(177q) Pixelated Physical Unclonable Functions through Capillarity-Assisted Particle Assembly

Enormous economic damage results worldwide from counterfeited or incorrectly attributed products. Moreover, as seen in the pharmaceutical industry, counterfeit products can pose severe dangers in addition to economic harm and intellectual property violations. These challenges have spurred the development of robust strategies to authenticate products. An ideal authentication technology must generate unique identification tokens that are impossible to replicate. Physical unclonable functions (PUFs) are an outstanding approach to ensure these requirements. A PUF is a physical object that, under specific conditions known as the challenge, produces a unique, physically defined output, known as the response, for a given input. Because PUFs are fabricated through unpredictable stochastic processes with random components, they cannot be reproduced within a viable timeframe. The performance of PUFs depend on the uniqueness of physical identifications induced by signals such as electrical, optical, and radio waves. Among various types of PUFs, optical PUFs, such as randomly dispersed fluorescent organic crystals and microparticles, liquid crystals, and optical random scattering patterns offer benefits due to their chaotic randomness and imperfections in microstructures and manufacturing processes, resulting in high entropy and uniqueness.

Typically, optical PUFs are generated on a micro-nanoscale to achieve high information density by isolating a specific region within a large random pattern, which serves as the unique token of interest. When reading these PUFs, the captured image must be authenticated by matching it with the original. Often, alignment markers are used to facilitate authentication, yet they inherently introduce a degree of mismatch between the original and subsequent readouts. Even minor alignment discrepancies result in the input and output images never being exactly identical, making them susceptible to changes over time and damage from mechanical or chemical factors. To address these limitations, various efforts have been made to enhance the robustness and stability of existing PUF systems. These efforts include encapsulation, improved system design, material development, amplification of response signals, and, most commonly, refinement of evaluation methods.

In this study, we propose a novel approach using a pixelated system of distinguishable single colloids (see Figure (a)). The token comprises a square array of randomly deposited fluorescent colloids, with each particle representing a single bit. Our method offers several advantages over previous approaches. Firstly, our pixelated PUFs can be easily obtained from commercially available particles, eliminating the need to synthesize specific fluorescent materials. Secondly, embedding the fluorescent dye inside microparticles, which are then encased in a matrix, enhances mechanical and chemical stability compared to direct exposure. Additionally, using fluorescent particles of uniform size simplifies image processing for obtaining the PUF key. Moreover, the stable and distinguishable nature of colloids results in a low false negative rate and an exponentially scalable encoding capacity. Furthermore, prearranging particles into pixels allows for safe token transfer across different materials without data changes.

We fabricate colloidal patterns using capillarity-assisted particle assembly (CAPA), a method where a droplet of a particle suspension is moved over a topographically patterned surface, depositing particles into microscopic cavities or traps. Patterning the cavities into square arrays simplifies token identification compared to previous continuous patterns. As particle deposition is a fully random process, the number of possible unique tokens for an equal mixture of M types of particles scales exponentially with the number of traps (N) in the pattern as M^N. At a laboratory scale, token deposition takes only a few seconds, and scaling up to an industrial setting becomes statistically unreasonable to reproduce within a viable timeframe.

We begin by outlining our fabrication methodology for pixelated PUFs, followed by an analysis of their randomness and uniqueness (see Figure (b)). We evaluate PUF performance by quantifying bit uniformity, entropy, intra-Hamming distance (Intra-HD), inter-Hamming distance (Inter-HD), and error rates. Additionally, we calculate the encoding capacity as a function of key size and particle types (see Figure (c)). We demonstrate that even with deviations from an ideal distribution, the encoding capacity remains largely unaffected. Our pixelated PUFs exhibit robustness and stability against various external stimuli, confirmed through authentication tests post-stimulation (see Figure (d)). Finally, we show the potential for transferring and embedding PUF tokens into transparent support materials for use in consumer products.