Simple Cell Segmentation Routine for Quantitative Electroluminescence Analysis of c-Si Modules

Electroluminescence (EL) testing of photovoltaic modules is a crucial step for quality inspection. As forward current is biased into the module, it emits light, allowing clear differentiation between defective and healthy regions. Beyond fault detection, performance parameters can also be accessed through this imaging technique. Many quantitative methods have already been developed in the literature providing metrics to estimate series and shunt resistances, dark saturation currents, diode ideality factors, etc. Nonetheless, these methods do not precise which pixels were considered for their approach. A single method to distinct one cell from the other on the whole EL image was found, yet it was focused on full-cell modules and no mention to its application on quantitative EL has been verified. The present work aims to establish an easy to apply cell segmentation algorithm for rectangular-shaped half-cell modules along with an alternative algorithm for corner-cut full-cell module segmentation. The proposed routines were applied to full and half-cell modules, providing satisfactory visual results. From the segmentation, the quantitative estimation of series resistances and dark saturation current densities were performed for each cell on a module. For a single cell within this module, the EL estimated parameters enabled to simulate its I-V curve which was then compared to its experimental one. A good agreement between the curves was verified despite a relative error for the fill factor of 7% which originates mainly from the assumption of an ideality factor being equal to 1. Furthermore, the relative errors for Pmpp, Impp, Vmpp and Voc and were all below 5%. Possible improvements are also presented, pointing out that this work is an initial step for more precise quantitative EL analysis.