Selecting and Optimizing threshold switching materials and devices for stochastic neuron

Rather than stiffly giving out spikes upon reaching a fixed threshold voltage, the biological neurons show quite dynamic excitation behaviors namely stochastic firing which is regarded to play a critical role in probabilistic computing. In this work we realize electronic stochastic neuron by developing and examining through three major kinds of threshold switching memristors (TSMs). The novel CuS/GeSe conducting-bridge threshold switching (CBTS) memristor is identified as the best candidate owing to the most desired stochastic switching behavior as well as the similarity with biological ion channels.