Embedded true random number generators enabled by hexagonal boron nitride memristors

Internet-of-things (IoT) devices require cheap, light, small and reliable true random number generators (TRNG) to encrypt data before transmission. Finding reliable randomness sources to seed the TRNG can be challenging in the presence of trade-offs such as power consumption, stability, and cost of the circuit implementation. We analyse the suitability of the random telegraph noise (RTN) signals of memristors made of two-dimensional (2D) layered hexagonal boron nitride (h-BN). After thorough evaluation of the noise characteristics, the h-BN memristor is connected to an analogue front-end and a microcontroller to fabricate a functional TRNG. Instead of using simulations as proof of potential functionality or bulky laboratory instrumentation, we provide an embedded TRNG which is scalable to the complementary-metal-oxide-semiconductor (CMOS) integration level and experimentally tested for randomness at the application level.