{"id":"febea201-8754-4b02-a682-690f9d4295cc","arxiv_id":"2506.00850","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Metallized silicon nitride membranes on a shared substrate are individually tunable, strongly coupled, and parametrically drivable, with coupled Arnold tongues observed in the hybridized regime.","lead":"A four-membrane silicon nitride chip is shown to act as a network of tunable, strongly coupled parametric oscillators, with each membrane individually addressable by voltage and read out with a single laser. The work is a step toward mechanical networks for analog computing and for studying collective nonlinear dynamics.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim of overlapping symmetric and antisymmetric KPO tongues in Fig. 4(ii) is not directly evidenced: the single-membrane readout cannot distinguish these states, and Appendix F's linear-mode symmetry check does not cover the nonlinear parametric states.","rationale":"Agree with the reader's weakest-assumption analysis. The core uncertainty is whether the measured single-membrane response in the hybridized regime uniquely identifies the symmetric and antisymmetric KPO states invoked to explain the overlapping Arnold tongues. The linear avoided-crossing data in Fig. 3 and the eigenmode symmetry check in Appendix F are strong but only constrain the linear normal modes. The nonlinear parametric states in the overlap region are inferred from the two-mode model and the theory of Refs. [41,43]. Because a measurement of x1 alone is invariant under exchanging the two membranes' phases up to a common sign, it cannot distinguish an in-phase from an anti-phase state; the only symmetry information available comes from the frequency ordering, which is a linear property and does not fully determine the nonlinear state. Alternative explanations for the observed amplitude jumps, such as optical readout nonlinearity or imperfect matching of parametric drive strengths, are not excluded by the present data. The proposed two-membrane readout test directly measures the relative phase and would settle whether the tongue assignment is correct. The platform claims—high Q, individual tuning, avoided crossings, and tunable parametric response—are well supported, so a conditional acceptance remains appropriate. No change to the reader's verdict is needed; the paper's central KPO-network claim should be considered contingent on the additional verification.","tokens_in":15843,"tokens_out":8905,"duration_ms":92220,"concrete_test":"Repeat the frequency sweep of Fig. 4(a)(ii) (U1 = 34.1 V, λ/λth = 3.1) while simultaneously measuring the displacement of membrane 2, e.g., with a second interferometer or a wide-field stroboscopic imaging setup. At each drive frequency, compute the relative phase Δφ = arg(x1)−arg(x2) and the amplitude ratio |x1|/|x2|. If the two overlapping tongues correspond to symmetric and antisymmetric KPO states, the higher-frequency tongue should show Δφ ≈ 0 and |x1|/|x2| ≈ 1, and the lower-frequency tongue should show Δφ ≈ π with |x1|/|x2| ≈ 1. If instead both tongues show the same relative phase, or the ratio deviates strongly from 1, the assignment in Fig. 4(ii) is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"At the avoided crossing, the symmetric and antisymmetric normal modes both have equal amplitude on membrane 1, and the parametric phase of x1 is equally likely to be 0 or π in either state. A measurement of x1 alone therefore cannot tell whether the network occupies the in-phase (x1≈x2) or anti-phase (x1≈−x2) KPO state. The two observed jumps in Fig. 4(a)(ii) and the overlapping tongue pattern in Fig. 4(b)(ii) are interpreted as a transition from the symmetric to the antisymmetric KPO state using the two-mode model of Eq. (1) and the theory of Refs. [41,43]. However, no direct observable—neither the relative phase x1−x2 nor the motion of membrane 2—is recorded in this regime. Appendix F only verifies the linear eigenmode symmetry with a wide-field stroboscopic interferometer; it does not verify that the nonlinear parametric states found in the overlap region have the assumed symmetry. Consequently, the central claim of 'strongly hybridized KPOs' and 'overlapping Arnold tongues' is underdetermined by the presented data, although it is consistent with theory. This is load-bearing because the paper's novelty relative to previous electrical-resonator work rests on demonstrating these coupled KPO states in a high-Q mechanical platform.