A method to produce unclonable cryptographic keys based on self-assembled carbon nanotubes (CNTs) has been developed by Shu-Jen Han and colleagues, as they report in Nature Nanotechnology. Information security is crucial in many aspects of our daily life: it protects private data from unauthorized access and manipulation, thereby enabling secure internet banking, protection of medical records, and the safe use of smart devices and applications that send sensitive information through the internet. Hard-to-forge data encryption is thus of foremost importance.
The perfect key is one that is easy to make but impossible to reproduce. Cryptographic keys are fed as inputs in encryption engines (along with other information) to generate encrypted outputs that can be safely transmitted. In most current technologies, the keys are stored in silicon memories, which are vulnerable to counterfeiting and information theft. “The lack of high-quality, on-chip cryptographic keys remains the major bottleneck for improving chip security and assurance,” explains Han. “Our work demonstrates a new ‘perfect’ cryptographic key constructed using CNTs, exploiting the inherent ‘imperfections’ of this nanomaterial as the code.”
“Our work demonstrates a new ‘perfect’ cryptographic key constructed using CNTs, exploiting the inherent ‘imperfections’ of this nanomaterial as the code”
The researchers placed CNTs, dispersed in an aqueous solution, on a patterned HfO2/SiO2 surface functionalized with a positively charged molecular monolayer that selectively self-assembles on the HfO2 trenches. In this approach, CNTs are coated in a negatively charged surfactant and thus feel a strong Coulombic attraction to the positively charged molecular monolayer in the HfO2 trenches; by contrast, a repulsive force is established between the CNTs and the negatively charged SiO2 regions. The competition between the attractive and repulsive interactions can be tuned by varying the trench dimensions, which the researchers optimized using numerical simulations to precisely control the randomness of CNT placement. If a trench is occupied by a CNT, current can flow through it and the trench can be seen as a ‘1’ bit; otherwise, the trench is not connected and constitutes a ‘0’ bit.
CNT synthesis typically leads to a mixture of metallic and semiconducting nanotubes. Han and colleagues demonstrated that it is possible to discriminate between trenches occupied by the two kinds of CNT; this way, the binary bits can be upgraded to ternary bits, which are rare in programmable devices. As a result, the number of possible bit combinations in the key is greatly increased, which provides a much higher level of security with a bit array of the same size.
Owing to their stability, these keys are insensitive to environmental noise and to changes in operation temperature. “Because CNTs are so small, it is almost impossible to reveal the information contained in the key without damaging it; moreover, this technology is highly reliable and prevents the alteration of bit information. This system is a perfect cryptographic key,” says Han. These CNT-based keys also have the advantage of being compatible with any substrate — for example, they could be used in flexible devices — and can be fabricated on a large scale using low-cost and low-temperature processing.
The demonstration of these nanotechnology-based, unclonable electronic security devices will lead to innovative data-protection technologies that go beyond traditional silicon approaches. “So far, we have demonstrated the assembly of static random structures using CNTs,” explains Han. “It is useful for constructing secure identification. We are now exploring dynamic random properties of nanomaterials. If we can find a way to control the dynamic randomness, it will be possible to use nanomaterials to construct random number generators — they would have a great impact, because they are at the heart of modern hardware security.”
Hu, Z. et al., Physically unclonable cryptographic primitives using self-assembled carbon nanotubes. Nat. Nanotechnol. http://dx.doi.org/10.1038/nnano.2016.1 (2016)