{"id":"e81d0d3d-3bba-4c1b-9dc8-6481bf595997","arxiv_id":"2501.12825","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"An adjoint topology-optimized SiN structure is predicted by FDTD simulations to give about 107% helical dichroism between +3 and -3 OAM beams at 800 nm, the highest reported HD value in simulation.","lead":"Researchers used an automated optimization routine to design a microscopic silicon-nitride pattern that reflects light carrying opposite 'twists' (orbital angular momentum) very differently, reaching a simulated helical dichroism of about 107%. The work is a design demonstration only, with no fabricated device, but it shows inverse design can push chiroptical contrasts far beyond earlier intuitive structures.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The adjoint update in Eq. (4) is not the gradient of the HD metric in Eq. (1), so the paper's central claim that the structure was inverse-designed to maximize HD is unsupported as written.","rationale":"The reader's concern about reflectance versus absorbance is real but secondary: the paper operationally defines HD from reflectance in Eq. (1), and the final number follows that definition, so the abstract wording can be corrected without invalidating the numerical claim. The Eq. (4) gradient error is more load-bearing because it attacks the claimed design mechanism itself. If the printed gradient is wrong, the paper does not demonstrate inverse design for HD, even though the final structure may still exhibit a large reflectance contrast. I still do not recommend rejection: the MEEP/Lumerical cross-check is a genuine independent check that the final structure's reflectance values are reproducible, and the gradient error may be a typographical slip that a corrected run or a released design file/code would resolve. The verdict stays conditional, with the conditions now including a corrected Eq. (4), an explicit statement of whether the implementation matches it, and a transmission/absorption analysis to reconcile the 'differential absorbance' wording.","tokens_in":7312,"tokens_out":12115,"duration_ms":128550,"concrete_test":"Independently compute dHD from Eq. (1) and compare with Eq. (4). Then, using the same forward and adjoint fields, run one optimization update from the reported converged design (or from the initial design for ~160 iterations) with the exact direction -R_- ∇R_+ + R_+ ∇R_- instead of Eq. (4). If the update directions differ by more than a positive scalar, or if the re-optimized design does not reproduce R_+ ≈ 6% and R_- ≈ 20% at |ℓ| = 3, the manuscript's optimization protocol is not the one that produced the claimed ~107% HD, and the method and results require revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central methodological claim is that the adjoint loop maximizes the paper's own HD metric. Eq. (4) breaks this. Differentiating HD = 200(R_- - R_+)/(R_- + R_+) gives dHD = [400/(R_- + R_+)^2](-R_- dR_+ + R_+ dR_-), so a gradient-ascent direction is -R_- ∇R_+ + R_+ ∇R_-. Eq. (4) instead uses w1 = -R_+/S and w2 = R_-/S (S = R_+ + R_-), i.e. -R_+ ∇R_+ + R_- ∇R_-. The coefficients weighting ∇R_+ and ∇R_- are swapped relative to the exact gradient, so the algorithm as written is not performing gradient ascent on HD. The final 107% reflectance contrast is then not evidence for the claimed HD-maximizing inverse design; it may be an incidental property of a freeform structure. No code or design file is provided to show that the implementation used the correct gradient despite the printed Eq. (4). A secondary issue: the abstract and Fig. 1 call HD 'differential absorbance' and attribute the weaker reflection to absorption, but SiN is stated to be nearly lossless and no transmission, scattering, or absorbed-power data are reported; the computed quantity is at most helicity-dependent reflectance, not demonstrated absorbance.