Poincaré's program for polarization by diffraction: a quantitative theory of edge rounding (arris) and finite conductivity reconciling Gouy's experimental laws — the equal-angle intensity maximum, the inter-polarization phase difference, and spectral coloration

When natural (unpolarized) light is focused onto the straight edge of an opaque metallic screen, the diffracted light becomes partially linearly polarized; L. G. Gouy established six empirical laws for this effect in the 1880s. H. Poincaré (1892) founded the theory by solving the perfectly conducting infinitesimal wedge (half-plane) problem for both polarizations via Bessel/Fourier series, explaining Gouy's first, second and fourth laws and the near-complete polarization at large deviations. But his theory predicts a minimum of the total diffracted intensity exactly where Gouy observed a maximum (when incident and diffracted rays make equal angles with the screen — law 3); it predicts no phase difference between the polarization components (law 6) and no spectral coloration (law 5); and it produces unphysical infinities at the boundaries of the direct and reflected beams. Poincaré attributed the failures to the unmodeled rounding of the edge (the arris), the finite conductivity of the metal, and the idealized incident beam, and announced a rigorous second part. The Seconde partie (1897) treated defocusing, a hyperbolic rounded edge, and spherical waves, but left its central eigenvalue problems and series kernels unevaluated and contained no quantitative comparison with Gouy's laws. Later literature resolved parts of the program — Sommerfeld's exact half-plane solution (1896) removed the infinities, and impedance-wedge theory (Malyuzhinets, Senior, Williams, 1959–60) incorporated finite conductivity for sharp edges — but no located work provides a quantitative theory of polarization by diffraction for a rounded, finitely conducting edge, or a reconciliation of Gouy's laws 3, 5 and 6 with experiment. The problem remains open.

Background

The phenomenon. When a beam of natural (unpolarized) light grazes the straight edge of an opaque metallic screen — in Gouy's experiments a blade on whose edge a lens focuses the light — the light scattered outside the directly transmitted and specularly reflected beams (the diffracted light) becomes partially linearly polarized. The two linear polarization components are distinguished by the direction of the electric vector: the parallel component has its electric vector along the diffracting edge, the perpendicular component has it perpendicular to the edge (these are Poincaré's components parallel and perpendicular to the plane of diffraction, the plane containing the edge and the ray). Gouy's six laws (known here through Poincaré's 1892 summary) state: (1) inside the geometric shadow the diffracted light is polarized perpendicular (electric vector perpendicular to the edge); (2) in the lit region outside the directly transmitted and reflected beams it is polarized parallel (electric vector along the edge); (3) for a fixed deviation (angle between incident and diffracted rays) the total diffracted intensity is greatest when the incident and diffracted rays make equal angles with the screen; (4) the intensity is the same for equal interior and exterior deviations; (5) the perpendicular component is strongly colored while the parallel component stays nearly white; (6) the two components acquire a phase difference that grows with the deviation.

Poincaré's theory (1892). Poincaré idealized the screen as a perfectly conducting wedge of infinitesimal angle (effectively a half-plane) whose edge is a mathematical line, illuminated by cylindrical waves focused on the edge (a two-dimensional problem). In two dimensions the two polarizations decouple: the electric field along the edge — the parallel component — satisfies the Dirichlet condition $Z=0$ on the metal (the field amplitude itself vanishes at the surface), and the magnetic field along the edge — the perpendicular component — satisfies the Neumann condition $\partial\gamma/\partial n=0$ (the derivative of the field along the surface normal vanishes). Separation of variables in polar coordinates about the edge (Fourier series in the angle, Bessel functions in the radius) gave closed-form diffracted-amplitude formulas for both components. These explain Gouy's laws 1, 2 and 4 and predict nearly complete polarization for large deviations, in agreement with observation; the same polarization pattern persists for wedges of arbitrary finite angle, and the perfect-conductor boundary conditions were argued to be approximated by highly reflective real metals.

What the idealized theory gets wrong (the source's own open issues). (i) Summing the two narrow-beam intensity formulas, the total diffracted intensity at fixed deviation $\delta$ from the direct beam (with incidence angle $\alpha$ measured from the screen face, the angular position $\omega$ of the diffracted ray measured from the same face, and $\delta=\omega-\alpha-\pi$) is proportional to $2[\csc^2(\delta/2)+\csc^2(\alpha+\delta/2)]$, which is minimal when the incident and diffracted rays make equal angles with the screen ($\omega+\alpha=2\pi$) — precisely where Gouy observed a maximum (law 3). Poincaré located the explanation in the rounding of the edge (the arris, the finite radius of curvature of a real blade), treated only qualitatively. (ii) Finite conductivity was handled by a rough argument (explicitly called too imperfect for numerical computation), so the theory yields no numerical values for the polarization degree and no phase difference between components — whereas Gouy's law 6 records a phase difference growing with deviation. (iii) The spectral coloration of law 5 was attributed only tentatively to the wavelength dependence of metallic reflectivity. (iv) The theory produces infinities in the theoretical intensity at the boundaries of the direct and reflected beams — artifacts of the assumed step-function incident beam. Poincaré acknowledged all of this and announced a second part.

