研究課題/領域番号 |
26400393
|
研究機関 | 大阪大学 |
研究代表者 |
SLEVIN KEITH 大阪大学, 理学(系)研究科(研究院), 准教授 (90294149)
|
研究期間 (年度) |
2014-04-01 – 2019-03-31
|
キーワード | semiconductor physics / Anderson localization / Anderson transition / Coulomb interaction |
研究実績の概要 |
The goal of the project is to understand how the long-range coulomb interaction affects the Anderson transition. In the first year of the project we have been focusing on the metal insulator transition in doped semiconductors. This transition occurs at zero temperature as a function of doping concentration. (Doping refers to the manipulation of a semiconductors properties by the controlled addition of impurities, a classic example being phosphor impurities in silicon.) We are studying this problem using a computer simulation that we have developed which approximates the semiconductor host as an effective medium but retains the disorder associated with the random positions of the impurities and the coulomb interaction between the resulting carriers. The major achievement of the first year was the publication in Physical Review B of the first results obtained using the simulation. In addition to introducing the new methodology used in the simulation, the most interesting result is that the critical exponent associated with the divergence of the correlation length at the transition may be different from that of the standard Anderson transition. If this result is confirmed in later simulations, it would mean that the coulomb interaction changes the universality class of the Anderson transition.
|
現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
Already during the first year we have been able to commence simulations of the metal insulator transition in a compensated semiconductor. This work was originally planned for later in the project. Analysis of the results is ongoing.
|
今後の研究の推進方策 |
In the interests of numerical tractability the simulations performed so far have been subject to the constraint of complete spin polarization. A major goal as the project proceeds is to remove this constraint by replacing the local density approximation with the local spin density approximation. To facilitate this it will be necessary to address the current bottlenecks in the computer simulation which involve the discretization of space using the finite difference approximation, and the diagonalization of large sparse matrices. We plan to investigate whether replacing the finite difference approximation with a pseudo-spectral method, and replacing the current matrix diagonalization scheme with the recently available FEAST algorithm would allow us to overcome these bottlenecks. A faster simulation would allow us to draw more reliable and definitive conclusions from the simulations.
|
次年度使用額が生じた理由 |
Development of the simulation code is still on going and it is not yet clear exactly what type of architecture is most appropriate. In particular, whether or not to use a GPU. Purchase of computer hardware is on hold until this is clear.
|
次年度使用額の使用計画 |
Computer hardware and parts. Renewal/upgrade of necessary software licenses including source control (PureCM), fortran compilers (Intel, PGI, NAG), computational tools (MATLAB, Mathcad), graphing tools (Origin), software development tools (Dreamspark subscription, Wing IDE, Enterprise Architect etc.), reference tools (EndNote), manuscript preparation tools (WinEdt) Travel expenses for participation in relevant conference and workshops.
|