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports an electromechanical platform based on four metallized silicon nitride membranes on a common chip, capacitively actuated and read out with a single laser interferometer. The authors characterize the linear response of individual membranes (Q1 ≈ 1.6×10^4, Q2 ≈ 5.2×10^3, Q3 ≈ 4.1×10^3), demonstrate electrostatic frequency tuning exceeding 2 kHz, and measure parametric Arnold tongues for single membranes. They then measure avoided crossings between membranes 1–2 and 1–3, extracting couplings J12/2π = (3.34 ± 0.03) kHz and J13/2π = (2.77 ± 0.03) kHz with normal-mode splittings comparable to the mechanical linewidth. Parametrically driving pairs of membranes, they observe tongue patterns that change as the membranes are tuned into and out of resonance, and interpret the near-resonant data as overlapping symmetric and antisymmetric KPO states. The claimed central result is a scalable, tunable high-Q mechanical platform for strongly coupled KPO networks.","tokens_in":16134,"tokens_out":9429,"duration_ms":90918,"significance":"If the coupled-state interpretation is accepted, the paper is a valuable step: it brings KPO network physics, previously demonstrated mainly in lower-Q electrical circuits, into a high-Q mechanical platform with individual tunability and substrate-mediated coupling. The paper has several genuine strengths: coupling values are extracted from avoided-crossing fits with quoted uncertainties; the parametric tongue outlines are compared with predictions using independently measured damping; and the sign of the linear coupling term is checked with a wide-field stroboscopic interferometer. The authors are also transparent about key limitations, including uncalibrated displacement amplitudes and drive-mismatch artifacts. However, the headline claim of strongly hybridized KPOs with overlapping symmetric and antisymmetric Arnold tongues is underdetermined by the presented measurements, because only one membrane is read out in the coupled nonlinear regime and the symmetry of the nonlinear parametric states is not directly verified.","major_comments":[{"comment":"The assignment of the second jump in Fig. 4(a)(ii) and the overlapping tongue pattern in Fig. 4(b)(ii) to a transition from the symmetric to the antisymmetric KPO state is not directly supported by the measurement. At the avoided crossing, the symmetric (x1+x2) and antisymmetric (x1−x2) normal modes both produce the same displacement amplitude on membrane 1, and the phase of x1 relative to the parametric drive does not distinguish the two spatial symmetries; distinguishing them requires the relative phase x1−x2 or the motion of membrane 2. Appendix F only verifies the linear eigenmode symmetry with a wide-field interferometer, not the symmetry of the nonlinear parametric states in the overlap region. Because the paper's novelty over previous electrical-resonator work rests on demonstrating strongly hybridized KPO states, this underdetermination is load-bearing. I recommend either measuring the second membrane simultaneously (or the relative phase), or explicitly presenting the state assignment as an inference from Eq. (1) and Refs. [41,43] rather than as a demonstrated observation.","section":"Coupled membranes, Fig. 4, Appendix F"},{"comment":"The sign convention for the normal-mode branches is inconsistent. In Eq. (4), Ω+ is the larger eigenfrequency, and Appendix F defines the splitting as Δij = Ω+ − Ω− and identifies the higher-frequency branch with x1+x2. The main text, however, defines Δ12 ≡ Ω− − Ω+ and writes Ω± = Ω1 ∓ J12^2/(2Ω1), which reverses the branch assignment. The fitted magnitude of the gap is unaffected, but the inconsistent notation makes it impossible to track which branch is called symmetric, which matters for the interpretation of the parametric tongues. Please correct the signs and define the branch assignment unambiguously.","section":"Eq. (4) and Appendix F"}],"minor_comments":[{"comment":"There are several typographical errors that should be corrected: 'solide lines' in the Fig. 2 caption, 'reduced laser absorption' in Appendix A, and 'orders of magnitudes' in the Summary and outlook.","section":"Fig. 2 caption and throughout"},{"comment":"The text 'symmetric coupling Jij = Jji0' contains a stray '0' and should read 'Jij = Jji'.","section":"Appendix F"},{"comment":"The abstract states that 'we read out multiple mechanical resonators using a single laser interferometer,' but in the coupled measurements only membrane 1 is read out and the laser is moved between membranes for individual characterization; please clarify that the readout is sequential unless the wide-field interferometer is used.","section":"Abstract"},{"comment":"The coupling enters as Jij^2 rather than as a linear coupling constant; since Jij has units of frequency and the resulting normal-mode splitting is J^2/Ω, this unconventional form should be stated explicitly where the coupling is introduced to avoid confusion with standard coupled-oscillator notation.","section":"Eq. (1)"},{"comment":"The inset mentioned in Fig. 4(i)(b) appears to show modeled x1 and x2 waveforms rather than measured data; the caption should state that the