Establishing three-dimensional Anderson localization of light in disordered dielectric particle media: asymptotic transition, microscopic theory, and experimental validation

Whether light can undergo Anderson localization in a fully disordered three-dimensional dielectric medium — Anderson's 1985 "white paint" question — remains unresolved. Full-vector Maxwell simulations by Grynko, Siebert, Sperling, and Förstner of dense packings of irregular dielectric particles (refractive index $n=3$, size parameter $kr\approx 1$, volume fraction $\rho\simeq 0.44$) reported converging localization signatures: non-exponential decay of time-resolved transmission $T(t)$, a time-dependent diffusion coefficient $D(t)$ approaching the predicted $t^{-1}$ law, spectrally isolated resonances with sub-unity Thouless conductance, and non-propagating near-field intensity clusters. The authors' 2026 follow-up reached the late-time $D(t)\sim 1/t$ regime in periodic slabs and resolved the real-space confinement morphology, but explicitly leaves open the thermodynamic-limit scaling and mobility edge, a predictive microscopic theory linking packing microgeometry to transport, and experimental validation. This problem asks the surviving open core: establish whether three-dimensional Anderson localization of light genuinely occurs in disordered dielectric media with realistic indices ($n\lesssim 3$) — the asymptotic localization transition and its parameter ranges, a quantitative theory deriving the transport scaling from microgeometry, and laboratory validation in real dielectric materials with absorption and fluorescence controls.

Background

Anderson localization is the halt of diffusive wave transport in a disordered medium caused by interference between multiply scattered wave paths. For light it is expected when the Ioffe–Regel criterion is met — when the wavenumber times the transport mean free path $l^{*}$ (the distance over which the propagation direction is randomized; related to the scattering mean free path by $l^{*}=l_{s}/(1-g)$, where $g$ is the single-scattering anisotropy factor) drops to order unity, $kl^{*}\lesssim 1$. Anderson's 1985 essay "The question of classical localization: a theory of white paint?" asked whether ordinary strongly scattering dielectric powders such as TiO$_2$ could localize visible light, and the question has remained open for four decades.

The experimental history explains why validation is the hard part. A claim of localization in semiconducting powder (Wiersma et al., 1997) was reinterpreted as classical diffusion with absorption (Scheffold, Lenke, Tweer, and Maret, 1999). Time-of-flight and transmission-profile measurements in TiO$_2$ (Störzer et al., 2006; Sperling et al., 2013) were subsequently questioned on grounds of inelastic scattering (Scheffold and Wiersma, 2013), and the original authors themselves showed in 2016 that their earlier data are consistently explained by a fluorescence process, doubting that the strong-localization regime can be reached in polydisperse TiO$_2$ at all. Any future experimental claim must therefore discriminate localization from absorption and fluorescence artifacts.

On the theory side, the self-consistent theory of Anderson localization in open three-dimensional media (Skipetrov and van Tiggelen, Physical Review Letters 96, 043902, 2006, DOI: 10.1103/PhysRevLett.96.043902) predicts that inside the localized regime the effective, time-dependent diffusion coefficient decays as $D(t)\sim t^{-1}$, so the time-resolved transmission $T(t)$ develops a non-exponential tail; observing this asymptotic law is a standard localization diagnostic. However, this theory was developed for dilute ensembles of point scatterers, and point-scatterer models are subject to no-go results: in random ensembles of resonant point scatterers the Ioffe–Regel criterion is never reached before absorption takes over (Skipetrov and Sokolov, 2014, 2018).

