Unveiling the properties of metagratings via a detailed analytical model for synthesis and analysis

I. INTRODUCTION

• Metasurface realization remains challenging because microscopic design often relies on full-wave lookup tables and may require tuning multiple meta-atom degrees of freedom.This burden grows for generally inhomogeneous designs.Source: §I · p. 1
• Metagratings can achieve optimal 100% conversion using a single polarizable particle per macro-period, reducing the microscopic design burden of metasurfaces.They cancel a finite number of unwanted propagating diffraction modes rather than implementing a prescribed field transformation that excludes all undesirable modes.Source: §I · p. 1
• Earlier metagrating design lacked a rigorous scheme for determining optimal geometry and relied mainly on physical intuition and numerical optimization.A prior analytical method addressed magnetically polarizable particles under TM excitation; this work instead treats electrically polarizable loaded wires under TE excitation, which are more practical for planar devices.Source: §I · p. 2
• The paper derives a semianalytical method for perfect wide-angle beam splitting that relates performance requirements to individual-wire loads and predicts printed-capacitor geometry with one simulation at the operating frequency.The model also analyzes conductor losses, reactance deviations, and frequency shifts, identifying working points with relatively low sensitivity to these non-idealities.Source: §I · p. 2
• The resulting analytical relations provide engineering tools and physical insight for synthesis and analysis of future metagratings.The authors present this as groundwork for practical realization and broader applications.Source: §I · p. 2

A. Formulation

• The model considers a two-dimensional, TE-excited periodic wire array a distance h from a PEC, with wires much narrower than the wavelength and loaded by a distributed impedance.In practice, lumped loads repeat with deep-subwavelength period L along the wires.Source: A. Formulation · p. 2
• The total fields are formed from external fields satisfying the PEC boundary condition and fields generated by identical induced currents on the periodic wires and their image sources.The wire-generated field is represented through Floquet–Bloch harmonics.Source: A. Formulation · pp. 2–3
• Ohm’s law at the wire position links the total electric field and induced current through the distributed impedance, enabling either current calculation for a given load or load assessment for a target current.To account for the local field singularity, the model uses the actual wire dimensions and an effective radius reff = w/4 when t ≪ w ≪ λ.Source: A. Formulation · p. 3

B. Eliminating specular reflection

• The desired output angles must correspond to the ±1 propagating Floquet–Bloch modes, while all modes with |m| ≥ 2 must be evanescent.With the available design freedoms, only one Floquet–Bloch mode can be eliminated.Source: B. Eliminating specular reflection · p. 4
• Specular reflection is eliminated by tuning the configuration so the fundamental wire-generated Floquet–Bloch mode destructively interferes with the specular reflection.The other propagating modes must therefore be the desired ±1 output modes.Source: B. Eliminating specular reflection · p. 4

C. Perfect beam splitting

• Vanishing net real power crossing a plane below the metagrating ensures that all incident power enters the remaining ±1 modes, which symmetry makes equal-power beams.The lossless-medium condition is independent of the chosen observation plane.Source: C. Perfect beam splitting · p. 4
• The wire–PEC separation h is obtained by solving a nonlinear equation, setting the first design degree of freedom.Its interference terms are trigonometric functions of kh and kh cos θout.Source: C. Perfect beam splitting · p. 5
• Once h is fixed, the required distributed impedance follows from the specular-reflection and perfect-splitting conditions, setting the second design degree of freedom.The resulting impedance is purely imaginary, so perfect coupling requires a purely reactive wire load; losses are the remaining obstacle once specular reflection is eliminated.Source: C. Perfect beam splitting · p. 5

A. Synthesis

• For each split angle, the smallest separation satisfying Eq. (14) is frequency-invariant and allows beam splitters thinner than one operating wavelength.The resulting separation curve is expressed in wavelength units.Source: A. Synthesis · p. 5, Figure 2 · p. 5
• At 10 GHz, the required reactance is negative across the considered angles, so the wires use capacitive loading; predicted capacitor widths agree excellently with simulated optima.The widths are estimated from the prescribed impedance and an analytical capacitance relation, with a correction factor assessed by simulation.Source: A. Synthesis · pp. 5–6, Figure 3 · p. 6
• Analytical and simulated field distributions agree closely except near the meta-atoms, where the uniformly loaded singular-wire model omits the finite copper-trace geometry.The exception is marked within a 0.1λ-diameter region around each element.Source: A. Synthesis · p. 6, Figure 4 · p. 7
• Simulated field amplitudes are lower than analytical predictions because the realistic conductors include losses omitted from the analytical model.The interference patterns still almost perfectly match, although amplitude differences are significant for some split angles, including 60.5°.Source: A. Synthesis · p. 6
• At 10 GHz, most designs couple more than 2 × 45% of incident power symmetrically to the ±1 FB modes, with losses rising near split angles of 30°, 60°, and 90°.Specular reflection is practically negligible; the loss trend is nonmonotonic with split angle.Source: A. Synthesis · p. 7, Table I · p. 7
• At 20 GHz, the width relation remains a good predictor, with some optimal dimensions deviating by about 10%; simulated losses are about 20% smaller than at 10 GHz.Losses are again more pronounced near split angles of 30°, 60°, and 90°.Source: A. Synthesis · pp. 7–8, Table II · p. 8, Figure 5 · p. 8

