ScholarWorks@숙명여자대학교

초록

With the advent of the Internet of Things (IoT), the demand for printing technologies such as inkjet or roll-to-roll (R2R) printing has also increased. In particular, carbon nanotubes (CNTs) and thin-film transistors (TFTs) have emerged as essential materials for R2R devices due to their electrical properties and low manufacturing costs. Since the performance of transistors strongly depends on the morphology of the CNT networks, analyzing the CNT networks is crucial. We use atomic force microscopy (AFM) images to analyze CNT networks because the surface height can be measured directly via AFM images. The conventional AFM image processing method is limited by the roughness of the substrate and particles on the substrate. To address these issues, we propose a novel unpaired learning method for extracting CNTs from the AFM images of CNTs deposited on the substrate. With the trained model, the profile of CNTs can be demonstrated from the AFM images, providing the possibility of quantitative analysis of the morphology of the CNT networks. The experimental results using the simulation dataset confirm that the proposed algorithm enables the accurate extraction of CNTs from the AFM images, allowing the analysis of the transistor performance.

키워드