Dynamic Control of Plasmonic Localization by Inverse Optimization of Spatial Phase Modulation

Abstract

We present a systematic inverse design approach to achieve digitally addressable plasmonic metasurfaces. Beyond existing literature, we adopt a variety of input phase profiles to control the local optical field distribution on the metasurface. Our inverse design approach relies on three building blocks. First, we model the spatial phase distribution of the incident field using a linear superposition of harmonic functions to generate smooth and flexible phase distributions. Second, we propose a localization scheme to quantify the local optical field concentration on the metasurface. Third, Bayesian optimization is employed to learn the underlying nonlinear mapping and to reduce the number of numerical simulations needed to reach the target hotspot arrangements. The obtained computational designs can be implemented with a spatial light modulator to enable dynamic reconfiguration of diverse high-contrast localization patterns on a metasurface without mechanical scanning. The advanced dynamic addressability could fuel relevant applications beyond proof-of-concept demonstrations.