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a topology-optimization approach, based on adjoint sensitivity analysis and FDTD simulations, to design a Si3N4 chiral nanostructure that maximizes helical dichroism (HD) between Laguerre-Gaussian beams carrying opposite orbital angular momentum topological charges. The optimized freeform structure is reported to exhibit a reflectance-based HD of about 107% for |ℓ|=3 at 800 nm, with R_- ≈ 20% and R_+ ≈ 6%, and the trend is reproduced with two independent FDTD solvers (MEEP and Lumerical). The authors also show an ℓ-scan from 0 to 10, demonstrating a peak at the target |ℓ|=3 and non-negligible HD at other charges.","tokens_in":7606,"tokens_out":2546,"duration_ms":27358,"significance":"If the central claims are correct, this would be a notable demonstration that adjoint-based inverse design can produce subwavelength dielectric structures with much larger OAM-dependent reflectance contrast than previous intuitive chiral geometries, and the use of two independent FDTD implementations is a genuine strength. The fabrication-conscious constraints (200 nm minimum feature size, binarization, C2 symmetry) also make the predicted structure more plausible as a real device. However, the load-bearing derivation of the adjoint update is incorrect as printed, and the manuscript does not establish that the computed reflectance contrast corresponds to differential absorption, despite the abstract and Fig. 1 using absorbance language. These two issues must be resolved before the quantitative claims can be accepted.","major_comments":[{"comment":"The weights in Eq. (4) do not form the gradient of the HD metric in Eq. (1). Differentiating HD = 200(R_- - R_+)/(R_- + R_+) with respect to the design variables gives ∇HD = [400/(R_-+R_+)^2](R_+ ∇R_- - R_- ∇R_+). A gradient-ascent direction is therefore proportional to R_+ ∇R_- - R_- ∇R_+, i.e., the coefficient of ∇R_+ should involve R_- and the coefficient of ∇R_- should involve R_+. Equation (4) instead uses w1 = -R_+/(R_-+R_+) for ∇R_+ and w2 = R_-/(R_-+R_+) for ∇R_-, which is the swapped weighting. As written, the optimization is not maximizing the HD defined in Eq. (1); it is maximizing a different objective. This directly affects the central claim that the structure was inverse-designed to maximize HD. The authors must either correct the derivation and weights, or explicitly state and derive the actual objective being optimized and reconcile it with the reported HD maximum.","section":"Eq. (4) and Fig. 2"},{"comment":"The manuscript describes HD as 'differential absorbance' and Fig. 1 attributes the weaker reflection for ℓ=+3 to stronger light absorption, but the only computed quantity is reflectance (Eq. (3)). No transmission, scattering, or absorbed-power data are reported, and the text states that SiN has 'nearly zero absorption' at the operating wavelength. A lower reflectance for one helicity could equally arise from higher transmission or scattering into other channels, and with a nearly lossless material it is questionable to interpret the reflectance contrast as absorption. The authors should report the full power budget (R + T + scattered power, or directly computed absorbed power) and either substantiate the absorbance claim or consistently describe the result as helicity-dependent reflectance contrast.","section":"Abstract, Fig. 1, and Eq. (3)"}],"minor_comments":[{"comment":"Reference 30 is listed twice with different entries (Longman and Fedosejevs; Hammond et al.), which disrupts the numbering and makes it difficult to map citations in the text to the bibliography.","section":"References"},{"comment":"The text states that the HD defined by Eq. (1) has a value range of 0 to 200%, but the expression can be negative if R_- < R_+; the sign convention and the intended range should be clarified.","section":"Eq. (1) and surrounding text"},{"comment":"The agreement between MEEP and Lumerical is shown only qualitatively; reporting numerical differences or a quantitative error metric between the two solvers would strengthen the claim of cross-platform robustness.","section":"Fig. 4(d)-(e)"},{"comment":"The phrase 'first-ever inverse-designed chiral structure' for HD enhancement is stronger than the literature survey supports; the authors should temper this to 'first, to our knowledge' and ensure the novelty claim is consistent with the cited prior work.","section":"Conclusion"}],"recommendation":"major_revision","confidential_remarks":"The Eq. (4) issue is central: if the implementation actually used the correct gradient, the printed derivation still needs correction; if it used the printed weights, the inverse-design claim is unsupported. The reflectance-versus-absorption gap is also important because the abstract and figure language overstates what is computed. These are fixable in revision, hence major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuine first—adjoint topology optimization aimed at helical dichroism—and the two-FDTD cross-check is a real strength. But the printed gradient in Eq. (4) is not the gradient of the HD metric, and the 'differential absorbance' framing overreaches what reflection-only simulations show.