{"id":"17f2fd7b-6826-44dc-80d8-f61f6175ac82","arxiv_id":"2501.04933","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":4,"one_line_summary":"A systematic inverse-design method for nonuniform pseudomagnetic fields is demonstrated in silicon photonic crystals, realizing a low-loss S-bend and a 50:50 splitter at 1520-1580 nm with a successful 140 Gb/s PAM-4 data transmission.","lead":"This paper designs and tests silicon photonic crystal devices that bend and split light using synthetic pseudomagnetic fields, and sends a 140 Gb/s signal through them. It offers a design recipe for routing light on a chip without sharp physical turns, which could make future photonic circuits more flexible and compact.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The inverse-design identity ψ ∝ exp(∫m dx/v) is exact only when ∂_y m = 0, so for the curved and splitting profiles the claimed 'arbitrary control' rests on an unquantified adiabatic approximation rather than on the exact relation in Eq. 2.","rationale":"The reader's verdict is CONDITIONAL and I agree; my concern is a sharpened version of the reader's weakest assumption. The paper's strongest claim—arbitrary, systematic control via m = v ∂_x ln ψ—depends on the ansatz ψ ∝ exp(∫ m dx/v) being an eigenstate of the position-dependent Dirac Hamiltonian. A direct substitution shows this is exact only when the mass is independent of the propagation coordinate. For nonuniform m(x,y), the y-derivative of the ansatz produces a position-dependent effective energy, i.e., a residual that couples the zeroth-order mode to massive transverse modes. No quantitative bound is given. This is not a criticism of the experiments; the measured S-bend and splitter performance and the 140 Gb/s transmission are credible evidence that the designed devices work in the demonstrated regime. But the paper's inference from two fabricated devices to 'arbitrary control' is broader than the evidence. A numerical diagonalization test would settle whether the ansatz is self-consistent for the splitter; if it passes, the conditional verdict can stand; if it fails, the design method needs an explicit adiabaticity criterion. Therefore I do not change the reader's CONDITIONAL verdict.","tokens_in":11943,"tokens_out":11449,"duration_ms":122488,"concrete_test":"Reconstruct the exact m(x,y) used for the 50:50 splitter from the inverse-design relation, then numerically diagonalize the 2D Dirac Hamiltonian in Eq. 1 on the same supercell and compare the lowest positive-energy eigenstate with ψ_target = exp(∫ m dx/v), computing both the normalized overlap and the residual norm ||(H − E)ψ_target||/||ψ_target||. If the overlap is below about 90% or the residual exceeds a few percent of the local bandgap, the design identity is not self-consistent and the 'arbitrary control' claim must be downgraded to a curvature-limited adiabatic approximation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central design identity, Eq. 2 and the general form ψ(x,y) ∝ exp(∫^x m(x′,y) dx′/v), is presented as the basis for arbitrary PMF control. Substituting this ansatz into Eq. 1 gives two consistency conditions: ∂_x ln ψ = −m/v (satisfied by construction) and E = −iv ∂_y ln ψ. The second condition can hold for all x only if ∂_y m = 0. For the S-bend Region II, m(x,y) = ax + by, and for the 50:50 splitter, m = v ∂_x ln ψ with a y-dependent two-peaked ψ, ∂_y m is nonzero, so the target field is not an exact eigenstate of Eq. 1. The residual is first order in ∂_y ln ψ and represents coupling to higher transverse (massive) modes; the paper supplies no quantitative adiabaticity bound, curvature limit, or error estimate. Thus the 'arbitrary control' claim rests on an untested slow-variation assumption rather than on the exact identity implied by Eq. 2. The two fabricated devices may well operate inside the adiabatic regime; the problem is that the claimed systematic method is not validated beyond that regime.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a method for designing nonuniform pseudomagnetic fields (PMFs) in silicon photonic crystals at telecommunication wavelengths. The effective mass term m(x,y), induced by locally breaking spatial inversion symmetry in the honeycomb unit cell, is identified with the z-component of a synthetic vector potential, so that light is guided along the line Az=0. For a linear mass profile this reproduces the zeroth-order Landau level and straight-waveguide transport; for nonuniform profiles the authors use the relation ψ ∝ exp(∫ m dx/v) to design an S-bend and a 50:50 power splitter. Both devices are simulated with 3D FDTD, fabricated on SOI, and characterized