The research teams of Kurt V. Mikkelsen at the University of Copenhagen, and Kasper Moth–Poulsen at the Technical University of Catalonia, Barcelona , have taken a closer look at the photoswitches best suited for this task. They studied molecules known as bicyclic dienes, which switch to a high-energy state when illuminated. The most prominent example of this bicyclic diene system is known as norbornadiene quadricyclane, but a vast number of similar candidates exist.
Screening a database of this size is typically done by machine learning, but this requires large amounts of training data based on real-world experiments, which the team did not have. Using a previously developed algorithm and a novel evaluation score,"eta," the screening and evaluation of the database molecules yielded a clear result: all six of the top scoring molecules differed from the original norbornadiene quadricyclane system at a crucial point in the structure.
The researchers' work demonstrates the potential for optimizing solar energy storage molecules. However, the new molecules must first be synthesized and tested under real conditions.