My Research

Research Interest

• Ab initio calculations of electronic, structural, and magnetic properties of materials

• Materials design from first principles

• Nanostructured materials and their applications

• Novel semiconductors for optoelectronic applications

• Electron-phonon coupling and phonon renormalization in metals

• Quasi-particle properties in strongly correlated systems

• High performance computing

Research Highlights

   (1) New formulation for calculating the phonon renormalization in metals from first principles

         In this work, a new method for calculating the phonon self-energy in metals arising from the coupling between phonons and electrons near the Fermi surface is developed. The essence of this scheme is the separation of the inter-and intra-band parts of the electron polarizability. The intra-band contribution provides an extra screening and is closely related to the electron-phonon coupling and phonon softening in metals. Since the electron-phonon matrix elements are not calculated explicitly in this new formulation, the computational cost is greatly reduced when compared with traditional approaches. Applications of this new technique to phonons in MgB₂ show enormous phonon renormalization and explain the experimental fact that both Li and Al substitutions of Mg result in reduced superconducting transition temperature (T_c).

   (2) Electronic, structural, and magnetic properties of Na_xCoO₂

        The recent discovery of superconductivity in hydrated Na_xCoO₂ has generated renewed interest in this technologically important material. Na_xCoO₂ has been known for several years as a potential thermoelectric material which exhibits an unexpectedly large thermoelectric power and at the same time a low resistivity. Due to the strong Coulomb interaction among the rather localized Co d electrons in this system, first principle calculations of its electronic and magnetic properties have been a great challenge. We employed a recently implemented LSDA+U method within the pseudopotential plane-wave formalism to study the electronic, magnetic and structural properties of Na_xCoO₂. Undoped CoO₂ is a charge transfer insulator within LSDA+U and a metal with a high density of states (DOS) at the Fermi level within LSDA. (CoO₂)^1.0- (corresponding to NaCoO₂), on the other hand, is a band insulator with a gap of 2.2 eV. Zone center optical phonon energies are calculated under the frozen phonon approximation and are in good agreement with measured values. The calculated Fermi surface of Na_0.7CoO₂ is in excellent agreement with experiments and, contrary to a recent suggestion, we find no violation of the Luthinger rule is in this system.

   (3) LSDA+U method within the pseudopotential plane-wave formalism

        Although the local density approximation (LDA) within the density functional theory (DFT) has been applied to various systems with great success, it is well known that the LDA fails in many aspects when applied to late transition metal oxides in which strong correlation between d electrons play an important role. The LSDA+U method attempts to incorporate the orbital specific screened Coulomb interaction while retaining the simplicity of LDA. In LSDA+U method, the Coulomb interaction among the localized electrons (e.g. transition metal d) is replaced by statically screened parameters U and J. Although LSDA+U is usually implemented in computational methods using atomic basis sets, we have successfully implemented this technique within the pseudopotential plane-wave formalism. This new development enables us to study the properties of a wide-range of interesting physical systems and materials in which strong interactions between electrons play an important role.

   (4) Quasi-particle properties in strongly correlated systems

        Quasi-particle calculations within the GW approximation usually start with LDA mean field solutions, which works well for weakly correlated materials. For strongly correlated systems, LDA typically gives qualitatively wrong ground states, making any further improvement difficult. By combining the LSDA+U mean field results and the GW approximation to the electron self-energy, we expect to have a better understanding of the quasi-particle properties in these systems.

   (5) Computational design of silicon compatible direct bandgap semiconductors

        Crystalline silicon is an indirect-bandgap semiconductor, making it an inefficient emitter of light. The successful integration of silicon-based electronics with optical components will therefore require optically active materials that can be grown on silicon with high-quality interfaces. Unfortunately, no direct-bandgap semiconductor has yet been produced that can lattice-match silicon. We have computationally designed two hypothetical direct bandgap semiconductor alloys, the synthesis of which should be possible through the deposition of specific group-IV precursor molecules, and which lattice-match silicon to 1.0±0.5%. This work promises a bright future of integrating silicon-based microelectronics with optoelectronics and was highlighted in Nature News Feature column [Nature 409, 974 (2001)] entitled ``Let there be light".

   (6) B₂O and BeB₂ nanotubes: new semiconductors and metals in one dimension

        Carbon nanotubes (CNTs) are metals or small-to-moderate gap semiconductors depending on the wrapping indices. Unfortunately, controlling the electronic properties of CNTs still poses a great challenge to experimentalists, and proposed electronics applications of CNT depend critically on our ability of doing so. In this work, new class of boron-based nanotubes, namely, BeB₂ and B₂O nanotubes were proposed. BeB₂ nanotubes are predicted to be all metallic whereas B₂O nanotubes are all moderate-gap semiconductors. This new class of tubular structures might offer electronic properties unavailable to the previously studied C, B-C-N and B-N based systems.

   (7) Extreme strength of carbon nanotubes: chirality dependence of the plastic deformation

        Structural applications proposed for carbon nanotubes rely upon the extreme elastic strength of the sp² bonding network. However, elastic response alone provides a limited picture of mechanical properties. Beyond a certain stress, the elastic limit, a material deforms plastically via changes in chemical bonding topology. Although the elastic properties of a carbon nanotube are nearly independent of wrapping indices, we show that the onset of plastic deformation depends sensitively on the wrapping index. In addition, the remnant bond rotations remaining after strain release strongly affect the electronic structure of the distorted nanotube. The predicted extreme strength of carbon nanotubes has been verified by several experiments. One recent review article [Physics Report, 390, 235 ( 2004)] dedicates an entire section reviewing this work.

   (8) O(N²) real-space ab initio electronic structure method: finite element methods with multigrid acceleration

        Traditional DFT-based ab initio methods have been successfully applied to calculate the electronic, optical, and structural properties of a vast array of materials. Unfortunately, most of these methods suffer from the so-called O(N³) scaling problem, which severely limits their applicability to large systems. We have successfully developed a O(N²) real-space ab initio package based on finite element methods with multigrid acceleration. Explicit orthogonalization between electronic states is eliminated through a multigrid projection and separation procedure thus achieving an O(N²) scaling. This newly developed technique enable us to study electronic, structural and optical properties of larger and more complex systems than traditional methods do.