Optically Stimulated Artificial Synapse with Ferroelectric Assistance for a Neuromorphic Visual System

Abstract

To address the challenges faced by a traditional computing system with von Neumann architecture, various artificial synapses and neurons have been developed. In this work, a twoterminal device was developed using an organic p-type semiconductor copper phthalocyanine (CuPc) and a molecular ferroelectric diisopropylammonium bromide (DIPAB). The incorporation of the ferroelectric layer DIPAB not only introduces an energy barrier for the migration of photogenerated carriers to enhance the device's synaptic performance but also enables the modulation of the device response to the light pulses by changing the polarization state of DIPAB. The device successfully mimics the biological synaptic functions including paired-pulse facilitation (PFF), spike-rate-dependent plasticity (SRDP), and spike-numberdependent plasticity (SNDP). The device is employed to recognize MNIST handwritten digits, achieving a recognition rate of over 75% for MNIST handwritten digits even with 50% noise interference. These results suggest the device's potential for neuromorphic computing.