{"id":"cb64504c-5070-40e8-929b-4be5ac066609","arxiv_id":"2607.26199","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"4H-SiC microdisks achieve sideband-resolved optomechanics (2πfm/κ ≈ 7.6) with optical Q > 10^6, mechanical Q_m = 15,100, and the first 4H-SiC OMIT.","lead":"A team built tiny 4H-silicon-carbide disk resonators that trap both light and mechanical vibration, reaching the sideband-resolved regime for the first time in this material. They also engineered the pedestal so the disks lose little vibrational energy without fragile deep etching, and observed optomechanically induced transparency.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Anchor-loss suppression claim rests on an inferred undercut axis and FEM simulation that the measured, surface-limited Qm cannot independently validate.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: the non-monotonic Qm data are mapped onto a simulated Qanchor curve using inferred undercut widths, and the FEM simulation depends on literature elastic constants and an assumed mechanism. I agree that this is the weakest link. The concern does not overturn the measured sideband-resolved and OMIT claims, which are internally consistent (three independent estimates of cooperativity agree). It does, however, mean the engineering claim of interference-engineered anchor-loss suppression and the yield advantage remains conditional on unverified geometry and simulation. Since the reader already issued CONDITIONAL, my read does not change the verdict; it sharpens the specific condition: direct undercut verification and yield statistics are required before the interference-engineered anchor-loss mechanism can be taken as established.","tokens_in":14286,"tokens_out":10143,"duration_ms":104804,"concrete_test":"Measure the undercut width/pedestal radius directly (e.g., SEM cross-sections or optical profilometry of sibling devices or after non-destructive imaging) for the same BOE etch steps used in Fig. 5(c), and compare the measured Qm peak position to the simulated Qanchor curve. Also report yield statistics (intact device counts for moderate vs aggressive undercut) from a fixed fabrication batch. If the Qm peak occurs at an undercut ratio differing from ~60% by more than the simulated tolerance, or if moderate-undercut yield is not substantially above the reported <20% aggressive-undercut yield, the interference-engineered anchor-loss suppression claim should be de-emphasized.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The engineering centerpiece—that a local minimum in anchor loss at undercut ratio ~60% enables high Qm and improved yield—is not independently established by the reported measurements. The measured Qm (≈15.1–16.4k) is an order of magnitude below the simulated Qanchor peak (~150k in Fig. 4b), and SI Sec. 2 reports that a deeply undercut device (undercut >85%) has essentially the same Qm (16.4k). Thus the measured Qm is dominated by a non-anchor dissipation channel (surface loss), and the Qm-vs-undercut data in Fig. 5(c) only show how anchor loss modulates a surface-loss-limited baseline. The horizontal axis of Fig. 5(c) is built from timed BOE etches (1.45→1.72 μm) rather than direct undercut measurement, so the alignment of the data with the simulated Qanchor peak is inferred, not verified. If the true undercut differs from the etch-time estimate by more than the ~50 nm stated tolerance of the peak, or if undercut-dependent surface losses shift the observed maximum, the 'interference-engineered anchor-loss suppression' mechanism and the claimed yield advantage are not supported by the data. This does not threaten the measured sideband-resolved and OMIT observations, which stand on direct spectra, but it is the weakest load-bearing link in the paper's central engineering narrative.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports 4H-SiC microdisk optomechanical resonators that operate in the sideband-resolved regime. The devices combine a compact radius (2.5–2.7 μm) with a designed undercut ratio around 60%, near a simulated local maximum of the anchor-loss-limited mechanical quality factor. Measured intrinsic optical Q exceeds 1×10^6, the fundamental radial breathing mode at 1.587 GHz has Q_m = 15,100 (f_m·Q_m ≈ 24 THz), and the sideband-resolution factor exceeds 7. The authors observe optomechanically induced transparency (OMIT) and extract a cooperativity C ≈ 0.88 at 0.85 mW on-chip power, a single-photon cooperativity C_0 ≈ 3.0×10^-4, and a vacuum coupling rate g_0/2π ≈ 40 kHz, consistent with their prior phonon-lasing results. They claim the first sideband-resolved 4H-SiC optomechanical resonator and the first OMIT observation in integrated 4H-SiC.","tokens_in":14587,"tokens_out":7026,"duration_ms":71561,"significance":"The optical and OMIT measurements are direct and internally consistent: the two independent OMIT extraction routes (Eqs. S2 and S4) agree with each other, and the extracted C is corroborated by the theoretical estimate. The sideband-resolution factor, the f_m·Q_m product, and the room-temperature demonstration place 4H-SiC microdisks