第7回 理論グループ共通セミナー （講師：押山 淳 教授）
|講 師||押山 淳 教授|
|題 目||Large-Scale Density Functional Calculations for Nanostructures|
I report first-principle total-energy electronic-structure calculations with unprecedentedly large scale performed for Si nanowires, Si nanodots, carbon nanotubes and Ge/Si interfaces. Our newly developed code named RSDFT (Real Space Density Functional Theory) has been designed and tuned so as to achieve the epoch making performance on ext-generation massively parallel computers. At present, electron states of the system consisting of 10,000 – 20,000 Si atoms can be calculated on a several Tera-FLOPS machine at the level of the local density approximation or the generalized gradient approximation in the density functional theory . In our scheme, the three dimensional grid is introduced in real space, and all the quantities are computed at each grid point. Each CPU or each core in a parallel architecture computer is assigned to treat a part of the whole grid. This makes our RSDFT code essentially free from FFT (Fast Fourier Transform) which is frequently used in conventional scheme and causes formidable communication task on the massively parallel computers. Accuracy of the calculations in our scheme is guaranteed by reducing separations among grid points systematically.
Si nanowires and nanodots are strong candidates for channel parts in transistors or memory cells in the next generation semiconductor devices. It is thus of principal importance to clarify current-voltage (I-V) characteristics in the wires and charging energies of the dots by reliable calculations. We have done electron-state calculations for Si nanowires with the diameters in a range of 4 nm – 10 nm, considering realistic sidewall roughness due to oxidation. The results clarify roles of the quantum confinement, the anisotropic effective mass and the sidewall roughness in determining the electronic structures. The obtained band structures show the expected I-V characteristics in the ballistic regime. As for the Si nanodots, we have calculated ionization energies, electron affinities and then the first excitation energies of the dot with the diameters in a range of 6 nm – 8 nm. The charging energy as a function of the dot diameter is unequivocally clarified. The relation between morphology and electron states in nanostructures is another interesting issue. I report an example in carbon nanotubes in which the wall-wall interaction is decisive in the morphology of the vacancy aggregates and the electron states are controlled by the morphology.
 J.-I. Iwata, D. Takahashi, A. Oshiyama, T. Boku, K. Shiraishi, S. Okada, K. Yabana,
J. Comp. Phys. 229, 2339 (2010).