Yablonovitch's statistical ray optics (J. Opt. Soc. Am. 72, 899, 1982) derives the celebrated light-trapping results for textured optical media — an internal intensity enhancement of 2n22n^2 (up to 4n24n^2 for the absorption path length in the weakly absorbing limit, nn being the refractive index) — from the assumption that the internal ray dynamics is ergodic, i.e. that a typical ray trajectory explores the accessible directions and positions uniformly over time; the separate requirement that this randomization occur before the light is absorbed is assumption (ii) below. The paper itself acknowledges that this rests on unproven ground: only a semiquantitative texturing criterion is given for ergodicity, the absorption formulas can underestimate absorption when absorption is strong (randomization may not occur before the light is absorbed), and wave-optical interference and polarization are neglected. This problem asks for a quantitative theory of light trapping and absorption in textured optical media that does not presuppose these assumptions: a criterion that decides, for a given surface texture or geometry, whether the internal ray dynamics is ergodic, together with absorption predictions (path-length statistics or equivalent) for the geometries and regimes these assumptions exclude, valid for strong absorption and for thicknesses comparable to the wavelength. Later work has resolved parts of this bundle — wave-optical and strong-absorption upper bounds now exist — but a rigorous ergodicity criterion and a nonergodic-geometry absorption theory remain open.

Background

Light trapping is the enhancement of optical absorption in a weakly absorbing slab by texturing its surfaces so that light inside is redirected by total internal reflection, with only the rays lying within the narrow escape cone (half-angle arcsin⁡(1/n)\arcsin(1/n) around the normal, a consequence of Snell's law) able to leave. Yablonovitch's 1982 paper 'Statistical ray optics' treated such media statistically: if the ray directions inside are randomized — the assumption of ergodic ray dynamics, borrowed from statistical mechanics, meaning that a typical ray trajectory explores the accessible directions and positions uniformly over time — then a detailed-balance (inflow–outflow) argument between the incident light and the internal 'photon gas' fixes the internal intensity at 2n22n^2 times the incident intensity, and the average absorption path length at 4n24n^2 times the slab thickness dd in the weakly absorbing limit. This 4n24n^2 'Yablonovitch limit' became the standard benchmark for solar-cell light management. The theory, however, relies on a chain of approximations the paper itself flags: (i) ergodicity is assumed, with only a semiquantitative criterion for the texturing angle (θ>12arcsin⁡(1/n)\theta > \frac{1}{2}\arcsin(1/n), about 8.3∘8.3^\circ for silicon) to ensure that rays totally internally reflected at the front surface are redirected into the escape cone; (ii) absorption must be weak enough that randomization occurs before the light is absorbed, so the lumped formulas may underestimate absorption in the strongly absorbing regime; (iii) the medium must be thick compared with λ/(2n)\lambda/(2n) so that wave-optical interference between internally reflected waves can be neglected, and polarization is ignored; (iv) the effective thickness ll and edge losses are treated approximately. Subsequent literature has rebuilt pieces of this foundation: the 'mean path length' of light in scattering media was shown to obey an invariance theorem (a Liouville-type result, meaning the average length of all trajectories entering and leaving a domain is set by the domain boundary alone), nonergodic geometries were shown to exceed the ergodic intensity limit, phase-space occupation was identified as the operative variable, and quasinormal-mode (resonance) calculations produced absorption bounds valid for arbitrary absorption strength. What has not been achieved is a criterion that predicts ergodicity of the ray dynamics from the geometry itself, or a general absorption theory for nonergodic, strongly absorbing, wavelength-scale media.

Problem Statement

For an optical medium of refractive index nn with textured (nonspherical, nonplanar) surfaces, Yablonovitch's statistical ray optics predicts an internal intensity enhancement of 2n22n^2 and, in the weakly absorbing limit, an average absorption path length of 4n24n^2 times the medium thickness — but only under the assumptions that the ray dynamics is ergodic (a typical ray trajectory explores the accessible directions and positions uniformly over time), that absorption is weak enough for randomization to precede absorption, and that the medium is thick compared with λ/(2n)\lambda/(2n) so wave interference and polarization can be ignored. Develop the theory beyond these assumptions: (1) give a criterion, rigorous or at least quantitative and general, that determines for a given surface texture or geometry whether the internal ray dynamics is ergodic — replacing the 1982 semiquantitative texturing condition θ>12arcsin⁡(1/n)\theta > \frac{1}{2}\arcsin(1/n) and settling the ergodicity assumption (i), on which the applicability of the statistical (ergodic) treatment rests together with assumptions (ii) and (iii) treated under (2); and (2) provide absorption predictions for the regimes these assumptions exclude — nonergodic geometries, the strong-absorption regime in which randomization does not occur before the light is absorbed, and media whose thickness is comparable to the wavelength, where interference and polarization must enter. These three regimes correspond to assumptions (i), (ii) and (iii) respectively, are mutually independent, and each is a required part of this deliverable; together with the criterion of (1) they make up the complete objective. For the criterion in (1), the minimum bar carried by 'quantitative and general' is a quantitative test stated in measurable geometric features and applying beyond an individual texture, with a rigorous general derivation the stronger target. Note that bounds are not predictions: rigorous upper bounds already exist for parts of this territory (the nanophotonic regime and broadband absorption of arbitrary strength), but a bound does not decide, for a given geometry, whether the ergodic treatment applies or what the actual absorption is; the demanded deliverables are geometry-resolved predictions together with validity conditions known a priori. The common objective is a quantitative light-trapping/absorption theory for textured optical media whose validity conditions are known a priori from the geometry, rather than assumed.