inset is an illustration.","section":"Fig. 4 caption"},{"comment":"The claim of a 'continuous crossover' between regimes is supported by only three voltage points along the avoided crossing; the wording slightly exceeds the displayed evidence, although the qualitative trend is clear.","section":"Summary and outlook"}],"recommendation":"major_revision","confidential_remarks":"The paper is well written and honest about limitations. The main risk is that the central claim of overlapping symmetric and antisymmetric KPO tongues rests on the two-mode model and on theory from the authors' group (Refs. [41,43]); a direct two-membrane measurement in the nonlinear regime would remove the ambiguity. The manuscript is within scope for the journal as a platform paper, but the interpretation should be either strengthened experimentally or explicitly softened."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear Colleague,\n\nThe short version: this is a solid platform paper that does what it claims. Four SiN membranes on one chip, individually tunable by a few kHz, Q up to ~1.6e4, substrate-mediated nearest-neighbor coupling at or above the linewidth, and parametric driving that shows Arnold tongues, including an overlapping-tongue pattern when the membranes are tuned into an avoided crossing. That combination is new for mechanical resonators, and the measurements are careful: avoided-crossing fits give J12/2pi = 3.34 +/- 0.03 kHz and J13/2pi = 2.77 +/- 0.03 kHz, with splittings comparable to the linewidth, and the linear mode symmetry is checked with a stroboscopic interferometer.\n\nThe main soft spot, which the authors partly acknowledge, is that the hybridized KPO state identification in Fig. 4 rests on a single-membrane readout. The symmetric and antisymmetric normal modes both have equal amplitude at membrane 1, so amplitude and phase of x1 alone cannot tell you which state the network is in. The authors infer the state from frequency ordering and from their own theory [41,43]. That is a plausible inference, since the linear higher mode is verified symmetric, but it is not a direct measurement. The paper would be stronger with a second readout or an explicit measurement of the relative phase. This is not fatal: the platform claim doesn't depend on that specific state assignment, and the tongue pattern itself is a reproducible experimental observation. Still, the phrase 'strongly hybridized KPOs' is a little ahead of the evidence.\n\nMinor issues: amplitudes are normalized, not calibrated, which limits quantitative claims about the Duffing nonlinearity or absolute amplitudes. The drive strength matching in Appendix H is described qualitatively. None of these are deal-breakers for a platform demonstration.\n\nThe citation pattern looks fine. It cites the relevant electrical-resonator KPO experiments and the theory it leans on, and it doesn't oversell its novelty. It does lean fairly heavily on [41,43] for the interpretation, but that is standard when theory is used to interpret experimental data.\n\nWho it's for: people working in nanomechanics, parametric oscillators, and analog computing with mechanical networks. It deserves a serious referee. I'd take it for review with the expectation that the hybridized-state identification be discussed honestly, possibly softened.\n\nBest,","headline":"A solid platform paper: tunable, coupled SiN membranes with parametric response, but the hybridized-state assignment is inferred from one membrane's readout and theory.","tokens_in":16639,"tokens_out":2777,"would_cite":true,"duration_ms":29247,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["85.85.+j","05.45.-a"],"model":"deepseek-v4-flash","headline":"Four silicon nitride membranes on a single chip, individually tuned and read out by one laser, form strongly coupled Kerr parametric oscillators whose collective parametric states appear as overlapping Arnold tongues.","keywords":["silicon nitride membranes","Kerr parametric oscillators","coupled mechanical resonators","parametric driving","Arnold tongues","avoided crossing","capacitive tuning","analog computing"],"falsifier":"Measure the displacement of both membranes in the hybridized parametric regime (e.g., with wide-field stroboscopic imaging). If the two membranes do not oscillate in the symmetric ($x_1 + x_2$) or antisymmetric ($x_1 - x_2$) combination predicted, or if the measured Arnold tongues do not match the two-mode model's overlap pattern, the central claim is refuted.","tokens_in":15666,"feed_emoji":"⚙️","tokens_out":9257,"duration_ms":76458,"temperature":0.7,"pith_summary":"This paper reports a chip-based electromechanical platform in which four silicon nitride membranes act as coupled, individually controllable nonlinear resonators. Each membrane can be tuned in frequency by several kilohertz with a DC voltage, shows quality