Numerically, brute-force finite-difference time-domain simulations demonstrated three-dimensional Anderson localization of vector electromagnetic waves in random aggregates of overlapping metallic spheres (Yamilov et al., Nature Physics 2023) — but, in sharp contrast, found an absence of localization for dielectric spheres with refractive index up to 10 in air. A companion study (Yamilov, Cao, and Skipetrov, 2025) located the Anderson transition for perfectly conducting spheres, identified a mobility edge (the disorder-strength/frequency boundary separating diffusive and localized transport), and found single-parameter scaling with a critical exponent of the orthogonal universality class (the symmetry class of time-reversal-invariant systems without spin–orbit coupling, as for electrons without magnetic fields). Localization signatures have also been reported in hyperuniform amorphous dielectric networks — hyperuniformity suppresses long-wavelength density fluctuations, producing a photonic band edge that helps localization — but that route uses spatially correlated, not fully disordered, media (Haberko, Froufe-Pérez, and Scheffold, 2020). An unambiguous experimental observation of three-dimensional Anderson localization of electromagnetic waves was reported only in 2026, for microwaves in disordered metal aggregates (Goïcoechea et al., 2026) — a metallic, non-optical system that leaves the dielectric optical case untouched.

Against this backdrop, Grynko, Siebert, Sperling, and Förstner (arXiv:2312.14393, 2023; published in Physical Review B in 2026 under the title "Signatures of light localization in three-dimensional disordered systems of dielectric particles") performed large-scale discontinuous Galerkin time-domain (DGTD) simulations — a full-vector Maxwell solver on unstructured meshes — of light transmission through dense random packings of irregular dielectric particles built from Gaussian random-field shapes, with size parameter $kr\approx 1$ ($k$ the wavenumber, $r$ the radius of the circumscribing sphere) and index $n=3$. They observed: (i) a transition from exponential to non-exponential $T(t)$ decay as the volume fraction $\rho$ rises above roughly $0.3$ and the index above roughly $2.5$; (ii) $D(t)$, extracted by local exponential fitting of $T(t)$, following a $t^{-1}$ law at late times in the densest samples; (iii) transmission spectra of isolated narrow resonances with mean Thouless conductance $\langle g_{Th}\rangle\approx 0.36$ — the Thouless conductance is the ratio $g_{Th}=\delta\omega/\Delta\omega$ of a resonance linewidth to the mode spacing, and $g_{Th}<1$ (spectral overlap weaker than spacing) is the Thouless criterion for localization; and (iv) late-time near-field maps showing non-propagating clusters of intensity hotspots. Smooth cylindrical sample boundaries support whispering-gallery modes — light trapped by total internal reflection along a curved boundary — which the authors used to extend the observation window, showing the localization signatures are caused by packing density rather than by the boundary trapping. The 2023 preprint identified as remaining open directions: the asymptotic long-time regime in larger samples ($N\gtrsim 10^{5}$ particles, called computationally prohibitive), quantification of boundary and finite-size leakage effects on the extracted $D(t)$, a predictive microscopic theory connecting the percolation-like transport mechanism (evanescent coupling through connected free-space cavities) to the observed scaling, experimental validation in real materials, and generality over $kr$, polydispersity, and shape irregularity.

The same authors' 2026 follow-up (Grynko and Förstner, arXiv:2605.25098) substantially advanced two of these items: using periodic slabs ($N=6000$–$12000$ particles) and analyzing late times after the diffusive component escapes, they observed the averaged $D(t)$ reaching the $1/t$ law at $t\gtrsim 100t_{a}$ (where $t_{a}$ is the transmitted pulse-peak arrival time), late-time spectra with $\langle g_{Th}\rangle=0.23$–$0.38$ and Poisson-like resonance spacings, and a real-space picture of compact non-propagating modal domains separated by persistent low-intensity channels — while emphasizing that this is finite-system evidence, that determining the asymptotic scaling behavior and locating the mobility edge would require ensemble averaging over many realizations and substantially larger systems, and that experimental studies in realistic dielectric systems remain to be done. What survives, and what this problem states, is the completion of the establishment of three-dimensional Anderson localization of light in disordered dielectric media.