• Splitting efficiency ηsplit measures incident power coupled to the ±1 modes combined; ηspec and ηloss measure power in specular reflection and absorbed in the wires.Source: B. Analysis · pp. 8–9
• The load is modeled as the optimal impedance plus δR and jδX, representing conductor resistance and deviation from the optimal reactance, respectively.The current response for a given load is used to calculate the coupling efficiencies.Source: B. Analysis · p. 9
• At the ideal point δR = δX = 0, ηsplit = 1 while ηspec = ηloss = 0.This matches the ideal coupling derived earlier in the paper.Source: B. Analysis · p. 9

Conductor loss

• For small losses, splitting efficiency falls mainly through absorption, reaching about 90% of its maximum when 10% of incident power is absorbed.The small-loss approximation is valid for δR/Rg ≪ 1; the exact expressions are needed for more significant losses.Source: §1 · pp. 9–10
• Effective grid resistance peaks near θout ≈ 57° and 78°, marking the least loss-sensitive working points; the trend is frequency invariant.Loss severity near the minima also depends on sin θout, so the approaches to 30°, 60°, and 90° are not equally lossy.Source: §1 · p. 9, Figure 6 · p. 10
• Destructive image-source interference near θout → 30°, 60°, and 90° drives high wire currents, making even slight conductor resistance produce significant dissipation.Constructive interference requires less current and leaves the device less susceptible to losses.Source: §1 · pp. 9–10
• At 10 GHz, extracted distributed resistance is (18.3 ± 1.2) × 10^-3 [η/λ], versus (14.5 ± 1.2) × 10^-3 [η/λ] at 20 GHz.These estimates come from matching analytical absorption to full-wave simulation; the approximately 20% resistance difference is consistent with the reported absorption difference.Source: §1 · p. 10, Figure 7 · p. 11
• A rounded-conductor approximation gives δR = 17.3 × 10^-3 [η/λ] at 10 GHz and 12.3 × 10^-3 [η/λ] at 20 GHz, reasonably matching the extracted values.The approximation uses a 3 mil (76.2 μm) wire width.Source: §1 · p. 10
• The analytical relations connect device losses to conductor geometry and indicate that absorption can be tuned by modifying copper features within the configuration’s constraints.The loss trend arises from interference between the wires and their PEC images.Source: §1 · pp. 10–11

Reactance deviation and frequency response

• The optimal load reactance remains the value given by Eq. (15) even when conductor losses are present.The maximum splitting efficiency occurs at zero reactance deviation, independently of wire resistance.Source: §2 · p. 11
• Reactance sensitivity is highest when sin θout sin^2(kh) is minimal, where destructive interference yields very high currents from weak exciting fields.The small effective fields and large currents make deviations from design parameters especially consequential for splitting efficiency.Source: §2 · p. 11
• Predicted capacitor-width tolerances agree well with full-wave simulations, although their maximum does not coincide with the reactance-tolerance maximum.The width tolerance also depends on nominal capacitor width, which Eq. (25) does not account for.Source: §2 · p. 12
• The analytical model predicts frequency bandwidths that agree remarkably with simulations, and bandwidth is smallest near the destructive-interference working points.Away from those points, performance is stable under moderate frequency shifts, though the split angle varies; bandwidth is also limited when ±1 modes become evanescent or higher modes propagate.Source: §2 · pp. 12–13

IV. CONCLUSION

• The model relates device performance to individual meta-atom loads, including realistic losses, and enables semianalytical prediction of the required geometry.It is verified with full-wave simulations of realistic physical structures.Source: §IV · p. 13
• Working points near zero effective grid resistance are especially sensitive to conductor losses, load-geometry inaccuracies, and frequency shifts.The analysis attributes this sensitivity to destructive interference that produces very high currents under very low effective fields.Source: §IV · p. 13
• Some of the best operating points occur at extremely wide splitting angles, unlike the large wave-impedance-mismatch conditions commonly associated with metasurface performance reduction.The comparison is specifically made for the investigated metagrating devices.Source: §IV · p. 13
• With suitable working points, metagratings can split a normally incident beam into equal-power beams at angles near 80° with minimal absorption, moderate bandwidth, and substantial fabrication resilience.The structure requires designing only a single meta-atom per macro-period.Source: §IV · p. 13
• The derivation and observations are not restricted to microwave frequencies, and the same meta-atom structures have been used for terahertz and optical applications.Source: §IV · p. 14