{"id":"39d5a68b-613f-4049-9dc9-6b4d2fd74aea","arxiv_id":"2501.18341","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A fully-etched Z-cut lithium niobate waveguide achieves 5.8 dB/m propagation loss and supports an all-normal-dispersion octave-spanning supercontinuum.","lead":"Researchers fabricated a Z-cut lithium niobate spiral waveguide with an ultra-low propagation loss of 5.8 dB/m and used it to generate an octave-spanning supercontinuum. The result could enable on-chip narrow-linewidth lasers, optical delay lines, and parametric amplifiers that need long low-loss waveguides.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Mode/polarization purity is assumed, not measured: the 5.8 dB/m loss (Fig. 3, §3) and the 30 cm ANDi supercontinuum (§4) both rely on the input TE00 mode remaining dominant in nominally multimode waveguides, with only a single simulated bend (Fig. 1(b)) as support.","rationale":"The reader's conditional verdict is already cautious. I agree with the thrust of the reader's weakest assumption, though I would sharpen it: the waveguide is not claimed to be single-mode—the low-loss devices are explicitly multimode, and the single-mode width (0.7 µm) has 67 dB/m loss (§3). The operative assumption is that the launched TE00 mode remains dominant and does not convert to other modes or polarizations along 15–30 cm. This assumption is what connects the two headline claims: the OFDR trace is not mode-resolved, so a multimode decay could bias 5.8 dB/m, and the ANDi spectrum is only attributable to TE00 if higher-order modes carry negligible power. A single bend simulation (Fig. 1(b)) addresses only the TE00-to-TM00 transfer for one geometry; it cannot certify the whole spiral, including the S-bend and 30 cm of sidewall roughness, against intermode coupling. There is also no uncertainty or independent loss estimate, which strengthens the need for a direct check. The concrete test—output polarization extinction and mode imaging over the full band—directly resolves whether the assumption holds. If it passes, the conditional acceptance should stand; if it fails, the loss number and the first coherent ANDi supercontinuum claim both need revision. I therefore leave the verdict unchanged at CONDITIONAL.","tokens_in":8559,"tokens_out":9821,"duration_ms":105383,"concrete_test":"Couple a tunable CW laser (1480–1620 nm) into the 15 cm spiral and the 30 cm SC waveguide through a polarization-maintaining launch, then at the output pass the light through a half-wave plate, a polarizer, and an InGaAs beam profiler or imaging spectrometer. Record the output spatial mode and polarization extinction while rotating the input polarization and sweeping wavelength. If the output stays a single lobe with >20 dB extinction and no higher-order lobes above noise at every wavelength, the mode/polarization-purity assumption is confirmed; if higher-order lobes or >5% cross-polarized power appear, the OFDR loss value and the single-mode ANDi interpretation must be re-analyzed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—5.8 dB/m over 15 cm—is extracted by OFDR on a spiral whose width (3 µm) the paper itself calls multimode (§3). The supercontinuum is generated in a 30 cm, 2.7 µm-wide guide, also likely multimode at 1.56 µm. OFDR returns backscattered power as a function of delay; it is not mode- or polarization-selective. If bends, the S-bend connector, or sidewall roughness converts even a small fraction of the launched TE00 power into higher-order TE/TM modes, the fitted decay mixes several modes with different losses and group delays, so 5.8 dB/m is not a single-mode propagation loss. Likewise, the claimed all-normal-dispersion supercontinuum is attributed to the designed TE00 mode; multi-mode excitation would introduce other dispersions and nonlinearities. The only evidence for mode/polarization purity is Fig. 1(b), a single simulated bend in Z-cut LN showing no TE00-to-TM00 transfer; no output polarization-extinction, mode-profile, or wavelength-resolved mode check is reported for any fabricated device. This unverified assumption is load-bearing for both headline results.