{"id":"2ed596c7-0c84-4953-beec-1aa98a98d45c","arxiv_id":"2501.10340","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Measured diffraction from sinusoidal and double-sinusoidal silver optical Fourier surfaces matches FDTD simulations quantitatively, while scalar diffraction models deviate by up to about 30 percent for deep structures and large angles.","lead":"This paper measures how light diffracts from precisely patterned wavy metal surfaces called optical Fourier surfaces, then compares the measurements with two types of theory. The experiments match full computer simulations of the light, but not the simpler analytical formulas many researchers use, especially for deep structures and large angles.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"PPA-to-Ag amplitude transfer is validated on only one sample and one parameter; FDTD agreement mitigates but does not fully close the gap.","rationale":"The reader identified the PPA-to-Ag transfer as the weakest assumption, and this is indeed the most load-bearing premise: both halves of the central claim depend on the surface profile input to FDTD and scalar models. However, the concern is partially mitigated by the paper's own evidence: FDTD simulations using PPA-derived amplitudes match experiment across 30 single-sinusoidal samples and 12 double-sinusoidal samples over a broad wavelength range. If template stripping had induced a significant systematic amplitude error, this agreement would almost certainly degrade, especially in the amplitude-dependent efficiency maps of Figure 4. The remaining gap is that the direct validation is limited to a period comparison on one sample, so the paper does not fully close the loop on amplitude and shape fidelity. This does not overturn the central claim, but it does justify the reader's conditional verdict rather than a full accept based on the current evidence. My proposed AFM-based check would settle the question directly. I therefore keep the verdict unchanged.","tokens_in":21672,"tokens_out":8933,"duration_ms":98568,"concrete_test":"Perform AFM topography on at least one deep single-sinusoidal Ag OFS (A approximately 125-150 nm) and one double-sinusoidal Ag OFS (phase approximately 0 or pi) over at least a 20 um by 20 um area (or several 10 um by 10 um regions), fit the designed sinusoids, and compare the fitted Ag amplitudes, phases, and RMSE with the TSPL PPA data used in Figures 2d and 2i. If the Ag amplitudes deviate from the PPA amplitudes by more than the roughly 5-10 percent experimental error budget, rerun the FDTD and Model 3 calculations with the Ag-measured amplitudes and re-evaluate Figures 4e/5e and 4f/5f; the central claim stands only if experiment-FDTD differences remain small while scalar-experiment differences remain large.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing premise is that the PPA topography recorded in-situ by the TSPL tool equals the final Ag OFS topography. The Methods section explicitly states this ('The topography of the PPA structures was taken directly to represent that of the corresponding Ag structures'), and all FDTD and scalar-model inputs use fitted amplitudes and phases from these PPA data. The only direct validation is an AFM measurement of the central 10 um by 10 um region of one single-sinusoidal Ag OFS, and the reported comparison is a ~1% difference in period; no amplitude, shape, or full-area comparison is given. If template stripping systematically altered the amplitude (e.g., by partial filling or deformation during evaporation or stripping), every model curve in Figures 4 and 5 would be evaluated at the wrong depth, potentially shifting both the FDTD agreement and the scalar-model disagreement. This concern is mitigated by the fact that FDTD matches experiment across a wide range of amplitudes, which would be unlikely if the PPA amplitude were systematically wrong; nevertheless, the paper's 'quantitative agreement' claim is formally anchored by a single-parameter, single-sample validation. The risk is moderate, not fatal.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a systematic comparison of measured diffraction efficiencies from reflective silver optical Fourier surfaces (OFSs) with FDTD simulations and four analytic scalar diffraction models. Thirty single-sinusoidal gratings (period 1 µm, fitted amplitudes 5–150 nm) and twelve double-sinusoidal gratings (periods 1 µm and 0.5 µm, amplitudes 90/45 nm, relative phase 0–2π) were fabricated by thermal scanning-probe lithography and template stripping. Diffraction efficiencies for the 0th, ±1st, and (indirectly) 2nd orders were acquired over 450–745 nm for s-polarized light. The central claim is that the measurements agree with FDTD within a few percent over most of parameter space, whereas scalar models deviate by up to roughly 30% for deep structures and long wavelengths. The paper also reports improved fabrication area, depth, and yield using PPA resist.","tokens_in":21791,"tokens_out":8300,"duration_ms":85896,"significance":"If the central claim holds, this is a valuable demonstration that OFSs are a precise experimental platform for Fourier-spectrum engineering and a rare, clean benchmark for the validity limits of scalar diffraction theory. The study's main strengths are its density (30+12 structures, wavelength sweeps, difference maps), the absence of fitted parameters in the scalar models, and the use of full electrodynamic simulations with an independently measured Ag permittivity rather than diffraction data. These features make the broad trends credible. However, the precise wording 'agree quantitatively' is currently supported by difference maps without per-point uncertainty estimates and by a topographic-transfer validation that covers only one sample and one parameter; these gaps need to be closed before the quantitative claim is fully established.","major_comments":[{"comment":"The load-bearing premise of the quantitative comparison is that the PPA topography recorded by TSPL is identical to the final Ag topography. The Methods states this directly ('The topography of the PPA structures was taken directly to represent that of the corresponding Ag structures'), and the fitted amplitudes and phases from Figures 2d and 2i are used as inputs for every FDTD and scalar-model curve in Figures 4 and 5. The only direct validation is an AFM measurement of the central 10 µm × 10 µm region of one single-sinusoidal Ag OFS, and the reported comparison is a ~1% difference in period; no amplitude, profile-shape, or full-area validation is given. A systematic amplitude change during evaporation or template stripping would bias all model/simulation comparisons, and while the broad FDTD agreement across many amplitudes makes a large systematic error unlikely, it does not quantify the remaining uncertainty. Please add AFM-based amplitude and roughness comparisons for multiple samples spanning the amplitude and phase ranges used, or otherwise demonstrate that the PPA-to-Ag transfer does not alter depths.","section":"Methods: Characterization of Surface Topography"},{"comment":"The paper repeatedly states agreement to within ~5% or discrepancies up to ~15%, but no per-point error bars, confidence intervals, or repeatability data are given. The only precision estimates are the measured +1/−1 asymmetry in Figure 4d, which reaches ~5%, and the amplitude and phase fit accuracies reported in the topography section. Because the scalar-model residuals in Figures 4f and 5f are of order 10–30%, the reader cannot determine whether the simulation-experiment residuals are within measurement uncertainty or whether particular deviations (e.g., near the Rayleigh anomaly or for the deepest structures) are real. Please provide a full uncertainty budget (laser power fluctuation, camera dark noise, reference normalization, integration-window choice, wavelength calibration, and sample-to-sample repeatability) and either plot error bars or mark confidence regions in the difference maps; this is necessary to support the abstract's 'agree quantitatively' claim.","section":"Methods: Evaluation and Analysis of Diffraction Efficiencies and Figures 4–5"}],"minor_comments":[{"comment":"The efficiency formula in Methods is presented as an unnumbered display equation; please number it and define C_s, C_ref, P_s, and P_ref in the preceding sentence.","section":"Methods"},{"comment":"The green vertical lines and pink integration regions are described in the caption but are difficult to identify in the reproduced panels; please make them more prominent or add labels in the figure itself.","section":"Figure 3"},{"comment":"The color scales in the difference maps use positive and negative deviations on the same color axis; a diverging colormap with an explicit zero contour would make the sign and magnitude of the deviations easier to read.","section":"Figures 4 and 5"},{"comment":"The caption of Figure S7 explains that the 2nd-order efficiency is determined indirectly from uncollected photons; because this quantity is used as complementary evidence when discussing the Rayleigh anomaly, please state the same caveat in the main text as well.","section":"Figure S7 and main text discussion of the Rayleigh anomaly"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal, and the concerns are fixable. The PPA-to-Ag transfer validation and the missing error bars are the reasons for major revision; I do not see circularity in the modeling, since no diffraction data are used to fit parameters. The overlapping authorship with Ref. 51 is worth noting, but the use of that Ag permittivity is appropriate. I expect the paper to be publishable after the revision if the requested data are added."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I have read the paper and the stress-test note. Here is my take.\n\nWhat is actually new: this is the first quantitative comparison of measured diffraction efficiencies from single- and double-sinusoidal reflective OFSs against FDTD simulations and scalar models. The dataset is dense—30 single-sinusoidal and 12 double-sinusoidal structures, wavelength sweeps 450–745 nm, and difference maps. The central finding is convincing: experiment and FDTD agree within ~5% over most of parameter space, while scalar Model 3 diverges by 10–30% in deep/high-angle regimes. That is a useful quantitative error map for the diffractive optics community, and it supports the OFS platform claim. The fabrication advance (PPA resist, 4x area, 50% deeper, higher yields) is a real, concrete improvement.