Time-Effective Simulation Methodology for Broadband Achromatic Metalens Using Deep Neural Networks.

Area of Science:

• Optics and Photonics
• Materials Science
• Computational Electromagnetics

Background:

• Metasurfaces offer novel optical properties but designing them, especially broadband achromatic metalenses, is time-consuming.
• Current design methods rely on extensive pre-calculated data libraries derived from electromagnetic simulations.
• The computational cost of generating these libraries limits the design of complex metasurfaces.

Purpose of the Study:

• To develop a time-effective and accurate method for designing complex metasurfaces, specifically broadband achromatic metalenses.
• To overcome the limitations of traditional design approaches that require huge data libraries.
• To improve the focusing efficiency of metalenses through an augmented design library.

Main Methods:

• Utilized deep neural networks to train a data library for metasurface design.
• Combined different numbers of nanofins to generate augmented data points.
• Employed an augmented data library for designing a broadband achromatic metalens.

Main Results:

• The deep neural network model predicted approximately ten times more data points than traditional methods.
• The augmented library led to a broadband achromatic metalens with up to 45% focusing efficiency in the visible spectrum.
• The proposed method demonstrated significant time savings and accuracy for complex electromagnetic problems.

Conclusions:

• Deep neural networks can effectively augment metasurface design libraries, reducing design time.
• The data-augmented approach enhances the performance of broadband achromatic metalenses.
• This method offers a promising, efficient, and accurate solution for designing complex electromagnetic devices.