Inverse Design of Nanohole Array Using Parameter-Based Genetic Algorithm for Optical Sensing Applications

The electromagnetic metasurface is a sub-wavelength structure consisting of meta-atoms, which modulates light behaviors. There are two approaches when designing metasurface: forward and inverse design. The forward design approach, the conventional method, is based on the physical understanding of metasurface. On the other hand, the inverse design approaches employ mathematical optimization algorithms or neural networks, which allow for achieving target optical properties without rigorous scientific understanding [1-2]. Most of the current studies that demonstrate the inverse design method start with the pixelating step of design space. Although the pixelating step can bring highly optimized structure based on a large degree of freedom, there are several limitations when pixelating nanophotonic metasurfaces. Since the fabrication technologies for nanoscale patterning are affected by physical limits, the size of a single pixel for inverse design space cannot be smaller than the technical resolutions. In addition, dealing with pixelated design space brings large amounts of calculation, thus requiring memory-efficient programming. Therefore, this research proposes a parameter-based genetic algorithm for the inverse design of materials as an alternative methodology for optimal design.<br/> For practical implementation, a gold nanohole array (NHA) whose reflectance spectrum has a sharp dip at 700 nm was optimized for optical sensing characteristics. Four geometrical parameters of NHA such as hole diameter, pitch, lattice structure, and thickness, were optimized through a genetic algorithm (GA). Mimicking the replication process of chromosomes, GA, a type of evolutionary algorithm (EA), operates by iterating through initialization, selection, crossover, and mutation steps. For each iteration, the reflectance spectrum of suggested NHA geometries was evaluated by finite-difference time-domain (FDTD) simulations. The NHA with optimized geometry was fabricated on a sputtered gold film with chromium as an adhesion layer. The optimized design was patterned using e-beam lithography (EBL) with polymethylmethacrylate (PMMA) as a resist material, followed by a directed etch using argon ion plasma. Then, the NHA structure was characterized using non-contact atomic force microscopy (NC-AFM) and scanning electron microscopy (SEM). The analyte sensing characteristic was also evaluated using the spectrometer in the visible-NIR range. This study demonstrates that the optimum structure can be efficiently obtained through the inverse design, leveraging a parameter-based genetic algorithm.<br/><br/>Keywords: Inverse Design, Genetic Algorithm, Parameter Optimization, Nanophotonic Metasurface, Nanohole Array<br/><br/>This research was supported by the National Research Foundation of Korea (NRF) under the “Korean-Swiss Science and Technology Program” (2019K1A3A1A1406720011), by the NRF grant funded by the Korea government (MSIT) (2023R1A2C2006811), and by the BK21 FOUR (Fostering Outstanding Universities for Research) funded by the Ministry of Education (MOE) of Korea and NRF.<br/><br/>References<br/><br/>[1] Li, Zhaoyi, et al. "Empowering metasurfaces with inverse design: principles and applications." ACS Photonics 9.7 (2022): 2178-2192.<br/>[2] Moon, Seunghwan, Jihun Kang, and Jong-Souk Yeo. "Review of Generative Models for the Inverse Design of Nanophotonic Metasurfaces." Applied Science and Convergence Technology 32.6 (2023): 141-150.