber-coupled source, and on a single best-performing device without reported uncertainties.","major_comments":[{"comment":"The paper's broadband claim and the factor-of-three closeness to the fundamental limit depend on the absorptance value α = 0.47, but this α is the unweighted average of the FTIR-measured absorptance only over the spectral window transmitted by the IR optical fiber (white area in Supp. Fig. 1a; grey areas explicitly excluded). The FTIR data in Fig. 2c and Supp. Fig. 1b show clear spectral dependence and, as the Supplementary states, the Pt thickness is approximately 3 nm rather than the 5 nm needed for flat 50% absorptance. No measurement supports α ≈ 0.47 at near-IR wavelengths below approximately 2 μm or above approximately 25 μm, so the asserted 'near-IR to far-IR' (or THz) range and the corresponding statement that D* is only about three times below the fundamental limit across that extended range are extrapolations, not observations. Please either provide absorptance data or validated optical calculations covering the full claimed spectral range, or restrict the central claims to the measured mid-IR window.","section":"Section IIIB and Supplementary 'Calculation of Platinum Absorptance'"},{"comment":"The headline values NEP = 27 pW/√Hz and D* = 3.8×10^9 cm√Hz/W are reported as point values without uncertainties, and the text in Section IIID states that 'performance varies strongly between different resonators' while the caption of Fig. 5 says 'minimal differences in performance between the different modes and dimensions.' If the best-performing device is selected from a set with strong device-to-device scatter, the factor-of-three closeness to the fundamental limit cannot be assessed without a quantitative measure of that scatter. Please provide the distribution of NEP and D* over the measured devices, or at least error bars propagated from the uncertainties in Sy, RP, and α through Eqs. (1) and (4).","section":"Section IIID and Fig. 5"},{"comment":"The responsivity is measured with a broadband IR source (Arclight-MIR) delivered through an optical fiber, with P = 7.5 μW stated as the impinging power. The absorptance used in the NEP calculation, α = 0.47, is an unweighted average over the fiber-transmitted spectral window. If the source spectrum is not flat within that window, the effective absorptance entering Eq. (1) differs from the unweighted average. The paper should specify the source spectral distribution and either weight α accordingly or show that the absorptance is sufficiently flat within the window that the unweighted average is a good approximation.","section":"Section IIIB"}],"minor_comments":[{"comment":"The statement in Section IIID that 'performance varies strongly between different resonators' appears to conflict with the Fig. 5 caption that 'minimal differences in performance between the different modes and dimensions.' Please clarify whether the variation is among individual devices of the same geometry or among different modes and membrane sizes.","section":"Section IIID and Fig. 5"},{"comment":"The symbol ε is used for emissivity in Eqs. (7) and (16), while α is used for absorptance in Eq. (1); the Supplementary and the text sometimes use these interchangeably. Please adopt consistent notation and state the assumption that the thermal emissivity equals the IR absorptance.","section":"Section IIIA, Eqs. (7) and (16)"},{"comment":"The phrases 'near-IR to far-IR' and 'to the terahertz regime' are used without quantitative wavelength bounds. Please specify the intended range or cite a reference for the FSIM bandwidth.","section":"Abstract and Section IV"},{"comment":"The caption of Fig. 2c does not explain the grey shaded regions; the Supplementary does, but the main-text figure should also state that these are wavelength ranges not transmitted by the IR optical fiber.","section":"Fig. 2c caption"},{"comment":"The '90-10 method' for extracting the thermal time constant is not defined; a brief sentence describing the method would improve readability.","section":"Section IIIA"}],"recommendation":"major_revision","confidential_remarks":"This is a well-executed device-characterization paper. The main technical concern is the spectral extrapolation: the broadband and near-limit claims are not yet supported by data outside the mid-IR fiber-transmitted window, and the paper's own Supplementary shows the Pt thickness is not the 5 nm needed for flat absorptance. The lack of error bars on the headline NEP/D* is also worth addressing, as the paper claims a factor-of-three margin from a fundamental limit while noting strong device-to-device variation. These items appear fixable within the manuscript's scope, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read the Martini et al. paper on uncooled IR detection with SiN membrane resonators. The punchline: they demonstrate a 1 mm resonator with NEP 27 pW/√Hz and D* 3.8e9 cm√Hz/W, within a factor of ~3 of the thermal-fluctuation limit for 50% absorptance, and they do it with a clean optical readout and a Pt FSIM absorber. The numbers look internally consistent: NEP is derived from measured responsivity, frequency stability, and absorptance, and the model curves match the data within uncertainties.