A Memristor-Based Silicon Carbide for Artificial Nociceptor and Neuromorphic Computing

With the advancement of artificial intelligence technology, more and more biological functions need to be imitated to complete more complex tasks and adapt to a complex external work environment. Memristors, as an excellent candidate for neuromorphic artificial electronic devices with many biological functions, have inspired the interest of researchers because of the advantages of scalability, good retention, and high operating speed. In this work, wide band gap semiconductor materials silicon carbide (SiC) films are prepared as a memristor medium. By adjusting the current compliance, both threshold character and bipolar resistive switching phenomenon are realized in one device with both lower powers for set operation. For the threshold characteristic, this device has mimicked the threshold, inadaptation, and relaxation features of a nociceptor, which will protect the artificial intelligence system to have stronger adaptability to the external environment. For the bipolar resistive switching characteristics, this device demonstrates good stability and retention time, with a switching speed of 18 ns. These bipolar resistance switching characteristics have simulated many synaptic functions. Pulses with hundreds of nanosecond time scale widths are conducive to fast learning and calculation. This device-based third-generation SiC semiconductor material will find a broad application in neuromorphic chip systems.

Automated summary

This study successfully achieved both threshold characteristics and bipolar resistive switching phenomenon in one device using silicon carbide films as the memristor medium. By adjusting the current compliance, the device has lower power consumption during set operations and demonstrates good stability and retention time.