{"id":"be285b09-88e1-4a80-a947-4a397c9a56f6","arxiv_id":"2506.04433","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Buried-electrode lithium niobate resonators excite S0 and SH0 lateral bulk acoustic waves, including the first reported overtones of this device class, with measured figures of merit of 437 at 673 MHz.","lead":"This paper reports acoustic resonators built in a 100-nanometer lithium niobate film with electrodes buried inside the film, and shows the first higher-order overtone operation for this resonator type. The devices reach figures of merit of 437 at 673 MHz and 53 at 1.05 GHz, which matters for multi-frequency radio front-ends in 5G/6G phones.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Thickness-decoupling claim is under-supported: only a single 100 nm film is simulated and fabricated, etch depth is unverified, and the best measured fundamental sits off the simulated optimum; if dispersion depends on thickness or recess depth, the key advantage over LVRs collapses.","rationale":"The reader's weakest_assumption correctly identifies the load-bearing premise: thickness-independent dispersion is the architectural advantage claimed for LBAW over LVRs. I agree that the support is insufficient: the COMSOL dispersion curves are computed at a single thickness, the recess depth is stated but not metrologically verified, and no device with a different film thickness is measured. My review adds one concrete observation that strengthens the evidence-sufficiency concern: the best measured fundamental SH0/S0 devices are at wm/wp=0.075, whereas the simulated coupling optima are at wm/wp=0.1125 and 0.1, so the model's predictive power at the operating point is not demonstrated by the reported data. This does not invalidate the device demonstration: the measurement chain is standard, the mBVD fits are plausible, and the overtone frequencies are at least consistent with the claimed lateral overmode concept. The paper deserves a conditional verdict rather than rejection because the central flaw is a missing experimental/computational check, not an internal contradiction. I found no more load-bearing concern: the headline FoM values are fit-dependent but not inherently implausible, and the 'first overtone demonstration' novelty claim, while needing a broader prior-art search, does not affect the physical operation of the devices. Therefore the reader's CONDITIONAL verdict should stand unchanged pending the proposed thickness-sensitivity check.","tokens_in":6468,"tokens_out":9578,"duration_ms":101174,"concrete_test":"Re-run the Section III COMSOL dispersion sweep at t=50 nm and t=200 nm (wp=2 µm, electrode depth=80% of t, same wm/wp grid) and overlay fr and kt^2 onto Figs. 2-3. If the resonance-frequency curves at fixed wm/wp shift by more than ~1% from the t=100 nm case, or if the kt^2 maxima move to different wm/wp values, the 'not sensitive to thickness' premise fails. The corresponding experimental check would be to co-fabricate the same wm/wp=1.1 SH0-overtone layout on 100 nm and 200 nm X-cut LN and compare fr; a thickness-independent fr would settle the claim, while a shift would refute it.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's core architectural claim is that recessing electrodes to 80% of the film thickness decouples the S0/SH0 dispersion relation from the piezoelectric thickness (Section III: 'the performed optimization is not sensitive to the thickness of the piezoelectric layer'), so frequency can be set purely by the lateral wm/wp ratio. This is the premise that distinguishes LBAW from LVRs and underwrites the on-chip multi-frequency scaling claim. The support is a COMSOL sweep at one thickness (t=100 nm) with no experimental thickness series and no etch-depth metrology. The premise is not self-evident: with 80 nm metal recessed into a 100 nm film, the composite unit cell contains a 20 nm residual LN plate beneath the electrodes and a full-thickness LN gap; a change in t at fixed wp=2 µm alters the vertical aspect ratio, the mass loading per unit area, and the residual-plate stiffness, and any finite t/p dispersion correction is not quantified. An additional sign that the model is not fully predictive is that the best measured fundamental devices are at wm/wp=0.075, outside the simulated optima (0.1 for S0, 0.1125 for SH0). If thickness or etch-depth variation shifts the dispersion curves materially, the device still resonates, but the central advantage—lithographic tuning independent of film thickness—is unsupported. This is an evidence-sufficiency concern, not an observed contradiction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the