Jun 2025
Ph.D. Dissertation Defense Committee
Abstract:
Resistive random-access memory (RRAM) stands as a pivotal technology for next-generation non-volatile storage and neuromorphic computing due to its complementary metal-oxide-semiconductor (CMOS) compatibility and scalability. However, conventional RRAM devices face critical limitations in endurance, stemming from stochastic defect dynamics in amorphous dielectrics. Atomristors—RRAM devices leveraging atomically thin two-dimensional (2D) insulators—offer a transformative solution, provided the crystalline dielectric monolayer exhibits ultralow defect density (or even defect-free), to achieve highly reproducible switching behaviour and enhanced endurance.
This thesis advances high-endurance atomristor development through three key contributions: (1) establishing a robust nanoscale electrical characterization protocol for ultra-thin insulating films, using conductive atomic force microscopy (CAFM), (2) benchmarking commercial 2D dielectrics—hexagonal boron nitride (h-BN), and (3) identifying and integrating a high-quality single-crystal monolayer h-BN into CMOS microchips. The resulting on-chip atomristors using the defect-free monolayer h-BN exhibit endurance exceeding 2.7 million cycles—over 100× higher than state-of-the-art RRAM. Overall, this work bridges nanoscale material engineering and practical device integration, positioning atomristors as a scalable and reliable memory solution for future computing technologies.