Design for a Sustainability Approach to Organic Solar Cell Design: the Use of Machine Learning to Quantify the Trade-off between Performance, Stability, and Environmental Impact

Organic photovoltaics (OPVs) are considered one of the best-performing photovoltaic (PV) technologies from an environmental point of view. Many of the constituent materials possess low embodied energies, which can generally be processed and disposed of in a less energy-intensive manner than other PV technologies. There has been an enormous range of materials used in OPVs; however, identification of the optimal materials and device architectures that provide the best environmental profile within this large search space has yet to have been conducted. This is a non-trivial task because the selection of these materials not only impacts the environmental profile but also on the solar cell efficiency and its operational stability. Here, we have developed a methodology that enables rapid assessment of the trade-off between efficiency, stability, and embodied energy of an OPV using machine learning. To achieve this, a database of OPV data was used, which has been acquired from the literature between 2011 and 2020 and consists of 1580 device data points. Our results highlight the importance of focusing activity on particular transport layers, substrates, and active layer materials, which are discussed further in the manuscript. We demonstrate that the trained and validated models can predict, with a high degree of confidence, the efficiency, stability, and embodied energy of an OPV. The methodology set out in this work provides a means of identifying optimum device configurations in a rapid manner such that the net energy production is maximized, whereas the environmental impact of OPVs is minimized. Materials which show promise toward delivering a positive net energy are PET + barrier layers, PET (substrates), NiOx, ZrOx, CrO2, ZnO, LiF, and MoO3 (transport layers). Active layer materials which show promise for delivering a positive net energy are DRCN7T, DR3TSBDT, ZnPc, PDPP4T-2F, PCDTBT (donors), and IT-4F, C61 (acceptors).