among competitive integrated optomechanical platforms. The central engineering claim, however, is weaker than the abstract suggests: the measured Q_m is an order of magnitude below the simulated anchor-loss peak, and the authors themselves state in SI Sec. 2 that surface-related losses likely dominate. The 'interference-engineered anchor-loss suppression' mechanism therefore is not established by the reported measurements.","major_comments":[{"comment":"The central anchor-loss narrative is not supported by the reported measurements. The simulated Q_anchor peak is ~150k in Fig. 4(a,b), whereas measured Q_m is 15.1–16.4k. SI Sec. 2 reports that a deeply undercut (>85%) device has Q_m = 16.4k, essentially the same; the text itself concludes that 'other dissipation mechanisms, such as surface-related losses, are likely to dominate'. Therefore Fig. 5(c) shows only how anchor loss modulates a surface-loss-limited baseline; it cannot validate the magnitude or the lateral-interference origin of the Q_anchor peak. To substantiate the title/abstract claim, the authors should either isolate the anchor-loss contribution (e.g., by comparing devices with different surface treatments or by extracting Q_anchor from the difference) or substantially soften the claim.","section":"§2.5, SI Sec. 2, Fig. 4"},{"comment":"The experimental undercut axis is inferred from timed BOE etches (1.45→1.72 μm) and optical-microscope monitoring rather than from direct measurement of each device's undercut width. The mapping of the three devices' maximum Q_m onto the simulated Q_anchor peak of Fig. 4(a) depends on this inferred axis, with a reported pedestal-radius tolerance of ±50 nm. If the true undercut differs by more than ~50 nm, or if undercut-dependent surface losses shift the measured maximum, the non-monotonic Q_m data cannot be attributed to anchor loss. Direct imaging (e.g., SEM of cross-sections or cleaved devices) and an uncertainty statement for the undercut widths are needed.","section":"§2.5, Fig. 5(c)"},{"comment":"Fig. 4(c) varies only the oxide thickness and shows that the Q_anchor peak is largely insensitive to it, which rules out a vertical quarter-wave condition. It does not, however, test the lateral-interference hypothesis. The interpretation that the peak arises from destructive interference of laterally propagating elastic waves is an assumption supported only by the FEM model. This is acceptable as a design hypothesis, but the phrase 'interference-engineered' in the title and abstract implies experimental confirmation that is not yet present. The authors should either provide a direct test of the lateral-wave mechanism or rephrase the claim as a simulation-guided design strategy.","section":"§2.4, Fig. 4(c)"}],"minor_comments":[{"comment":"The terminology is inconsistent: the abstract and text refer to a 'local minimum in anchor loss', while Fig. 4(a) shows a local maximum in Q_anchor. Please use a consistent description, e.g., 'local maximum of Q_anchor' or 'local minimum of the anchor-loss rate'.","section":"Abstract, §2.4"},{"comment":"The footnote indicates that some C_0 and Q_m values were measured under cryogenic or vacuum conditions. The text should explicitly caution readers that a direct comparison of C_0 across different measurement conditions is not meaningful, especially when benchmarking the room-temperature ambient value reported here.","section":"Table 1"},{"comment":"The cooperativity is extracted from only four optical powers, with no error bars or fit residuals shown. Please include uncertainties in the extracted C values and, if possible, additional power points to demonstrate the linear C-versus-P_in scaling.","section":"Fig. 6(d)"},{"comment":"Fig. 1(b) assumes a device-layer thickness of 450 nm, but the fabrication process results in a final SiC thickness of approximately 470 nm. Please clarify whether the simulations use 450 or 470 nm, and comment on the sensitivity of Q_rad and RBM frequency to this 20 nm difference.","section":"§2.1, §2.2"},{"comment":"The two-photon detuning Ω is defined, but the sign convention for the probe-pump detuning is not explicitly given. Defining the sign convention and the exact relation Ω = ω_probe − ω_pump − ω_m would help readers reproduce the fit.","section":"SI Sec. 3, Eq. S1"},{"comment":"The agreement between the OMIT-extracted C and C_theory uses g_0 ≈ 40 kHz from Ref. [5], a prior paper by the same group. This is a consistency check rather than an independent confirmation. Please state explicitly that the OMIT fits themselves (Eqs. S2 and S4) do not use g_0, and provide the uncertainty in g_0 from Ref. [5] so the reader can assess the agreement.","section":"§3, SI Table S1"}],"recommendation":"major_revision","confidential_remarks":"The direct optical, mechanical, and OMIT measurements appear sound and would likely survive independent scrutiny. The main weakness is the mismatch between the headline 'interference-engineered anchor-loss suppression' and the authors' own admission that the measured Q_m is surface-loss limited; the measured data do not independently validate the