design, fabrication, and characterization of laterally excited bulk acoustic wave (LBAW) resonators on X-cut LiNbO3 with interdigitated electrodes recessed into the piezoelectric film. The authors demonstrate fundamental S0 and SH0 modes and, for the first time for this resonator type, higher-order overtones (LOBAW). The headline results are mBVD-fitted figures of merit of 437 at 673 MHz for the SH0 fundamental and 53 at 1.05 GHz for the SH0 overtone. The central architectural claim is that recessed electrodes decouple the dispersion relation from film thickness, so that frequency can be tuned lithographically via the metal-to-piezoelectric width ratio wm/wp, enabling on-chip multi-frequency scaling. Design optimization is performed with COMSOL Multiphysics, and the fabricated devices are measured in vacuum with a 2-port VNA and fitted to a single-tone mBVD model.","tokens_in":6677,"tokens_out":3855,"duration_ms":36198,"significance":"If the thickness-decoupling claim holds, the proposed LBAW architecture would offer a meaningful advantage over conventional Lamb-wave resonators, whose dispersion depends sensitively on film thickness, and would support the vision of single-chip multi-frequency RF front-ends. The demonstration of overtones with FoMs of 437 and 53 is a substantive experimental advance, and the measurement methodology (2-port VNA in vacuum, standard mBVD fitting) is sound. However, the strongest claims rest on a single 100-nm fabrication run and on simulation results that are only partially connected to the measured devices. The paper would be substantially stronger with direct evidence for thickness insensitivity and with transparent reporting of fit quality and device statistics.","major_comments":[{"comment":"The claim that 'the performed optimization is not sensitive to the thickness of the piezoelectric layer' is load-bearing for the whole architecture, but its support is only COMSOL dispersion curves at a single thickness (t=100 nm) with electrodes recessed to 80% of the film. No simulation sweep over thickness, no experimental thickness series, and no etch-depth metrology are reported. Because the fabricated unit cell contains a 20 nm unmetallized LiNbO3 plate beneath the electrodes, the dispersion relation could still depend on t and on the recess depth in ways that are not quantified. This is an evidence-sufficiency concern: the device still works, but the core advantage over LVRs—lithographic frequency tuning independent of film thickness—is not established. Please provide FEA results at several film thicknesses and recess depths, or an explicit quantitative bound on the resulting frequency shift and coupling variation.","section":"Section III, 'Design and Optimization'"},{"comment":"The headline FoM values of 437 and 53 are extracted from mBVD fits that appear only as overlaid curves in Figures 6 and 7. The manuscript reports no fit residuals, no number of characterized devices, and no device-to-device statistics. Because these values anchor the abstract and conclusions, the reader cannot judge whether the fits are representative or whether the FoM values are robust. Please provide fit residuals, the number of devices measured for each geometry, and the spread in the fitted parameters (or state explicitly that a single best device is reported).","section":"Section IV and Figs. 6-7"},{"comment":"The statement that 'no net trade off between coupling and quality factor is to be expected' is based on the simulated energy confinement ratio η in Fig. 4, not on measured quality factors. The energy confinement ratio is a proxy for one acoustic loss mechanism, and it does not directly establish the Q-FoM trade-off. Please report measured or fitted Q (and FoM) as a function of wm/wp for the fabricated devices, or explicitly restrict the claim to a simulation-based prediction about η.","section":"Section III, last paragraph"},{"comment":"The measured optimum for the fundamental tones is wm/wp=0.075, which lies outside the simulated optima (wm/wp=0.1 for S0 and 0.1125 for SH0). This discrepancy is not discussed. Since the purpose of the COMSOL optimization is to guide design, the predictive accuracy of the model at the fabricated geometry needs to be reconciled, for example by investigating sensitivity to recess depth, thickness, or fabrication tolerances.","section":"Section IV, 'FABRICATION AND CHARACTERIZATION'"}],"minor_comments":[{"comment":"The sentence 'The excited SH0 modes exhibits Figures of Merit' has a subject-verb agreement error: 'modes exhibit'.","section":"Abstract"},{"comment":"The paragraph