\n\nWhat's genuinely new: the circular clearance in the Pt absorber for the readout laser, which suppresses photothermal back-action from laser intensity noise. They show a direct comparison of Allan deviation with laser on Pt vs on the clearance; the improvement is clear and the model matches. That's a practical, transferable trick. The thermal time constant and responsivity characterization across three membrane sizes and modes is also careful, with FEM and analytical models.\n\nSoft spots, in order of importance. First, the broadband claim: abstract and conclusions say \"near-IR to far-IR\" and \"broadband spectral absorptance on average 47%,\" but the absorptance is measured only over the window transmitted by the IR fiber (~2–25 μm, roughly mid-IR). The grey regions in Fig. 2c are excluded from α, and the Pt came out ~3 nm instead of the 5 nm needed for flat 50% FSIM behavior, so the spectra show real wavelength dependence. The D* value of 3.8e9 is measured with that fiber-delivered source, so for that window the result holds. But the \"less than three times below the fundamental limit\" across the full claimed range is an extrapolation, not something they measured. The authors do flag the grey ranges, so they're not hiding it, but the abstract overstates.\n\nSecond, NEP and D* are reported without error bars. There's visible device-to-device spread in Fig. 5, and the best device is selected. A single number with no uncertainty makes it hard to compare against the commercial pyroelectric they cite. This is a fixable reporting issue.\n\nThird, the noise budget: the measured frequency stability is dominated by additive phase noise (thermomechanical + detection), not temperature-fluctuation noise—the ultimate limit. That's fine for D*, but the \"near fundamental limit\" phrasing in the title leans on D* rather than on being at the actual noise floor. They do show the phase-noise model matches, so the analysis is honest.\n\nOverall: this is a competent experimental paper with a genuinely useful design improvement. The central D* claim is likely correct for the measured mid-IR band. The broadband extension is plausible but unverified outside the fiber window. I'd take it for peer review—it will survive referee scrutiny if the error bars and spectral qualification are added. Worth citing if you work on NEMS thermal detectors.\n\nRecommendation: engage; accept with minor-to-moderate revisions.","headline":"Solid experimental demonstration of a nanomechanical IR detector with a clever readout-laser clearance; the headline D* is internally consistent, but the broadband absorptance claim outruns the measurement.","tokens_in":10678,"tokens_out":2715,"would_cite":true,"duration_ms":24610,"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":"A nanomechanical IR detector reaches 27 pW/√Hz at room temperature","keywords":["infrared detection","nanomechanical resonator","thermal detector","FSIM absorber","noise equivalent power","specific detectivity","silicon nitride membrane","photothermal back-action"],"falsifier":"Measure the absorptance spectrum of the same 3 nm Pt-on-SiN film from 1 micrometre out to terahertz wavelengths with a broadband source; if the spectrally averaged absorptance drops well below 0.47 outside the measured mid-IR window, or if a calibrated NEP measurement at those wavelengths degrades by more than the stated factor, the extended-range near-fundamental-limit claim is falsified.","tokens_in":9685,"feed_emoji":"🌡️","tokens_out":9130,"duration_ms":73288,"temperature":0.7,"pith_summary":"The paper aims to show that a nanomechanical silicon nitride drum, coated with a thin platinum film, can detect infrared light at room temperature almost as well as physics allows. The key result is a noise equivalent power of $27\\,\\mathrm{pW}/\\sqrt{\\mathrm{Hz}}$ and a specific detectivity of $3.8\\times10^9\\,\\mathrm{cm}\\sqrt{\\mathrm{Hz}}/\\mathrm{W}$ in the best 1 mm resonator, within a factor of three of the thermal-fluctuation floor for a detector that absorbs 50% of the light. The broadband platinum absorber gives the device a flat response from near-IR to far-IR, unlike narrow-band metamaterial absorbers. A circular clearance in the platinum film lets the readout laser hit bare silicon nitride, removing the photothermal back-action that otherwise degrades frequency stability. If the claims hold, the device offers a practical route to uncooled IR sensing near the fundamental limit.","feed_headline":"Nanomechanical IR detector hits 27 pW/√Hz at room temperature","feed_subtitle":"Platinum-coated silicon nitride drum senses infrared from near-IR to far-IR at three times the physics limit.","key_machinery":"The load-bearing object is a square 50 nm SiN membrane (1–3 mm side) coated with a roughly 3 nm Pt film that acts as a free-space impedance-matched (FSIM) absorber: a thin metal film with a nominal 50% absorptance over a wide infrared range. Absorbed IR changes the membrane temperature, which changes its stress and therefore its resonance frequency; that frequency shift is read out optically with a laser Doppler vibrometer and a phase-locked loop. The platinum film has a circular clearance where the readout laser hits the bare SiN, suppressing photothermal back-action noise from laser intensity fluctuations. The performance model combines the thermal time constant $\\tau_{\\mathrm{th}} = C/G$, the temperature responsivity $R_T = -\\alpha_{\\mathrm{th}}/(2(1-\\nu))\\,E/\\sigma$, and noise terms (additive phase noise, temperature-fluctuation noise, photothermal back-action) into $\\mathrm{NEP} = \\sqrt{S_y}/(R_P\\alpha)$.","core_discovery":"The paper's central claim is that a 50 nm square silicon nitride membrane resonator, coated with a roughly 3 nm platinum free-space impedance-matched (FSIM) absorber, works as an uncooled thermal infrared detector whose sensitivity approaches the fundamental thermodynamic limit. In the best 1 mm device, operated in the (2,2) mode, the measured noise equivalent power is $27\\,\\mathrm{pW}/\\sqrt{\\mathrm{Hz}}$ and the specific detectivity is $D^* = 3.8\\times10^9\\,\\mathrm{cm}\\sqrt{\\mathrm{Hz}}/\\mathrm{W}$, less than a factor of three below the theoretical room-temperature limit $D^* \\approx 1.0\\times10^{10}\\,\\mathrm{cm}\\sqrt{\\mathrm{Hz}}/\\mathrm{W}$ for an ideal detector with 50% absorptance. The detector keeps this performance while covering an extended spectral range from near-infrared to far-infrared, because the metal-film absorber gives a nominally flat ~50% absorptance rather than a narrow resonant peak. The authors position the device among the most sensitive room-temperature IR detectors reported, on par with state-of-the-art optomechanical detectors using subwavelength metamaterial absorbers.","pith_inferences":["If the absorptance remains flat beyond the measured fibre window, the same detector could serve as a broadband reference standard for IR power metrology without spectral calibration.","The mode-shape dependence of responsivity suggests that engineering the temperature field, for example placing the absorber where thermal isolation and displacement overlap, could improve sensitivity beyond what this geometry achieves.","The clearance trick could be transferred to other optomechanical detectors with absorbing coatings, wherever readout-light absorption creates back-action.","One testable extension is to repeat the NEP measurement with band-pass filters inside the claimed range; a flat D* versus wavelength would confirm the broadband claim."],"forward_implications":["The same detector geometry should work as a broadband spectrometer element, because the flat 50% absorptance avoids spectral shaping by the absorber.","Only a factor of 1.4 in ultimate sensitivity is traded for broadband operation compared with a perfect 100% absorber.","Pointing the readout laser at the clearance suppresses photothermal back-action, leaving additive phase noise as the dominant noise source.","Smaller membranes respond faster ($\\tau_{\\mathrm{th}} = 14$ ms) and give the best NEP, so footprint and sensitivity align.","Trampoline resonators with the same FSIM absorber should push the detector closer to the fundamental limit."],"supporting_citations":[{"why":"Supplies the NEP definition, the thermal-fluctuation noise model, and the fundamental detectivity limit that sets the paper's target.","marker":"[21]"},{"why":"Establishes the 50% maximum absorptance of a thin metal film, the theoretical basis of the FSIM