anchor-loss mechanism. A major revision that either provides direct evidence for the anchor-loss contribution or softens the engineering claim would be appropriate. The scope of the journal is suitable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead this one. It reports something genuinely new: the first sideband-resolved 4H-SiC microdisk optomechanical resonator, with a sideband ratio around 7.6, and the first OMIT in integrated 4H-SiC. The optical Q values are direct fits, the mechanical spectrum shows a clean thermal peak at 1.586 GHz with Qm = 15,100, and the OMIT analysis is internally consistent — three independent routes to cooperativity land at 0.88–0.91, with single-photon cooperativity ~3e-4, comparable to other platforms. That part is solid and deservedly citable.\n\nThe soft spot is the engineering claim in the title. The measured Qm (~15k) sits an order of magnitude below the simulated Q_anchor peak (~150k), and the paper's own SI shows a deeply undercut device (undercut >85%) with essentially the same Qm (~16.4k). So the observed Qm is surface-loss-limited, not anchor-loss-limited. The non-monotonic Qm-vs-undercut data in Fig. 5c do show anchor loss modulating a surface-loss baseline, which is a reasonable interpretation; but the horizontal axis is built from timed BOE etches rather than direct undercut measurement, and the FEM anchor-loss peak cannot be validated by data that sits far below it. The paper honestly concedes surface-loss dominance in the SI, but the main text still sells anchor-loss suppression as the enabling mechanism. The weaker, still useful claim — that moderate undercut (~60%) gives the same Qm as aggressive undercut (>85%) with better robustness — is supported and is worth making. The overreach is attributing the absolute Qm to anchor-loss suppression.\n\nThe circularity in using their own prior g0 for the theoretical C is minor; it is labeled, and the two experimental extraction routes agree independently. I would not reject over that.\n\nThis paper is for people working on SiC, silicon, or microdisk optomechanics, and anyone wanting to reproduce the yield improvement. It deserves a serious referee, but the referee should push for direct undercut geometry (e.g., SEM cross-sections) and yield statistics before the anchor-loss mechanism is accepted. My recommendation: send to peer review, conditional on the authors toning down the mechanism claim or adding the missing geometry data.","headline":"The sideband-resolved and OMIT results are real and worth citing; the anchor-loss-engineering story is not nailed down by the data.","tokens_in":15207,"tokens_out":2060,"would_cite":true,"duration_ms":21002,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"First sideband-resolved optomechanical resonators on 4H-SiC are demonstrated, with optical Q above one million and mechanical Q up to 15,100 at room temperature.","keywords":["4H-silicon carbide","cavity optomechanics","sideband-resolved","optomechanically induced transparency","anchor loss","microdisk resonator","mechanical quality factor","integrated photonics"],"falsifier":"Measure the actual undercut width of devices after each etch step using cross-sectional electron microscopy and correlate it with the measured Q_m values; if the Q_m peak occurs at an undercut ratio substantially different from the simulated ~60%, or if vacuum measurements show that surface loss rather than anchor loss sets the room-temperature Q_m, the interference-engineered anchor-loss explanation is undermined.","tokens_in":14090,"feed_emoji":"🔬","tokens_out":2681,"duration_ms":29263,"temperature":0.7,"pith_summary":"The paper reports the first sideband-resolved cavity optomechanical resonators built on the 4H-SiC platform. The devices combine million-level intrinsic optical quality factors, gigahertz-frequency mechanical modes, and room-temperature mechanical quality factors up to 15,100, giving a sideband-resolution factor above seven. The central engineering insight is that anchor loss has a local minimum at a moderate undercut ratio, so high mechanical quality can be achieved without the fragile, aggressive undercutting that usually limits yield. The paper also reports the first observation of optomechanically induced transparency in integrated 4H-SiC, confirming coherent photon–phonon coupling in this material. If correct, this establishes 4H-SiC as a practical, scalable platform for integrated cavity optomechanics.","feed_headline":"4H-SiC microdisks clear the sideband-resolved bar","feed_subtitle":"Million-level optical Q plus interference-engineered anchor loss give fm·Qm ~24 THz and the first OMIT in integrated 4H-SiC.","key_machinery":"The central object is a 4H-SiC on insulator microdisk supporting a whispering-gallery optical mode and a fundamental radial breathing mechanical mode. The mechanism that carries the argument is interference-engineered anchor-loss suppression: finite-element simulations show a local maximum of the anchor-loss-limited mechanical quality factor (Q_anchor) at an undercut ratio around 60%, attributed to destructive interference of