{"id":"0f960f75-4151-46c1-99fa-890880205115","arxiv_id":"2411.14101","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A shallow-etched silicon pair-rod metasurface supports a quasi-bound state in the continuum with a measured quality factor of 101,486 at 1560 nm, a record for BIC metasurfaces.","lead":"Researchers built tiny paired silicon rods on a silicon-on-insulator chip and measured an optical resonance quality factor of 101,486, roughly ten times higher than any previously reported BIC metasurface. The shallow, low-contrast etching reduces roughness scattering, a design idea that could make ultra-sensitive sensors and low-power nanoscale lasers more practical.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The record Q value is plausible, but the paper's mechanistic claim that shallow etching reduces scattering loss rests on a fitted α-independent Qscat; that assumption is untested and could conflate sidewall scattering with measurement floors.","rationale":"The reader's weakest-assumption analysis identifies exactly the load-bearing point: the extracted Qscat(d) trend is the only quantitative evidence for the paper's central mechanism, and it depends on an untested assumption of α-independent scattering loss. I agree with that assessment. The record Q=101,486 measurement is independent of the Qscat model and can survive even if the model is wrong, so this concern does not overturn the headline number. It does, however, make the stated design principle — reduced scattering losses in shallow-etched low-contrast metasurfaces — conditional rather than established. The proposed nested-model refit is a concrete, low-cost check that uses the existing data to test whether the constant-Qscat assumption actually holds. The paper has independent supporting elements, including FEM-computed radiative Q factors, FDTD spectra, multiple etching depths, and sensing data, but none of these validates the scattering-loss extraction. Therefore the appropriate verdict remains CONDITIONAL, with no adjustment from the reader's verdict.","tokens_in":12201,"tokens_out":15645,"duration_ms":169377,"concrete_test":"Using the data behind Fig. 3b, re-fit each d dataset with a nested model in which Qscat is allowed to vary with α, e.g., Qscat(α)=Qscat,0(1+β α), and compare against the constant model by AIC or an F-test. If nonzero β is preferred, the α-independence assumption is falsified and the extracted Qscat(d) cannot be interpreted as a pure sidewall-scattering loss; if β≈0, the constant model is internally consistent, though independent checks of aperture and angular floors would still be needed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central mechanistic claim — that shallower etching raises Q by reducing scattering loss — is supported quantitatively only by the Q-analysis in Fig. 3b–d, which decomposes Q^-1 = Qr^-1 + Qscat^-1 and fits each etching depth with a single α-independent Qscat. That decomposition is assumed, not tested. Sidewall-roughness scattering is not automatically α-independent: as α changes, the qBIC mode profile changes, so the field overlap with the rough sidewalls changes; if Qscat has any α-dependence, the single fitted value is a weighted average that can bias the comparison across d. The small-α saturation that fixes Qscat could also be produced by experimental floors excluded from the model, such as the 100×100 μm pattern aperture, the NA=0.26 focused illumination, or spectral-resolution limits, none of which is characterized. Consequently, the trend in Fig. 3d may reflect fit artifacts or measurement floors rather than a real reduction of sidewall scattering. The record Q value itself is not at issue, but the stated design principle is not established by the present analysis.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports silicon-on-insulator metasurfaces supporting higher-order quasi-bound states in the continuum (qBIC) in shallow-etched pair-rod unit cells. The authors experimentally demonstrate a Q factor of 101,486 at λ=1560.3 nm for an etch depth of 82.7 nm and asymmetry α=1%, and they attribute the ultrahigh Q to a combination of high radiative Q of the qBIC2 mode and reduced scattering loss in shallow-etched designs. The interpretation is supported by a Q-analysis that decomposes the measured Q as Q^{-1}=Qr^{-1}+Qscat^{-1}, with Qr taken from FEM simulations and Qscat fitted independently for three etch depths (82.7, 116.1, and 149.5 nm). The paper also demonstrates refractometric sensing in water with a limit of detection near 3×10^{-5} and a FOM up to 829.","tokens_in":12447,"tokens_out":5332,"duration_ms":53045,"significance":"If confirmed, the reported Q factor would be a substantial advance over previous BIC metasurface demonstrations, which have typically remained in the thousands or low tens of