At the core of the work is a metasurface, an ultrathin optical structure composed of microscopic elements that can precisely control incident light. When illuminated, these engineered surfaces reconstruct holographic images in free space, but traditional configurations have generally stored only a single piece of information per device. To overcome that constraint, the POSTECH team designed a modular diffractive deep neural network in which each metasurface layer plays the role of a neural network layer that processes light as it propagates through the stack.

In this all optical architecture, diffraction and interference act as the computational engine, enabling the system to process and transform information without electronic chips or external power beyond the light source. Each metasurface layer is trained to generate a distinct holographic image at a chosen wavelength, so that one layer can produce an ID pattern while another reconstructs a QR code when illuminated with specific colors of light. As a result, multiple independent data channels can coexist within the same physical platform, with each layer operating as a standalone optical module.

The security capability becomes more powerful when layers are combined at carefully defined separations. When two metasurface layers are aligned with a precise interlayer spacing and illuminated with a designated wavelength, the system reconstructs an encrypted hologram corresponding to a password or other sensitive data. If either the wavelength or the distance between layers is even slightly off target, the encrypted pattern does not appear and the information remains concealed.

In this scheme, the color of the light and the exact spacing between layers jointly serve as a physical password that is not represented in digital form. The researchers note that the number of possible information channels grows rapidly as the design scales, because in theory the number of channels increases as m(2^n-1), where m is the number of available wavelengths and n is the number of metasurface layers. This exponential growth implies that a single stacked metasurface device can host a very large set of independent and combinational holographic outputs.

The team highlights that the same platform supports standalone layer wise functions, multiwavelength operation, and combinational cipher outputs in a single integrated configuration. This flexibility makes the technology attractive for secure identity cards, anti counterfeiting labels on passports and high value products, and protected military or diplomatic documents. It could also underpin new schemes for optical communications in which critical information is encoded in wavelength combinations and layer configurations rather than in conventional electronic signals.

By exploiting the intrinsic properties of light as a security key, the researchers argue that the approach can help address fundamental weaknesses in software based encryption that remains vulnerable as long as it exists as code. "By using the physical properties of light itself as a security key, this study could fundamentally reshape the paradigm of conventional digital security," said Professor Junsuk Rho of POSTECH. "As digital technologies become more advanced, our results highlight that physical security can ultimately provide the strongest solution."

The work was carried out by Professor Junsuk Rho and colleagues at POSTECH across the Department of Mechanical Engineering, Department of Chemical Engineering, Department of Electrical Engineering, and the Graduate School of Convergence Science and Technology. The research appears in the journal Advanced Functional Materials under the title "Recomposable Layered Metasurfaces for Wavelength-Multiplexed Optical Encryption via Modular Diffractive Deep Neural Networks" and received support from the POSCO POSTECH RIST Convergence Research Institute and the National Research Foundation of Korea.

Research Report:Recomposable Layered Metasurfaces for Wavelength-Multiplexed Optical Encryption via Modular Diffractive Deep Neural Networks

Related Links
Pohang University of Science and Technology (POSTECH)
Cyberwar - Internet Security News - Systems and Policy Issues