Optimizing the Nonlinear Optical Response of Plasmonic Metasurfaces

Controlling the nonlinear optical response of nanoscale metamaterials opens new exciting applications such as frequency conversion or flat metal optical elements. To utilize the already well-developed fabrication methods, a systematic design methodology for obtaining high nonlinearities is required. In this paper we consider an optimization-based approach, combining a multiparameter genetic algorithm with three-dimensional finite-difference time domain (FDTD) simulations. We investigate two choices of the optimization function: one which looks for plasmonic resonance enhancements at the frequencies of the process using linear FDTD, and another one, based on nonlinear FDTD, which directly computes the predicted nonlinear response. We optimize a four-wave-mixing process with specific predefined input frequencies in an array of rectangular nanocavities milled in a thin free-standing gold film. Both approaches yield a significant enhancement of the nonlinear signal. Although the direct calculation gives rise to the maximum possible signal, the linear optimization provides the expected triply resonant configuration with almost the same enhancement, while being much easier to implement in practice.