Dual-function switchable terahertz metamaterial device assisted by a deep neural network.

Area of Science:

• Metamaterials
• Terahertz (THz) technology
• Phase-transition materials

Background:

• Conventional THz devices often lack multifunctionality.
• Vanadium dioxide (VO2) exhibits a tunable insulator-to-metal phase transition.
• Optimizing metamaterial designs traditionally requires extensive computational resources.

Purpose of the Study:

• To propose a novel dual-functional switchable THz metamaterial (MM) device.
• To achieve broadband absorption and polarization conversion capabilities.
• To explore the application of deep neural networks (DNNs) in optimizing metamaterial design.

Main Methods:

• Utilizing the insulator-to-metal phase transition of VO2.
• Designing a metamaterial structure for THz applications.
• Employing deep neural networks (DNNs) for efficient parameter space navigation and optimization.

Main Results:

• The device demonstrates broadband absorption (3.93 THz bandwidth, ≥90% rate) in its metallic state.
• In its insulating state, it provides linear-to-cross-polarization (LTX) conversion (1.17-3.63 THz) and linear-to-circular polarization (LTC) conversion (1.04-1.14 THz and 3.71-3.87 THz).
• DNNs significantly reduced computational time compared to traditional optimization methods.

Conclusions:

• The proposed switchable THz MM device offers dual functionality, overcoming single-device limitations.
• The integration of VO2 and DNNs presents a novel and efficient approach to designing advanced functional materials.
• The device shows potential for applications in THz stealth, energy harvesting, communication, and imaging.