Self-Limiting Oxide Dielectric Enables Interface-State Engineering in p-Type m-GaTe Transistors Toward Bio-Inspired Visual Recognition.

The development of efficient artificial intelligence hardware has spurred interest in neuromorphic systems that unify sensing, memory, and computing. Optoelectronic synaptic transistors are appealing due to their rapid response times, broad spectral range, and low energy consumption. However, most 2D material-based systems require complex structural and interfacial engineering, which limits scalability and stability. Moreover, the lack of air-stable p-type 2D semiconductors hinders the creation of energy-efficient architectures. In this study, we present monoclinic GaTe (m-GaTe), an intrinsically p-type layered semiconductor, as a robust alternative. Through chemical vapor deposition, we achieve orientation-controlled epitaxial growth of high-quality m-GaTe, and reveal a unique self-limiting surface oxidation. This native oxide serves as both an encapsulation layer and a dielectric, introducing interfacial charge-trapping states that facilitate effective modulation of hole transport in p-type m-GaTe transistors. By exploiting these oxide-mediated properties and engineered contacts, we achieve tunable memristive switching and synaptic plasticity in a single p-type transistor. More importantly, the strong ultraviolet photoresponse of m-GaTe supports a device-array-based sensing system that, when paired with convolutional neural networks, excels in ultraviolet feature extraction and bio-inspired visual perception. This work offers a scalable and reliable pathway to integrated opto-neuromorphic systems.