Proceedings of Asia-Pacific International Conference on Perovskite, Organic Photovoltaics and Optoelectronics (IPEROP20)
DOI: https://doi.org/10.29363/nanoge.iperop.2020.042
Publication date: 14th October 2019
The field of photovoltaics has seen significant developments in recent years. Among the available technologies, dye sensitised solar cells (DSSCs) have attracted considerable attention[1]. Here, the majority of the research efforts are centred on the dye sensitizer, since it influences many of the key electron transfer processes that impact photovoltaic performance. Designing new dyes requires various aspects (wide absorption in the visible region, high molar extinction coefficients, photochemical stability, stable oxide surface anchoring etc.) to be taken into account. The traditional routes adopted thus far involve the determination of relevant properties of a large number of potential candidates, via high-throughput experiments or computations[2]. This approach, however, is laborious and time-consuming.
With a view to streamline the design process, we have combined data-driven approaches with Darwin-inspired evolutionary algorithms[3,4,5]. While the former makes use of statistics and machine learning to provide on-demand property estimates, the latter optimises constituent fragments of the dye molecule in a synthetically tractable manner, leading to the design of dyes with given target properties[6]. In this presentation, we demonstrate how data-driven molecular engineering can be gainfully employed to accelerate materials discovery. We will also present other applications of the method for designing dye-based optical filters for low cost fluorescence detection.