Taking Supercomputing to the Limit

Because detailing electronic structure requires complicated quantum-mechanical calculations, Freeman's research demands supercomputing. About 18 person-years -- by Freeman and three post-doctoral fellows -- went into development of a very precise computational approach known as the full-potential linearized augmented plane wave (FLAPW) method. With FLAPW, Freeman and his collaborators are able to study 75 atoms in a unit cell on the C90. Fifteen years ago, even using less precise methods, the number was only 10. But he's still not satisfied.

"I'm still overwhelmed that we can deal with so much complexity. The ability to calculate spin and orbital magnetic moments layer-by-layer and atom-by-atom is remarkable. But we want to answer questions that are more complex. We need every increase in computing power. We never get enough for our simulations of real materials."

Currently Freeman and his colleagues are further developing their ability to calculate perpendicular magnetism in overlayer and multilayer systems. They also are studying the single-layer magnetism of metals from the next row in the periodic table -- the so-called 4d elements, such as rhodium and ruthenium. While these elements have no magnetism in the bulk, a magnetic moment arises when a monolayer is deposited on a substrate. "We're creating magnetism," says Freeman, "where none existed before."

Calculated spin density for a ruthenium monolayer deposited on magnesium oxide. The large positive spin area (light blue through pink contours) of the ruthenium atom corresponds to a magnetic moment of 2.1 bohrs. Since ruthenium is not magnetic as a bulk metal, this is an important prediction.

go back to the main page