Spiking neural networks are attracting significant attention because they can perform cognitive tasks with relatively low power. In addition, we proposed a tunnel field-effect transistor (TFET)-based charge trapping memory to reduce the power consumption of the flash memory-based neural network circuit. The current-voltage characteristics of the fabricated TFET based memory cell were typical of the charge trapping memory. We then measured the symmetric and asymmetric spike-timing-dependent plasticity (STDP) characteristics of the fabricated TFET-based memory cell. The obtained characteristics reproduce the STDP of a biological synapse. These results indicated that there is a possibility of applying the proposed devices to neural network circuits.
ASJC Scopus subject areas
- Physics and Astronomy(all)