\n\nWhat's new: No previous work in the cited literature applies inverse design specifically to maximize HD under OAM beams. The optimized freeform Si3N4 structure is new, and the l-scan from |ℓ|=0 to 10 shows a peak at 3 while retaining ~30% HD at |ℓ|=7, which is a nice robustness check. Agreement between MEEP and Lumerical gives confidence that the 107% figure is not a single-tool artifact.\n\nThe main technical problem is Eq. (4). Differentiating HD = 200(R_- − R_+)/(R_- + R_+) gives a gradient direction proportional to R_+ ∇R_- − R_- ∇R_+. The paper instead weights the two reflectance gradients as −R_+ ∇R_+ + R_- ∇R_-, i.e., the coefficients on ∇R_+ and ∇R_- are swapped. So the algorithm as written is not performing gradient ascent on the paper's own HD metric. The structure may have been optimized with the correct gradient in the actual code, but no code or design file is provided to confirm that. This is load-bearing because the central claim is that the structure was inverse-designed to maximize HD.\n\nSecond, the abstract and Fig. 1 describe HD as 'differential absorbance' and attribute the weaker reflection to absorption, but SiN is nearly lossless and the paper reports no transmission, scattering, or absorbed-power data. The computed quantity is helicity-dependent reflectance, not demonstrated absorbance. That language should be corrected or supplemented.\n\nThird, the 107% peak occurs at the exact |ℓ| the design was optimized for, so it is partly a measure of the optimizer's success rather than an independent prediction. The l-scan partially mitigates this, but calling it 'record-breaking' without experimental or independent confirmation is too strong.\n\nThe simulation result is plausible and cross-checked, but the methodological claim as written is unsupported. The fix is straightforward: correct the gradient derivation, share the design file or code, and either measure the device or clearly label the result as a simulation prediction. I would send this to peer review—a good referee will catch the gradient issue, and the authors can address it in revision.","headline":"A genuine first application of adjoint inverse design to helical dichroism, with a solid two-FDTD cross-check, but the printed gradient in Eq. (4) is swapped and the absorbance framing overreaches.","tokens_in":8172,"tokens_out":4785,"would_cite":false,"duration_ms":42100,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"An inverse-designed chiral structure reaches ~107% helical dichroism for vortex beams at 800 nm.","keywords":["helical dichroism","orbital angular momentum","inverse design","topology optimization","adjoint method","chiral nanostructure","FDTD simulation","silicon nitride"],"falsifier":"Place power monitors for transmission and scattering in the same FDTD geometry and compute the absorbed power directly from the Poynting vector divergence, or from incident-minus-reflected-minus-transmitted power, for $\\ell=+3$ and $\\ell=-3$. If the absorption contrast is much smaller than the reflectance contrast, the $107\\%$ HD claim would not represent differential absorption. Alternatively, an experimental measurement of both reflected and transmitted powers of the two vortex beams on a fabricated sample would settle the question.","tokens_in":7082,"feed_emoji":"🌀","tokens_out":8338,"duration_ms":76077,"temperature":0.7,"pith_summary":"Helical dichroism (HD) is the differential response of a structure to light beams that carry opposite orbital angular momentum (OAM), and it is a candidate tool for sensing chirality without relying on circular polarization. This paper claims that a topology-optimized silicon nitride nanostructure can be made to distinguish OAM beams with topological charges $+3$ and $-3$ so strongly that its reflectance-based HD reaches roughly $107\\%$ at 800 nm, with the negative-charge beam reflected near $20\\%$ and the positive-charge beam near $6\\%$. The design is obtained by adjoint-based inverse design rather than by intuitive spiral or helical geometry, and it is verified in two independent FDTD implementations. If the