thousands, and the low-contrast / shallow-etch design is a simple and CMOS-compatible route to ultrahigh-Q operation. The systematic variation of etch depth and asymmetry, the combined FDTD/FEM simulations, and the comparison table with prior work are useful contributions. The sensing demonstration adds practical value. However, the headline record and the mechanistic claim that shallow etching reduces scattering loss currently rest on a single Fano fit and on an assumed form of the loss model, so the significance can only be fully realized after the statistical and modeling concerns below are addressed.","major_comments":[{"comment":"The central mechanistic claim that shallower etching raises Q by reducing scattering loss is supported only by fitting the experimental Q factors with Q^{-1}=Qr^{-1}+Qscat^{-1}, where Qscat is assumed independent of α and is fitted as a single value per etch depth. This assumption is not tested, and the qBIC mode profile changes with both α and d, so sidewall-roughness scattering can in principle depend on α. If Qscat has any α-dependence, the fitted value is a weighted average and the trend in Figure 3d may be biased. Please provide the fit residuals, report confidence intervals for Qscat, and compare the quality of fit with an α-dependent Qscat or with disorder-incorporating simulations. Without this, the stated design principle is not quantitatively established.","section":"Figure 3b-d and Q-analysis paragraph"},{"comment":"The headline Q factor of 101,486 is extracted from a single Fano fit of a single device, with no fitting uncertainty, no repeated-device statistics, and no discussion of how the sidebands mentioned for Figure 2c or the choice of fitting background affect the extracted linewidth. Because the record claim is a central result, please report the standard deviation or spread obtained from multiple nominally identical devices and multiple independent fits, show representative fit residuals, and state the criteria used to select spectra for Q extraction.","section":"Figure 2d and abstract record claim"},{"comment":"The small-α saturation in Figure 3b is interpreted as an inherent Qscat, but the experimental setup may impose a measurable Q ceiling: the tunable laser sweep speed, spectral resolution, NA=0.26 focused illumination, the 100×100 μm patterned aperture, and polarizer-based background suppression are not characterized. The fact that Qscat appears to saturate near 10^5 for d=49.3 nm (Figure S3) is also consistent with an instrument-limited floor rather than a material or fabrication limit. Please characterize the minimum measurable linewidth of the setup and include this contribution explicitly in the Q budget before concluding that the saturation originates from scattering loss.","section":"Figure 3b and Supporting Information S3"},{"comment":"Reference [18] (Huang et al., Nature Communications 2023) reports a Q factor of 2.4×10^5 in an all-dielectric metasurface, yet Table 1, which is captioned 'all-dielectric metasurfaces,' does not include it, and the text states that the present Q factor is one order of magnitude higher than the highest reported value of 18,511. If the record is intended to be restricted to BIC metasurfaces, this must be stated explicitly in both the abstract and Table 1; otherwise, the comparison is inaccurate and the 'record-high' claim is not supportable.","section":"Table 1 and abstract claim of record-high Q"}],"minor_comments":[{"comment":"There is a typo in the sentence describing Figure 1d: 'low-contrast matesurfaces' should be 'low-contrast metasurfaces'.","section":"Figure 1d caption/paragraph"},{"comment":"The caption lists 'troidal dipole' as a resonance type; this should be 'toroidal dipole'.","section":"Table 1 caption"},{"comment":"The text reports that small sidebands occasionally appear in the spectra and attributes them to structural variations or interference, but it does not describe how the Fano fitting treats these sidebands or how spectra with sidebands were excluded. Please specify the fitting procedure and any selection criteria.","section":"Figure 2c and 2d"},{"comment":"The sub-picometer wavelength fluctuations are evaluated over approximately one minute; the authors note this is to minimize drift, but it would be useful to state explicitly that longer-term stability is not claimed.","section":"Figure 4c"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline number is the reason to read this: an experimentally measured Q of 101,486 in a silicon BIC metasurface, about a factor of five above the previous best in this class, from a simple shallow-etch pair-rod design. That is a real, concrete advance. The design idea is sensible: etch only partway into the silicon layer so the qBIC mode's field stays away from the roughest sidewall region, and use a higher-order mode that is already well confined. The Q-versus-etch-depth trend in Fig. 3 is consistent with that picture, and the sensing data (sub-picometer wavelength noise, LOD ~3e-5) are a reasonable supporting demo.