Scientists identify molecules for potential use in redox flow batteries


In recent years, scientists have looked at a growing number of electroactive compounds for their potential use in high-performance redox flow batteries for grid-scale energy storage. The Dutch Institute for Fundamental Energy Research (DIFFER) has built on these efforts by setting up RedDB, a database of 31,618 potentially interesting molecules for redox flow batteries.

The database includes information on the miscellaneous physicochemical properties of the compounds, including their redox potential and water solubility. They described the development of RedDB in a recent paper in Scientific Data. The development steps included chemical library generation, molecular property prediction based on quantum chemical calculations, aqueous solubility prediction using machine learning, and data processing and database creation.

They used a desktop computer and smart algorithms to create thousands of virtual variants for two prominent classes of organic electroactive compounds: quinones and aza-aromatics. They fed the computer with backbone structures of 24 quinones and 28 aza-aromatics, plus five different chemically relevant side groups. From that, the computer created 31,618 different molecules.

In the next step, the researchers used supercomputers to calculate nearly 300 different properties for each molecule. Then, they used machine learning to predict whether the molecules would be dissolvable in water. Finally, they created a human- and machine-readable database.

Popular content

“When you work with theoretical models and machine learning, you obviously want to be confident in the results,” says Süleyman Er, the leader of DIFFER's Autonomous Energy Materials Discovery research group. “This is why we used computer programs that have proven their excellence. For this purpose, we also implemented dedicated validation procedures.”

Now that the database is live, the researchers, including academics outside of DIFFER, can use it to improve their machine-learning models used when they design redox-flow batteries. Scientists can also use it to search for interesting molecules that they could potentially synthesize for further study.

This content is protected by copyright and may not be reused. If you want to cooperate with us and would like to reuse some of our content, please contact: