A topological microstructure-based Bi/Al-Cu memristor

Memristor has emerged as a potential for neuromorphic computation to overcome von Neumann bottleneck for computing in memory. However, the resistance states of the conventional filament-type memristors modulated by conductive filaments have stochasticity in devices and cycles, and interface-type memristors need a complex design to achieve a low ON/OFF ratio via the Schottky barrier. This study introduces a multi-level storage memristor based on a topological microstructure of semimetal bismuth (Bi) and aluminum copper alloy (Al-Cu). The electrical measurements under an applied sweeping voltage, cause Bi particles to migrate in the interfacial microstructure and subsequent modulation of the resistive states between high-resistance (HRS) and low-resistance states (LRS). The memristor exhibits performances, including consistency, repeatability, ON/OFF ratio of >3 × 10 3 , endurance >2000 cycles, retention, and multi-state storage. This research provides an alternative solution for designing simple-structured memristors and advancing hardware for artificial intelligence applications. • Bi/Al-Cu memristor with topological microstructure shows Bi-migration-enabled switching. • Depositing Bi before Al-Cu forms the topological microstructure with an initial high-resistance state. • Bi migration modulates resistance, enabling potential multilevel data storage.