factors around $10^4$, and is coupled to its neighbors strongly enough that the normal-mode splitting is comparable to, or larger than, the mechanical linewidth. The authors demonstrate a continuous crossover from a detuned regime of mostly single-membrane modes to a hybridized normal-mode regime, where parametric driving yields overlapping Arnold tongues attributed to symmetric and antisymmetric collective states. If correct, the device is a scalable, controllable setting for coupled nonlinear resonator physics, with applications to analog computing and the study of collective phenomena.","feed_headline":"On-chip membranes become strongly coupled parametric oscillators","feed_subtitle":"Each membrane tunes individually; coupling is strong enough to hybridize pairs into collective states.","key_machinery":"The central object is the coupled, parametrically driven membrane network of Eq. (1): each membrane is a Duffing (Kerr) oscillator whose spring constant is modulated at twice its resonance frequency, damped, and linearly coupled to its neighbors through the shared substrate. Three mechanisms carry the demonstration. First, capacitive voltage tuning shifts each membrane's frequency quadratically, $\\delta\\Omega_i \\propto -U_i^2$, enabling individual in-situ control. Second, the two-mode eigenvalue problem gives the avoided-crossing frequencies and a normal-mode splitting $\\Delta_{ij} = J_{ij}^2/\\Omega_i$ at resonance, so the coupling strength is read directly from the gap. Third, parametric driving produces Arnold tongues—regions in drive-strength versus frequency space where the oscillator locks into oscillating states—whose boundary follows the threshold condition $\\lambda_{\\rm th} = 2/Q_i$. The paper combines these to interpret the measured response of one membrane as the phase diagram of the coupled network.","core_discovery":"On its own terms, the paper demonstrates that metallized silicon nitride membranes on a common silicon substrate, driven capacitively at twice their resonance frequency, form a network of Kerr parametric oscillators with individually tunable frequencies and nearest-neighbor coupling through the substrate. The key quantitative results are a frequency tunability exceeding 2 kHz, a quality factor up to about $1.6\\times10^4$, a coupling $J_{12}/2\\pi \\approx 3.34$ kHz corresponding to a normal-mode splitting $\\Delta_{12}/2\\pi \\approx 55$ Hz that is on the order of the combined linewidth, and the observation of avoided crossings that allow the system to be swept from a detuned resonator regime into a hybridized normal-mode regime. In the hybridized regime, parametric driving produces a pattern of overlapping Arnold tongues attributed to symmetric and antisymmetric collective parametric states, with jumps between states visible in the amplitude of a single read-out membrane. The paper claims this is the first high-quality-factor mechanical system in which strong coupling and in-situ tuning are combined in a scalable architecture, giving access to physics previously studied only in lower-quality electrical resonator networks.","pith_inferences":["We infer that a direct measurement of the relative phase $x_1 - x_2$ in the driven hybridized regime would harden the central interpretation; the paper identifies symmetric and antisymmetric states from single-membrane data combined with theory.","We infer that the platform, if it scales to six or nine membranes as argued, is a plausible testbed for KPO-based Ising machines and Boltzmann sampling, but multi-membrane imaging would be needed to resolve the state of each membrane.","We infer that applying soft clamping to raise $Q$ by orders of magnitude could push the system into a regime where thermal switching between KPO phase states is slow enough to observe directly.","We infer that since coupling runs through the shared substrate, the connectivity graph is fixed by geometry; engineering the substrate could open paths to non-reciprocal or programmable couplings."],"forward_implications":["The same platform can be extended to larger membrane arrays (six or nine) without major technical changes, yielding programmable networks of coupled KPOs.","Because the normal-mode splitting matches the linewidth, the system can enter the strongly coupled KPO regime, which theoretical work connects to ghost states and mixed-symmetry states previously seen only in electrical resonator networks.","With the detuning between membranes controllable in situ, the network can implement asymmetric Ising models relevant to neural-network emulation.","The slow ringdown times (tens of milliseconds) make the platform well suited for studying activated fluctuations and interstate transitions in multi-stable systems.","A single-laser interferometric readout of one membrane suffices to infer the state of the whole coupled network, simplifying