Broadband reflectors with a disordered layered structure: statistical properties of high performing configurations selected via genetic algorithm

Disordered multilayers consisting of alternating layers of two lossless dielectric materials with random thicknesses can behave as good reflectors in wide wavelength ranges except for narrow bands where the transmittance is significative. We use a dedicated genetic algorithm to select different configurations (thickness sequences) of such structures which exhibit very low transmittance in the entire visible wavelength range, showing that broadband disordered reflectors with very high performance can be designed. A statistical analysis of the thickness sequences selected by the genetic algorithm reveals that such sequences are characterized by correlated disorder and that the degree of autocorrelation is a key parameter in determining the reflection performance.