Multigenerational Crumpling of 2D Materials for Anticounterfeiting Patterns with Deep Learning Authentication

Abstract

Summary Physical unclonable function (PUF) is a cornerstone of anticounterfeiting. However, conventional PUF key-based secure tags encounter several bottlenecks, such as complicated manufacturing, specialized and tedious readout, long authentication time, and insufficient stability. Here, we utilize various two-dimensional materials (2DMs), including Ti3C2Tx MXene and graphene oxide, to construct multigenerational microstructures as PUF patterns. Two intermediate treatments, cation intercalation and moisture-induced lubrication, are introduced in between sequential contractions to engineer the multiscale patterns in a transfer-free and scalable fashion. A deep learning (DL)-facilitated software is developed to pre-categorize the hierarchical topographies with classifiable features. Thereafter, the search-and-compare is conducted within a smaller database to shorten the overall authentication time. The synergy between 2DM tags and DL-facilitated software enables a reliable and environmentally stable anticounterfeiting technology, DeepKey, showing superior encoding capacity (>10144,494) and short authentication time (∼3.5 min). Our 2DM anticounterfeiting tag is finally integrated with QR codes to provide two-layer information security.