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a fabrication process for fully etched Z-cut lithium niobate (LN) waveguides and claims a record-low propagation loss of 5.8 dB/m measured by optical frequency-domain reflectometry (OFDR) in a 15 cm spiral waveguide. It also demonstrates supercontinuum generation in a 30 cm, 2.7 µm-wide all-normal-dispersion (ANDi) waveguide, reaching an octave-spanning spectrum at ~207 pJ on-chip pump energy, and labels the supercontinuum as coherent. The central design idea is that Z-cut LN with TE polarization avoids the direction-dependent birefringence that causes intermode coupling in X-cut bent waveguides, supported by a finite-element simulation of a single bend (Fig. 1(b)). The paper includes detailed fabrication steps, loss measurements versus width and wavelength, and a dispersion simulation.","tokens_in":8861,"tokens_out":3239,"duration_ms":34891,"significance":"If the claims hold, this would be a notable advance for LN photonics: a decimeter-long, tightly confined waveguide with dB/m-level loss would benefit delay lines, narrow-linewidth lasers, and parametric devices, and an ANDi octave-spanning supercontinuum in an integrated LN waveguide would be a first. The fabrication methodology (multipass EBL, thermal management, single-writing-field spiral design) and the direct OFDR loss extraction are useful contributions. However, the current evidence is incomplete in a load-bearing way: the waveguides used for both headline results are explicitly multimode, yet no experimental characterization of mode or polarization purity along the long spirals is provided, and the supercontinuum coherence is inferred rather than measured. The significance is therefore conditional on additional verification.","major_comments":[{"comment":"","section":"Section 3, Fig. 1(b), Fig. 3(a)"},{"comment":"","section":"Section 3, Fig. 3(a)–(d)"},{"comment":"","section":"Section 4, Fig. 4(c)"},{"comment":"","section":"Section 4, experimental setup"}],"minor_comments":[{"comment":"","section":"Abstract/Introduction"},{"comment":"","section":"Section 4"},{"comment":"","section":"Funding"},{"comment":"","section":"Section 3, Fig. 3(c),(d)"},{"comment":"","section":"Section 3, Fig. 3(a)"},{"comment":"","section":"References"},{"comment":"","section":"Section 2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript has a strong fabrication story and the loss and supercontinuum results are plausible, but the absence of experimental mode/polarization characterization for the long multimode spirals is a load-bearing gap that affects both headline claims. I would support publication after the authors provide either direct mode/polarization measurements or a convincing analysis that the OFDR and supercontinuum measurements are single-mode, together with uncertainty estimates and a coherence measurement or simulation. The 'first' claims should also be checked against the existing LN supercontinuum literature during revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The two headline results are real enough to take seriously. A 5.8 dB/m propagation loss in a 15 cm LN spiral is a new mark for long LN waveguides, and the octave-spanning all-normal-dispersion supercontinuum in a 30 cm LN guide is, as far as I can tell, a first. The fabrication work is careful: the Si3N4 adhesion layer, multipass EBL, and thermal management during IBE are all described with enough detail to reproduce, and the width- and wavelength-dependent loss data add credibility. The loss is a direct OFDR measurement, not a fit to a derived model, so I do not see circularity.\n\nThe soft spots are real but not fatal. The loss figure has no uncertainty and no independent cross-check (no ring-Q or cutback). More importantly, OFDR is not mode-selective, and the 3 µm and 2.7 µm waveguides are explicitly multimode. So 5.8 dB/m is likely a mode-averaged loss, not the TE00 propagation loss. The paper does not hide this—it openly calls the wide guides multimode—but the practical claims about delay lines and narrow-linewidth lasers depend on single-mode operation, and that is not demonstrated. The only mode-purity evidence is a simulated single bend; there is no measured polarization extinction or mode profile on any fabricated device. The stress-test worry is therefore valid, with the caveat that the paper never claims these wide guides are single-mode.\n\nThe supercontinuum claim has a similar gap. The spectrum reaching one octave at 207 pJ is shown, but the word \"coherent\" in the abstract is justified only by reference to prior ANDi work, not by a direct coherence measurement. Given that the pump mode composition is uncharacterized, the coherence statement is stronger than the evidence.