experimentally, with insertion/excess losses below about 2 dB and splitter imbalance below ±0.5 dB. A 140 Gb/s PAM-4 transmission experiment is reported, and qualitative defect-robustness tests are described.","tokens_in":12193,"tokens_out":7064,"duration_ms":78878,"significance":"If the general design methodology is valid, it would advance PMF photonics from straight Landau-level waveguides toward flexible routing and splitting in integrated photonic circuits, which is a genuine step beyond prior demonstrations of Landau levels and chiral states. The paper's strengths are its concrete experimental validation, good agreement between 3D FDTD and measured transmission spectra, low loss and balanced splitting, and the high-speed data-transmission experiment. The inverse-design relation is used in a constructive way and is not circular, because the final devices are validated by FDTD and measurement. Nevertheless, the central 'arbitrary control' claim currently rests on an unquantified adiabatic approximation, and the evidence covers only two relatively mild nonuniform structures. These issues affect the generality of the method more than the correctness of the two demonstrated devices.","major_comments":[{"comment":"The identity ψ(x,y) ∝ exp(∫^x m(x′,y) dx′/v), used both to describe the Landau-level state and to invert a desired field into a mass profile, is an exact solution of Eq. (1) only when ∂_y m = 0. Substituting this ansatz into Eq. (1) gives, in addition to ∂_x ln ψ = m/v, a consistency condition involving ∂_y ln ψ = (1/v)∫^x ∂_y m dx′; this condition can be satisfied for all x only if ∂_y m vanishes identically. Both nonuniform devices violate this condition: Region II of the S-bend has m = ax + by, and the splitter has m = v ∂_x ln ψ with a y-dependent two-peaked ψ. The residual is first order in ∂_y ln ψ and represents coupling to higher transverse (massive) modes, but the paper supplies no adiabaticity bound, no curvature limit, and no error estimate. Because the universal design method is the paper's central claim, the authors should quantify the validity range of the approximation, for example by projecting the designed state onto eigenstates of the local Hamiltonian or by computing the overlap with higher Landau levels, and should compare those estimates with FDTD for stronger y-gradients and larger curvatures.","section":"§Results, Eq. (2) and inverse-design paragraph"},{"comment":"The phrase 'arbitrary control' is supported by only two device classes, both with mild nonuniformity. The S-bend is a single smooth curve with b/a = 1/10, and the splitter target is a pair of Gaussian wave packets whose separation grows linearly with propagation distance. The paper should state the class of propagation paths and target field distributions for which the inverse-design method is expected to work, and demonstrate at least one case with larger curvature or a non-separable target field. Without such a demonstration, the 'systematic and universal' methodology remains a conjecture rather than an established property of the design procedure.","section":"§Device design and experimental demonstration, Figs. 2–3"}],"minor_comments":[{"comment":"The manuscript contains several typographical and grammatical errors, including 'an universal', 'the flo w of light', and 'synthe sizing'; a careful copyedit is needed.","section":"Throughout"},{"comment":"Equation (2) is rendered with garbled symbols in the submitted text, and the relation between the scalar factor exp(∫ m dx/v), the spinor components, and the k_y dependence is not fully defined. Please provide a clean equation with all variables (v, a, k_y) and state explicitly which component of the spinor is being plotted.","section":"Eq. (2) and surrounding text"},{"comment":"The identification Az = m(r) and the statement kz = 0 for the 2D PhC are made quickly; the gauge convention and the sign relating m to Az should be stated explicitly, since the inverse-design relation changes sign if the opposite valley or gauge is used.","section":"§Results, after Eq. (1)"},{"comment":"The key mapping from the hole side lengths d1 and d2 to the effective mass m is only referenced to supplementary text S2; the main text should include at least the functional form or a table of the calibration so that the transferability of the mapping from band structure to fabricated devices can be assessed by the reader.","section":"§Device design and experimental demonstration"},{"comment":"The robustness claim is qualitative: the reader is told only that 'no severe discrepancies' were observed in supplementary text S7. Please report the defect geometry, the number and