the measurement of future networks."],"supporting_citations":[{"why":"Predicts overlapping Arnold tongues and state jumps for strongly coupled KPOs, the pattern the hybridized-regime data are interpreted with.","marker":"[41]"},{"why":"Supplies the analytical and numerical phase diagram of coupled Kerr parametric oscillators used to identify which parametric state the network occupies.","marker":"[43]"},{"why":"Provides the parametric threshold condition whose boundary defines the measured Arnold tongue.","marker":"[18]"},{"why":"Demonstrates capacitive actuation and tuning of mechanical membranes, the scheme used for individual control.","marker":"[49]"},{"why":"Explains mechanical dissipation via substrate-mode coupling, the mechanism assumed to couple the membranes.","marker":"[50]"},{"why":"Earlier experiment on two coupled KPOs in the weak-coupling regime that this platform extends.","marker":"[40]"},{"why":"Shows strongly coupled KPO physics (biased Ising model) in electrical resonators, the comparison baseline for this mechanical system.","marker":"[32]"},{"why":"Demonstrates a small strongly coupled KPO network acting as a Boltzmann machine, motivating the strongly coupled regime.","marker":"[34]"}],"fun_headline_variants":["Tunable silicon nitride membranes form strongly coupled network","Parametric membranes couple via substrate into hybridized states","Membrane pair shows avoided crossings and collective Arnold tongues","On-chip membrane network achieves strong coupling with individual tuning","Silicon nitride membranes: parametric oscillators with tunable coupling"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The interpretation of the hybridized parametric response as overlapping symmetric and antisymmetric Kerr-oscillator states assumes that the two-membrane model with a fixed symmetric coupling describes the dynamics, and that measuring only one membrane's amplitude and phase is enough to identify which collective state the network occupies.","fun_headline_variants_meta":{"raw":{"variants":["Tunable silicon nitride membranes form strongly coupled network","Parametric membranes couple via substrate into hybridized states","Membrane pair shows avoided crossings and collective Arnold tongues","On-chip membrane network achieves strong coupling with individual tuning","Silicon nitride membranes: parametric oscillators with tunable coupling"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000173,"raw_usage":{"total_tokens":1247,"prompt_tokens":883,"completion_tokens":364,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":499,"completion_tokens_details":{"reasoning_tokens":286}},"tokens_in":499,"tokens_out":364,"duration_ms":3842,"temperature":1.0,"reasoning_tokens":286,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T11:56:47.478863+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the displacement of both membranes in the hybridized parametric regime (e.g., with wide-field stroboscopic imaging). If the two membranes do not oscillate in the symmetric ($x_1 + x_2$) or antisymmetric ($x_1 - x_2$) combination predicted, or if the measured Arnold tongues do not match the two-mode model's overlap pattern, the central claim is refuted.","supporting_citations":[{"cited_title":"Margiani, J","cited_arxiv_id":null,"evidence_quote":"Predicts overlapping Arnold tongues and state jumps for strongly coupled KPOs, the pattern the hybridized-regime data are interpreted with."},{"cited_title":"Proliferation of unstable states and their impact on stochastic out-of-equilibrium dynamics","cited_arxiv_id":"2307.13718","evidence_quote":"Supplies the analytical and numerical phase diagram of coupled Kerr parametric oscillators used to identify which parametric state the network occupies."},{"cited_title":"Dykman, Fluctuating Nonlinear Oscillators (Oxford University Press, 2012)","cited_arxiv_id":null,"evidence_quote":"Provides the parametric threshold condition whose boundary defines the measured Arnold tongue."},{"cited_title":"Bagci, a","cited_arxiv_id":null,"evidence_quote":"Demonstrates capacitive actuation and tuning of mechanical membranes, the scheme used for individual control."},{"cited_title":"Puglia, R","cited_arxiv_id":null,"evidence_quote":"Explains mechanical dissipation via substrate-mode coupling, the mechanism assumed to couple the membranes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Earlier experiment on two coupled KPOs in the weak-coupling regime that this platform extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows strongly coupled KPO physics (biased Ising model) in electrical resonators, the comparison baseline for this mechanical system."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates a small strongly coupled KPO network acting as a Boltzmann machine, motivating the strongly coupled regime."}],"review_version":1}