absorber.","marker":"[28]"},{"why":"Demonstrates ultrathin metal films as impedance-matched absorbers for infrared light, the design the present absorber extends.","marker":"[19]"},{"why":"Prior SiN trampoline resonator with FSIM absorber and the thermal model that the present work builds on and benchmarks against.","marker":"[12]"},{"why":"Shows that photothermal back-action from readout light limits frequency stability in SiN resonators, motivating the clearance design.","marker":"[22]"},{"why":"Provides the analytical responsivity and noise models used for the NEP calculation and the trampoline-resonator outlook.","marker":"[25]"},{"why":"State-of-the-art frequency-noise-optimized nanomechanical THz detector used as a performance comparison.","marker":"[13]"},{"why":"Thermodynamically limited uncooled detector with an ultra-low-mass subwavelength absorber used as a comparison baseline.","marker":"[15]"}],"fun_headline_variants":["Nanomechanical IR detector nears fundamental limit","Uncooled IR sensor hits 27 pW/√Hz sensitivity","Broadband IR detection in a platinum-coated nanodrum","Room-temperature IR detector within 3x of ideal","Nanoresonator IR sensing from near- to far-infrared"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The detection claims rest on the platinum film absorbing about 47% of infrared across the whole near-IR-to-far-IR range, but the paper only measures absorptance in the mid-infrared window that its optical fibre transmits.","fun_headline_variants_meta":{"raw":{"variants":["Nanomechanical IR detector nears fundamental limit","Uncooled IR sensor hits 27 pW/√Hz sensitivity","Broadband IR detection in a platinum-coated nanodrum","Room-temperature IR detector within 3x of ideal","Nanoresonator IR sensing from near- to far-infrared"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001114,"raw_usage":{"total_tokens":4699,"prompt_tokens":1065,"completion_tokens":3634,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":681,"completion_tokens_details":{"reasoning_tokens":3552}},"tokens_in":681,"tokens_out":3634,"duration_ms":23766,"temperature":1.0,"reasoning_tokens":3552,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:53:50.950356+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the absorptance spectrum of the same 3 nm Pt-on-SiN film from 1 micrometre out to terahertz wavelengths with a broadband source; if the spectrally averaged absorptance drops well below 0.47 outside the measured mid-IR window, or if a calibrated NEP measurement at those wavelengths degrades by more than the stated factor, the extended-range near-fundamental-limit claim is falsified.","supporting_citations":[{"cited_title":"Fundamentals of Nanomechanical Resonators","cited_arxiv_id":null,"evidence_quote":"Supplies the NEP definition, the thermal-fluctuation noise model, and the fundamental detectivity limit that sets the paper's target."},{"cited_title":"Comparative Analysis of Nanomechanical Resonators: Sensitivity, Response Time, and Practical Considerations in Photothermal Sensing","cited_arxiv_id":"2406.03295","evidence_quote":"Establishes the 50% maximum absorptance of a thin metal film, the theoretical basis of the FSIM absorber."},{"cited_title":"Thermal ir de- tection with nanoelectromechanical silicon nitride trampoline resonators","cited_arxiv_id":null,"evidence_quote":"Prior SiN trampoline resonator with FSIM absorber and the thermal model that the present work builds on and benchmarks against."},{"cited_title":"Ultrathin 2 nm gold as impedance-matched ab- sorber for infrared light","cited_arxiv_id":null,"evidence_quote":"Shows that photothermal back-action from readout light limits frequency stability in SiN resonators, motivating the clearance design."},{"cited_title":"Frequency fluctuations in nanomechanical silicon nitride string resonators","cited_arxiv_id":null,"evidence_quote":"Provides the analytical responsivity and noise models used for the NEP calculation and the trampoline-resonator outlook."},{"cited_title":"Nanoelectromechanical infrared spectroscopy with in situ separation by thermal desorption: Nems-ir-td.ACS sensors, 8(4):1462–1470, 2023","cited_arxiv_id":null,"evidence_quote":"State-of-the-art frequency-noise-optimized nanomechanical THz detector used as a performance comparison."},{"cited_title":"Thermodynamically limited uncooled infrared detector using an ultra-low mass perforated subwavelength absorber","cited_arxiv_id":null,"evidence_quote":"Thermodynamically limited uncooled detector with an ultra-low-mass subwavelength absorber used as a comparison baseline."}],"review_version":1}