The Department of Energy (DOE) funded Giustino’s latest work with polarons under the DOE Office of Science, Basic Energy Sciences, Computational Materials Sciences award no. DE-SC0020129.

“The National Energy Research Scientific Computing Center (NERSC) has provided the main resources that we use for this DOE project. We ran a significant fraction of our calculations on the Cori and Perlmutter supercomputers at NERSC,” Giustino said.

The calculations pertain to the formation energies, wave functions, and atomic displacements of polarons, a quantum wave packet consisting of an electron “dressed” by a cloud of atomic vibrations.

Electrons alone behave like delocalized waves, and polarons behave differently in that the wave packet jumps from one lattice site to another. “This ‘hopping transport’ regime confers the material with novel properties and has implications on the design of materials for electronics,” Giustino said.

The packet extends over 10 nanometers, encompassing about 30,000 atoms of boron and nitrogen, considering all the interactions between atoms. This kind of calculation isn’t possible currently using standard density functional theory methods.

“We recast this problem into the solution of a very large nonlinear eigenvalue problem. Then we used supercomputers to solve this gigantic linear algebra problem,” Giustino said.

“We believe that with these new methods that we developed through the EPW code, we are now capitalizing on and supporting experimental data and see new directions in materials design,” he added.

Said Giustino: “The materials that we know today are just a tiny fraction of what is possible. Supercomputers are key to exploring this valuable and gigantic space without having first to invest billions into experimental synthesis and material characterization. Computation gives scientists a first pass to more easily see what is possible, something which I strongly advocate for and want the U.S. to maintain leadership in materials research through sustained support for high performance computing.”

Source: TACC