ALD-Driven Ultrathin IGZO Neuromorphic Transistors with Defect-Engineered 1/ <i>f</i> Noise for Stable Multimodal Information Encryption

Ultrathin neuromorphic transistors with sub-10 nm indium-gallium-zinc oxide (IGZO) channels and 20 nm Al₂O₃ gate dielectrics have been fabricated via low-temperature (250 °C) atomic layer deposition (ALD), achieving conformal interfaces critical for reliable synaptic functionality. Low-frequency noise analysis reveals that the ALD-derived atomically flat interface (γ = 0.934) generates tunable shallow-level defects, acting as transient trapping sites and forming a controllable 1/<i>f</i> noise source. The devices exhibit exceptional stability, with threshold voltage drift below 330 mV under 4500 s of bias/illumination stress and minimal variation (<10 mV) after 500 mechanical bending cycles. Leveraging these properties, we implement hybrid audio-digit labels for information concealment, achieving 89.58% recognition accuracy, and demonstrate a three-level image encryption system. This work provides a scalable, energy-efficient platform for back-end-of-line-compatible monolithic 3D memory integration, highlighting the interplay between defect engineering and interfacial dynamics in ultrathin oxide semiconductors.