claim holds, it would mean that freeform dielectric nanostructures can be custom-tuned for helicity-selective interactions, opening a route to more sensitive chiral sensing and spectroscopy.","feed_headline":"Inverse-designed chiral film hits 107% helical dichroism","feed_subtitle":"A topology-optimized silicon nitride film separates +3 from -3 vortex beams, reflecting 20% versus 6% at 800 nm.","key_machinery":"The engine of the design is adjoint-based topology optimization with a weighted two-channel figure of merit. At each iteration, forward and adjoint FDTD simulations are run for the $+3$ and $-3$ Laguerre-Gaussian modes, and the gradient of each reflectance $R_{\\ell^\\pm}$, defined in Eq. (3) as the ratio of reflected to incident power flux, is computed from the matrix product of forward and adjoint fields. The two gradients are combined with weights $w_1 = -R_{\\ell^+}/(R_{\\ell^+}+R_{\\ell^-})$ and $w_2 = R_{\\ell^-}/(R_{\\ell^+}+R_{\\ell^-})$, so that the optimizer simultaneously suppresses $R_{\\ell^+}$ and raises $R_{\\ell^-}$, i.e., maximizes the HD numerator $R_{\\ell^-}-R_{\\ell^+}$. A fabrication constraint enforces a 200 nm minimum feature size and a $C_2$ (two-fold rotational) symmetry, and the final binarized Si$_3$N$_4$ geometry is evaluated with both MEEP and Lumerical FDTD solvers. The Laguerre-Gaussian source from Eq. (2) supplies the helical phase front $e^{-i\\ell\\phi}$ that gives the beams their orbital angular momentum.","core_discovery":"The paper's central claim is that inverse design can create a chiral structure with a helical dichroism response far larger than the intuitive chiral geometries previously studied. Specifically, the optimized Si$_3$N$_4$ freeform structure, designed to maximize the reflectance difference between incident Laguerre-Gaussian beams with topological charges $+3$ and $-3$, exhibits a peak HD of approximately $107\\%$ at $\\lambda=800$ nm, computed from the reflected power contrast $R_{\\ell^-}\\approx 20\\%$ versus $R_{\\ell^+}\\approx 6\\%$. The same structure retains substantial HD at other topological charges, about $75\\%$ for $|\\ell|=4$ and about $30\\%$ for $|\\ell|=7$, so the response is not confined to the exact design mode. The authors take this as evidence that adjoint topology optimization is a viable strategy for engineering OAM-selective chiroptical responses, where no intuitive design rule previously existed.","pith_inferences":["If absorption, rather than reflection/transmission partitioning, is later confirmed as the origin of the contrast, then the unbounded topological charge $\\ell$ could be exploited as a spectroscopic axis: a family of inverse-designed structures, each tuned to a different $\\ell$, would map a sample's chiral response across OAM orders in a way that circular polarization cannot.","The persistence of HD at $|\\ell|=4$ and $7$ hints that the freeform structure acts as a broadband helicity filter rather than a mode-matched resonator; measuring its response to fractional or superimposed OAM modes would test whether the mechanism is genuinely topological or a scalar overlap effect.","The $107\\%$ number is reflectance-based, so a full electromagnetic energy balance would reveal whether the contrast is due to absorption or to asymmetric scattering; that distinction determines whether the device is best used as a chiral absorber, a chiral reflector, or a chiral scatterer."],"forward_implications":["Topology optimization can be targeted at a specific OAM topological charge, producing a structure whose peak HD occurs at the design value $|\\ell|=3$ and degrades smoothly away from it.","The optimized structure is a CMOS-compatible dielectric (Si$_3$N$_4$) with minimum features of 200 nm and two-fold rotational symmetry, so it is compatible with e-beam lithography and free of metal quenching.","The HD value of roughly $107\\%$ at $\\lambda=800$ nm is about three times the previous reported value of $50\\%$ for intuitive chiral structures, suggesting inverse design reaches a different regime of chiroptical response.","The structure's response remains strong at nearby topological charges, about $75\\%$ at $|\\ell|=4$ and about $30\\%$ at $|\\ell|=7$, indicating a broadband OAM-selective behavior rather than a sharp resonance.","Because $R_{\\ell^-}$ and $R_{\\ell^+}$ are validated in two independent FDTD solvers (MEEP and Lumerical), the