Encryption and Privacy Safeguarding via True Random Number Generation Using Structurally Metastable Solution-Processed 1T’ MoTe₂

Cryptography is of critical importance in the modern electronics era when the exponentially growing data is at risk of being attacked and sabotaged. True random numbers play a critical role in secure cryptography, where the generation relies on a stable and readily extractable entropy source. Here, from solution-processed structurally metastable 1T’ MoTe₂, we prove stable output of featureless, stochastic yet stable conductance noise at a broad working temperature with minimal power consumption for true random number generation. We prove cryptographic applications of the random numbers in secure encryption and privacy safeguarding.<br/><br/>We start with MoTe₂ exfoliation vie electrochemical exfoliation (ECE). This gives scalable few-layer thick monoclinic 1T’ MoTe₂ nanosheets that are structurally metastable. This structural metastability can potentially induce volatile, stochastic polarization of the underlying ferroelectric dipoles that can lead to stochastic noise in the electronic properties of the 1T’ MoTe₂ nanosheets., With this understanding, we develop simple vertically structured devices where the 1T’ MoTe₂ nanosheets are sandwiched between gold electrodes, and then probe the current noise from the devices as the entropy for true random number generation. Notably, we prove stable conductance noise probing with an ultra-low voltage (0.05V) and an ultra-low power consumption (0.05µW) at a broad working temperature from 15K to 370K. Our detailed characterizations and statistical analysis of the conductance noise characteristics indeed suggest that the noise arises from the volatile stochastic polarization of the underlying ferroelectric dipoles in the 1T’ MoTe₂ nanosheets. Further, as proved in our experiments and indicated by our Monte Carlo simulation, the ferroelectric dipole polarization is a reliable entropy source with the stochastic polarization persistent and stable over time. This allows us to design a simple circuit to extract this stable entropy noise for the true random number generation. Particularly, the circuit is designed without compromising the randomness of the noise such that the generated true random numbers can successfully pass the NIST test. Exploiting the conductance noise, we achieve high throughput generation (1M bit/s) of random numbers.<br/><br/>Using the random numbers, we prove common cryptographic applications, for example, password generation and data encryption/decryption (including pictures, audios, and videos). Besides, particularly, we show a privacy safeguarding approach to sensitive data that can be critical for the cryptography of neural networks. Briefly, to prove privacy safeguarding, the random numbers are injected as noise to the target sensitive data for masking. We demonstrate that the noise can disrupt the feature detection capability of well-trained Resnet variant models while the perturbations are not discernible to the human eyes. Our safeguarding approach therefore can serve as a novel privacy protection measure to avoid the leakage of the critical data without causing destructive interferences to the neural networks.