Evolutionary multi-objective optimization for multi-resonant photonic nanostructures

The tailoring of optical properties of photonic nanostructures is usually based on a reference design. The target optical behavior is obtained by variations of an initial geometry. This approach however can be of limited versatility, in particular if complex optical properties are desired. In order to design double-resonant photonic nano-particles, we attack the problem in the inverse way: We mathematically define an optical response and optimize multiple of such objective functions concurrently, using an evolutionary multi-objective optimization algorithm coupled to full-field electro-dynamical simulations. We demonstrate that this approach is extremely versatile, that it allows the consideration of technological limitations and that it yields a correct prediction of the optical response of nano-structures fabricated by state-of-the-art electron beam lithography. We demonstrate the technique on multi-resonant photonic nanoparticles made from silicon, which belong to the emerging class of high refractive index dielectric nanostructures with applications such as field-enhanced spectroscopies.