beginning 'electrodes are initially etched via timed Deep Oxide Etching' appears to start in the middle of a thought; a leading subject (e.g., 'The electrodes') is missing, and the transition from Section III is abrupt.","section":"Section IV"},{"comment":"The notation 'σxx stress distributions' should use a subscript format (σ_xx) for clarity, and the caption could specify that the stress is along the plate midline at resonance.","section":"Fig. 1 caption"},{"comment":"The phrase 'first ever demonstration of viable overtones for this type of resonator' is strong. Consider positioning it more carefully relative to prior overtone work in related laterally excited platforms (e.g., OBARs and Refs. [21] and [22]) and to the B-IDT work in Ref. [22], which may already report overtone-like modes.","section":"Section V"}],"recommendation":"major_revision","confidential_remarks":"The main risk is that the thickness-decoupling claim, which differentiates this work from LVRs, is supported only by simulations at a single thickness and by a single fabrication run without etch-depth metrology. This is fixable with additional FEA sweeps over thickness and recess depth, plus a careful presentation of fit residuals and device statistics. The 'first ever' claim for overtones should also be checked against Ref. [22] and related literature before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nYou should know about this paper if you care about RF MEMS resonator scaling. It reports the first higher-order overtone operation (they call it LOBAW) of laterally excited bulk acoustic wave resonators on X-cut LiNbO3 with recessed interdigitated electrodes. The headline results are mBVD-fitted Figures of Merit of 437 at 673 MHz for the SH0 fundamental and 53 at 1.05 GHz for the overtone. The overtone result is genuinely new and rests on fresh VNA measurements, not on prior literature. The recessed-electrode architecture is a continuation of the XBAR and buried-electrode B-IDT line, and the self-aligned release is the group's own earlier method, so the architecture is not the novelty; the overtone demonstration plus the X-cut S0/SH0 optimization is.\n\nWhat the paper does well: the measurement path is standard and credible—2-port VNA in vacuum, mBVD fits, frequencies consistent with the 2 µm pitch. The COMSOL design sweeps are clearly laid out, and the authors are candid about known limitations like low Q at antiresonance from incomplete plate release and parasitic capacitance from pad oxidation.\n\nThe soft spots are real but not fatal. The FoM values come from fits shown only in figures, with no residuals, no device count, and no statistics. The paper defines FoM only loosely, so the headline numbers are hard to reproduce from the text. More substantively, the claim that the recessed-electrode design decouples dispersion from film thickness—the premise that gives LBAW its stated edge over LVRs—rests entirely on the authors' COMSOL curves at a single 100 nm thickness. No experiment varies thickness, and the etch depth is not verified by metrology. The best measured fundamental sits at wm/wp=0.075, off the simulated optima of 0.1 and 0.1125, which suggests the model is not fully predictive. That does not invalidate the overtone demonstration, but the thickness-decoupling claim is under-supported and needs a dedicated experiment.\n\nThis is a specialized device paper for the RF MEMS community. A serious referee should engage it: the central overtone result deserves scrutiny, and the thickness claim deserves to be tested. I'd send it to peer review, asking for full fit disclosures and statistical reporting, and ideally an experimental thickness series. I'd bring it to a reading group and would cite the overtone result if I worked in this area.","headline":"First LOBAW overtone demonstration is real and worth refereeing, but the thickness-decoupling claim needs experimental support before it carries weight.","tokens_in":7391,"tokens_out":1866,"would_cite":true,"duration_ms":16198,"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":"This paper demonstrates that recessed interdigitated electrodes in X-cut lithium niobate excite S0 and SH0 lateral bulk waves and, for the first time in this resonator class, make their higher-order overtones usable.","keywords":["lateral bulk acoustic wave resonators","LBAW","lithium niobate","X-cut","recessed electrodes","S0 mode","SH0 mode","overtone resonators"],"falsifier":"Fabricate two resonators with identical electrode geometry and recess depth but different film thicknesses, then compare their $S_0$ or $\\mathrm{SH}_0$ resonance and