Detection of plastic OLED micro-crack using laser thermography and deep-learning

Abstract

The study proposes a continuous-wave laser thermography (CLST) system for rapid inspection of micro-cracks within dam areas beneath the thin-film barrier of plastic OLED. The proposed CLST system consists of a continuous-wave (CW) laser driver, high-speed infrared (IR) camera with a close-up lens, motion control stage, and computer. The CW laser generates thermal waves on the dam areas of the plastic OLED and the IR camera records the corresponding thermal responses, while the motion control stage continuously moves the plastic OLED. Then, the micro-crack detection algorithm detects micro-cracks within the enhanced images using detail enhancement and convolutional neural network (CNN). In addition, the proposed CLST system can overcome the limitations of the optical microscopy which is mainly used for the micro-crack inspection in the dam areas as follows: (1) the optical microscopy are difficult to distinguish micro-crack from surface scratches and depends on the subjective judgment of the inspector, but the CLST system can objectively and consistently diagnose micro-cracks leading abnormal thermal propagation

and (2) when a large number of surface scratches are generated by the friction on a large area, optical microscopy cannot inspect the dam areas beneath thin-film barrier due to surface scratches, but CLST system can detect micro-cracks irrespective of presence of surface scratches. For the proposed CLST system reliability evaluation, blind tests using 100 plastic OLED specimens were performed. In this study, the CLST system achieved 92% inspection reliability, and the test results indicate that the system can successfully detect the micro-cracks wider than width of 4 μm as well as the inspection time can take within around 63 seconds for the specimen with a size of 120 (±5) mm x 70 (±5) mm x 0.3 (±0.1) mm at a scanning speed of 5.0 mm/sec.