Self-Contained Analog Switching in AgHfO Memristor with Lattice-Sourced Ag Filament for Reliable Synaptic Behavior and Associative Learning.

HfO is one of the most widely studied materials for resistive switching, owing to its CMOS compatibility and scalable performance. However, HfO-based memristors suffer from stochastic filament formation, high device-to-device variability, and limited analog tunability due to abrupt and uncontrolled conductive filament growth, which limits their suitability for neuromorphic computing applications. Moreover, they often require external active electrodes like Ag or Cu to induce reliable switching, limiting material flexibility and integration. To overcome these challenges, AgHfO is investigated, a silver-containing perovskite that inherently supports uniform and stable analog switching without relying on active metal electrodes. Oxygen vacancies formed during deposition, together with lattice silver, create hybrid filaments that drive resistive switching. To enhance reliability, oxygen vacancies are suppressed by introducing additional oxygen during deposition, leading to more stable switching and an improved ON/OFF ratio. Beyond reliable memory behavior, the AgHfO memristor demonstrates synaptic functionalities essential for neuromorphic computing. The device exhibits analog modulation of conductance in response to voltage pulse protocols, effectively emulating key biological learning processes such as potentiation, depression, and paired-pulse facilitation. Furthermore, the device shows associative learning through a Pavlovian conditioning protocol, highlighting its potential as a hardware-implemented artificial synapse in next-generation brain-inspired systems.