This work introduces a method for screening potential hotspots in monolithic interconnected thin-film silicon modules using injection-dependent electroluminescence (EL) imaging. The fraction of dark area of the cell in the low- and high-injection EL images, respectively, is used to extract the severity and localization information associated with a defect. For the first time, a factor, namely, severity-to-localization (SL), is introduced for each defect as the ratio of severity to localization. Further, defects are broadly classified as A, B, AB, and C modes. Mode A and Mode B are severe, where the former is a distributed defect across the cell, and the latter is a localized defect. In contrast, Mode C is a localized trivial defect. The severe defects that are neither entirely distributed within the cell area nor localized are classified as Mode AB. The SL factor values associated with A, B, AB, and C modes are ≈ 1, > 4, between 1 and 4, and ≈ 1, respectively. Furthermore, the potential of four modes of defects for hotspot formation is tested following the IEC61215 standard. The hotspot endurance test results reveal that high SL factor defects, such as Mode B, always lead to hotspots, and low SL factor defects, such as Mode A and C, do not produce distinguishable hotspots. Similarly, Mode AB with a higher SL formed clear hotspots, and with a lower SL factor ( < 1.5) never formed hotspots. The proposed method applies to all thin-film technologies with monolithic interconnects and is, therefore, expected to gain significant attention. • A method to screen the hotspots in flexible thin-film silicon modules is presented. • Injection-dependent electroluminescence (EL) imaging is employed for defect screening. • Severity to localization (SL) factor is introduced for defects based on their luminescence pattern at low- and high-injection EL images. • SL factor is a reliable predictor for hotspot formation during shading. • The proposed method is applicable to all thin-film technologies that employ monolithic interconnection.
Sreejith et al. (Fri,) studied this question.