Researchers from the United States Department of Energy’s (DOE) Argonne National Laboratory developed a computational model for identifying low-cost catalysts that convert biomass into fuels and valuable chemicals with a minimal carbon footprint in milliseconds rather than months. The computational model was created through simulations on Argonne’s Theta supercomputer. Theta is part of the Argonne Leadership Computing Facility, a U.S. Department of Energy (DOE) Office of Science user facility.
Scientists have devised an artificial intelligence-based model to accelerate the development of a low-cost molybdenum carbide catalyst. The researchers built a library of 20,000 configurations for oxygen binding energies to doped molybdenum carbide.
”The ultimate goal is to cheaply transform that carbon into valuable products for society, such as biofuel and chemicals such as biodegradable plastic. These goods eliminate the need for fossil fuel,” explained MSD group head Rajeev Assary.
Carbon-rich materials, such as corn, soybeans, sugar cane, switchgrass, algae, and other plant matter, can be turned into liquid fuels and chemicals with various applications. In the United States, for example, there is enough biomass to provide renewable jet fuel for all air traffic. However, a key impediment is the need for suitable, low-cost catalysts to convert biomass into biofuel or other valuable products.
Scientists can make pyrolysis oil, a petroleum-like substance, by heating raw biomass to high temperatures. However, the final product has a significant oxygen concentration. That oxygen is undesired and thus eliminated via a process facilitated by employing a molybdenum carbide catalyst. However, the surface of this catalyst absorbs oxygen atoms, which aggregate on the surface and decrease catalyst function.
One potential remedy is to add a small quantity of a new element, such as nickel or zinc, to the molybdenum carbide. This dopant element weakens the bonding of the oxygen atoms on the catalyst surface, preventing it from becoming poisonous.
“The issue is finding the correct combination of dopant and surface structure,” Doan noted. “The structure of molybdenum carbide is quite intricate. As a result, we used supercomputing and theoretical calculations to predict the behaviour of not only surface atoms associating with oxygen, but also atoms nearby.”
The researchers used an artificial intelligence-based model to create simulations considering several dozen dopant elements and over a hundred different placements for each dopant on the catalyst surface. The database was then utilised for training a deep learning model. Deep learning is a type of machine learning in which the computer first analyses a vast sample set of data before learning to solve issues. “Rather than being limited to evaluating a few thousand catalyst structures over months with conventional computational methods, we can now do exact and cost-effective measurements for tens of thousands of configurations in milliseconds with our deep learning model,” said Hieu Doan, an assistant scientist in Argonne’s Materials Science Division (MSD). “It’s materials screening on crack.”
Their atomic-scale simulations and deep-learning model results were submitted to the Chemical Catalysis for Bioenergy Consortium. Next, they will conduct tests to examine several potential catalysts.
“We plan to handle over a million configurations and other binding atoms, such as hydrogen, shortly,” Assary said. “We also aim to apply this same computational technique to catalysts for other decarbonisation technologies, such as water-to-clean hydrogen fuel conversion.”
The team’s findings were published in Digital Discovery, the Royal Society of Chemistry journal. Aside from Assary and Doan, Chenyang Li, Logan Ward, Mingxia Zhou, and Larry Curtiss contributed.
This study was financed by the DOE Office of Energy Efficiency and Renewable Energy, Bioenergy Technologies Office. The research was conducted due to the team’s participation in the DOE’s Consortium for Computational Physics and Chemistry, which aimed to develop bioenergy technology as an alternative to fossil fuels.