position of defects, and the measured transmission differences, ideally compared with a non-topological control structure, so that the claim can be evaluated.","section":"§Results, defect-robustness paragraph"},{"comment":"The assertion that this is 'the first time' a systematic method is proposed for synthesizing PMFs to control light should be tempered, because refs. 29, 41, and 44 already address nonuniform PMFs and Landau-level transport; the novelty claim should be limited to the specific inverse-design procedure demonstrated here.","section":"§Discussion and Introduction"}],"recommendation":"major_revision","confidential_remarks":"The paper is likely to be of interest to the photonics community, and the experimental work appears solid. The main risk is that the headline claim of 'arbitrary control' is broader than what is mathematically established; the authors should add a quantitative analysis of the adiabatic approximation and either soften the claim or provide additional simulations. The supplementary material is heavily relied upon for the Dirac model, the d1–d2-to-mass calibration, and the defect experiments; please ensure it is available to reviewers and contains the requested quantitative details."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nThe one thing to know: this is a credible experimental demonstration of PMF-based routing and splitting in silicon at telecom wavelengths, with a simple inverse-design recipe. The S-bend and 50:50 splitter work, the measured losses and imbalance agree with FDTD, and the 140 Gb/s PAM-4 test below the HD-FEC threshold is a nice practical touch. The Dirac model and Landau-level formalism are standard, and the relation between hole-size asymmetry and the mass term appears properly calibrated. The paper deserves a serious referee.\n\nWhat is genuinely new: using the ψ ∝ exp(∫m dx/v) relation as an inverse-design tool, and validating it with fabricated devices. Previous work (including ref. 29) treated nonuniform PMFs, but the specific silicon demonstrations and the data-transmission test are absent there. The two devices are well chosen as proof-of-concept.\n\nThe soft spots, in order. First, the stress-test concern is legitimate: the relation is exact only when m is independent of y. For the S-bend, m=ax+by with b/a=0.1; for the splitter, m varies in y by construction. The devices evidently operate inside the adiabatic regime, but the paper gives no quantitative adiabaticity bound, no curvature limit, and no error estimate. So the headline claim of 'arbitrary control' is stronger than the evidence. I'd ask for a paragraph that quantifies the regime of validity, perhaps by testing increasingly sharp bends or splitter angles. Second, the paper does not benchmark against ref. 29, which already addresses nonuniform PMFs. The authors should state explicitly what their method adds beyond that work. Third, the robustness claim rests on qualitative comparisons of spectra with and without defects; no statistics or quantitative metrics. For a proof-of-concept, that's acceptable, but it should be labeled as such.\n\nOverall: the central physics holds up, the approximation is not an error in the demonstrated devices, and the overclaims are addressable. I'd send this to review, with the request that the authors quantify the adiabatic approximation and situate the novelty against ref. 29. It's a useful paper for the applied photonics and topological transport communities.","headline":"A solid experimental demonstration of PMF-based S-bend and splitter in silicon, with an overclaimed 'arbitrary control' narrative that needs an adiabaticity bound and a benchmark against ref. 29.","tokens_in":12791,"tokens_out":5145,"would_cite":true,"duration_ms":43232,"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":"A systematic method uses nonuniform pseudomagnetic fields in silicon photonic crystals to bend and split light along designer paths, demonstrated by a low-loss S-bend and a 50:50 power splitter at telecom wavelengths.","keywords":["pseudomagnetic fields","photonic crystals","Landau levels","silicon photonics","optical routing","power splitter","telecommunication wavelength","PAM-4"],"falsifier":"Fabricate a PMF waveguide whose prescribed path has a curvature radius comparable to the lattice constant (or a mass gradient that changes sign over a few unit cells) and measure whether the transmitted beam follows the predicted $A_z=0$ curve; a systematic deviation, beam broadening, or a sharp rise in insertion loss would show the adiabatic zeroth-Landau-level model no longer applies.","tokens_in":11664,"feed_emoji":"💡","tokens_out":5308,"duration_ms":48840,"temperature":0.7,"pith_summary":"This