Crystallographic Parity and Inverse Area Scaling in Two-Dimensional Perovskite Ferroelectric Tunnel Junctions

Ferroelectric tunnel junctions (FTJs) are promising for next-generation high-density non-volatile memory; however, conventional polycrystalline oxide FTJs face issues with defect-induced leakage and degraded ferroelectricity at nanometer scales. Here, we demonstrate atomically defined nanocrossbar FTJs based on single-crystalline two-dimensional (2D) Dion–Jacobson perovskite nanosheets designed to address these scaling challenges. Electrical transport measurements reveal crystallographic-parity-dependent switching behavior: odd-layered (m = 3, 5) homologs exhibit hysteretic switching consistent with ferroelectric tunnel barriers, whereas even-layered (m = 4, 6) homologs show non-switching behavior consistent with paraelectric controls. Simmons-model analysis supports direct tunneling as the dominant transport mechanism through the atomically uniform nanosheet barriers, even at the smallest junction area of 520 nm². Notably, the tunneling electroresistance (TER) ratio increases as the junction area decreases, reaching approximately 35, consistent with the suppression of defect-induced OFF-state current inhomogeneity under nanoconfinement. These FTJs show strong non-volatile retention (extrapolated to 10 years) and reliable switching endurance (100 cycles) at room temperature. These results highlight that crystallographic layer parity and nanoconfinement-enabled defect avoidance serve as key design principles for scalable 2D ferroelectric tunnel devices and facilitate pathways toward ultra-dense post-Moore memory and neuromorphic computing architectures. This favorable scaling behavior aligns lateral miniaturization with improved memory contrast and, together with the metal-electrode nanocrossbar architecture, provides a promising basis for high-density integration if wafer-scale nanosheet assembly and reproducible interface control can be established.