All-optical diffractive neural networked terahertz hologram

Area of Science:

• Optics and Photonics
• Artificial Intelligence
• Imaging Technology

Background:

• Holography enables full-wave light reconstruction but conventional design methods are time-consuming.
• Artificial intelligence approaches for hologram design are underexplored.
• Terahertz (THz) imaging offers unique applications but requires efficient holographic elements.

Purpose of the Study:

• To propose and demonstrate a novel 3D-printed hologram for THz imaging using a diffractive neural network (DNN).
• To accelerate the design process of holographic elements through AI.
• To achieve high-quality THz imaging with a compact and efficiently fabricated device.

Main Methods:

• A diffractive neural network (DNN) was trained to compute phase profiles for holographic imaging.
• The DNN design process, including target image and input field pattern training, was completed in 60 seconds.
• The resulting hologram, a 2D array of dielectric posts, was fabricated using 3D printing technology.

Main Results:

• The DNN successfully generated quantified phase profiles for high-quality THz imaging.
• Full-wave simulations and experimental results validated the hologram's performance in the THz regime.
• The fabricated hologram achieved clear imaging of the target letters 'THZ' with uniform amplitude.

Conclusions:

• The proposed DNN-based design strategy offers a rapid and effective method for creating advanced holographic elements.
• 3D printing enables the fabrication of complex diffractive structures for THz applications.
• This approach holds potential for future innovations in display technology, optical data storage, and optical encryption.