Teaching computers to predict efficient catalysis

Researchers from Aarhus and Berlin have developed a new algorithm that can teach computers to predict how complex molecules will bind to the surface of catalysts. This is important when you need to produce synthetic fuels, for example. And it’s almost like playing extreme Tetris.

Imagine a game of Tetris where you not only have to stack the pieces in three dimensions, but the pieces are also much more complicated than the seven geometric shapes you normally use in the game.

In this case, the pieces are large, complex molecules that need to bond with another material in a chemical reaction.

To make matters even more difficult, the molecules and the other material have several places on the surface where they can bond to each other – and it is crucial that the bond is neither too weak nor too strong.

The binding must be perfectly straight, otherwise the other material cannot act as a catalyst (see box at the end of the text).

Such an extreme game of Tetris perfectly illustrates the challenges faced by researchers around the world when working to develop new and better catalysts for a wide range of technical-chemical processes.

Computers learn to play extreme Tetris

Researchers from the University of Aarhus and the Fritz Haber Institute in Berlin have developed a machine learning algorithm that can teach computers to predict how molecules will bind on the surface and how well they will bind.

The results have just been published in the journal Computational science of nature.

The algorithm can help predict efficiently and accurately whether a given material is amenable to catalyzing chemical reactions involving many different complex molecules.

This is important because the first step in developing new and better catalysts is determining whether a new material is even a promising catalyst for a desired reaction. The next step, material synthesis, is often too expensive and difficult to simply experiment with different materials through trial and error in the lab.

… for the green transition

The aim of the collaboration between Aarhus and Berlin was to find a quick and efficient way to screen for potential catalysts. Specifically catalysts that can be used to synthesize green fuels such as methanol, ethanol, and liquid methane.

To make these chemical processes economically viable, there is an ever-increasing need for new and better catalysts.

One of the challenges is that these reactions typically involve so many different and complex molecules that even a supercomputer wouldn’t be able to pick out the candidates. A supercomputer performs a quantum mechanical calculation of the bond strength for each possible position of the molecule on the surface of the catalyst, and that would simply take way too long.

An experienced Tetris player defeats a supercomputer

However, using machine learning, we can train algorithms with the results of previous calculations for similar molecules and catalyst surfaces to make reliable predictions about bonding without the need for additional supercomputer time, making the much faster process.

“It’s comparable to an experienced Tetris player developing the ability to intuitively place the pieces without analyzing them individually as they fall,” says Wenbin Xu, a doctoral student at the Fritz Haber Institute.

The new algorithm already provides precise binding information for large and complex molecules that are at the heart of reactions that can produce synthetic fuels.

Learning requires information

Artificial intelligence and machine learning are not new inventions, but so far existing algorithms have only been able to predict the binding of small molecules that bind with an obvious orientation on the surface, just like a simple piece of square Tetris.

So what makes this new algorithm different from others?

“The problem with existing algorithms was that they lacked the information about the geometric location of the individual atom in the molecule, or how it related to other atoms in the molecule and to atoms on the surface. Our new machine learning algorithm, which is based on mathematical graph theory, can extract this information and use it effectively,” says Mie Andersen, Fellow of the Aarhus Institute of Advanced Studies (AIAS) and Associate Professor in the Department of Physics and Astronomy at Aarhus University as well as a member of the Department’s Center for Interstellar Catalysis.

Can also be used for origins of life research

The new algorithm could even impact our understanding of how life began on Earth. Indeed, it may help determine how nano-sized interstellar dust grain particles function as catalysts in the chemical processes that cause gases in space to form complex molecules such as DNA bases and amino acids. on which life on Earth depends.

Catalyst Facts:

A catalyst is a substance that can facilitate a chemical reaction without itself being consumed by it.

In many cases, a catalyst is required for a chemical reaction to even begin, and in all cases catalysts increase the rate of chemical reactions.

You may remember that you were taught in chemistry class at school how to make hydrogen by placing zinc in a container with acid. And how to animate the reaction by placing a catalyst in the form of a piece of copper above the zinc.

The electrons move from the zinc to the copper where they react with the hydrogen ions in the acid. The hydrogen formed bubbles to the surface of the copper, but the material that decays is zinc.