Low-Energy Ion-Implanted Nanometer-Thick Metal Oxide Memristor for Random Number Generation at the Nanoscale

Abstract

Resistive random-access memory (ReRAM) devices have emerged to be a promising source for ample information storage in next-generation nonvolatile memories (NVMs) and true random number generation (TRNG)-based robust cybersecurity. However, cycle-to-cycle performance degradation at the nanoscale results in deviation of the normal probability distribution, limiting ReRAM applicability to micron-size devices. This work demonstrates the efficacy of low-energy Au-ion implantation in nanometer-thick HfO_x films in achieving a superior ReRAM performance at the nanoscale toward TRNG application. The role of Au-ion implantation in a single amorphous HfO_x layer is demonstrated to create high-performance memory cells at the nanoscale. Local probe microscopy analysis provides a spatial identification of homogeneously distributed reproducible conductive filament locations and its one-to-one correlation with surface topography. The Au-implanted films exhibit a highly stable, forming-free bipolar resistive switching, having well-separated set and reset transitions with significantly small fluctuations (as low as 3.6% and 3.2%, respectively). The films display a negligible cycle-to-cycle degradation in low and high resistive states, thereby resulting in an outstanding cumulative probability distribution. The chi-square tests further reveal that the low resistive state current is an excellent physical source of six distinct random states. Hence, the ion implantation on economically viable and fabrication-friendly HfO_x material is found to be a robust technique to fabricate high-performance ReRAM devices for future ultrahigh-integration density NVMs and TRNG sources for cybersecurity applications.