Second harmonic generation enhanced with inversely designed plasmonics for nonlinear metamaterials

Traditional processes for the design of metamaterial structures are often computational heavy, time-consuming, and occasionally does not lead to the desired optical response. Deep learning can quickly optimize structures through inverse design, and create new geometries for devices. This research uses a deep learning framework for the inverse design of an optimal plasmonic structure to maximize the second-order nonlinear response from a nonlinear metamaterial. The thinfilm nonlinear metamaterial employed is a nanolaminate, and the optimal plasmonic structure is fabricated to establish the validity of the deep learning algorithm.