The Netherlands – A library of 31,618 compounds that might be utilized in upcoming redox-flow batteries has been produced by researchers from the Dutch Institute for Fundamental Energy Research (DIFFER).

These batteries have a lot of potential as energy storage devices. The researchers, among other tools, employed supercomputers and artificial intelligence to determine the characteristics of the molecules.

Chemists have created hundreds of compounds in recent years that might be effective for energy storage in flow batteries. Researchers at DIFFER in Eindhoven, the Netherlands, envisaged how amazing it would be if the characteristics of these molecules could be rapidly and readily found in a database. But many molecules’ characteristics are unknown, which is a problem. Redox potential and water solubility are a few of examples of molecular characteristics. These are significant because they relate to the redox flow batteries’ capacity for power output and energy density.

Four steps

The researchers used four actions to uncover molecules’ as-yet-unknown characteristics. To start, scientists generated hundreds of virtual variations of two different types of molecules using a desktop computer and clever algorithms. Quinone and aza aromatic chemical families are adept at reversibly accepting and donating electrons. It’s crucial for batteries. The backbone structures of 24 quinones, 28 aza-aromatics, and five different chemically significant side groups were provided to the computer by the researchers. The algorithm then generated 31,618 distinct compounds from that.

The second step involved the researchers computing roughly 300 different attributes for each molecule using supercomputers. This is accomplished by the computer using quantum chemistry equations. Powerful supercomputers are useful in use since these formulas are challenging to answer.

To determine whether the molecules would dissolve in water, the researchers employed machine learning in the third phase.

The fourth and last phase involved establishing a database that was readable by both humans and machines. The molecules and their properties are included in the database, which goes by the name RedDB (from Redox DataBase).

Since the database is now accessible to all researchers, including those outside of DIFFER, it is simple to look for compounds that could be useful for redox flow batteries.