Study of Threshold Switching behavior of Pt/HfO2 /TiN cell and its application in True Random Number Generator

Abstract

The novel technologies like the Internet of Things (IoT) raise the security concerns because of the sensitive information they will handle. That makes the random number generator become one important role in the protection of privacy, which could create one unique signature for each party. And the reliability of this kind of authentication highly depends on how random number generator is. Usually pseudo-random number generator (PRNG) strongly depends on its algorithms or codes, which results in the easy attack. Especially for the sequences they generate could be predictable from their seed value that is fed into generator at the start. That makes true-random number generator (TRNG) become the main role in information security. Memristor can show resistive switching behavior under proper bias conditions, however, the nonuniformity has troubled people to push the next generation memory. This kind of chaos in switching performance results from the stochastic physical characteristics and other complex mechanisms. The Pt/HfO2/TiN memristor shows a large change scale in switching parameters, which is a big problem for memory application, but a great superiority in the area of the random number generator. The electron transportation of this Pt/HfO2/TiN memristor, which is explained by trapping / detrapping mechanism, is mainly attributed to trap-assisted-tunneling process. The cell shows threshold switching behavior under a low compliance current due to limited the number of total injected electrons. And the trap-assisted-tunneling mechanism is related with temperature and distance between intrinsically existed traps in oxide layer, resulting in its stochastic delay time and relaxation time. One true random number generator based this Pt/HfO2/TiN memristor is proposed. And in this proposed TRNG, the Pt/HfO2/TiN memristor is used as seed provider and plays the role of the entropy source. Its stochastic physical characteristics are fully used by transformation of Linear Feedback Shift Registers feedback function in the new circuit. The 4-stage LFSR structure is used and restructured to achieve one more complex RNG circuit to avoid attacking. The output from 4-stage LFSR has been collected and digitized by MATLAB coding. The binary output bits have been run in NIST randomness tests. And the data collected from proposed TRNG has passed all 15 National Institute of Standards and Technology randomness tests, indicating Pt/HfO2/TiN memristor would be perfect candidate for hardware security application.