Chelikowsky Awarded $1.2M NSF Grant To Find New Magnetic Materials
But the powerful, permanent magnetics in most high-tech products are not the same magnets stuck on your fridge. They contain rare-earth metals, which help boost magnetic strength to about 250 times that of the front-of-the-refrigerator variety.
On a global distribution scale, these metals are not all that rare. (Neodymium, the most widely used rare-earth, is about as common in the earth’s crust as copper.) However, their continued accessibility and potential environmental effects are a reason for concern: China produces 97 percent of all rare earths, and the rare-earth mining process can release dangerous radioactive elements, such as uranium, into the environment.
To ensure access to powerful magnets, and to make technology more environmentally friendly, we need a way to make magnets strong without resorting to rare-earths, says Jim Chelikowsky, chemical engineering professor and researcher with the Institute for Computational Engineering and Sciences (ICES).
Chelikowsky is also director of the ICES Center for Computational Materials, and is working to identify the next generation of powerful magnetic materials by computationally analyzing compounds that could offer the same properties as rare-earth magnets without containing any rare-earths themselves.
His work is funded by a three-year, $1.2 million National Science Foundation grant from the agency’s Designing Materials to Revolutionize and Engineer our Future (DMREF) program, and is conducted in collaboration with Iowa State University computational materials researchers Kai-Ming Ho and Cai-Zhuang Wang.
Two codes combined
Scientists estimate millions of possible magnetic compounds have yet to be explored—and that’s just materials made up of three to four elements.
Chelikowsky and his collaborators are investigating such tertiary and quaternary compounds for exceptional magnetic properties by applying two computer codes with complimentary properties. One code, called the Genetic Algorithm, generates compound variations. The other, called PARSEC, analyzes their electronic structure for signs that could indicate good magnetic properties.
Developed by Iowa State researchers, the Genetic Algorithm computer program borrows from biology to find new compound structures by imitating the process of genetic recombination. But instead of recombining different parental genes, the algorithm breaks apart and recombines “parent” material clusters into new, yet related, combinations.
“The Genetic Algorithm literally takes things and cuts them up, so you generate all sorts of things,” said Chelikowsky, mentioning that the code can even introduce chemical “mutations” from time to time to add variety.
Multiple generations of “children” are created with the algorithm until a stable compound with promising properties results, Chelikowsky said.
Identifying such a compound, and evaluating all the other compounds along the way, is the job of PARSEC.
The PARSEC advantage
PARSEC (an acronym for pseudopotential algorithm for real-space electronic calculations) is a code that Chelikowsky has been developing for more than 20 years to evaluate chemical compounds on high performance computing systems.
It was designed in collaboration with computer scientists to run chemical calculations efficiently on modern high performance computing systems—a key step when analyzing large quantities of data.
“It’s efficient because it’s one of the few codes that was written specifically for computing many types of materials on state-of-the-art computers,” Chelikowsky said.
In addition, these calculations, which describe electronic and magnetic properties of materials, were programmed to handle a variety of different material structures, such as atoms in localized clusters or atoms arranged periodically in perfect crystals, said Chelikowsky. This flexibility is especially well suited to the research project, because it maintains computational accuracy when analyzing complex materials.
Using PARSEC to comb through the Genetic Algorithms results will hopefully identify some promising magnetic materials. But learning more about the connection between electron structure and chemical property is a benefit for material science as a whole, said Chelikowsky.
“That’s one of the strengths of what we do,” Chelikowsky said. “The things that we learn about examining magnetic materials should have implications when examining different types of materials, anything from, complex organic materials, to hard materials like semiconductors, insulators, and dielectrics, with a wide variety of properties.”
Back to the lab
The computational power of PARSEC and the Genetic Algorithm will together identify candidate materials. But the research still requires a traditional chemistry lab to test them. To do that, University of Nebraska professor Dave Sellmyer will be synthesizing and analyzing the most promising of the computational predictions to see how they hold up outside of the computational realm.
“We need an experimentalist, for verification and validation, which is very important to us, for this program and for the NSF,” Chelikowsky said. “So, we scoured the field to find a superb experimentalist to work with us.”
By the end of the grant, the goal is to have identified and synthesized powerful magnets without rare-earths. However, the process of the research, which will include compiling an open access database of magnetic materials and their properties, will likely be valuable in and of itself.
“At the minimum, we’ll have a great improvement in magnetic materials, how we look at them, and calculate their properties,” Chelikowsky said. “And in the best scenario, we will not only get that, but we will also get the ability to predict new materials and to have actually made new materials.”
Tags: chemical calculations, chemical engineering, computational codes, DMREF program, genetic algorithm, ICES Center for Computational Materials, Jim Chelikowsky, Magnetic materials, Magnets, National Science Foundation, NSF Grant, PARSEC, rare-earth magnets, UT Austin