// Digitalization in photovoltaics with the self-learning solar factory (SelFab)

Solar cells are high-tech products and their manufacturing is a complex sequence of many very different process steps and materials. It produces a vast amount of data in the lab and on the shop floor. There is enormous potential slumbering in that data. If all of it were to be processed and put to productive use, this could fast-track engineers’ efforts to advance the state of the art and optimize production lines. But that requires processes and methods that can mine this mountain of high dimensional data.

A consortium of Baden-Württemberg-based research institutes has been investigating the smart solar factory in an Industry 4.0 context since February 2019 as part of a project funded by the state of Baden-Württemberg. These researchers are focusing on digital models, digital twins, machine learning methods for data analysis, and on mapping self-learning processes. ZSW is on board to develop digital models and digital twins of CIGS thin-film solar cells. Its scientists are applying typical methods such as semiconductor simulations as well as combined methods from the field of machine learning. The latter also serve to control processes and the composition of layers in CIGS coatings with an even finer touch, the object being to further increase yield.