Edge detection and surface inspection for roll-to-roll and slot-die coating based on machine vision approach

Roll to roll (R2R) processes are critical in the mass production of thin film products especially in flexible electronics. Slot-die techniques is widely used in R2R as a simple but accurate method of applying conducting thin film inks onto a large area flexible substrate. For example, organic conducting inks can be printed directly in the form of tracks with high homogeneity and uniformity. Furthermore, slot-die coating process can potentially be optimised if a sufficient control system is provided, which provides a direct link between the slot-die head and the operating conditions, such as flow-rate of the ink and the speed of the substrate. In this work, the authors are aiming to provide the basis of a feedback system through a machine vision surface inspection system. A method is presented for edge detection to extract quantitative information related to the printed ink edge consistency and hence track width regularity. Monochrome cameras illuminated by a broadband light source are used to captures image frames while the substrate is moving. Different printed ink samples on polymer based substrates have been considered for developing the test protocol. The images are captured at full-bright field, where the substrate is positioned coaxially. A Sobel operator has been used to analyse the captured images in order to inspect coating quality and detect defects, whilst an Otsu's thresholding approach is used to define and extract the edge line of the coating. In this paper, the authors use the lateral arithmetic average (equivalent to Ra parameter in surface metrology) to describe the edge regularity along the printed track. It was found that for the measured tracks, the edge straightness can vary by 0.51 μm (Ra(LAT)). This metric value can be used as a feedback signal to the control system to adjust the operational conditions during the printing operation.