\n\nWho gets value from this? Groups working on low-loss LN waveguides, delay lines, and parametric nonlinear devices will want the fabrication details and the loss data. The paper deserves a serious referee. My recommendation: send it out, and use the review to push for error bars on the loss, a mode/polarization check on the fabricated guides, an independent loss confirmation, and a measured coherence trace for the SC. None of these are deal-breakers, but they would turn a good experimental letter into a solid one.","headline":"Genuine fabrication advance with a record loss figure and the first ANDi octave-spanning supercontinuum in integrated LN, but the loss is not mode-selective and the coherence is inferred rather than measured.","tokens_in":9390,"tokens_out":2482,"would_cite":true,"duration_ms":26230,"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":"Z-cut lithium niobate waveguides hit 5.8 dB/m loss and host the first octave-spanning all-normal-dispersion supercontinuum in an integrated LN waveguide.","keywords":["lithium niobate","LNOI waveguides","ultralow propagation loss","Z-cut","all-normal dispersion","supercontinuum generation","optical frequency-domain reflectometry","integrated photonics"],"falsifier":"Measure the output mode profile and polarization extinction of the 15 cm and 30 cm spirals: if a measurable TM component or a second transverse mode appears after propagation, the claim that Z-cut TE avoids birefringence-induced intermode coupling would be contradicted.","tokens_in":8417,"feed_emoji":"🔬","tokens_out":9489,"duration_ms":79104,"temperature":0.7,"pith_summary":"This paper reports a fabrication process for fully etched, tightly confined Z-cut lithium niobate (LN) waveguides and claims two advances enabled by it. The first is a propagation loss of 5.8 dB/m measured in a 15 cm spiral, the lowest loss directly measured in a decimeter-long LN waveguide. The second is the first all-normal-dispersion octave-spanning supercontinuum generated in an integrated LN waveguide, obtained in a 30 cm device pumped at 1560 nm. These results matter because they combine low loss, strong confinement, and dispersion control in one chip-scale platform, removing a barrier for nonlinear and phase-sensitive LN photonics.","feed_headline":"Z-cut lithium niobate waveguides hit 5.8 dB/m loss","feed_subtitle":"Fully etched Z-cut LN spirals also produce the first octave-spanning all-normal-dispersion supercontinuum on a chip.","key_machinery":"The load-bearing object is a fully etched strip waveguide in a 600 nm Z-cut lithium-niobate-on-insulator film, operated in the quasi-TE mode so that bent sections do not convert TE into TM energy. The fabrication combines a thin adhesion layer, negative-tone electron-beam lithography with multipass exposure for smooth sidewalls, and argon ion-beam etching with thermal management to reach a sidewall angle around 70 degrees. Long spirals are built from Archimedean units each contained in a single writing field, joined by S-bends whose curvature is a cubic polynomial of arc length, $\\kappa(s)=a_0+a_1s+a_2s^2+a_3s^3$, which keeps mode coupling low. Loss is read out with optical frequency-domain reflectometry, and dispersion is engineered through the 2.7 $\\mu$m by 0.6 $\\mu$m cross-section so that the waveguide is all-normal-dispersion, which preserves coherence during supercontinuum generation.","core_discovery":"The central claim is that choosing the Z-cut crystal orientation and operating in the quasi-TE mode eliminates the material-birefringence penalty that distorts modes in bent X-cut LN waveguides, so a fully etched strip waveguide can be both tightly confining and ultralow loss. On this platform, optical frequency-domain reflectometry yields 5.8 dB/m over a 15 cm spiral, and 207 pJ pulses produce a spectrum spanning more than an octave in a 30 cm all-normal-dispersion waveguide. The authors argue this makes the waveguide a practical building block for on-chip delay lines, narrow-linewidth lasers, parametric amplifiers, and $\\chi^3$ nonlinear devices.","pith_inferences":["If the 5.8 dB/m loss persists at multi-decimeter lengths, Z-cut LN could compete with silicon nitride in delay-line and narrow-linewidth applications while adding strong quadratic and cubic nonlinearity; the paper does not quantify those trade-offs.","A direct measurement of output mode purity and polarization extinction along the 30 cm waveguide would convert the single-bend simulation argument into an experimental guarantee that birefringence is fully avoided.","The