Hybridization state transition-driven carbon quantum dot (CQD)-based resistive switches for bionic synapses

As an emerging carbon-based material, carbon quantum dots (CQDs) have shown unstoppable prospects in the field of bionic electronics with their outstanding optoelectronic properties and unique biocompatible characteristics. In this study, a novel CQD-based memristor is proposed for neuromorphic computing. Unlike the models that rely on the formation and rupturing of conductive filaments, it is speculated that the resistance switching mechanism of CQD-based memristors is due to the conductive path caused by the hybridization state transition of the sp(2) carbon domain and sp(3) carbon domain induced by the reversible electric field. This avoids the drawback of uncontrollable nucleation sites leading to the random formation of conductive filaments in resistive switching. Importantly, it illustrates that the coefficient of variation (C-V) of the threshold voltage can be as low as -1.551% and 0.083%, which confirms the remarkable uniform switching characteristics. Interestingly, the Pavlov's dog reflection as an important biological behavior can be demonstrated by the samples. Finally, the accuracy recognition rate of MNIST handwriting can reach up to 96.7%, which is very close to the ideal number (97.8%). A carbon-based memristor based on a new mechanism presented provides new possibilities for the improvement of brain-like computing.

Automated summary

In this study, a novel carbon quantum dots (CQDs)-based memristor is proposed for neuromorphic computing. Unlike other models, the resistance switching mechanism of CQD-based memristors is speculated to be due to the conductive path caused by the hybridization state transition of carbon domains. This mechanism avoids the formation of random conductive filaments and shows remarkable uniform switching characteristics. The accuracy recognition rate of MNIST handwriting using this new carbon-based memristor can reach up to 96.7%, which provides new possibilities for brain-like computing.