Black silicon textures are widely used as broadband antireflection layers, yet their optical response is difficult to interpret directly from their stochastic nanoscale morphology. Here, we show that black silicon can be understood as an ensemble of adiabatic photonic nanotapers that progressively inject and guide light into the bulk substrate. Black silicon samples are fabricated by reactive ion etching using a metallic island mask, and their specular and diffuse reflectance spectra are measured and accurately reproduced by three-dimensional finite-difference time-domain simulations based on a stochastic effective morphology model fitted to scanning electron microscopy statistics. To connect these experimental and stochastic full-wave results to a simpler physical picture, the disordered morphology is then distilled into an idealized periodic model consisting of a compact hexagonal array of identical silicon nanotapers, analyzed using the Fourier Modal Method combined with a biorthogonal Bloch-mode decomposition. Within this framework, we introduce an angle-dependent eikonal parameter as a quantitative measure of adiabaticity. At normal incidence, light transport is dominated by a single downward Bloch branch that evolves quasi-adiabatically along the taper, whereas at grazing incidence the driven mode becomes weakly confined and couples non-adiabatically to leaky upward channels. This modal picture explains the low experimental reflectance in terms of progressive light injection, transient confinement inside the tapered silicon layer, and transmission into the substrate and is finally used to optimize the nanotaper profile. A quadratic-convex shape is identified as an optimal one, yielding a mean specular reflectance as low as 0.2% over the 400–800 nm spectral range for 400 nm-tall nanotapers with a 150 nm base diameter arranged in a hexagonal lattice. Overall, this combined experimental, stochastic, and periodic-modal framework provides a physically transparent route for interpreting black-silicon antireflection and for designing ultra-low-reflectance dielectric interfaces based on silicon nanotapers.
Rumyantseva et al. (Fri,) studied this question.