C2SEPEM - Center for Computational Study of Excited-State Phenomena in Energy Materials

Mei-Yin Chou

mc34@gatech.edu
(404) 894-4688
Georgia Institute of Technology, Atlanta, Georgia
Georgia Tech Profile

Education
Ph.D. in Physics, University of California, Berkeley, 1986
M. A. in Physics, University of California, Berkeley, 1983
B. S. in Physics, National Taiwan University, Taipei, 1980

Research Interests
The objectives of our research have been to investigate the electronic structure of condensed matter, and to study its effects on the structural and dynamical properties of materials. For systems with translational symmetry, calculations are performed from first principles with no adjustable parameters. This has been made possible by advances in computational methods developed in the past decades. When these first-principles calculations are not feasible in some larger and more complicated systems, model calculations are used to identify the important features. The purposes of these studies are to provide unambiguous explanations for various interesting phenomena observed experimentally in clusters, solids, and surfaces, and to make reliable predictions of new material properties from microscopic quantum theories.

Our theoretical efforts can be classified into two categories: (1) the study of the electronic and dynamical properties of technologically important materials, and (2) the development of new algorithms and calculational methods in studying materials properties using quantum mechanics. Detailed descriptions of specific projects are:

(1) Hydrogen in Metals
Hydrogen is the smallest and lightest interstitial impurity in transition metals. It interacts strongly with the metal host and behaves like a three-dimensional lattice gas with extremely high mobility. The diversity of phenomena exhibited by these systems (such as structural, order-disorder, and metal-semiconductor phase transitions) are closely linked to electronic structure changes induced by the presence of hydrogen. We are interested in examining the fundamental interaction of hydrogen with the metal lattice in this strongly coupled system.

Theoretical studies using the state-of-the-art computational methods have been carried out for hydrogen in yttrium, palladium, and niobium, each with a different lattice structure. The hydrogen-yttrium system is a prototype of rare-earth-metal hydrides. We have thoroughly examined a variety of intriguing properties over the whole composition range. Topics investigated include the nature of hydrogen pairing in the solid solution phase, the mechanism of lattice contraction and the (420)-plane ordering of hydrogen in the cubic phase, and the identification of Peierls distortion in the trihydride which is related to the metal-insulator transition. We have also studied the ordering of hydrogen in close-packed palladium, and in niobium which has a relatively open lattice.

(2) Hydrogen on Metal and Semiconductor Surfaces
Hydrogen on surfaces modifies the physical properties dramatically and has drawn attention from researchers in many different fields. This is the simplest chemisorption system on a surface, with connections to the catalytic synthesis of hydrocarbons and homoepitaxial growth of semiconductors. Interesting features we have studied include the hydrogen ordering and hydrogen induced giant phonon anomaly on surfaces of certain transition metals, and the hydrogen induced change in the surface energy anisotropy on semiconductor surfaces and its effect on stability of different adsorption phases.

(3) Development of Theoretical Techniques
We have also carried out developmental work on computational techniques in studying real materials. A new method to obtain the phonon spectrum has been developed by reconstructing real-space force-constant matrices from the projection along several symmetry directions. The dynamical matrix at any arbitrary wave vector can then be determined and the full phonon spectra can be obtained completely from first principles for the material. We have applied this method to Si and Ge, and obtained excellent phonon dispersions and polarization vectors. The volume-dependent phonon frequencies were then used to investigate thermodynamical properties of silicon, including the abnormal (negative) thermal-expansion coefficient at low temperature.

Recently, we have explored the possibility of using the wavelet basis in self-consistent electronic structure calculations. Wavelet analysis is a newly developed mathematical tool that has found a unique niche in many scientific and engineering fields. This basis set is capable of describing localized features of different scales and at different locations. We have successfully carried out the first implementation of orthonormal wavelets in three-dimensional self-consistent calculations, and demonstrated the feasibility and advantages of this basis in modern electronic-structure calculations.

(4) Quantum Monte-Carlo Study of Real Materials
The electronic many-body problem has been a long-standing challenge in condensed matter physics. Difficulties arise mainly from the number of electrons involved and the special statistical properties required. For lack of feasible theoretical tools to attack the problem directly, most researchers have been working with various simplified models in the past. On the other hand, the widely adopted local-density approximation within the density functional theory maps the many-body problem to a series of one-particle equations. However, it has not been possible to assess the error and to make improvements in a systematic fashion. We have therefore decided to investigate the quantity of central importance in density functional theory, the exact exchange-correlation energy density, using quantum Monte-Carlo techniques. In collaboration with Dr. Richard Needs’ group at Cambridge through a NATO grant, we have evaluated the pair correlation function and the exchange-correlation hole in crystalline silicon using a variational many-body wave function obtained by the Monte-Carlo method. Through the coupling-constant integration technique, we have then computed the almost exact exchange-correlation energy density in bulk silicon, and identified the errors in the results obtained from the local density and other approximations.