{"id":"95bcea44-81c0-4910-a62c-09042841f924","arxiv_id":"2507.13004","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Sidewall-poled x-cut thin-film lithium niobate waveguides with 215 nm periods produce counter-propagating and backward-propagating light, including the first backward-propagating spontaneous parametric down-conversion.","lead":"Researchers poled thin-film lithium niobate waveguides with domain periods as small as 215 nanometers, enabling light-conversion processes where generated light travels backwards. The advance gives integrated photonics a new tool for making photon pairs with tailored direction, which matters for quantum communication and computing.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Backward SHG/SPDC in the 215 nm device may be mimicked by reflected forward-generated light; a direction-resolving time-delay or reflectivity test is needed before the central claim is secure.","rationale":"The paper reports a plausible and impressive fabrication advance: 390 nm and 215 nm poling periods in x-cut TFLN, with SEM and wet-etch evidence of sub-100 nm domain widths, high measured counter-propagating SHG efficiency, SFG maps that match simulations, and SPDC with high CAR. The 390 nm counter-propagating device is relatively well supported. The weaker point is the 215 nm backward device, where the authors themselves note partial poling depth and bottom merging, and where the central novelty depends on the propagation direction of the generated light. The reader's weakest-assumption analysis points to exactly this combination of partial domain geometry and possible reflection artifacts, and I agree. This is not an accusation of fabrication; it is the standard experimental ambiguity in backward-wave nonlinear optics, and the paper's own SFG-shape argument is suggestive but not uniquely direction-resolving. Because the issue is concrete and testable, and because the rest of the evidence is substantial, the correct disposition is to keep the conditional verdict and require a direction-resolving check before full acceptance.","tokens_in":10390,"tokens_out":9992,"duration_ms":123284,"concrete_test":"Use a pulsed 1550 nm pump (or a pump with a fast intensity step) on the same 215 nm backward device and time-resolve the 775 nm output at the pump-input port. In true backward SHG the output should follow the local pump intensity with only the short transit from its generation point to the port; a component delayed by about 2nL/c ~ 7 ps (L = 0.5 mm, n ~ 2.2) would indicate forward SHG reflected at the output end. Complement this by directly measuring the 775 nm reflectivity of the output grating/demultiplexer and the forward 775 nm power with the same input; if R_775 times P_forward is below the measured backward power by orders of magnitude, reflection cannot explain the observation. The same time-delay test can be applied to the SPDC signal.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central novelty is the first backward-propagating SPDC and 215 nm backward phase matching. The load-bearing condition is that the 775 nm light collected from the pump-input port is genuinely generated propagating backward inside the waveguide, not forward-generated 775 nm reflected back by the output grating coupler, demultiplexer, or facet. The paper's explicit exclusion of reflections rests on the SFG map shape (Fig. 3f) matching simulation, but a phase-matching map produced by forward SFG followed by a round-trip reflection can share the same spectral locus because the same grating momentum is involved; spectral shape alone does not uniquely identify propagation direction. This ambiguity is aggravated by the fabrication characterization in Fig. 2d/f: the 215 nm domains reach only about half the film thickness and are merged at the bottom, so the true three-dimensional duty cycle of the poled grating is unknown and no depth-resolved ferroelectric measurement is provided. The nonlinear overlap integral governing backward SHG/SPDC could therefore be much weaker than assumed, making the reported 44 %/W/cm2 and 11 kHz/mW harder to reconcile with an intrinsic backward process. Without a direction-discriminating measurement, the paper's principal claim is not fully settled.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports periodic poling of x-cut thin-film lithium niobate with periods down to 215 nm using a sidewall-electrode technique, and uses these gratings to demonstrate counter-propagating and backward-propagating phase-matched nonlinear interactions. The authors characterize second-harmonic generation (efficiencies 1474 %/W/cm2 and ~45 %/W/cm2), map sum-frequency generation and compare with simulations, and measure spontaneous parametric down-conversion with on-chip brightness values of 89 kHz/mW and 11 kHz/mW, claiming the first backward-propagating SPDC in an integrated platform.","tokens_in":10674,"tokens_out":10192,"duration_ms":112825,"significance":"If the direction assignment is correct, this is an important step for nanodomain engineering and for integrated quantum sources with tailored photon propagation. The sidewall-poling method is a scalable approach to sub-300 nm periods in the technologically important x-cut TFLN platform. The paper provides clear fabrication details, SEM and wet-etch domain characterization, and nonlinear optical data with consistent power scalings and high CAR, which strengthens confidence in the core observations.","major_comments":[{"comment":"The claim that reflections are excluded is not quantitatively justified. The SFG map shape is compared to simulation for the intended backward process, but no calculation is shown