Experimental study demonstrates monolithic five-layer ferroelectric diodes using atomic layer deposition, indicating a viable selector-free path for high-density 3D cross-point memory.
3D cross-point (XPoint) memory is attractive for high-density integration because it retains compact two-terminal cells and bit-level accessibility during vertical scaling. However, further layer scaling of 3D XPoint architectures remains constrained by two coupled challenges: sneak-path leakage and uniform multilayer process integration. Here, we experimentally demonstrate a monolithic five-layer sequential XPoint-compatible integration route co-enabled by hafnia-based ferroelectric diodes (FeDs) and an all-ALD ultrathin stack. In this route, the intrinsic nonlinearity of FeDs suppresses half-select leakage without an external selector, while the ~10 nm-per-tier all-ALD device stack supports repeated monolithic multilayer integration through conformal growth and precise thickness control. Across all layers, the devices exhibit consistent polarization-modulated I–V hysteresis, with a high ON/OFF ratio of ~103, intrinsic nonlinearity up to ~600, programming down to 50 ns, and endurance beyond 107 write/erase cycles. These results establish a selector-free integration route based on FeDs and all-ALD ultrathin stacks for sequential monolithic 3D XPoint memory.
No takes yet. Share an insight, caveat, or question.
Zhou et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: