Stochastic-Resilient NEMS: A Discretized Architecture for Phase-Change Frequency Tuning

Phase-change nanoelectromechanical systems (NEMS) resonators have long promised a breakthrough in telecommunications: a single device capable of tuning across the entire Global System for Mobile Communications (GSM) frequency band. However, this promise has been stalled by a fundamental material conflict. Existing designs rely on continuous thin films, treating the phase-change material as a smooth, tunable variable. In reality, at the nanoscale, crystallization is chaotic, dominated by random, lightning-bolt-like filaments rather than uniform growth. This ``analog'' stochasticity creates unpredictable frequency jitter, rendering the devices unusable for precise filtering. In this study, we propose a solution that embraces, rather than fights, this physical reality: the Discretized Nanodot Array (DNA). By replacing the continuous film with a high-density grid of isolated nanodots, we effectively convert the device from an unpredictable analog system into a reliable digital one. Using Monte Carlo simulations (\((N=2000)\)), we demonstrate that this architectural shift forces statistical averaging, taming the random nucleation noise. The result is a 79.28% reduction in frequency variance and a 4.8x improvement in stability compared to traditional thin-film designs. These findings suggest that the future of tunable NEMS lies not in perfecting materials, but in patterning them, moving from chaotic films to ordered, digital pixels.