Grain boundary control for high-reliability HfO₂-based RRAM

Recently, neuromorphic computing has emerged as a promising solution to the limitations of conventional von Neumann computing architectures. Two-terminal memristors, particularly resistive random-access memory (RRAM), are gaining attention because of their structural resemblance to biological synapses, enabling the emulation of neuromorphic synaptic operations. Metal oxide-based RRAM leverages the formation and rupture of conductive filaments based on oxygen vacancies for resistive switching. Despite extensive research on conductive filament formation in amorphous and crystalline configurations, understanding of the impact of grain sizes and boundaries on RRAM properties remains limited. In this study, we investigate the influence of grain conditions on addressing challenges such as high operating voltages and large resistance variations during switching operations using a Ti/HfO₂/Pt structure. Additionally, this study extends the application of HfO₂-based RRAM to neuromorphic computing, demonstrating linear synaptic weight updates, which are essential for constructing accurate neuromorphic systems. Our device has better reliability than amorphous HfO₂-based RRAM, which we achieve by precisely manipulating grain sizes and boundaries depending on the annealing conditions to solve cycle-to-cycle and device-to-device variations. Our experimental results suggest the importance of precise grain control for fabricating highly reliable and robust RRAM and artificial synaptic devices.