coherence of the octave-spanning spectrum is assumed from the all-normal-dispersion regime rather than measured; a pulse-to-pulse interferometric measurement would test that assumption directly."],"forward_implications":["Decimeter-long ultralow-loss LN spirals become practical for on-chip delay lines and laser cavities, since a 15 cm path adds under 1 dB of propagation loss.","Z-cut TE geometry removes the single-direction layout restriction caused by birefringence, so dispersion-engineered and phase-sensitive components can be arranged freely on a chip.","An octave-spanning all-normal-dispersion supercontinuum in an integrated LN waveguide offers a path toward chip-scale self-referenced frequency combs without the coherence degradation seen in anomalous-dispersion supercontinuum.","The same fabrication flow is expected to carry over to X-cut LN, preserving access to the largest electro-optic and second-order nonlinear coefficients."],"supporting_citations":[{"why":"Establishes the fully etched LN waveguide fabrication basis that the present work extends toward ultralow loss.","marker":"[15]"},{"why":"Supplies the single-writing-field spiral strategy used to reduce stitching errors in long waveguides.","marker":"[33]"},{"why":"Gives the cubic-polynomial curvature S-bend design that connects spiral units without mode coupling.","marker":"[39]"},{"why":"Provides the optical frequency-domain reflectometry method used to measure the 5.8 dB/m propagation loss.","marker":"[41]"},{"why":"Underpins the claim that all-normal-dispersion supercontinuum generation stays coherent.","marker":"[42]"},{"why":"Demonstrates coherent all-normal-dispersion supercontinuum in Si3N4, the benchmark the LN waveguide result extends.","marker":"[44]"},{"why":"Documents birefringence-induced intermode crosstalk in bent LN waveguides that Z-cut TE geometry is meant to avoid.","marker":"[37]"}],"fun_headline_variants":["Z-cut LN waveguides slash loss to 5.8 dB/m","First integrated octave-spanning supercontinuum in Z-cut LN","Z-cut LN: 5.8 dB/m loss and octave-spanning comb","Z-cut LN kills birefringence, hits 5.8 dB/m"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The platform relies on the assumption that the quasi-TE mode stays single-mode and polarization-pure throughout the 15-30 cm spiral, yet the paper tests this only in a simulation of one bend rather than on a real long waveguide.","fun_headline_variants_meta":{"raw":{"variants":["Z-cut LN waveguides slash loss to 5.8 dB/m","First integrated octave-spanning supercontinuum in Z-cut LN","Z-cut LN: 5.8 dB/m loss and octave-spanning comb","Z-cut LN kills birefringence, hits 5.8 dB/m"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001351,"raw_usage":{"total_tokens":5481,"prompt_tokens":937,"completion_tokens":4544,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":553,"completion_tokens_details":{"reasoning_tokens":4460}},"tokens_in":553,"tokens_out":4544,"duration_ms":30232,"temperature":1.0,"reasoning_tokens":4460,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T23:50:50.001145+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the output mode profile and polarization extinction of the 15 cm and 30 cm spirals: if a measurable TM component or a second transverse mode appears after propagation, the claim that Z-cut TE avoids birefringence-induced intermode coupling would be contradicted.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the fully etched LN waveguide fabrication basis that the present work extends toward ultralow loss."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the single-writing-field spiral strategy used to reduce stitching errors in long waveguides."},{"cited_title":"Kazama, T","cited_arxiv_id":null,"evidence_quote":"Gives the cubic-polynomial curvature S-bend design that connects spiral units without mode coupling."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the optical frequency-domain reflectometry method used to measure the 5.8 dB/m propagation loss."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Underpins the claim that all-normal-dispersion supercontinuum generation stays coherent."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates coherent all-normal-dispersion supercontinuum in Si3N4, the benchmark the LN waveguide result extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents birefringence-induced intermode crosstalk in bent LN waveguides that Z-cut TE geometry is meant to avoid."}],"review_version":1}