A general optimization framework for nanofabrication using shadow sphere Lithography: A case study on chiral nanohole arrays

Shadow sphere lithography (SSL) offers unparalleled advantages in fabricating complex nanostructures, yet optimizing these structures remains challenging due to vast parameter spaces. This study presents a general optimization framework for SSL-fabricated nanostructures, demonstrated through chiral metamaterials. The approach combines a custom SSL program, a novel mathematical model for eliminating redundant structures, and machine learning (ML) analysis of finite-difference time-domain (FDTD) simulations. Applied to rotated nanohole arrays (RHAs), this framework efficiently navigates a 7200-structure parameter space, identifying optimal configurations with circular dichroism (CD) and g-factor up to 3.23˚ and 0.28, respectively. Experimental validation of optimized RHAs shows good agreement with predictions, exhibiting twice the chiral response of random configurations. Notably, the framework reduces the dataset by 86%, significantly decreasing computational costs. This optimization framework enables faster, more systematic, and more efficient optimization of structures manufactured using SSL, potentially accelerating discoveries in nanophotonics, plasmonics, and chiral sensing applications.