Physically Unclonable Function by an All-Printed Carbon Nanotube Network

The emerging Internet of things (IoT) demands not only new device technologies but also strengthening the security of things in order to prevent tampering and hacking. Humans have unique characteristics such as fingerprints, irises, and facial features, allowing the use of biornetric authentication technology. IoT related devices in the future would likewise need similar identification methods for security. We demonstrate an analog version of physically unclonable function (PUF) using a carbon nanotube (CNT) network to implement a unique identification for this purpose. A plurality of electrodes dispersed around a single CNT network, formed by printing on a flexible substrate, allows local segmented resistance to be read. A pattern of resistances across an arbitrary pair of electrodes in the CNT network is naturally random as the resistance depends strongly on the type, diameter, and length of the nanotubes and the link and angle between them, and consequently it is nearly impossible to predict or duplicate them. The relative difference from the internal resistance of the CNT network is standardized and visualized, which is demonstrated here to be robust and stable against tampering attacks such as high-temperature baking, light illumination, and radiation exposure. Therefore, the present all-printed analog version of CNT PUF can be implemented by simple and low-cost fabrication to provide identification of things on-demand for the IoT era.