Inverse design of lateral hybrid metasurfaces structural colour: an AI approach

Area of Science:

• Nanophotonics
• Materials Science
• Artificial Intelligence

Background:

• Metasurface structural color design traditionally relies on simulations and human intuition, limiting optimization efficiency.
• Developing active metasurfaces with tunable colors presents significant design challenges.

Purpose of the Study:

• To introduce an AI-assisted design process for metasurfaces that bypasses complex simulations.
• To demonstrate a unified model for generating multiple colors from a single geometry via strain, facilitating active metasurface development.

Main Methods:

• Developed an AI-assisted design process for metasurface structural color.
• Established a swift and precise mapping between metasurface parameters and color coordinates.
• Utilized varying levels of strain on a single geometry to generate multiple colors.

Main Results:

• Achieved AI-assisted design, eliminating the need for extensive simulations.
• Demonstrated that a single metasurface geometry can produce multiple colors by applying strain.
• Created a unified AI model for designing active metasurfaces considering geometry and dynamic responses.

Conclusions:

• The AI-assisted approach accelerates metasurface design and optimization.
• The ability to generate multiple colors from a single geometry under strain enables novel active metasurface functionalities.
• This work paves the way for practical applications of active metamaterials.