\n\nSoft spots, in proportion. The stress-test concern about PPA-to-Ag amplitude transfer is real but moderate. The Methods state that PPA topography is taken to represent the Ag, and the validation is AFM of the central 10 um x 10 um of one single-sinusoidal sample, comparing only the period (~1% difference), not amplitude or full shape. So formally, the amplitude input to every FDTD/scalar curve rests on a single-sample check. However, the FDTD agreement across the full range of amplitudes up to 150 nm is strong circumstantial evidence that the amplitudes are not systematically wrong; if they were, the match would break in a depth-dependent way. So I would not call this fatal, but I would ask the authors to provide amplitude comparison on at least a few structures, and to state explicitly what was measured. The other soft spots are minor: no per-point error bars (though the paper gives a ~10% estimated error for the deepest samples and discusses chromatic aberration), and no data/code release. The SI is thorough; the four scalar models are clearly derived and the choice of Model 3 as representative is justified. The citation pattern is fine; using McPeak permittivity from the same group is normal practice.\n\nBottom line: this is a solid, careful experimental study. The central claim stands. It deserves a serious referee, and I would accept it for peer review, with the PPA-to-Ag validation as the main point to address. I would cite it if I worked on diffractive surfaces.","headline":"A careful, dense experimental study that delivers the first quantitative FDTD validation of OFSs and a useful error map for scalar models, with a modest but real caveat about PPA-to-Ag topography transfer.","tokens_in":22423,"tokens_out":2150,"would_cite":true,"duration_ms":20420,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["42.25.Fx","42.79.Dj"],"model":"deepseek-v4-flash","headline":"Measured diffraction from sinusoidally pure silver surfaces matches full electrodynamic simulations, while scalar models diverge for deep gratings and large angles.","keywords":["optical Fourier surfaces","sinusoidal phase gratings","scalar diffraction theory","FDTD simulations","thermal scanning-probe lithography","Rayleigh anomaly","diffraction efficiency","template stripping"],"falsifier":"Measure the full 40 µm × 40 µm silver topography of the deepest single-sinusoid and double-sinusoid samples with an independent AFM or interferometer and compare fitted amplitudes and phases to the PPA-derived values used in this paper; if the stripping step changes amplitude or phase by more than the stated ~1–2%, the quantitative agreement with electrodynamic simulations would need to be reassessed.","tokens_in":21423,"feed_emoji":"🌊","tokens_out":5582,"duration_ms":53863,"temperature":0.7,"pith_summary":"This paper claims that 'optical Fourier surfaces'—silver mirrors patterned with pure sinusoidal height profiles—diffract light exactly as full electrodynamic simulations predict, but not as common analytical scalar diffraction models predict. That claim matters because single- and double-sinusoidal profiles are the most fundamental phase gratings, so they provide a clean platform for designing diffractive optics by Fourier-spectrum engineering. The authors fabricate 30 single-sinusoid and 12 double-sinusoid surfaces with thermal scanning-probe lithography and template stripping, measure wavelength-dependent diffraction efficiencies, and compare them with simulations and four scalar models. If correct, the work both validates a manufacturing route to precise wavy surfaces and draws a quantitative boundary for when simple scalar formulas can be trusted.","feed_headline":"Sinusoidal mirrors match full simulations, not scalar models","feed_subtitle":"Precisely wavy silver surfaces confirm where simple diffraction formulas break down.","key_machinery":"The central object is the optical Fourier surface (OFS), a phase grating whose height profile contains only the spatial frequencies intended in the design: for a single sinusoid $h(x)=A\\sin(gx)$ with $g=2\\pi/\\Lambda$, and for a double sinusoid $h(x)=A_1\\sin(g_1x)+A_2\\sin(2g_1x-\\varphi)$. The surface converts to a phase modulation through the optical path difference, $OPD=-\\gamma h(x,y)$, with $\\gamma=2$ in the paraxial approximation and $\\gamma=\\cos\\theta_i+\\cos\\theta_m$ in the non-paraxial treatment. Diffraction-order amplitudes follow from the Jacobi–Anger expansion of the grating transparency, giving Bessel-function coefficients $J_m(A_b)$ with $A_b=(2\\pi/\\lambda)A\\gamma$; the four scalar models differ in whether $\\gamma$ is paraxial or non-paraxial and whether evanescent orders are renormalized away. The comparison is made quantitative by feeding the measured fitted amplitudes of every structure into both the electrodynamic simulation and the scalar models, so the experimental match isolates the physics of the diffraction process rather than the fabrication design.","core_discovery":"The paper establishes that reflective optical Fourier surfaces—silver gratings whose profiles are exactly one sinusoid or two superposed sinusoids—diffract light in quantitative agreement with full electrodynamic simulations, while analytical scalar diffraction models fail for deep profiles and large angles. For the single-sinusoid series (period 1000 nm, amplitudes 5–150 nm) the measured