The Polymer Society of Korea

Abstract

This study introduces a transistor device designed to implement the Gaussian function, one of the activation functions in artificial neural networks, by depositing a twisted electrode configuration. This "twist contact" structure consists of a channel layer, dinaphtho [2,3-b:2,3-f] thieno [3,2-b] thiophene (DNTT), positioned between two electrodes. This design enables the device to exhibit both bottom contact and top-contact characteristics, leading to a different saturation behavior compared to conventional devices. To evaluate the performance, bottom contact, top contact, and twist contact transistors were fabricated, and their I-V characteristics were measured. The extracted data were fitted to the Gaussian function, and their similarity was analyzed. The research results demonstrated that the twist contact device can implement Gaussian function.