for the phase-matching condition of forward-generated light followed by reflection. Since the backward device is the principal novelty, please add a quantitative estimate of the phase mismatch for forward SHG/SFG with the 215 nm grating (showing that it is far from phase matching), or provide a direction-resolving measurement such as a time-delay or reflectivity test. Without this, the reader cannot independently verify the direction assignment.","section":"Nonlinear spectrum and efficiency measurement"},{"comment":"The 215 nm domains reach only about half of the film thickness and are merged at the bottom. The reported backward efficiencies (45 %/W/cm2 and 11 kHz/mW) are attributed to this imperfection, but no quantitative model of the nonlinear overlap integral for the actual domain shape is provided. Please include a calculation of the expected efficiency reduction from the measured domain depth and duty cycle, to confirm that the observed backward signals are consistent with the partially poled grating.","section":"Fabrication and characterization, Fig. 2d/f"}],"minor_comments":[{"comment":"In the paragraph following Fig. 2, the text refers to the 215 nm period sample as '(Fig.2e)', but the correct subfigure is Fig. 2f; please fix this citation.","section":"Fabrication and characterization"},{"comment":"The reported conversion efficiencies and SPDC brightness values are given without error bars or confidence intervals; please report uncertainties from the fits, as these numbers are central to the quantitative claims.","section":"Nonlinear spectrum and efficiency measurement"},{"comment":"The abstract states an efficiency of 45 %/W/cm2 for the backward device, while the introduction says 45 %/W/cm2 and the measurement section reports 44 %/W/cm2; please harmonize these values.","section":"Abstract / Introduction"},{"comment":"The terms 'backward-propagating' and 'back-propagating' are used interchangeably; please define the terminology once and use it consistently.","section":"Introduction / throughout"},{"comment":"The statement that the theoretical conversion efficiency is the same for all three phase-matching schemes is nontrivial, since the nonlinear overlap integrals for forward, counter-propagating, and backward geometries may differ; please provide a derivation or reference.","section":"Nonlinear spectrum and efficiency measurement"}],"recommendation":"major_revision","confidential_remarks":"The experimental work is impressive and likely correct, but the principal novelty rests on the backward-direction assignment. My requested quantitative phase-mismatch argument or a direct time-domain measurement is important for a broad optics audience. The partial poling depth is honestly disclosed but deserves a quantitative link to the measured efficiency. These are fixable within the manuscript's scope, hence major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: this is a meaningful advance. The authors push sidewall poling to 390 nm and 215 nm periods in x-cut TFLN, and they report the first backward-propagating SPDC in an integrated platform. The central evidence is internally consistent: SHG power sweeps are quadratic, SFG maps match the simulated phase-matching shapes, and SPDC rates scale linearly with pump power with high CAR. The fabrication characterization is careful, including wet-etch domain depth inspection, and the authors are upfront that the 215 nm domains invert only half the film thickness.\n\nWhat makes this novel: sidewall poling itself is not new, but applying it to reach 215 nm first-order poling, and using it to demonstrate counter- and backward-propagating phase matching, is new. The 1474 %/W/cm2 counter-propagating efficiency is a solid number, and the 44 %/W/cm2 backward number is plausible given the reduced nonlinear overlap from partial-depth domains. The counter-propagating brightness of 89 kHz/mW also compares favorably with earlier work.\n\nSoft spots: the biggest is the backward directionality claim. The paper argues that reflections are excluded because the SFG map shows strong self-SHG lines only for the backward case, while the counter-propagating map shows weak self-lines. That is a decent argument, but it is not a direct discriminator: a forward-generated SHG signal that is reflected back through the same grating could, in principle, produce a similar spectral map. A time-resolved or spatially resolved measurement that directly distinguishes the direction of the generated light would fully settle it. The lack of error bars on the efficiencies and brightness, no yield statistics, and no deposited raw data are minor but real. The partial-depth poling of the 215 nm device is acknowledged, but the quantitative effect on the mode overlap is not independently verified.\n\nOverall, the paper deserves a serious referee. The fabrication advance is useful, the claims are specific, and the authors are honest about limitations. I would encourage the editor to send it out, with a request for a direction-resolving check of the backward process and for error bars on the headline numbers.","headline":"A well-executed fabrication and nonlinear optics paper that likely delivers the first backward-propagating SPDC in TFLN, but the directionality claim would benefit from a direct time-reversal check.","tokens_in":11190,"tokens_out":3748,"would_cite":true,"duration_ms":43853,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["42.65.Ky","42.79.Nv"],"model":"deepseek-v4-flash","headline":"The paper reports scalable periodic poling of x-cut thin-film lithium niobate with periods down to 215 nm, demonstrating counter-propagating second-harmonic generation