Towards outstanding resistive switching and synaptic behaviors using high-entropy alloys-based memristors for neuromorphic computing

Neuromorphic computing, inspired by biological neural networks, integrates data storage and processing to overcome the von Neumann bottleneck. Despite their potential, memristive devices—key components of neuromorphic systems—face significant challenges, including high operating voltages, limited cycling stability, and variability. This study explores high-entropy alloys (HEAs) as a functional layer in Ag/HEAs (Ag, Ge, Te, Pb, Sb)/TiN-based memristors. The proposed device exhibits exceptional performance, including ultra-low SET/RESET voltages (0.21 V/−0.15 V), high DC cycling endurance (>900 cycles), and extended retention times (>2 × 10⁵ s). It also demonstrates advanced synaptic behaviors like long-term potentiation, long-term depression, paired-pulse facilitation, and spike-timing-dependent plasticity, essential for neuromorphic applications. The device’s synaptic properties were successfully applied to static image recognition and dynamic speech signal recognition. Mechanism studies, combining experimental characterization and first-principles calculations, reveal that the migration of Ag ions controls the behavior of resistance switching. The high-entropy effects of the HEAs stabilize conductive filaments, significantly enhancing device performance. These findings highlight HEAs’ potential in neuromorphic computing and offer valuable insights into their electronic and ionic behaviors, positioning HEAs as a groundbreaking material for reliable, scalable systems in next-generation computing architectures.