Thickness-dependent resistive switching in engineered TiN/Ti/HfO2/W memristors

We systematically investigated TiN/Ti/HfO 2 /W memristive architectures by varying the HfO 2 and the Ti layer thickness ratios to optimize memory window characteristics and switching stability. Devices with balanced thickness configurations between the Ti and the HfO 2 layers achieved optimal performance with memory windows exceeding two orders of magnitude and superior cycle-to-cycle stability. Kinetic Monte Carlo simulations revealed that optimal memory window performance correlates directly with controlled oxygen vacancy-ion dynamics during forming, set, and reset processes by the thickness of both layers. A Ti/HfO 2 thickness ratio close to one provides an optimal balance between oxygen ion generation in the HfO 2 layer and ion storage in the Ti reservoir, which spontaneously becomes TiO x previously. When the TiO x layer is relatively thin, compared to the HfO 2 layer, a large number of oxygen ions are confined near the HfO 2 /TiO x interface. As a consequence, during the reset process, a large number of oxygen ions can recombine simultaneously with oxygen vacancies, significantly increasing cycle-to-cycle variability. On the contrary, a thicker Ti layer allows oxygen ions to migrate excessively far from the interface, reducing the effective recombination rate during the reset process. These dynamics also depend on the number of oxygen ions available, which is correlated to the HfO 2 layer thickness. This balance explains the superior stability and enhanced memory window observed in devices with near-unity thickness ratios. Our findings establish quantitative design principles for bilayer oxide memristors, demonstrating that thickness ratio optimization enables precise control over memory window characteristics and switching stability for practical memory applications. • Optimal Ti/HfO 2 1:1 ratio achieves optimal memory window with superior cycle-to-cycle stability. • KMC simulations reveal oxygen vacancy dynamics correlate with thickness ratio optimization. • Ti scavenging layer controls oxygen storage; HfO 2 thickness governs ion generation. • Thickness engineering reduces reset variability, enhancing the memory window for practical devices.