at 1474 %/W/cm² and the first backward-propagating spontaneous…","keywords":["thin-film lithium niobate","periodic poling","nanodomain","counter-propagating phase matching","backward second harmonic generation","spontaneous parametric down-conversion","quasi-phase matching","integrated quantum photonics"],"falsifier":"Image a cross-section of the 215 nm-period waveguide with a depth-resolved technique such as piezoresponse force microscopy to map the actual inverted-domain depth, then compute the backward second-harmonic and SPDC efficiencies from that true domain profile and compare with the measured values. If the measured backward signal matches a model in which the inverted region has no overlap with the guided mode, or if preventing any forward-generated light from returning (e.g., by placing an absorber at the far end of the waveguide) leaves the backward count unchanged, the intrinsic-backward-generation claim is disproven.","tokens_in":10217,"feed_emoji":"🔬","tokens_out":8403,"duration_ms":88159,"temperature":0.7,"pith_summary":"Periodic poling of x-cut thin-film lithium niobate has been limited to micrometre-scale periods because adjacent domains merge when poled at sub-micron pitch. This paper claims to break that limit by placing poling electrodes on the etched sidewalls of the waveguide, achieving first-order gratings at 390 nm and 215 nm periods. In these devices the authors measure counter-propagating second-harmonic conversion at $1474\\,\\%/\\mathrm{W}/\\mathrm{cm}^2$ and backward conversion at $45\\,\\%/\\mathrm{W}/\\mathrm{cm}^2$, with sum-frequency phase-matching maps that match simulation, and they report the first backward-propagating spontaneous parametric down-conversion, with source brightness of 89 kHz/mW (counter) and 11 kHz/mW (backward). If correct, this turns propagation direction into an engineerable property of integrated nonlinear sources, enabling high-purity, spectrally separated photon pairs and potential mirrorless parametric oscillators on the standard x-cut platform.","feed_headline":"First backward photon pairs from thin-film lithium niobate","feed_subtitle":"Sidewall poling reaches 215 nm periods and sends signal and idler in opposite directions for new quantum sources.","key_machinery":"The central object is the sidewall-poled ferroelectric domain grating: a periodic inversion of the crystal's nonlinearity whose period encodes the propagation directions of the interacting waves. For degenerate signal and idler, counter-propagating phase matching needs $\\Lambda = \\lambda_p / n(\\lambda_p)$ and backward-propagating phase matching needs $\\Lambda = \\lambda_p / [n(\\lambda_p)+n(\\lambda_s)]$, which in this platform dictates periods near 390 nm and 215 nm, respectively. The fabrication method, depositing titanium electrodes on the etched waveguide sidewalls and applying a single 1 ms voltage pulse, is what makes these gratings possible, by preventing the lateral domain merging that occurs when electrodes are placed on the top surface far apart, and by avoiding the shallow inversion that occurs when they are placed close together.","core_discovery":"The central claim is that ultrashort poling periods for counter- and backward-propagating quasi-phase matching can be achieved in x-cut thin-film lithium niobate by etching the waveguide before poling and patterning the high-voltage electrodes on the waveguide sidewalls. This 'sidewall poling' confines the lateral growth of the ferroelectric domains and starts nucleation along the full etched depth, yielding first-order gratings at 390 nm (full film inversion, roughly 50% duty cycle) and 215 nm (domains reaching about half the film thickness). On these gratings the authors demonstrate counter-propagating second-harmonic generation at $1474\\,\\%/\\mathrm{W}/\\mathrm{cm}^2$, backward second-harmonic generation at $45\\,\\%/\\mathrm{W}/\\mathrm{cm}^2$, sum-frequency maps whose line shapes match the predicted phase-matching functions, and the first backward-propagating spontaneous parametric down-conversion with an internal brightness of 11 kHz/mW. The paper treats the 215 nm device's partially inverted domains as a remaining limitation and suggests a second poling step could complete the inversion.","pith_inferences":["Completing the inversion of the 215 nm domains with a second poling step after thinning the sidewall electrodes would likely raise the backward efficiency far above the current 45 %/W/cm².","The same sidewall-poling geometry should scale to even shorter periods, potentially below 200 nm, by narrowing the waveguide top width, opening other directional phase-matching configurations.","A direct check of the backward origin of the photon pairs, such as distinguishing backward-generated from reflected forward light via time-resolved detection after short-pulse pumping, would convert the indirect SFG-map evidence into a definitive demonstration.","Direction-engineered phase matching of this kind might be used to build nonlinear isolators or add-drop elements, since the sign of the phase-matching condition selects which propagation direction couples."],"forward_implications":["Counter- and backward-propagating photon-pair sources become feasible on the x-cut TFLN platform, inheriting its high-speed electro-optic tunability and established fabrication ecosystem.","Backward-propagating SPDC gives deterministic spatial separation of signal and idler, plus natural pump filtering, which simplifies quantum routing and time-bin entanglement experiments.","The measured counter-propagating SHG efficiency of 1474 %/W/cm² at roughly half the theoretical value indicates that first-order 390 nm