Revealing switching statistics and artificial synaptic properties of Bi2S3 memristor

초록

Complex information processing in neuromorphic systems relies on artificial neurons and synapses as fundamental components. Frequently, memristors are utilized as artificial synapses due to their straightforward configurations, capacity for gradual conductance modulation, and compatibility with high-density integration. The present study reports a novel Ag/Bi2S3/FTO memristor, fabricated using the arrested precipitation technique (APT) based solution processable method. Characterization techniques, including XRD, Raman scattering, and FESEM with EDS, were employed to evaluate the properties of Bi2S3. The device exhibited robust, forming-free, non-volatile resistive switching at low voltages (SET:-0.58 V and RESET: 0.42 V) with an endurance exceeding 6 x 103 cycles and retention times greater than 1.5 x 104 s. Moreover, switching variability was modeled using different statistical distribution techniques. It can mimic learning and forgetting behaviors and different forms of spike-timing-dependent plasticity, akin to its biological counterpart. The trap-filled SCLC mechanism dominated the charge transport in the device. This work introduces new material for investigating low-power consuming electronic devices which holds significant potential for future applications in artificial intelligence electronics and neuromorphic computing systems.

키워드