\n\nThe soft spots are not fatal but they are real. The 101,486 figure comes from one Fano fit of one spectrum, with no uncertainty, no repeated-device statistics, and no discussion of what the fit background or the visible sidebands do to the linewidth. That is a reporting gap, not evidence the number is wrong; the order of magnitude is credible.\n\nThe larger issue is the mechanistic claim. The paper attributes the improvement to reduced scattering loss and extracts Qscat from Q^-1 = Qr^-1 + Qscat^-1 with Qscat assumed independent of alpha. That assumption is never tested. As alpha changes, the mode profile changes, so the field overlap with the sidewalls can change; the fitted alpha-independent Qscat could be absorbing alpha-dependent variations or experimental floors such as the 100x100 um aperture, the NA=0.26 focusing, or the sweep speed. The saturation below d=50 nm in Fig. S3 hints at a floor, but that floor is not characterized. So the design principle is plausible and likely correct, but the paper does not nail it with the present analysis.\n\nI would send this to peer review. The measurement alone is worth refereeing, and the shallow-etched design is simple enough that others can reproduce it. The referees should push for error bars, multiple devices, and either a direct test of the alpha-independence of Qscat or a softened mechanistic wording. With those changes, this becomes a solid, citable record.","headline":"Record Q measurement is credible and worth publishing; the scattering-loss mechanism is suggestive but not established.","tokens_in":12959,"tokens_out":2502,"would_cite":true,"duration_ms":24632,"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":"Shallow-etched silicon pairs reach a record quality factor of 101,486 under normal-incidence light.","keywords":["bound states in the continuum","metasurfaces","quality factor","silicon photonics","refractometric sensing","nanofabrication","telecom wavelength"],"falsifier":"Fabricate metasurfaces with the same etching depth but deliberately varied sidewall roughness (for example, by changing the Bosch process passivation time), and measure Q at fixed small asymmetry; if Q does not change with roughness, or if the fitted Q_scat varies systematically with α, the paper's scattering-loss conclusion would be falsified.","tokens_in":11997,"feed_emoji":"🔬","tokens_out":7099,"duration_ms":65938,"temperature":0.7,"pith_summary":"The paper sets out to show that the quality factor of all-dielectric bound-state-in-the-continuum metasurfaces is no longer limited to the thousands: by making the silicon etching shallow, both the radiative quality factor and the scattering-limited quality factor increase, and the authors measure a resonance with Q = 101,486 at λ = 1560.3 nm under normal incidence. A sympathetic reader would care because such ultrahigh Q values were previously confined to photonic-crystal slabs or fragile patterned-photoresist designs, whereas this device is a monolithic silicon metasurface compatible with standard CMOS fabrication. The paper further shows that the same shallow design gives sub-picometer wavelength stability in water, with a refractive-index limit of detection near 10⁻⁵. If the claim holds, it removes a fabrication bottleneck for high-Q nanophotonics and makes strong light-matter interactions accessible in planar devices.","feed_headline":"Silicon pair array hits record Q factor of 101,486","feed_subtitle":"The low-contrast design beats prior all-dielectric metasurfaces by an order of magnitude, opening high-Q nanophotonics.","key_machinery":"The load-bearing identity is the loss decomposition $Q^{-1} = Q_{\\mathrm{r}}^{-1} + Q_{\\mathrm{scat}}^{-1}$, in which the radiative part follows the quasi-BIC scaling $Q_{\\mathrm{r}} = Q_0 \\alpha^{-2}$ (with $\\alpha$ the geometric asymmetry of the silicon pair) and the scattering part $Q_{\\mathrm{scat}}$ is assumed to be a constant independent of $\\alpha$ and extracted by fitting the measured Q. The mechanism that carries the argument is the low-contrast pair-rod design: etching the 400 nm silicon layer only 82.7 nm deep keeps the qBIC2 mode's field concentrated away from the sidewalls, so both $Q_{\\mathrm{r}}$ and $Q_{\\mathrm{scat}}$ are large. The critical coupling condition $Q_{\\mathrm{r}} = Q_{\\mathrm{scat}}$ is used in the sensing demonstration.","core_discovery":"The central claim is that shallow-etched (low-contrast) silicon pair-rod metasurfaces supporting a higher-order quasi-BIC mode reach an experimental Q of 101,486, exceeding the previous all-dielectric BIC-metasurface record by about a factor of