Dataset of aluminum nanostructures for the inverse design of optical few-layer metasurfaces

This dataset contains 46000 sets of data, describing the relationship between the structural configuration of single layer aluminum nanostructures (embedded in the medium composed of SU-8 photoresist) and their scattering matrix under normal incidence. The dataset was collected by using numerical simulation based on Finite Integration Technique (FIT) method. Each dataset consists of two files: one file describes the structural configuration of the aluminum nanostructure using a 1×25 matrix, while the other file characterizes the elements of the corresponding scattering matrix (150-300 THz) under normal illumination by a 151×33 matrix. The dataset provides a more detailed description in the "readme.doc" file within the dataset. This dataset can be utilized for training Convolutional Neural Networks to enable the prediction of optical spectra of aluminum nanostructures or the inverse design of aluminum nanostructures based on given optical spectra. Furthermore, it can be utilized to achieve the inverse design of few-layer metasurfaces with diverse scattering functionalities. A specific example along with a detailed description can be found in the relevant published paper and its supplementary information.