City University of Hong Kong Scientists Develop 3D Metastructures for Time-Programmable Optical Encryption and Self-Destructing Data

Scientists develop 3D metastructures that enable time-programmable structural colors and self-destructive "burn-after-reading" optical encryption.

By: AXL Media

Published: Mar 26, 2026, 4:56 AM EDT

Source: Information for this report was sourced from the Light Publishing Center, Changchun Institute of Optics, Fine Mechanics and Physics, CAS.

Addressing the Vulnerabilities of Electronic Encryption

As the era of quantum computing approaches, traditional electronic encryption—which relies heavily on mathematical algorithms—is becoming increasingly vulnerable to brute-force attacks. In response, scientists are turning to optical encryption grounded in physical principles. By exploiting high-dimensional variables such as wavelength, polarization, and phase, optical systems can reduce decryption risks at the source. The City University of Hong Kong team has advanced this field by moving beyond simple binary "on-off" color switching to a continuous color-gamut control system that uses micro- and nanostructures to manipulate light with extreme precision and stability.

Geometric Programming and Structural Color

The core of the innovation lies in the design of "meta-atoms"—the individual building blocks of the 3D metastructure. By accurately programming the geometric parameters of these meta-atoms, the researchers can predictably tune structural colors across the entire visible spectrum. Unlike traditional pigments that fade over time, these structural colors are produced by the physical interaction of light with the nanostructure's geometry. This allows for high-resolution monochromatic and multicolor printing that remains consistent and reproducible, providing a robust physical substrate for complex information encoding.

Deep-Learning Recognition for Anti-Counterfeiting

To enhance the practicality of these metastructures, the team developed a deep-learning-based recognition scheme. Using a convolutional neural network (CNN), the system can automatically extract spatial-texture and color-distribution features from structural-color labels. This allows for rapid and accurate authentication, even under challenging real-world conditions such as defocus, rotation, background variations, or local contamination. This AI-driven approach offers a significant improvement over manual threshold settings, making the technology highly reliable for high-security anti-counterfeiting applications in commercial and military sectors.