A computerised data handling procedure for defect detection and analysis for large area substrates manufactured by roll-to-roll process

摘要

The development of optical on-line/in-process surface inspection and characterisation systems for flexible roll to roll (R2R) thin film barriers used for photo-voltaic (PV) modules is a core research goal for the EU funded NanoMend project. Micro and nano scale defects in the ALD (atomic layer deposition) Al_2O_3 barrier coating produced by R2R techniques can affect the PV module efficiency and lifespan. The presence of defects has been shown to have a clear correlation with the water-vapour-transmission-rate (WVTR). Hence, in order to improve the PV cell performance and lifespan the barrier film layer must prevent water vapour ingress. One of the main challenges for the application of in process metrology is how to assess large and multiple measurement data sets obtained from an in process optical instrument. Measuring the surface topography over large area substrates (approximately 500 mm substrate width) with a limited field-of-view (FOV) of the optical instrument will produce hundreds/thousands of measurement files. Assessing each file individually to find and analyse defects manually is time consuming and impractical. This paper reports the basis of a computerised solution to assess these files by monitoring and extracting areal surface topography parameters. Comparing parameter values to an experimentally determined threshold value, obtained from extensive lab-based measurement of Al_2O_3 ALD coated films, can indicate the existence of the defects within a given FOV. This process can be repeated automatically for chosen parameters and the existence of defects can be indicated for the entire set of measurement files spontaneously without interaction from the inspector. A running defect log and defect statistics associated with the captured set of data files can be generated. This paper outlines the implementation of the auto-defect logging using advanced areal parameters, and its application in a proof of concept system at the Centre for Process Innovation (UK) is discussed.