antiresonance frequencies: a shift beyond the simulation error would falsify the thickness-independence claim. A supporting check is to measure the actual recess depth by cross-section or profilometry, since uncontrolled depth variation would make the geometry-only conclusion untestable.","tokens_in":6113,"feed_emoji":"📡","tokens_out":9024,"duration_ms":78090,"temperature":0.7,"pith_summary":"The paper aims to show that a laterally excited bulk acoustic wave resonator works better when its interdigitated electrodes are recessed into the piezoelectric film rather than laid on top of it. With 100 nm X-cut lithium niobate and electrodes sunk to 80 percent of the thickness, the device excites both the $S_0$ and $\\mathrm{SH}_0$ lateral bulk modes, and for the first time in this resonator class it makes their higher-order overtones usable. The best measured results are a figure of merit of 437 at 673 MHz for the $\\mathrm{SH}_0$ fundamental and 53 at 1.05 GHz for the $\\mathrm{SH}_0$ overtone. The larger claim is that the recessed layout puts the resonance frequency under lithographic control, so several frequencies could be fabricated on a single chip without changing the film stack.","feed_headline":"First viable overtones for lithium niobate bulk-wave resonators","feed_subtitle":"Embedded electrodes free the frequency from film thickness, so many RF bands can share one chip.","key_machinery":"The central object is the recessed interdigitated electrode pair: metal fingers sunk 80 nm into a 100 nm X-cut lithium niobate film, leaving a thin unmetallized plate under the fingers. The design variable is the ratio $w_m/w_p$ of metal finger width to piezoelectric width; sweeping this ratio sets the dispersion of the $S_0$ and $\\mathrm{SH}_0$ lateral bulk modes and their first odd overtones. Finite-element dispersion curves put the peak electromechanical coupling at narrow fingers for the fundamentals (about $w_m/w_p = 0.1$) and at wide fingers for the overtones (about $w_m/w_p = 1.1$). The measured admittance is then reduced to an equivalent-circuit model so that quality factors and figures of merit can be extracted.","core_discovery":"On the paper's own terms, the discovery is that a recessed-electrode geometry efficiently launches lateral bulk acoustic waves in a thin X-cut lithium niobate plate, in both $S_0$ and $\\mathrm{SH}_0$ polarizations, and that the first odd overtones of those modes are strong enough to use. The measured $\\mathrm{SH}_0$ fundamental reaches a figure of merit of 437 at 673 MHz, and the $\\mathrm{SH}_0$ overtone reaches 53 at 1.05 GHz, which the authors read as the first viable overtone demonstration for this type of resonator. They also claim that because the electrodes sit inside the film, the dispersion relation is set by the lateral geometry, namely the ratio $w_m/w_p$ of metal width to piezoelectric width, rather than by film thickness, so that the operating frequency can be tuned lithographically.","pith_inferences":["Extending beyond the paper: a direct thickness sweep at fixed $w_m/w_p$ would test whether the decoupling claim survives fabrication reality; if it does, the design can move to thicker films for higher power handling.","The same recessed-electrode trick could transfer to other high-coupling piezoelectric films, where removing the thickness from the dispersion problem would simplify multi-band filter design.","Overtones could push the operating frequency upward without shrinking lithographic pitch, which is the lever that matters for 5G/6G front ends."],"forward_implications":["The resonator works in overtone mode, not only at its fundamental tone, giving designers an extra frequency band from the same plate.","Shifting the mask changes the frequency: adjusting $w_m/w_p$ selects the operating band without changing the piezoelectric film, which enables on-chip multi-frequency scaling.","Recessed electrodes increase static capacitance density and reduce series resistance, lowering ohmic losses in a filter.","Because both $S_0$ and $\\mathrm{SH}_0$ modes are excited on the same substrate, one chip offers two polarizations with different coupling and frequency behaviour."],"supporting_citations":[{"why":"introduced the original laterally excited bulk-wave (XBAR) concept that this device builds on","marker":"[21]"},{"why":"demonstrated suspended lithium niobate resonators with buried electrodes, the immediate predecessor of the recessed-electrode design","marker":"[22]"},{"why":"provided the single-chip multifrequency X-cut lithium niobate resonator benchmark that LBAW aims to improve on","marker":"[18]"},{"why":"supplied