AI helps narrow down candidates for liquid hydrogen carriers from billions to around 40

In a computational study leveraging artificial intelligence (AI), scientists at the U.S. Department of Energy’s (DOE) Argonne National Laboratory evaluated 160 billion molecules, a number exceeding those born in the entire history of humanity. Their goal was to determine whether the molecules were suitable as liquid hydrogen carriers.

The results are published in the journal Digital discovery.

The sun is essentially a huge ball composed mostly of hydrogen gas, releasing energy in a form that warms the Earth and the rest of our solar system. Due to its energy content and abundance, hydrogen also shows great promise as an energy source on Earth. This could include powering cars, trucks, buses, trains and ships and generating electricity for consumers. While solar energy comes from the fusion of hydrogen atoms, the process the team aims to use on Earth involves burning hydrogen.

Hydrogen in its pure form exists as a gas under normal conditions. For its use as fuel, one of the challenges is getting this gas safely to refueling stations and storing it. Hydrogen-carrying compounds in liquid form, however, have several advantages. They have a much better safety profile because they are not as prone to leaks and explosions. They also have a much higher energy content per unit volume, making storage and transportation much easier.

“The liquid compound form would essentially eliminate some of the issues associated with pure hydrogen, especially since there is a well-established infrastructure to store and transport gasoline and other liquid chemicals safely,” said Rajeev Surendran Assary, chemist and head of the Argonne group. in the Materials Science division.

The most visible form of a liquid hydrogen-carrying compound is water: two atoms of hydrogen and one of oxygen. Another form is organic molecules, essentially an infinite number of possible combinations of hydrogen and carbon atoms, in addition to other atoms such as nitrogen and oxygen.

“Aided by AI, we are looking for organic liquid molecules that, through a low-cost chemical reaction with a catalyst, could alternatively add or release hydrogen for use as fuel,” said computer scientist Logan Ward in the Data Science and Learning study. division. It would be essential that this reaction does not add carbon to the atmosphere. In other words, it must be carbon neutral.

“We were looking for organic liquid molecules that retain hydrogen for a long time, but not so strongly that they couldn’t be easily removed on demand,” said Hassan Harb, a postdoctoral fellow in the Materials Science division. “They must also have the capacity to store enough hydrogen for practical use.” After the hydrogen was removed, replacement hydrogen would be added to the liquid for reuse.

Among the billions of possible liquid hydrogen carriers, common examples include chemicals like ammonia and methanol. However, the few candidates tested so far in the laboratory have suffered from chemical instability and unwanted side reactions.

The team screened candidate molecules based on four factors. One of them was the structural similarity to known liquid hydrogen carriers. Another issue concerned desirable physical properties, such as melting and boiling points: the liquid should remain liquid when hydrogen was added or extracted. Third, the liquid must be able to store a large amount of hydrogen per unit volume. Finally, the amount of energy required to release the hydrogen from the liquid must be sufficiently low.

“We started by accessing chemical databases that had data on organic molecules,” said Sarah Elliott, a postdoctoral fellow in the Division of Chemical Sciences and Engineering. “We have found more than 160 billion such molecules, and combining AI with the latest theoretical computational methods is essential to select the best from this enormous army of molecules.”

The team’s calculations required access to supercomputers available in a few locations around the world. One of them is Argonne, home to the Argonne Leadership Computing Facility, a DOE Office of Science user facility. The team also relied on Bebop, a computing cluster operated by the Argonne Laboratory Computer Resource Center.

Even with these powerful resources available, if we allocate one millisecond of computing time per molecule, that translates to five years of computing time for 160 billion molecules. For this reason, the team developed an AI-based screening approach that accelerated calculations to three million molecules per second, or about 14 hours for the 160 billion molecules.

“This transforms the calculations from something we might only do once in an entire project, if at all, to something we can do overnight and repeat as we get feedback on the calculations and experiments,” Ward said.

Using their unique approach, the team narrowed down the number of candidates from over 160 billion to just 41. Now the task is given to experimenters to test the most promising ones. The team’s IT approach is paving the way for a new era of innovation in sustainable energy solutions.

In addition to Assary, Elliott, Ward, and Harb, authors include Ian Foster, Stephen Klippenstein, and Larry Curtiss.