In our understanding of the propagation of light, a central role is played by the dispersion relations. These are relations between the frequency and the wave vector (or inverse wavelength) of the waves. In Nanophotonic metamaterials – such as photonic crystals, cavities, complex random media, or plasmonic matter – the dispersion relations are strongly modified compared to the known behavior in everyday optics. As a result, we are beginning to understand how light behaves in extreme situations, when it is nearly standing still, when it is modified at the nanoscale, or when it is strongly absorbed.

In complex media with or without long-range order, it is instructive to consider typical length scales for light in comparison to the wavelength. When light performs a random walk in such a medium, the average step size is given by the scattering mean free path. This length is probed in the transmission of the direct beam, also known as the Lambert-Beer law (or more completely the Beer-Lambert-Bouguer law). The average distance after which light has become diffuse (that is, “light has forgotten its initial direction”) is given by the mean free path, also known as transport mean free path.This length is probed in total transmission or a coherent backscatter cone. The average distance after which light is absorbed to 1/e is the absorption mean free path, and typically probed by total transmission.

Figure 1: Universal dependence of extinction length on interaction strength S in photonic crystals. Dashed curves: extinction calculated for fcc inverse opals (assuming 26% high index material). Symbols: our data (triangles), and literature data analyzed by us (diamonds). Observed losses are consistent with δR > 4%. If ell/a is shorter than the length needed for Bragg diffraction, structures are essentially disordered (shaded red). Complete band gaps are expected for S > 15% (shaded green). Photonic crystal integrated circuits require ell/a ≥ 10^4 at S > 15%, beyond the current state-of-the-art (click here).

Remarkably, photonic crystals reveal not only their anticipated gap behavior but also light scattering and extinction of coherent beams. The latter are caused by the unavoidable variations in size and position of the crystal’s building blocks. COPS scientists first derived a model for both two- (2D) and three-dimensional (3D) photonic crystals that relates the extinction length to the magnitude of the structure variations. The predicted lengths agree well with our experiments on high-quality opals and inverse opals, and with literature data analyzed by us. As a result, control over photons is limited to distances up to 50 lattice parameters (about 15 micron) in state-of-the-art structures (see Figure), thereby impeding applications that require large photonic crystals, such as photonic integrated circuitry. Meanwhile, nanofabrication has definitely improved, but centimeter-size extinction lengths are the realm of, e.g., SiN waveguides (in absence of surrounding photonic crystal). Nevertheless, scattering in photonic crystals also has favorable consequences namely new physics such as Anderson localization and nonclassical diffusion.

To probe dispersion relations, one uses phase-sensitive measurements and interferometry. In this way, COPS scientists have determined the dispersion relations of waveguides in 2D photonic crystal slabs. They picked up light that propagated through a tiny “W1” waveguide (1 row of pores missing) and interfered it with unperturbed light. As a result, they could determine the dispersion relations all the way to the edge of the band gap, and also detect Anderson localized modes and a Lifshitz tail (click here).

Thanks to the COPS invention of wavefront shaping, it has become feasible to radically control the propagation of light in nanophotonic waveguides between highly transmitting or completely localized. Using an adaptive holographic method, COPS scientists managed to “smoothen” the refractive index profile in a coupled-cavity waveguide. They observed the dynamically formation or breaking of highly transmitting necklace states, which is an essential step toward photonic-crystal-based quantum networks and photonic integrated circuits (PICs), click here.