An artificial synapse based on organic heterojunction of conducting polymer and molecular ferroelectrics

Abstract

To address the challenges posed by the rapid development of artificial intelligence and big data, various artificial synaptic devices have been developed to overcome the limitations of traditional computing systems. In this study, a two-terminal device using a conducting polymer, poly(hexylthiophene) (P3HT), and a molecular ferroelectric, diisopropylammonium bromide (DIPAB), is fabricated. The incorporation of the ferroelectric layer not only establishes an energy barrier for charge carriers generated by light pulses in the P3HT layer but also allows the device's response to be modulated by altering the ferroelectric polarization state of DIPAB. This device successfully simulates some basic biological synaptic functions by modulating light stimuli. Furthermore, the light logic functions of “AND” and “OR” are realized by using light pulses with different wavelengths, as well as the simulation of associative learning. The device is used to recognize MNIST handwritten digits based on a convolutional neural network (CNN), achieving a recognition accuracy of over 90 %. These results highlight the device's potential for neuromorphic computing.