Forschungs­zentrum Jülich GmbH

At the Institute of Energy and Climate Research (IEK-5) – Photovoltaics of the Forschungszentrum Jülich (FZJ), novel materials and innovative component architectures for sustainable photovoltaics based on thin films are being researched. On the one hand, the physical fundamentals of mainly disordered material systems are investigated, such as amorphous and microcrystalline silicon and their alloys, organic and hybrid structures as well as various functional oxides. On the other hand, pioneering technological applications are being developed, e.g. passive and contact layers for high-efficiency silicon heterostructure solar cells and optimized silicon thin-film stacked cells on flexible substrates or for applications in the field of solar water splitting. At the same time, IEK-5 is developing the necessary manufacturing processes with special attention to industrial applicability. An essential prerequisite for a targeted material and technology development is the comprehensive characterization and modeling of the optoelectronic properties of thin-film materials and components at IEK-5. The activities cover all relevant length scales and range from the atomistic resolution of nanostructures and elementary processes to defect and yield analysis at the module level.


Within the PEARL Project the FZJ is setting up and maintaining a database to collect and sort luminescence images and performance data gathered by the project partners of thin-film modules. Consequently this database is be used for an extended statistical analysis of the data. The focus lies on the determination of a comprehensive catalog that includes information about the impact on the performance and development over time of individual defects that are visible in electroluminescence images. To achieve that, FZJ is also working on the automatic evaluation of electroluminescence images via various image analysis methods. These methods include among other: image segmentation methods based on deep neural networks.


Bart Pieters,


Figure 1: Typical defects visible in EL images.
Figure 2: The appearance of more shunts in an EL image leads to a performance loss.
Figure 3: Heat map of shunt appearances for an industrial produced module type. Shunts often appear at two spots on the bottom edge of this kind of module.
Figure 4: An example of a module with the automatically identified shunts (labeled by the purple color), and “droplets” (labeled by the yellow color) defects.
Figure 5: Module image (top image) decomposition on 3 components: cell components (bottom left image), global intensity trend (bottom middle image) and residual image (bottom right image)