All-MoS2 double floating-gate synaptic transistors for non-volatile memory and brain-inspired computing

초록

In recent years, the demand for efficient data processing has led to a surge of interest in neuromorphic computing based on emerging two- dimensional (2D) materials. Nonetheless, floating gate devices have drawn widespread attention by virtue of their great potential for non-volatile memory. Herein, we present a van der Waals (vdW) multi-heterostructure of MoS2/MoS2/MoS2 for floating gate transistor to emulate its synaptic features. This architecture is established by surface oxidation via oxygen plasma which confirm by transmission electron microscopy (TEM). However, our devices are distinctly reconfigured to function as without (SiO2/MoS2), single-floating gate (SiO2/MoS2/MoOx/MoS2) and double-floating gate (SiO2/MoS2/MoOx/MoS2/MoOx/MoS2) transistors. While on comparative study, our double-floating gate (DFG) devices exhibit promising non-volatile memory performance metrics of memory window (> 150 V), high current ON-OFF ratio (similar to 10(6)), long retention time (> 6000 s), and excellent endurance (> 5000 cycles) which attributed to the increased charge-storage capacity and spatial redistribution. Moreover, DFG devices are explored for artificial synaptic behavior such as LTP, LTD, STDP, SNDP, SADP, and PPF, enabling its applications in brain-inspired computing. In addition, by using the MNIST dataset, we achieved identification accuracy similar to 94% under ANN simulations. Our work provides the avenue for multi-mode neuromorphic computing devices to address the recent challenges of complex integration.

키워드