Achievement

Statistically-unbiased random number generation is a key component for secure communications, cryptography, and computational chemistry, medicine and material science. Pseudo-random number generation from a single seed with deterministic algorithms have wide implications in electronic information systems. Truly random number generators, however, are typically statistically unbiased, and achieved through physical entropy processes such as oscillator frequency jitter, with limited rates of tens of Mbps or less. In this IGERT cross-disciplinary training we demonstrate for the first time random number generation on-chip, through trapped defect states in nanoclusters at oxide-nitride interfaces at 20 uW resonator powers. The stochastic bistable optical states demonstrate inherently unbiased and uncorrelated random bit generation, validated by a battery of statistical test suites, for chip-scale all-optical secure encryption and computational processing modules.