Ames Laboratory scientists use supercomputers to beat the clock on new magnet discovery

Ames Laboratory is taking advantage of Titan, one of the world’s most powerful computers, to discover substitutes for rare-earth magnets. In the race to find substitutes, supercomputers are the lead-off runner, ensuring that scientists can rapidly target the best possibilities for materials discovery. Titan, located at the DOE’s Oak Ridge National Laboratory in Oak Ridge, Tenn., uses a combination of traditional central processing units and graphics processing units that were first created for computer gaming.

ImageU.S. Department of Energy (DOE) Ames Laboratory’s scientists Bruce Harmon, Kai-Ming Ho, and Cai-Zhuang Wang use the enormous capability of supercomputing to identify promising compositions of new magnetic materials that do not contain rare earth elements, elements that are increasingly in short supply. Stronger non-rare earth permanent magnets are critical to replace the current rare earth magnets for energy efficient electric drive motors (used in hybrid and electric vehicles) and more powerful electrical generators (used in wind turbines).

Harmon, Ho, and Wang are using a highly efficient genetic algorithm (GA) developed in the 1990s at the Ames Laboratory by Ho and David Deaven to search the enormous phase space of stable crystal structures and compositions.

It’s a very visual model that looks at possible compositions in terms of “parents” splitting and joining to produce “offspring,” with each successive generation being evaluated and selected to optimize specific traits, explained Harmon.

 “Back in the ‘90s it would take 65,000 iterations to get a result using this algorithm, and that would take several months. But paired with increased abilities in supercomputing, this algorithm has been the foundation for what we’re trying to do in solids right now. It’s powerful, and it works,” said Harmon.

Titan’s capabilities offer scientists the tools to complete computational research work in mere hours, said Wang.

“To get the new materials we are looking for with the right magnetic properties, we are most likely looking at ternary structures with a larger number of atoms. This is a much more complex problem, and very difficult to explore,” said Wang. “The increased computer capabilities paired with an efficient algorithm will open up new areas for us to search for new alloys.”

Harmon believes supercomputing will be useful in reducing the need for both the painstaking experimental work and often sheer luck involved in finding a useful magnetic material in the huge number of yet undiscovered ones.

See how far
calculating power
has come.

Slide Rules to Petaflops

“For thousands of years, humans have discovered useful alloys through hit-or-miss metallurgy,” said Harmon. “Now, we finally have the computing capability to explore greater complexities using quantum mechanics-- greater numbers of atoms, the possible crystal structures, and their properties. We can determine which will be optimal, hand it to experimentalists, and say ‘this alloy composition may have a ‘sweet spot’ with ideal properties.’ In turn, experimentalists can return to us and say, ‘this did not work, but maybe the addition of this other element can help.’”

In this way, Harmon said, theorists and experimentalists will be able to work closely together to speed the discovery process for alternatives to rare earth alloys.

The computer time was granted through the DOE’s Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program, which promotes scientific discoveries and technological innovations by awarding time on supercomputers annually on a competitive basis. A total of 61 projects, including the Ames Laboratory’s, were granted computer time for 2013.

The computational search for the crystal structures and new materials for permanent magnet applications is supported by the Beyond Rare Earth Materials project, led by Iver Anderson at the Ames Laboratory and funded by the DOE’s Energy Efficiency and Renewable Energy (EERE) Office, Vehicle Technology Program. The basic science, algorithms, and computational code development for the structure prediction and phase diagram exploration is supported by the Exploratory Development of Theoretical Methods project funded by Materials Science and Engineering Division of Basic Energy Sciences, U.S. Department of Energy Office of Science through the Ames Laboratory.

The Ames Laboratory is a U.S. Department of Energy Office of Science national laboratory operated by Iowa State University. The Ames Laboratory creates innovative materials, technologies and energy solutions. We use our expertise, unique capabilities and interdisciplinary collaborations to solve global problems.

DOE’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States, and is working to address some of the most pressing challenges of our time. For more information, please visit