Problem Statement

Establish whether three-dimensional Anderson localization of light occurs in disordered media of densely packed dielectric particles with realistic refractive indices ($n\lesssim 3$), converting the currently available finite-system numerical signatures into a verified and understood phenomenon. The following directions correspond to the open issues the source study left and later work did not close; together they support one question — is the dielectric localization transition real, what causes it, and can it be observed in the laboratory? All three directions are jointly required for a full affirmative resolution.

(1) Asymptotic transition and parameter ranges. By full-vector Maxwell simulations with ensemble averaging over many disorder realizations and substantially larger system sizes than the current $N\sim 10^{4}$ evidence, determine the asymptotic long-time behavior of the time-resolved transmission $T(t)$ and of the effective time-dependent diffusion coefficient $D(t)$ (extracted by local exponential fitting), including whether and over what parameter range $D(t)$ attains the predicted $t^{-1}$ localized-regime law; locate the mobility edge separating diffusive and localized transport; and establish over what ranges of volume fraction $\rho$, refractive index $n$, particle size parameter $kr$, polydispersity, and shape-irregularity statistics the localization signatures persist — taking the source's computed onset observations (non-exponential decay emerging for $\rho\gtrsim 0.3$ and $n>2.5$, with the densest samples at $\rho\simeq 0.44$, $n=3$) as the current finite-system evidence to be confirmed, or refuted as a finite-size or boundary artifact.

(2) Microscopic mechanism. Provide a quantitative, predictive theory that derives the observed transport scaling — the non-exponential $T(t)$ tail, $D(t)\sim t^{-1}$, and sub-unity Thouless conductance — from the packing microgeometry (void-size distribution, connectivity, and particle-shape statistics, including near-field evanescent coupling between particles).

(3) Experimental validation. Demonstrate the predicted localization signatures in real dense dielectric materials with $n\lesssim 3$ (for example TiO$_2$-class powders) via time-resolved, and where feasible polarization-resolved, transmission measurements that explicitly control for absorption, fluorescence, and inelastic scattering.

An answer may resolve the problem negatively: a rigorous demonstration that the transition does not occur in this class of dielectric media resolves it on its own, while a rigorous demonstration that the reported signatures are caused by finite-size leakage, boundary trapping, or an artifact answers only the finite-system-evidence part of direction (1) and leaves the rest of the program open.

