Accelerating the Development of Organic Solar Cells: A Standardized Protocol with Machine Learning Integration

Abstract

Organic Photovoltaic (OPV) devices show a large gap between laboratory-recorded cells with over 20% efficiency and commercial roll-to-roll printed modules reaching a maximum half that efficiency. A novel OPV material not only needs high efficiency, but must be processable in architectures suited for large-scale applications and provide sufficient stability under various stress factors. We present a holistic screening protocol to cover all relevant aspects of OPV material development. Using machine learning techniques together with systematic experimental protocols, only a minimum amount of a novel semiconductor is necessary. We utilize process parameters, optical features, and IV data to explore the processing window, benchmark process stability, and enable structure-property predictions. We implement a combinatorial degradation protocol that investigates key stress factors, like temperature, oxygen, and illumination, at different stages of device fabrication. Testing partially finished devices, conventional and inverted architectures, as well as hole-only and electron-only devices, enables the identification of individual layers responsible for degradation. The protocol includes a solvent test to investigate processability with green solvents. The systematic data collected in this protocol provides a general and reliable basis for material development and the imminent creation of digital twins for OPV.