predicted contrast is not an artifact of a single simulation setup."],"supporting_citations":[{"why":"This reference defines the helical dichroism formula used in Eq. (1) and provides the earlier 20% HD measurement that the new design outperforms.","marker":"[21]"},{"why":"This reference reports the preceding 50% HD result on additively manufactured copper helices, which is the baseline for the claimed improvement.","marker":"[22]"},{"why":"This reference supplies the adjoint shape-optimization technique on which the inverse-design loop is built.","marker":"[24]"},{"why":"This reference provides the adjoint sensitivity foundation, including the Born approximation and Lorentz reciprocity, used to compute gradients.","marker":"[25]"},{"why":"This reference is MEEP, the open-source FDTD solver used for both optimization iterations and final reflectance evaluation.","marker":"[29]"},{"why":"This reference gives the Laguerre-Gaussian beam expression used to build the OAM sources with the desired topological charges.","marker":"[30]"}],"fun_headline_variants":["Inverse-designed chiral film attains 107% helical dichroism","Adjoint topology optimization enables 107% helical dichroism","107% helical dichroism in inverse-designed chiral film","Inverse design yields chiral film with 107% helical dichroism","Chiral film optimized to 107% helical dichroism via inverse design"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the computed reflectance difference is a faithful measure of differential absorption of the two OAM beams: if the structure transmits or scatters the $+3$ and $-3$ beams differently, the reported reflectance-based HD could overstate the material's chiral absorption contrast.","fun_headline_variants_meta":{"raw":{"variants":["Inverse-designed chiral film attains 107% helical dichroism","Adjoint topology optimization enables 107% helical dichroism","107% helical dichroism in inverse-designed chiral film","Inverse design yields chiral film with 107% helical dichroism","Chiral film optimized to 107% helical dichroism via inverse design"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001527,"raw_usage":{"total_tokens":6102,"prompt_tokens":924,"completion_tokens":5178,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":540,"completion_tokens_details":{"reasoning_tokens":5086}},"tokens_in":540,"tokens_out":5178,"duration_ms":33660,"temperature":1.0,"reasoning_tokens":5086,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T16:44:48.374234+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Place power monitors for transmission and scattering in the same FDTD geometry and compute the absorbed power directly from the Poynting vector divergence, or from incident-minus-reflected-minus-transmitted power, for $\\ell=+3$ and $\\ell=-3$. If the absorption contrast is much smaller than the reflectance contrast, the $107\\%$ HD claim would not represent differential absorption. Alternatively, an experimental measurement of both reflected and transmitted powers of the two vortex beams on a fabricated sample would settle the question.","supporting_citations":[{"cited_title":"ACS nano, 2021","cited_arxiv_id":null,"evidence_quote":"This reference defines the helical dichroism formula used in Eq. (1) and provides the earlier 20% HD measurement that the new design outperforms."},{"cited_title":"ACS nano, 2023","cited_arxiv_id":null,"evidence_quote":"This reference reports the preceding 50% HD result on additively manufactured copper helices, which is the baseline for the claimed improvement."},{"cited_title":"Optics express, 2013","cited_arxiv_id":null,"evidence_quote":"This reference supplies the adjoint shape-optimization technique on which the inverse-design loop is built."},{"cited_title":"2012: University of California, Berkeley","cited_arxiv_id":null,"evidence_quote":"This reference provides the adjoint sensitivity foundation, including the Born approximation and Lorentz reciprocity, used to compute gradients."},{"cited_title":"Computer Physics Communications, 2010","cited_arxiv_id":null,"evidence_quote":"This reference is MEEP, the open-source FDTD solver used for both optimization iterations and final reflectance evaluation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"This reference gives the Laguerre-Gaussian beam expression used to build the OAM sources with the desired topological charges."}],"review_version":1}