Tunable grating surfaces with high diffractive efficiency optimized by deep neural networks.

Area of Science:

• Optics and Photonics
• Artificial Intelligence in Engineering
• Materials Science

Background:

• Diffractive gratings are essential optical components for manipulating light.
• Achieving high diffractive efficiency across the visible spectrum remains a challenge.
• Traditional design methods are often limited in optimizing complex grating structures.

Purpose of the Study:

• To predict and design novel grating surfaces with high diffractive efficiency using deep neural networks (DNNs).
• To achieve >99% diffractive efficiency for the -1st order within a 100 nm bandwidth in the visible light spectrum.
• To demonstrate the tunability of grating performance by altering the grating period.

Main Methods:

• Utilized deep neural networks (DNNs) for the inverse design of grating surfaces.
• Simulated and analyzed the diffractive efficiency of the predicted grating structures.
• Investigated the effect of grating period on the operational wavelength band.

Main Results:

• Predicted grating surfaces exhibit over 99% diffractive efficiency for the -1st order within a ~100 nm visible light bandwidth.
• The DNN-aided design strategy proved efficient and feasible for optical device development.
• Achieved >90% diffractive efficiency across the entire visible light spectrum by tuning the grating period.

Conclusions:

• Deep neural networks offer a powerful approach for designing high-performance optical devices like diffractive gratings.
• The developed AI-aided design strategy significantly surpasses previous grating structures in efficiency and bandwidth.
• The tunability of the predicted gratings opens possibilities for broadband visible light applications.