laterally propagating elastic waves. This local maximum is what lets the device keep a large pedestal and high mechanical Q simultaneously, and it is probed experimentally by measuring Q_m across multiple timed undercut steps.","core_discovery":"On the paper's own terms, the discovery is that pedestal-supported 4H-SiC microdisks can operate in the sideband-resolved regime (mechanical frequency exceeding optical cavity linewidth) by exploiting a local minimum in anchor loss versus undercut ratio. Simulated and measured mechanical quality factors show a pronounced peak near 60% undercut, where anchor loss drops by over an order of magnitude. This allows the pedestal to remain relatively large and mechanically robust while achieving Q_m up to 15,100 at 1.587 GHz, an f_m·Q_m product of about 24 THz. The same devices exhibit optical Q_i above 10^6, and a coherent optomechanical interaction is confirmed through a transparency window whose","pith_inferences":["The same undercut-engineering principle may transfer to other pedestal-supported microdisk platforms (e.g., silicon or diamond), where anchor loss also shows non-monotonic behavior, potentially improving yield without sacrificing mechanical Q.","If surface losses indeed cap Q_m around 1.5–1.6×10^4, then further gains in mechanical Q would require surface passivation or vacuum operation, not just deeper undercuts; the paper's data suggest this direction implicitly.","The measured g0 of about 40 kHz is modest compared to optomechanical crystals, so future work could focus on mode-shape engineering (e.g., slot or phononic shields) to raise single-photon cooperativity while preserving the fabrication-friendly anchor design.","A direct testable extension: varying the oxide thickness or disk thickness should shift the undercut ratio at which the Q_anchor peak occurs, allowing a predictive design rule for other wavelengths or mechanical frequencies."],"forward_implications":["Sideband-resolved operation unlocks coherent optomechanical effects in 4H-SiC, including ground-state cooling, optomechanically induced transparency, and coherent wavelength conversion, which were previously out of reach in this material.","The anchor-loss engineering strategy removes the need for extreme undercut ratios, so fabrication yield improves from below 20% to a regime where many devices survive; this makes the platform scalable to arrays and integrated photonic circuits.","The f_m·Q_m product of roughly 24 THz is among the highest reported for undercut microdisk optomechanical resonators, making the device competitive with optomechanical crystals for both classical and quantum applications.","With a Purcell factor estimated above 5000, the same microdisks can enhance emission from embedded 4H-SiC color centers, pointing toward hybrid quantum systems that couple photons, phonons, and spins in one device.","The demonstrated optomechanically induced transparency provides a standard, power-tunable tool for measuring optomechanical cooperativity, which can be used to benchmark future improvements in coupling."],"fun_headline_variants":["First sideband-resolved 4H-SiC optomechanical resonators","Interference-engineered anchors boost SiC optomechanics to sideband-resolved","SiC microdisks achieve sideband-resolved via anchor-loss suppression","First OMIT in integrated 4H-SiC microdisks","High-Q SiC optomechanics without aggressive undercut"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The central claim rests on the assumption that the observed non-monotonic mechanical quality factor versus undercut step is actually caused by the simulated anchor-loss peak, but the undercut widths are inferred from timed etches rather than measured directly, and the simulation depends on assumed elastic constants and on anchor loss dominating other dissipation at the peak.","fun_headline_variants_meta":{"raw":{"variants":["First sideband-resolved 4H-SiC optomechanical resonators","Interference-engineered anchors boost SiC optomechanics to sideband-resolved","SiC microdisks achieve sideband-resolved via anchor-loss suppression","First OMIT in integrated 4H-SiC microdisks","High-Q SiC optomechanics without aggressive undercut"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000392,"raw_usage":{"total_tokens":1927,"prompt_tokens":805,"completion_tokens":1122,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":549,"completion_tokens_details":{"reasoning_tokens":1037}},"tokens_in":549,"tokens_out":1122,"duration_ms":8244,"temperature":1.0,"reasoning_tokens":1037,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T00:31:07.268571+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the actual undercut width of devices after each etch step using cross-sectional electron microscopy and correlate it with the measured Q_m values; if the Q_m peak occurs at an undercut ratio substantially different from the simulated ~60%, or if vacuum measurements show that surface loss rather than anchor loss sets the room-temperature Q_m, the interference-engineered anchor-loss explanation is undermined.","supporting_citations":[],"review_version":1}