Quasi‐digital memristor with self‐rectifying and synaptic functions in crossbar array architectures

The advancement of edge neuromorphic computing hinges on developing memory devices that combine high integration density with intelligent functionality. However, high‐density crossbar arrays, a conventional architecture for such systems, are often plagued by inherent sneak‐path currents. Furthermore, achieving multifunctional integration of analog and digital bipolar resistive switching (RS) within a single device remains a significant challenge. Here, we report a quasi‐digital memristor (QDM) based on an Ag/AgI/PbI 2 /ITO architecture, which integrates tri‐functional attributes: diode‐like rectification, digital switching, and analog synaptic modulation. High self‐rectifying ratio and selectivity are achieved through the strategic design of a p‐n junction at the AgI/PbI 2 Interface. The switching mechanism is governed by the field‐driven migration of dopant ions (e.g., Ag + , iodine vacancies) within the functional layers. In the analog mode, a synaptic array constructed from QDMs demonstrates a recognition accuracy of 97.2% for 8 × 8‐pixel letter images (A–Z) after only six training epochs. In the digital mode, the QDM exhibits a high ON/OFF ratio, robust data retention, and excellent cycling endurance. Notably, the device also features a record‐low set voltage, enabling superior system‐level energy efficiency. This work presents a novel and effective pathway toward precise, high‐density information processing in neuromorphic computing systems.