Enhancing the stability of organic photovoltaics through machine learning. (December 2020)

• Record Type: Journal Article
• Authors: David, Tudur Wyn; Anizelli, Helder; Jacobsson, T. Jesper; Gray, Cameron; Teahan, William; Kettle, Jeff
• Is Part Of: Nano energy. Volume 78(2020)
• Publication Date: 2020-12
• Subjects: Organic photovoltaics -- Data-analytics -- Machine learning -- Stability -- Performance; Nanoscience -- Periodicals; Nanotechnology -- Periodicals; Nanostructured materials -- Periodicals; Power resources -- Technological innovations -- Periodicals; Nanoscience; Nanostructured materials; Nanotechnology; Power resources -- Technological innovations; Periodicals; 621.042
• DOI: 10.1016/j.nanoen.2020.105342
• Languages: English
• ISSNs: 2211-2855

Abstract

Abstract: A machine learning approach for extracting information from organic photovoltaic (OPV) solar cell data is presented. A database consisting of 1850 entries of device characteristics, performance and stability data is utilised and a sequential minimal optimisation regression (SMOreg) model is employed as a means of determining the most influential factors governing the solar cell stability and power conversion efficiency (PCE). This is achieved through the analysis of the acquired SMOreg model in terms of the attribute weights. Significantly, the analysis presented allows for identification of materials which could lead to improvements in stability and PCE for each thin film in the device architecture, as well as highlighting the role of different stress factors in the degradation of OPVs. It is found that, for tests conducted under ISOS-L protocols the choice of light spectrum and the active layer material significantly govern the stability, whilst for tests conducted under ISOS-D protocols, the primary attributes are material and encapsulation dependent. The reported approach affords a rapid and efficient method of applying machine learning to enable material identification that possess the best stability and performance. Ultimately, researchers and industries will be able to obtain invaluable information for developing future OPV technologies so that can be realised in a significantly shorter period by reducing the need for time-consuming experimentation and optimisation. Graphical abstract: Image 1 Highlights: Machine learning techniques used to predict stability and efficiency of OPV technologies. The techniques allow the optimum device structures to be determined which enhance T80 and E0 . Materials which enhance and deteriorate T80 and E0 have been identified. Sequential minimal optimisation regression employed for predicting T80 and E0 . SMOreg model analysed for determining significant attributes.