RESEARCH SUMMARY

B. N. Harmon

Nearly all of my research falls into the category of computational physics.  With the doubling of computing power every 18 months for more than 30 years, the field has not only maintained its vitality, but is entering a new era where calculations can be performed accurately enough and fast enough to have an immediate impact on the processing and synthesis of new materials.  The reliable and timely prediction of material properties may require detailed solutions to the Schrödinger or Dirac equations for systems containing hundreds to thousands of interacting particles.  The key for real materials is to understand quantitatively the role of defects in determining many bulk physical properties.  The modern goal is to find methods to bridge the length scales from atomic level defects to mesoscopic lengths where phenomena dependent on those defects are manifested.  Over the years, methods have been developed and extensively utilized to solve this many-body problem; although in systems having strongly correlated electrons (some magnetic, high Tc, and heavy fermion materials) there are still many challenging questions.  A few projects are very briefly outlined below.

1.   Martensitic Phase Transformations

First principles calculations of energies and forces are used to evaluate the stability, transformation paths, and the role of alloying and defects in materials undergoing structural phase transformations.  Some of this work is sponsored by the Materials and Engineering Physics Division of Ames Laboratory.  Most recently these investigations have involved complex magnetic materials where the application of an external field can induce the martensitic transformation.  Gd₅Si₂Ge₂ and Ni₂MnGa are examples.

2.   High Tc Superconductors

Electronic structure calculations have been used to study the magnetic properties of the parent (undoped) phases of high Tc materials.  We were the first group to find that conventional band theory is inadequate to obtain a stable antiferromagnetic ground state in these systems, and we have helped pioneer a new method which is better able to take into account the strong correlations in the Cu-d bands.  (This research area is on hold…the field is crowed and new ideas are needed.)

3.   Circular Magnetic X-ray Dichroism (CMXD)

This is a new technique made possible by the advent of highly intense synchrotron facilities.  The difference in absorption between left and right circularly polarized X-rays at a core edge gives information about the size of the local magnetic moment and the percent of orbital (vs. spin) components which arise from spin-orbit coupling.  The results are element and angular momentum specific, so one may learn a great deal about microscopic magnetic interactions even in complicated magnets (e.g. Nd2Fe14B with 68 atoms per unit cell).  We are the leaders in the United States in calculating the expected theoretical profiles and are working with Alan Goldman's group in developing the method for general utility.  Our most recent effort (in 2004) is developing similar methods and analysis for resonant magnetic x-ray scattering.

4.   Optical Properties - and M.O.M.

In the last five years several of Dave Lynch's students have worked with the theory group in making detailed calculations of the optical properties of various materials.  This has been a successful operation.

With the undertaking of CMXD (above) an opportunity arose to combine the two efforts and I wrote a large proposal to make a concerted study of magneto-optic materials (M.O.M.).  The proposal was funded at $500K/year.  It combines experimental efforts in crystal synthesis, thin film fabrication, optical measurements, neutron and x-ray scattering, CMXD experiments, and theory.  Paul Canfield was brought to Ames on this program, as was Vladimir Antropov.  With Paul’s success, he has taken over leadership of the majority of the project with the emphasis now more on correlated electron materials.

5.  Spin Dynamics:

Arising from our work in Magneto-Optics we realized the potential for extending first principles magnetic calculations to include temperature and dynamical properties.  We wrote a proposal in collaboration with Oak Ridge National Laboratory (ORNL) and it was accepted in FY96.  $300K/year for Ames Lab and $200K/year for ORNL.  This funding has sponsored a number of postdocs, with Slava Dobrovitsky being the most prominent.  Work in this area continues, and our colleagues at ORNL ran our first principles spin dynamics method for a system of 3000 atoms and reached an all time world record of a sustained 4.51 Teraflops on the NERSC SP supercomputer.  These and other recent advances are allowing very complex materials (e.g., magnetic thin film interfaces) to be analyzed.

6.   Magnetic Molecules / Magnetic Tunneling

In the last few years we have studied the magnetic interactions in small magnetic molecules.  An example is the so called Mn12 molecule which has a net magnetic moment of 20 Bohr magnetons.  Most preliminary work that focused on the magnetic tunneling at low temperatures considered the “single spin” model.  We have investigated what the exchange and anisotropy interactions among the 12 Mn spins are and how a many spin model differed from the single spin model in the prediction of various phenomena.  We were able to show that inelastic neutron scattering spectra are only explainable in the many spin model, and that some of the tunneling splitting values differed by orders of magnitude for the two models (experiments are not yet precise enough to allow quantitative comparison).  Finally we have analyzed the tunneling and decoherence in weakly anisotropic magnetic molecules (particularly V15) and have shown that these systems may be considered a qubits for quantum computing applications. Army Office of Research funding (about $200k/year) was obtained (with V. Dobrovitski) to study decoherence in quantum computing models.

7.   Other

In collaboration with scientists from ORNL and many universities we have formed a collaborative research team as part of the DOE computational materials sciences network (CMSN).  The goal is to create methods to accurately model important phenomena in real (defected) magnetic materials.  We have developed a coarse graining method that is able to average over atomic scale interactions and obtain accurate dynamics for processes like the nucleation of magnetic domains and the pinning of domain walls.  (with Dobrovitsky)

From time to time interesting topics arise in the many experimental groups in the lab and we get involved to see if first principles calculations can help clarify the underlying cause of the phenomena being studied.  A recent example is the discovery by Alan Russell and Bruce Cook in the Materials and Engineering Physics Division of the superhard properties of AlMgB14.  We calculated the full set of elastic constants for both the ideal 64 atoms/cell and the actual 62 atom/cell material where two metal atom vacancies are present.  While our results indicated a very hard material, it was not “the world’s second hardest material”.  To reach such a level of hardness it seems that Si or TiB2 inclusions are required.  We have not analyzed the material with this level of complexity.