five and typical values by one to two orders of magnitude. The authors attribute this to two effects working together: the higher-order mode's electric field stays confined inside the silicon, keeping a large radiative Q even as the asymmetry parameter α grows, and the shallow etching (depth 82.7 nm on a 400 nm silicon layer) shrinks the sidewall area, cutting scattering loss from fabrication roughness. They support this by decomposing the measured Q as Q⁻¹ = Q_r⁻¹ + Q_scat⁻¹ and showing the extracted scattering term decreases as etching depth decreases, while the radiative term obeys Q_r = Q_0 α⁻².","pith_inferences":["One natural extension the authors leave implicit: the same shallow-etch strategy should suppress scattering in other materials (e.g., silicon nitride or TiO2) where sidewall roughness is the dominant loss, so Q above 10⁵ should be reproducible in other systems.","A testable consequence: if Q_scat is truly a geometric property of the etch depth, then two devices with the same d but different roughness levels (e.g., from different Bosch process parameters) should show measurable differences in Q; this would verify that the Q gain comes from reduced scattering rather than from a changed radiative mode.","The paper measures only one incident polarization; checking whether the ultrahigh Q persists for arbitrary linear polarization would determine whether the design can be used in unpolarized or polarization-agnostic sensing environments."],"forward_implications":["The experimental Q factor of 101,486 is one to two orders of magnitude above typical qBIC metasurface values and about five times the prior all-dielectric metasurface record.","The shallow-etch approach applies to other BIC metasurface configurations and other wavelengths, not just silicon pairs at telecom wavelengths.","A BIC metasurface in water can track the resonance peak with sub-picometer wavelength fluctuations, yielding a limit of detection around 10⁻⁵ for refractive index changes.","The optically measured Q in the small-asymmetry limit is set by the fabrication-quality floor Q_scat, so further Q gains require reducing sidewall roughness rather than only increasing radiative Q."],"supporting_citations":[{"why":"Provides the prior topologically engineered ultrahigh-Q approach that the low-contrast design seeks to simplify.","marker":"[15]"},{"why":"Demonstrates an ultrahigh-Q guided-mode resonance using a thin patterned photoresist on silicon, which the new monolithically etched design aims to replace.","marker":"[18]"},{"why":"Reports an earlier shallow-etched BIC metasurface with Q around 6061, the direct precursor that this paper extends to record values.","marker":"[21]"},{"why":"Underlies the statement that a thick buried-oxide layer suppresses substrate leakage, isolating scattering as the main loss channel.","marker":"[22]"},{"why":"Supplies the inverse-square scaling Q_r = Q_0 α⁻² for qBIC radiative quality factors used in the Q decomposition.","marker":"[23]"},{"why":"Provides the previous experimental record Q of 18,511 for all-dielectric metasurfaces, the explicit baseline that the new Q factor surpasses.","marker":"[27]"}],"fun_headline_variants":["Low-contrast metasurfaces reach Q of 101,486","Shallow-etch silicon pairs hit ultrahigh Q of 101,486","Record Q: 101,486 in low-contrast BIC metasurfaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire quantitative argument depends on the assumption that the measured quality factor is the sum of a radiative rate that scales as α⁻² and a scattering rate that stays constant as α varies; if scattering loss actually depends on asymmetry, wavelength, or fabrication disorder tied to the pattern, the inferred reduction of scattering loss with shallower etching is not established.","fun_headline_variants_meta":{"raw":{"variants":["Low-contrast metasurfaces reach Q of 101,486","Shallow-etch silicon pairs hit ultrahigh Q of 101,486","Record Q: 101,486 in low-contrast BIC metasurfaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001005,"raw_usage":{"total_tokens":4242,"prompt_tokens":932,"completion_tokens":3310,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":548,"completion_tokens_details":{"reasoning_tokens":3244}},"tokens_in":548,"tokens_out":3310,"duration_ms":21731,"temperature":1.0,"reasoning_tokens":3244,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:31:55.530821+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fabricate metasurfaces with the same etching depth but deliberately varied sidewall roughness (for example, by changing the Bosch process passivation time), and measure Q at fixed small asymmetry; if Q does not change with roughness, or if the fitted Q_scat varies systematically with α, the paper's scattering-loss conclusion would be falsified.","supporting_citations":[],"review_version":1}