Scientists at Ames National Laboratory are currently expanding a new artificial intelligence tool to improve predictions regarding how materials behave during the operation of fusion energy systems.

This expansion involves integrating new data and models to better understand material performance under intense heat, radiation, and mechanical stress.

The project is supported by the ARPA-E CHADWICK program and laboratory investments and aligns with the Department of Energy’s Genesis mission to speed up the discovery and deployment of advanced materials for energy technologies.

The tool, called DuctGPT, combines artificial intelligence with physics-based modeling to help researchers identify materials with properties suitable for the interior of fusion reactors.

“Now when you ask it, ‘I want to design a material for fusion that has all x, y, z properties that are critical for use in fusion reactors,” Tell me the combination of elements which satisfy the criteria,’ it will give you those combinations of elements with properties,” said Ames Lab Scientist Prashant Singh.

Exploring potential alloy compositions

A significant challenge in this field is the need to explore a wide range of potential alloy compositions that maintain strength at high temperatures while retaining the ductility required for manufacturing. 

The project is led by Ames Lab scientist Prashant Singh and demonstrates how AI tools can assist in finding materials capable of handling extreme environmental conditions.

The research team developed DuctGPT by modifying an existing model called AtomGPT, which was created by the National Institute of Standards and Technology.

They fine-tuned this model using established material science data to make it applicable to fusion systems. DuctGPT can analyze a vast number of element combinations in seconds. 

It features a conversational interface that allows researchers to pose questions and define specific parameters using standard text. Researchers can ask the tool to design a material for fusion with specific properties, and the AI will provide a combination of elements that satisfy those criteria.

Identifying alloys that maintain high melting temp

One material of particular interest is tungsten. It is effective at withstanding high heat, has a relatively short cooling period, and remains radioactive for a shorter duration than other materials after fusion exposure. 

“Tungsten’s main limitation is its lack of low-temperature tensile ductility, which makes it difficult to form into complex shapes,” concluded Singh. 

“With DuctGPT, we can now query compositions within a desired space, such as tungsten-titanium-zirconium-hafnium, to identify alloys that maintain tungsten’s strength and high melting temperature while improving ductility.”

The goal is to identify alloys that maintain the high melting temperature of tungsten while improving its ductility for practical use.

These material queries can be performed on a standard desktop computer rather than requiring expensive supercomputer calculations. This accessibility reduces the time needed for material discovery from several months to a few days or hours. 

Ames Lab has shown that ductile refractory alloys can be designed through predictive modeling. The laboratory also has the resources to synthesize and test these predicted materials to confirm they exhibit the properties required for fusion applications.