The Seconde partie (1897). Poincaré analyzed the offset focus (defocus) of the incident cylindrical wave — which removes the beam-edge singularities for a realistic continuous beam and connects his results to Sommerfeld's plane-wave solution — and set up diffraction by a wedge whose cross-section is a hyperbola (a rounded edge of curvature parameter $c$): separating variables in elliptic coordinates he reduced the problem to eigenvalue equations (his (A1)–(A4)) for four solution types, but solved them only in the limiting cases $c=0$ and very large $\alpha^2 c^2$, leaving the kernels $\Lambda$ of his diffracted-field series unevaluated; he also treated spherical waves on a half-plane under idealizing assumptions. His own open questions state that the results are incomplete, and no quantitative comparison with Gouy's laws 3, 5, 6 was made.

Later literature. Sommerfeld's exact solution of the perfectly conducting half-plane (1896), via the contour representation now called the Sommerfeld integral, resolves the beam-edge infinities and provides the exact sharp-edge benchmark for both polarizations. For finite conductivity, the impedance (Leontovich) boundary condition — expressing the tangential field relation at the surface of a good conductor through a surface impedance, thereby avoiding solving inside the metal — allowed Malyuzhinets, Senior and Williams (1959–60) to solve diffraction by sharp imperfectly conducting wedges and half-planes; the diffraction coefficients become complex, so phase differences between polarizations appear in principle. High-frequency engineering practice uses the Uniform Theory of Diffraction (Kouyoumjian–Pathak 1974), which uniformizes edge diffraction for curved edges but only in the limit of curvature radii large compared with the wavelength. The canonical modern review of analytical methods for perfect (sharp, perfectly conducting) wedge diffraction (Nethercote–Assier–Abrahams 2020) covers exact, spectral, transform and asymptotic methods and their impedance and penetrable extensions, but contains no theory of edge rounding at curvature comparable to the wavelength — which is the regime of Gouy's optical experiments. Microwave experiments (Einstein–Juliano–Pine 1971, 1973, 1979) quantitatively verified Sommerfeld's exact sharp-edge solution and found screen-thickness effects insubstantial, but did not address the optical polarization laws 3, 5, 6.

Problem Statement

Construct a quantitative theory of polarization by diffraction at a real metallic edge and confront it with Gouy's experimental laws. Specifically: for a metallic screen whose edge is rounded with a finite radius of curvature (the arris) — including the regime where this radius is comparable to the optical wavelength — with finite conductivity described by realistic optical constants, illuminated by a focused beam of finite width and soft angular profile (cylindrical, and where relevant spherical, waves), compute: (i) the total diffracted intensity as a function of the deviation $\delta$ between incident and diffracted rays and of the incidence angle $\alpha$ — in particular determine whether it is maximum (as Gouy observed) or minimum (as the sharp perfect-conductor theory predicts) when the incident and diffracted rays make equal angles with the screen ($\omega+\alpha=2\pi$, where $\omega$ is the angular position of the diffracted ray about the edge, measured from the same screen face as $\alpha$), and identify quantitatively which physical effect (edge rounding, finite conductivity, focus) is responsible for the reversal of Gouy's third law; (ii) the degree of polarization and the phase difference between the two linear polarization components as functions of the deviation (Gouy's law 6); (iii) the spectral distribution of each component (Gouy's law 5: coloration of the perpendicular component, near-whiteness of the parallel component). The theory must reduce, in its limiting cases, to the known sharp perfect-conductor results (Poincaré's formulas and Sommerfeld's exact half-plane solution) and to the sharp-edge impedance-wedge results (Malyuzhinets, Senior, Williams), and it must be compared quantitatively with Gouy's experiments or an equivalent modern replication.