References

1. Y. Grynko, D. Siebert, J. Sperling, J. Förstner, "3D Anderson localization of light in disordered systems of dielectric particles", arXiv:2312.14393 (2023; v5 April 2026), DOI: 10.48550/arXiv.2312.14393, https://arxiv.org/abs/2312.14393
2. Y. Grynko, D. Siebert, J. Sperling, J. Förstner, "Signatures of light localization in three-dimensional disordered systems of dielectric particles", Physical Review B (2026), DOI: 10.1103/tf5z-jytz, https://doi.org/10.1103/tf5z-jytz (published journal version of arXiv:2312.14393)
3. Y. Grynko, J. Förstner, "Resolved Anderson localization of light in dense three-dimensional dielectric disorder", arXiv:2605.25098 (2026), DOI: 10.48550/arXiv.2605.25098, https://arxiv.org/abs/2605.25098
4. A. Yamilov, S. E. Skipetrov, T. W. Hughes, M. Minkov, Z. Yu, H. Cao, "Anderson localization of electromagnetic waves in three dimensions", Nature Physics 19, 1308–1313 (2023), DOI: 10.1038/s41567-023-02091-7, https://doi.org/10.1038/s41567-023-02091-7
5. A. Yamilov, H. Cao, S. E. Skipetrov, "Anderson Transition for Light in a Three-Dimensional Random Medium", Physical Review Letters 134, 046302 (2025), DOI: 10.1103/PhysRevLett.134.046302, https://doi.org/10.1103/PhysRevLett.134.046302
6. S. E. Skipetrov, B. A. van Tiggelen, "Dynamics of Anderson Localization in Open 3D Media", Physical Review Letters 96, 043902 (2006), DOI: 10.1103/PhysRevLett.96.043902, https://doi.org/10.1103/PhysRevLett.96.043902
7. S. E. Skipetrov, I. M. Sokolov, "Absence of Anderson Localization of Light in a Random Ensemble of Point Scatterers", Physical Review Letters 112, 023905 (2014), DOI: 10.1103/PhysRevLett.112.023905, https://doi.org/10.1103/PhysRevLett.112.023905
8. S. E. Skipetrov, I. M. Sokolov, "Ioffe-Regel criterion for Anderson localization in the model of resonant point scatterers", Physical Review B 98, 064207 (2018), DOI: 10.1103/PhysRevB.98.064207, https://doi.org/10.1103/PhysRevB.98.064207
9. S. E. Skipetrov, I. M. Sokolov, "Anderson localization of light by impurities in a solid transparent matrix", Physical Review B 112, 064206 (2025), DOI: 10.1103/t5nj-fz3p, https://doi.org/10.1103/t5nj-fz3p
10. P. W. Anderson, "The question of classical localization: a theory of white paint?", Philosophical Magazine B 52, 505–509 (1985), DOI: 10.1080/13642818508240619, https://doi.org/10.1080/13642818508240619
11. D. S. Wiersma, P. Bartolini, A. Lagendijk, R. Righini, "Localization of light in a disordered medium", Nature 390, 671–673 (1997), DOI: 10.1038/37757, https://doi.org/10.1038/37757
12. F. Scheffold, R. Lenke, R. Tweer, G. Maret, "Localization or classical diffusion of light?", Nature 398, 206–207 (1999), DOI: 10.1038/18347, https://doi.org/10.1038/18347
13. M. Störzer, P. Gross, C. M. Aegerter, G. Maret, "Observation of the Critical Regime Near Anderson Localization of Light", Physical Review Letters 96, 063904 (2006), DOI: 10.1103/PhysRevLett.96.063904, https://doi.org/10.1103/PhysRevLett.96.063904
14. T. Sperling, W. Bührer, C. M. Aegerter, G. Maret, "Direct determination of the transition to localization of light in three dimensions", Nature Photonics 7, 48–52 (2013), DOI: 10.1038/nphoton.2012.313, https://doi.org/10.1038/nphoton.2012.313
15. F. Scheffold, D. Wiersma, "Inelastic scattering puts in question recent claims of Anderson localization of light", Nature Photonics 7, 934 (2013), DOI: 10.1038/nphoton.2013.210, https://doi.org/10.1038/nphoton.2013.210
16. T. Sperling, L. Schertel, M. Ackermann, G. J. Aubry, C. M. Aegerter, G. Maret, "Can 3D light localization be reached in 'white paint'?", New Journal of Physics 18, 013039 (2016), DOI: 10.1088/1367-2630/18/1/013039, https://doi.org/10.1088/1367-2630/18/1/013039
17. J. Haberko, L. S. Froufe-Pérez, F. Scheffold, "Transition from light diffusion to localization in three-dimensional amorphous dielectric networks near the band edge", Nature Communications 11, 4867 (2020), DOI: 10.1038/s41467-020-18571-w, https://doi.org/10.1038/s41467-020-18571-w
18. A. Goïcoechea, A. Yamilov, C. Ferise, S. E. Skipetrov, H. Cao, M. Davy, "Experimental observation of three-dimensional Anderson localization of electromagnetic waves", arXiv:2606.04897 (2026), DOI: 10.48550/arXiv.2606.04897, https://arxiv.org/abs/2606.04897
19. F. Delmotte, T. Brunet, J. Leng, J. H. Page, "3D Anderson localization of classical scalar waves", Science Advances (2026), DOI: 10.1126/sciadv.aed1319, https://doi.org/10.1126/sciadv.aed1319
20. C. Day, "Localizing Light", Physics 18, s10 (2025), DOI: 10.1103/Physics.18.s10, https://doi.org/10.1103/Physics.18.s10