the energy-confinement-ratio method used to relate electrode geometry to acoustic loss","marker":"[25]"},{"why":"provided the self-aligned release method and prior $S_0$-mode lithium niobate resonator demonstration","marker":"[26]"},{"why":"defined the equivalent-circuit model used to fit admittance and extract figures of merit","marker":"[27]"},{"why":"showed overmoded bulk acoustic resonance at 33 GHz, the overtone behaviour the LOBAW design mirrors","marker":"[23]"},{"why":"compared against scandium aluminum nitride overmoded resonators to inform the overtone design rules","marker":"[24]"}],"fun_headline_variants":["First viable overtones in LiNbO3 lateral bulk-wave resonators","Embedded electrodes tune resonator frequency independent of thickness","Recessed electrodes enable multi-frequency LiNbO3 resonators","X-cut LiNbO3 overtones reach FoM 53 at 1.05 GHz","Lithographic resonator tuning opens multi-band RF on one chip"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the resonance frequency is set by the lateral electrode geometry $w_m/w_p$ alone, so the dispersion relation is decoupled from the 100 nm film thickness; the paper's support is finite-element simulation at one thickness with electrodes recessed to 80 percent, with no experiment varying the film thickness and no reported etch-depth metrology.","fun_headline_variants_meta":{"raw":{"variants":["First viable overtones in LiNbO3 lateral bulk-wave resonators","Embedded electrodes tune resonator frequency independent of thickness","Recessed electrodes enable multi-frequency LiNbO3 resonators","X-cut LiNbO3 overtones reach FoM 53 at 1.05 GHz","Lithographic resonator tuning opens multi-band RF on one chip"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000611,"raw_usage":{"total_tokens":2819,"prompt_tokens":897,"completion_tokens":1922,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":513,"completion_tokens_details":{"reasoning_tokens":1830}},"tokens_in":513,"tokens_out":1922,"duration_ms":14356,"temperature":1.0,"reasoning_tokens":1830,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T10:43:52.425336+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fabricate two resonators with identical electrode geometry and recess depth but different film thicknesses, then compare their $S_0$ or $\\mathrm{SH}_0$ resonance and antiresonance frequencies: a shift beyond the simulation error would falsify the thickness-independence claim. A supporting check is to measure the actual recess depth by cross-section or profilometry, since uncontrolled depth variation would make the geometry-only conclusion untestable.","supporting_citations":[{"cited_title":"5 ghz laterally-excited bulk -wave resonators (xbars) based on thin platelets of lithium niobate,","cited_arxiv_id":null,"evidence_quote":"introduced the original laterally excited bulk-wave (XBAR) concept that this device builds on"},{"cited_title":"Suspended lithium niobate acoustic resonators with buried electrodes for radiofrequency filtering,","cited_arxiv_id":null,"evidence_quote":"demonstrated suspended lithium niobate resonators with buried electrodes, the immediate predecessor of the recessed-electrode design"},{"cited_title":"2 –16 ghz multifrequency x -cut lithium niobate nems resonators on a single chip,","cited_arxiv_id":null,"evidence_quote":"provided the single-chip multifrequency X-cut lithium niobate resonator benchmark that LBAW aims to improve on"},{"cited_title":"Characterization of acoustic losses in interdigitated vhf to mmwave piezoelectric m/nems resonators,","cited_arxiv_id":null,"evidence_quote":"supplied the energy-confinement-ratio method used to relate electrode geometry to acoustic loss"},{"cited_title":"Sub -ghz x -cut lithium niobate s mode mems resonators,","cited_arxiv_id":null,"evidence_quote":"provided the self-aligned release method and prior $S_0$-mode lithium niobate resonator demonstration"},{"cited_title":"Modified Butterworth- Van Dyke circuit for FBAR resonators and automated measurement system,","cited_arxiv_id":null,"evidence_quote":"defined the equivalent-circuit model used to fit admittance and extract figures of merit"},{"cited_title":"33 GHz Overmoded Bulk Acoustic Resonator,","cited_arxiv_id":null,"evidence_quote":"showed overmoded bulk acoustic resonance at 33 GHz, the overtone behaviour the LOBAW design mirrors"},{"cited_title":"Scandium aluminum nitride overmoded bulk acoustic resonators for future wireless communication,","cited_arxiv_id":null,"evidence_quote":"compared against scandium aluminum nitride overmoded resonators to inform the overtone design rules"}],"review_version":1}