References

1. H. Poincaré, 'Sur la polarisation par diffraction', Acta Mathematica 16, 297–339 (1892). DOI: 10.1007/BF02418992. URL: https://doi.org/10.1007/BF02418992
2. H. Poincaré, 'Sur la polarisation par diffraction: Seconde partie', Acta Mathematica 20, 313–355 (1897). DOI: 10.1007/BF02418036. URL: https://doi.org/10.1007/BF02418036
3. H. Poincaré, 'Sur la diffraction des ondes électriques: à propos d'un article de M. Macdonald', Proceedings of the Royal Society of London 72, 42–52 (1904). DOI: 10.1098/rspl.1903.0013. URL: https://doi.org/10.1098/rspl.1903.0013
4. A. Sommerfeld, 'Mathematische Theorie der Diffraction', Mathematische Annalen 47, 317–374 (1896). DOI: 10.1007/BF01447273. URL: https://doi.org/10.1007/BF01447273
5. C. J. Bouwkamp, 'A note on singularities occurring at sharp edges in electromagnetic diffraction theory', Physica 12, 467–474 (1946). DOI: 10.1016/S0031-8914(46)80061-2. URL: https://doi.org/10.1016/S0031-8914(46)80061-2
6. C. J. Bouwkamp, 'Diffraction Theory', Reports on Progress in Physics 17, 35–100 (1954). DOI: 10.1088/0034-4885/17/1/302. URL: https://doi.org/10.1088/0034-4885/17/1/302
7. G. D. Malyuzhinets (published as G. D. Malyughinetz), 'Das Sommerfeldsche Integral und die Lösung von Beugungsaufgaben in Winkelgebieten', Annalen der Physik 461, 107–112 (1960). DOI: 10.1002/andp.19604610111. URL: https://doi.org/10.1002/andp.19604610111
8. T. B. A. Senior, 'Diffraction by an imperfectly conducting wedge', Communications on Pure and Applied Mathematics 12, 337–372 (1959). DOI: 10.1002/cpa.3160120209. URL: https://doi.org/10.1002/cpa.3160120209
9. W. E. Williams, 'Diffraction of an E-polarized plane wave by an imperfectly conducting wedge', Proceedings of the Royal Society of London A 252, 376–393 (1959). DOI: 10.1098/rspa.1959.0159. URL: https://doi.org/10.1098/rspa.1959.0159
10. T. B. A. Senior, 'Diffraction by an imperfectly conducting half-plane at oblique incidence', Applied Scientific Research, Section B 8, 35–61 (1960). DOI: 10.1007/BF02920043. URL: https://doi.org/10.1007/BF02920043
11. R. G. Kouyoumjian and P. H. Pathak, 'A uniform geometrical theory of diffraction for an edge in a perfectly conducting surface', Proceedings of the IEEE 62, 1448–1461 (1974). DOI: 10.1109/PROC.1974.9651. URL: https://doi.org/10.1109/PROC.1974.9651
12. T. B. A. Senior, 'Diffraction tensors for imperfectly conducting edges', Radio Science 10, 911–919 (1975). DOI: 10.1029/RS010i010p00911. URL: https://doi.org/10.1029/RS010i010p00911
13. F. S. Einstein, R. A. Juliano, Jr., and C. Pine, 'Far-Field Diffraction by a Semi-Infinite Plane: Parallel Polarization', Journal of the Optical Society of America 61, 1487–1491 (1971). DOI: 10.1364/JOSA.61.001487. URL: https://doi.org/10.1364/JOSA.61.001487
14. F. S. Einstein, R. A. Juliano, and C. Pine, 'Far-field diffraction by a semi-infinite plane: perpendicular polarization', Journal of the Optical Society of America 63, 419 (1973). DOI: 10.1364/JOSA.63.000419. URL: https://doi.org/10.1364/JOSA.63.000419
15. F. S. Einstein, R. A. Juliano, L. A. DeAcetis, and I. Lazar, 'Experimental investigation of the far-field diffraction by a semi-infinite plane of variable thickness: Parallel polarization', Journal of the Optical Society of America 69, 24 (1979). DOI: 10.1364/JOSA.69.000024. URL: https://doi.org/10.1364/JOSA.69.000024
16. P. Siegmann, L. M. Sánchez-Brea, J. C. Martínez-Antón, and E. Bernabeu, 'Diffraction in wide slits with semi-cylindrical edges', Optik 113, 57–62 (2002). DOI: 10.1078/0030-4026-00115. URL: https://doi.org/10.1078/0030-4026-00115
17. J.-M. Caron, 'Diffraction d'une onde électromagnétique par un dièdre infini absorbant', Comptes Rendus Mathématique 337, 387–392 (2003). DOI: 10.1016/S1631-073X(03)00350-9. URL: https://doi.org/10.1016/S1631-073X(03)00350-9
18. M. A. Nethercote, R. C. Assier, and I. D. Abrahams, 'Analytical methods for perfect wedge diffraction: A review', Wave Motion 93, 102479 (2020). DOI: 10.1016/j.wavemoti.2019.102479. URL: https://doi.org/10.1016/j.wavemoti.2019.102479 (preprint arXiv:1905.10562)