total first-order efficiency reproduces the simulated wavelength and depth trends; discrepancies stay small except near the Rayleigh anomaly and for the deepest structures, where experimental imperfections of about 10% are estimated. For the double-sinusoid series the measured +1st-order efficiency matches simulation within about 5% over most of the phase–wavelength map, whereas the best scalar model deviates by up to about 30%. The authors conclude that the measured optical response of OFSs is accurately predicted by electrodynamics and that commonly used scalar formulas should be used with caution outside shallow, small-angle regimes.","pith_inferences":["Beyond the paper: if the PPA-to-silver transfer is as faithful as assumed, the same platform should support arbitrary superpositions of Fourier components, making full-wave simulation the practical design tool for complex diffractive devices.","Beyond the paper: the growing discrepancy with the number of sinusoids suggests that scalar-model error is cumulative; holograms and optical neural networks with many Fourier components may need error maps such as Figure 4f to correct designs or full electrodynamic simulation.","Beyond the paper: the phase-asymmetry discrepancy around $\\varphi=\\pi$ hints at shadowing or multiple-scattering effects that scalar phase-only models cannot capture; a test would be to measure the same double-sinusoid series at oblique incidence or with p-polarized light and compare the asymmetry ratio to simulation."],"forward_implications":["Sinusoidal OFSs can serve as quantitative references for Fourier-spectrum engineering: measured diffraction efficiencies across wavelength and depth match full electrodynamic simulations, so a designed Fourier content can be translated into a measured response.","Scalar diffraction models, even the best tested non-paraxial model with renormalization, should not be trusted for deep profiles or large diffraction angles; deviations up to about 30% are mapped as a function of amplitude and wavelength.","Adding one extra sinusoid with controlled relative phase produces strongly asymmetric diffraction, with up to 18–24 times more light in one first order than the other, tunable by phase.","Near the Rayleigh anomaly, simulation–experiment discrepancies trace mainly to finite-size and experimental effects rather than to the OFS fabrication itself, since the simulations assume an infinite grating."],"supporting_citations":[{"why":"Introduces optical Fourier surfaces fabricated by thermal scanning-probe lithography, establishing the platform this work extends to larger, deeper, and quantitatively tested structures.","marker":"[41]"},{"why":"Supplies the scalar diffraction theory and Fourier optics framework from which all four analytical models are derived.","marker":"[22]"},{"why":"Provides the grating behavior treatment, including conical diffraction and Rayleigh anomaly renormalization, that underlies the best-performing scalar model.","marker":"[23]"},{"why":"Gives the parametric sinusoidal reflection grating model that is the direct basis for the analytical efficiency formulas used in the comparison.","marker":"[32]"},{"why":"Supplies the silver permittivity data used as input for the electrodynamic simulations, so the simulation reproduces the actual metal response.","marker":"[51]"},{"why":"Establishes template stripping as the method for producing ultrasmooth patterned metals, the fabrication step that yields the final silver OFSs.","marker":"[52]"}],"fun_headline_variants":["Wavy mirrors match simulations, not scalar models","Sinusoidal gratings align with full wave simulations","Measured diffraction: electrodynamics beats scalar theory","Optical Fourier surfaces validated by experiment","Deep wavy gratings challenge simple diffraction formulas"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the polymer-template topography measured during lithography is still the true topography of the final silver surface after template stripping; only a small central region of one sample was directly checked.","fun_headline_variants_meta":{"raw":{"variants":["Wavy mirrors match simulations, not scalar models","Sinusoidal gratings align with full wave simulations","Measured diffraction: electrodynamics beats scalar theory","Optical Fourier surfaces validated by experiment","Deep wavy gratings challenge simple diffraction formulas"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000533,"raw_usage":{"total_tokens":2578,"prompt_tokens":976,"completion_tokens":1602,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":592,"completion_tokens_details":{"reasoning_tokens":1542}},"tokens_in":592,"tokens_out":1602,"duration_ms":12781,"temperature":1.0,"reasoning_tokens":1542,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T19:11:12.574637+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the full 40 µm × 40 µm silver topography of the deepest single-sinusoid and double-sinusoid samples with an independent AFM or interferometer and compare fitted amplitudes and phases to the PPA-derived values used in this paper; if the stripping step changes amplitude or phase by more than the stated ~1–2%, the quantitative agreement with electrodynamic simulations would need to be reassessed.","supporting_citations":[],"review_version":1}