gratings in x-cut TFLN are already close to practical quality.","With periods near 215 nm available, the mirrorless degeneracy-locked optical parametric oscillator and backward-wave devices previously limited to bulk crystals can be considered for integrated implementation.","Because the method is based on standard electron-beam lithography and a single voltage pulse, it is scalable to millimetre-long poled regions, directly increasing photon-pair brightness with length."],"supporting_citations":[{"why":"Supplies the sidewall-poling method (electrodes on etched waveguide sidewalls) that the paper adapts to ultra-short periods.","marker":"[32]"},{"why":"Demonstrates backward-wave SPDC in bulk periodically poled crystals; the paper compares its counter-propagating brightness to this prior result.","marker":"[28]"},{"why":"Reports counter-propagating photon-pair generation in a nonlinear waveguide, providing the integrated baseline that this work advances to first-order x-cut poling.","marker":"[30]"},{"why":"Achieves first-order counter-propagating SHG in z-cut TFLN, the efficiency benchmark the paper compares its x-cut result against.","marker":"[38]"},{"why":"Reports backward SHG in silicon nitride via optically induced quasi-phase-matching, the only prior integrated backward-generation result, at three orders of magnitude lower efficiency.","marker":"[39]"},{"why":"Documents the lateral domain merging that limits sub-micron poling in TFLN, the fabrication obstacle the sidewall method is designed to overcome.","marker":"[24]"},{"why":"The authors' preceding demonstration of a counter-propagating SPDC source in LNOI, the direct precursor of the devices reported here.","marker":"[29]"},{"why":"Establishes the wafer-scale periodic poling process for conventional micrometre periods, defining the state of the art that the paper pushes to the nanoscale.","marker":"[12]"}],"fun_headline_variants":["Sidewall poling hits 215 nm periods for backward light conversion","Nanodomain poling yields backward photon pairs in thin-film lithium niobate","Ultra-small poling gates backward nonlinear light in TFLN","Backward spontaneous down-conversion achieved with 215 nm gratings","Thin-film niobate reaches 215 nm poling for reverse light"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The backward-propagating demonstration assumes that in the 215 nm-period device, the partially inverted domains that reach only about half the film thickness and are merged at the bottom still provide enough nonlinear overlap in the guided optical mode, and that the backward-detected light is genuinely generated backward by the grating rather than being forward-generated light that is reflected somewhere in the circuit.","fun_headline_variants_meta":{"raw":{"variants":["Sidewall poling hits 215 nm periods for backward light conversion","Nanodomain poling yields backward photon pairs in thin-film lithium niobate","Ultra-small poling gates backward nonlinear light in TFLN","Backward spontaneous down-conversion achieved with 215 nm gratings","Thin-film niobate reaches 215 nm poling for reverse light"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000353,"raw_usage":{"total_tokens":1946,"prompt_tokens":993,"completion_tokens":953,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":609,"completion_tokens_details":{"reasoning_tokens":857}},"tokens_in":609,"tokens_out":953,"duration_ms":9215,"temperature":1.0,"reasoning_tokens":857,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:32:23.733310+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Image a cross-section of the 215 nm-period waveguide with a depth-resolved technique such as piezoresponse force microscopy to map the actual inverted-domain depth, then compute the backward second-harmonic and SPDC efficiencies from that true domain profile and compare with the measured values. If the measured backward signal matches a model in which the inverted region has no overlap with the guided mode, or if preventing any forward-generated light from returning (e.g., by placing an absorber at the far end of the waveguide) leaves the backward count unchanged, the intrinsic-backward-generation claim is disproven.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates backward-wave SPDC in bulk periodically poled crystals; the paper compares its counter-propagating brightness to this prior result."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports counter-propagating photon-pair generation in a nonlinear waveguide, providing the integrated baseline that this work advances to first-order x-cut poling."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Achieves first-order counter-propagating SHG in z-cut TFLN, the efficiency benchmark the paper compares its x-cut result against."},{"cited_title":"Yakar, E","cited_arxiv_id":null,"evidence_quote":"Reports backward SHG in silicon nitride via optically induced quasi-phase-matching, the only prior integrated backward-generation result, at three orders of magnitude lower efficiency."},{"cited_title":"Ayhan, M","cited_arxiv_id":null,"evidence_quote":"Documents the lateral domain merging that limits sub-micron poling in TFLN, the fabrication obstacle the sidewall method is designed to overcome."},{"cited_title":"Kellner, A","cited_arxiv_id":null,"evidence_quote":"The authors' preceding demonstration of a counter-propagating SPDC source in LNOI, the direct precursor of the devices reported here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the wafer-scale periodic poling process for conventional micrometre periods, defining the state of the art that the paper pushes to the nanoscale."}],"review_version":1}