A Novel Leaky-FeFET Based True Random Number Generator with Ultralow Hardware Cost for Neuromorphic Application

In this work, a novel Leaky-FeFET (L-FeFET) based true random number generator (TRNG) with only one transistor is proposed and experimentally demonstrated for neuromorphic application. By utilizing intrinsic stochastic dynamic process of ferroelectric domains nucleation, and accelerating polarization degradation, the L-FeFET can generate controllable random bits without peripheral circuits. Based on proposed TRNG, a stochastic neural network is efficiently implemented for solving NP-hard problems. This work provides a promising highly-integrated TRNG solution for neuromorphic systems.