Laser interaction with solids is routinely used for functionalizing materials' surfaces. In most cases, the generation of patterns/structures is the key feature to endow materials with specific properties like hardening, superhydrophobicity, plasmonic color-enhancement, or dedicated functions like anti-counterfeiting tags. A way to generate random patterns, by means of generation of wrinkles on surfaces resulting from laser melting of amorphous Ge-based chalcogenide thin films, is presented. These patterns, similar to fingerprints, are modulations of the surface height by a few tens of nanometers with a sub-micrometer periodicity. It is shown that the patterns' spatial frequency depends on the melted layer thickness, which can be tuned by varying the impinging laser fluence. The randomness of these patterns makes them an excellent candidate for the generation of physical unclonable function tags (PUF-tags) for anti-counterfeiting applications. Two specific ways are tested to identify the obtained PUF-tag: cross-correlation procedure or using a neural network. In both cases, it is demonstrated that the PUF-tag can be compared to a reference image (PUF-key) and identified with a high recognition ratio on most real application conditions. This paves the way to straightforward non-deterministic PUF-tag generation dedicated to small sensitive parts such as, for example, electronic devices/components, jewelry, or watchmak.
Keywords: chalcogenide materials; laser-matter interaction; neural networks; physical unclonable functions; silicon nitride.
© 2020 Wiley-VCH GmbH.