paper proposes a systematic method for designing nonuniform pseudomagnetic fields (PMFs) in silicon photonic crystals, treating the local spatial-inversion-symmetry-breaking strength as an effective mass that acts like a magnetic vector potential. Because the field profile of the zeroth-order Landau level is determined by this mass profile through $\\psi(x,y)\\propto\\exp(\\int m(x,y)\\,dx/v)$, any chosen light path or target field distribution can be translated into a concrete pattern of hole sizes in the crystal. The authors demonstrate the approach with a low-loss S-bend, a highly efficient 50:50 power splitter, and 140 Gb/s PAM-4 data transmission through both devices at telecom wavelengths. If the method holds, it would turn PMFs from a curiosity that produces straight chiral guiding or flat Landau levels into a practical tool for programmable routing and splitting in on-chip optical circuits.","feed_headline":"Pseudomagnetic fields route light through arbitrary curves on a chip","feed_subtitle":"A design recipe turns any desired path into an effective mass profile, demonstrated by a low-loss S-bend and 50:50 splitter.","key_machinery":"The load-bearing object is the position-dependent effective mass term $m(x,y)$ in the Dirac Hamiltonian, realized by the asymmetry of the two triangular holes (sizes $d_1$, $d_2$) in each honeycomb unit cell. The paper identifies $m(x,y)$ with the pseudomagnetic vector potential $A_z$, and derives the governing design identity $\\psi(x,y)\\propto\\exp(\\int m(x,y)\\,dx/v)$, so the desired field profile fixes the mass distribution as $m(x,y)=v\\,d(\\ln\\psi)/dx$. The zeroth-order Landau-level state, whose field is localized on the line $A_z=0$ and propagates along it, carries the routing and splitting; the relation between hole sizes and the mass term translates this abstract mass profile into a concrete photonic-crystal geometry.","core_discovery":"The central claim is that by varying the side lengths of the two triangular holes in every unit cell of a honeycomb photonic crystal, one can synthesize an arbitrary position-dependent mass term $m(x,y)$ in the Dirac Hamiltonian near the K and K' points. This mass term plays the role of the z-component of a magnetic vector potential, and its zero line $A_z=0$ is where the zeroth-order Landau-level wavefunction localizes; the wavefunction's transverse profile is $\\psi(x,y)\\propto\\exp(\\int m(x,y)\\,dx/v)$. Inverting this relation gives $m(x,y)=v\\,d(\\ln\\psi)/dx$, so specifying a desired field distribution determines the mass profile, and hence the hole sizes, everywhere. The paper proves the concept by experimentally demonstrating a low-loss S-bend (insertion loss < 1.83 dB) and a 50:50 power splitter (excess loss < 2.11 dB, imbalance < ±0.5 dB), and by transmitting 140 Gb/s PAM-4 signals with bit error rates below the HD-FEC threshold. The method does not break real time-reversal symmetry, and the transport is topologically protected because intervalley coupling is negligible.","pith_inferences":["Inference: The inverse relation $m=v\\,d(\\ln\\psi)/dx$ suggests a general inverse-design recipe: any separable target field $\\psi(x,y)$ could in principle be synthesized by independently tuning the mass profile in $x$ and $y$, potentially enabling wavefront shaping and mode converters beyond simple waveguide-style paths.","Inference: The adiabatic assumption constrains practical curvature; a bend sharper than those tested should cause coupling to higher Landau levels or intervalley scattering, and mapping that failure threshold would delineate the true design envelope of the method.","Inference: The same mass-to-hole-size mapping could be transferred to other symmetry-broken lattices or to acoustic and mechanical wave systems, extending the design method beyond photonics.","Inference: The demonstrated devices operate in a single pass; combining the mass-profile design with reconfigurable elements (e.g., thermo-optic tuning) could lead to dynamically programmable light paths, a step the paper does not explore."],"forward_implications":["Any continuous light path can be turned into a PMF design by choosing the zero-$A_z$ line, enabling arbitrarily shaped routing in a planar photonic circuit.","Light field distributions can be engineered directly: taking the $x$-derivative of a target $\\psi$ gives the mass profile, allowing splitters with tailored splitting ratios and wavefront shapes.","PMF-based devices can be fabricated in CMOS-compatible silicon-on-insulator at telecom wavelengths with low excess loss, making them relevant for practical photonic integration.","The