Researchers publish 31,618 molecules with potential for energy storage in batteries

The Dutch Institute for Fundamental Energy Research (DIFFER) has created a database of over 30,000 compounds that could be used in future batteries. The batteries hold promise. The researchers used artificial intelligence and computers to find the molecule's properties. Their findings are published in the journal.

Hundreds of compounds have been designed by chemists that could be useful in flow batteries. Researchers from DIFFER in the Netherlands imagined that the properties of the molecule could be easily found in a database. Many of the properties of the molecule are not known. There are examples of water and redox potential in the molecule. Power generation capability and energy density are related to those.

The researchers performed a series of steps to find the properties of the molecule. They created thousands of virtual versions of two types of molecule with the help of a computer. The quinones and aza aromatics are good at reversibly giving and taking electrons. It's important for batteries. The researchers fed the computer with a variety of side groups. 31,618 different molecule were created from that.

The researchers used the power of computers to calculate hundreds of different properties of each molecule. This is done using equations from quantum chemistry. A powerful supercomputer is a useful tool.

The researchers used machine learning to determine if the molecule would be dissolvable.

The final step was the creation of a database. The RedDB database contains the molecule and their properties with easy to remember names and descriptions.

Sleyman Er says that when working with theoretical models and machine learning, you want to be confident in the results. We used computer programs that proved their excellence. Dedicated validation procedures were implemented.

Now that the database is public, researchers can easily search for potentially interesting compounds. They can buy or synthesise the molecule and research it further. The researchers may use the database to improve their machine- learning models to speed up the design of high-quality molecule for energy storage.

Scientific Data has more information about Elif Sorkun and his RedDB database.

Journal information: Scientific Data