Silk Protein Based Volatile Threshold Switching Memristors for Neuromorphic Computing

Memristors based neuromorphic devices can efficiently process complex information and fundamentally overcome the bottleneck of traditional computing based on von Neumann architecture. Meanwhile, natural biomaterials have attracted significant attention for biologically integrated electronic devices due to their excellent biocompatibility, mechanical flexibility, and controllable biodegradability. Thus, biomaterial-based memristors may have a transformative impact on bridging electronic neuromorphic systems and biological systems. However, the working voltage in biological system is low, but the operation voltages of conventional memristors are high, violating the energy-efficient biological system. Here, high-performance silk fibroin-based threshold switching (TS) memristors are demonstrated, which reveal an on-current of 1 mA, a low threshold voltage (V-th) of 0.17 V, a high selectivity of 3 x 10(6), and a steep turn-on slope of <2.5 mV dec(-1). Meanwhile, the silk TS devices depict outstanding device uniformity and stability even at high humidity (80%) and temperature (70 degrees C) environments. The silk TS devices exhibit typical short-term plasticity (STP) of biological synapses including pair-pulse facilitation (PPF). More importantly, a leaky integrate-and-fire (LIF) artificial neuron is successfully realized based on the volatile characteristics of silk TS memristors. These achievements pave the way for utilizing silk biomaterials in advanced bioelectronics and neuromorphic computing.

Automated summary

High-performance silk fibroin-based threshold switching (TS) memristors with outstanding device uniformity and stability, even under high humidity and temperature conditions, are demonstrated. These memristors exhibit short-term plasticity similar to biological synapses and successfully realize a leaky integrate-and-fire (LIF) artificial neuron based on their volatile characteristics, paving the way for advanced bioelectronics and neuromorphic computing.