demonstrated robustness against intentionally introduced defects supports deployment of PMF devices in large-scale photonic circuits.","The successful 140 Gb/s PAM-4 transmission indicates PMF-based components are compatible with high-speed on-chip optical communication.","The method generalizes the PMF concept from straight chiral transport and Landau-level physics to a flexible design principle for functional nanophotonic devices."],"supporting_citations":[{"why":"Establishes that Landau levels and Landau rainbows can be induced in 2D photonic crystals by synthetic strain or local inversion-symmetry breaking, providing the phenomena the paper generalizes from straight chiral transport to arbitrary routing.","marker":"[41-44]"},{"why":"Reports direct observation of Landau levels in silicon photonic crystals, confirming the material platform used here.","marker":"[42]"},{"why":"Reports observation of Landau levels and chiral edge states in photonic crystals via synthetic strain, representing the prior state of the art that this method extends.","marker":"[43]"},{"why":"Supplies the chiral Landau level formalism for 2D Dirac cone systems with inhomogeneous effective mass, including the zeroth-order Landau level dispersion used throughout the paper.","marker":"[44]"},{"why":"Provides the k·p approximation that yields the effective Dirac Hamiltonian with the mass term and the wavefunction relation $\\psi\\propto\\exp(\\int m\\,dx/v)$.","marker":"[49-54]"},{"why":"Establishes the silicon-on-insulator slab platform for topological valley transport, supporting the choice of materials and geometry for the fabricated devices.","marker":"[54]"},{"why":"Represents the spin- and valley-Hall photonic waveguides and cavities that the paper contrasts with its PMF-based approach to arbitrary routing and splitting.","marker":"[45-48]"}],"fun_headline_variants":["Pseudomagnetic fields let light take any path on a chip","Synthetic magnetic fields steer light arbitrarily on silicon chips","Arbitrary light flow via pseudomagnetic fields at telecom wavelengths","Photonic crystals guide light along arbitrary curves with synthetic fields","Arbitrary light paths from photonic pseudomagnetic fields"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The design assumes the zeroth-order Landau-level wavefunction remains an accurate description when the effective mass varies in both directions with substantial gradients and curvature, so that light stays locked to the predefined $A_z=0$ line.","fun_headline_variants_meta":{"raw":{"variants":["Pseudomagnetic fields let light take any path on a chip","Synthetic magnetic fields steer light arbitrarily on silicon chips","Arbitrary light flow via pseudomagnetic fields at telecom wavelengths","Photonic crystals guide light along arbitrary curves with synthetic fields","Arbitrary light paths from photonic pseudomagnetic fields"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00127,"raw_usage":{"total_tokens":5220,"prompt_tokens":989,"completion_tokens":4231,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":605,"completion_tokens_details":{"reasoning_tokens":4147}},"tokens_in":605,"tokens_out":4231,"duration_ms":26309,"temperature":1.0,"reasoning_tokens":4147,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:21:43.021697+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fabricate a PMF waveguide whose prescribed path has a curvature radius comparable to the lattice constant (or a mass gradient that changes sign over a few unit cells) and measure whether the transmitted beam follows the predicted $A_z=0$ curve; a systematic deviation, beam broadening, or a sharp rise in insertion loss would show the adiabatic zeroth-Landau-level model no longer applies.","supporting_citations":[{"cited_title":"Barsukova, F","cited_arxiv_id":null,"evidence_quote":"Reports direct observation of Landau levels in silicon photonic crystals, confirming the material platform used here."},{"cited_title":"Barczyk, L","cited_arxiv_id":null,"evidence_quote":"Reports observation of Landau levels and chiral edge states in photonic crystals via synthetic strain, representing the prior state of the art that this method extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the chiral Landau level formalism for 2D Dirac cone systems with inhomogeneous effective mass, including the zeroth-order Landau level dispersion used throughout the paper."},{"cited_title":"He, E.-T","cited_arxiv_id":null,"evidence_quote":"Establishes the silicon-on-insulator slab platform for topological valley transport, supporting the choice of materials and geometry for the fabricated devices."}],"review_version":1}