Triple-layer unclonable anti-counterfeiting enabled by huge-encoding capacity algorithm and artificial intelligence authentication

Abstract As a fundamental security problem, counterfeits pose a tremendous threat to public health and social economy. Herein, we exploit multi-functional nanoinks made of one-dimensional silicon-based nanohybrids for constructing fluorescent and plasmonic security tags. Of particular significance, the presented security solution exhibits triple-layer authentication model, simultaneously featuring the advantages of physical unclonable functions (PUFs), huge-encoding capacity algorithm and artificial intelligence technique. In macroscale, the multi-color fluorescence security signals are used as the first layer, which can be verified through portable smartphone. In the second security layer, the unclonable surface-enhanced Raman scattering (SERS) security signals at low-level magnification could be visualized using confocal Raman system. Taking advantages of coarse grained and quaternary encrypting of signals from Raman at each pixel, the encoding capacity reaches 6.43 × 1024082, which is much higher than the value (i.e., 3 × 1015051) ever reported. In the third layer, the aggregated SERS signals at high-level magnification Raman mapping produce unrepeatable patterns with shape-specific information. By further applying specifically artificial intelligence (AI), faint features of different SERS images are extracted and trained, allowing 98–100% of recognition accuracy after 1000 learning cycles. Such triple-layer security solution ensures the PUFs, huge encoding capacity and AI authentication simultaneously, providing newly high-performance platform of unbreakable anti-counterfeiting.