References

  1. Eli Yablonovitch, "Statistical ray optics", Journal of the Optical Society of America 72(7), 899 (1982). DOI: 10.1364/josa.72.000899. URL: https://doi.org/10.1364/josa.72.000899
  2. Zongfu Yu, Aaswath Raman, and Shanhui Fan, "Fundamental limit of nanophotonic light trapping in solar cells", Proceedings of the National Academy of Sciences 107(41), 17491–17496 (2010). DOI: 10.1073/pnas.1008296107. URL: https://doi.org/10.1073/pnas.1008296107
  3. Emily D. Kosten, Emily L. Warren, and Harry A. Atwater, "Ray optical light trapping in silicon microwires: exceeding the 2n2 intensity limit", Optics Express 19(4), 3316 (2011). DOI: 10.1364/OE.19.003316. URL: https://doi.org/10.1364/OE.19.003316
  4. Ian Marius Peters, "Phase space considerations for light path lengths in planar, isotropic absorbers", Optics Express 22(S3), A908 (2014). DOI: 10.1364/OE.22.00A908. URL: https://doi.org/10.1364/OE.22.00A908
  5. Rajeshkumar Mupparapu, Kevin Vynck, Tomas Svensson, Matteo Burresi, and Diederik S. Wiersma, "Path length enhancement in disordered media for increased absorption", Optics Express 23(24), A1472 (2015). DOI: 10.1364/oe.23.0a1472. URL: https://doi.org/10.1364/oe.23.0a1472 (preprint arXiv:1507.03484, https://arxiv.org/abs/1507.03484)
  6. Marina Mariano, Gregory Kozyreff, Luis G. Gerling, Pablo Romero-Gomez, Joaquim Puigdollers, Jorge Bravo-Abad, and Jordi Martorell, "Intermittent chaos for ergodic light trapping in a photonic fiber plate", Light: Science & Applications 5(12), e16216 (2016). DOI: 10.1038/lsa.2016.216. URL: https://doi.org/10.1038/lsa.2016.216
  7. Romolo Savo, Romain Pierrat, Ulysse Najar, Rémi Carminati, Stefan Rotter, and Sylvain Gigan, "Observation of mean path length invariance in light-scattering media", Science 358(6364), 765–768 (2017). DOI: 10.1126/science.aan4054. URL: https://doi.org/10.1126/science.aan4054 (preprint "Mean path length invariance in multiple light scattering", arXiv:1703.07114, https://arxiv.org/abs/1703.07114)
  8. Guillermo Martínez-Denegri, Silvia Colodrero, Quan Liu, Johann Toudert, Gregory Kozyreff, and Jordi Martorell, "Ergodic Light Propagation in a Half-Cylinder Photonic Plate for Optimal Absorption in Perovskite Solar Cells", Advanced Optical Materials 7(10), 1900018 (2019). DOI: 10.1002/adom.201900018. URL: https://doi.org/10.1002/adom.201900018
  9. Matthieu Davy, Matthias Kühmayer, Sylvain Gigan, and Stefan Rotter, "Mean path length invariance in wave-scattering beyond the diffusive regime", Communications Physics 4, 85 (2021). DOI: 10.1038/s42005-021-00585-5. URL: https://doi.org/10.1038/s42005-021-00585-5 (preprint arXiv:2011.07146, https://arxiv.org/abs/2011.07146)
  10. Matt Majic, Walter R. C. Somerville, and Eric C. Le Ru, "Mean path length inside nonscattering refractive objects", Physical Review A 103(3), L031502 (2021). DOI: 10.1103/PhysRevA.103.L031502. URL: https://doi.org/10.1103/PhysRevA.103.L031502 (preprint "Mean path length inside non-scattering refractive objects", arXiv:2107.05308, https://arxiv.org/abs/2107.05308)
  11. Stéphane Collin and Maxime Giteau, "Upper Bounds on Broadband Absorption and Application to Solar Cells", PRX Energy 5(2), 023006 (2026). DOI: 10.1103/1tzg-hgqx. URL: https://doi.org/10.1103/1tzg-hgqx (preprint "Upper bounds on broadband absorption", arXiv:2407.19559, 2024, https://arxiv.org/abs/2407.19559)
  12. Eli Yablonovitch and Zunaid Omair, "There Are Two Distinct Photon Gases Present Inside Every Solar Cell", arXiv:2405.08835 (2024). URL: https://arxiv.org/abs/2405.08835