Computer Simulation Of New Nanostructure Semiconductor Devices Based On Adaptive Parallel Methods

With the development of computer technology, continuously improvement of the processor speed, maturity of network communication technology, and the improvement of software tools, provides parallel processing technology into the rapid development. It has been recognized that parallel processing technology is the road to achieve high-performance computing. Adaptive technology is a future-oriented technology, it can get the economic, security and comfort, as well as to reduce and avoid distortion and noise. It describes a new class of "smart components / intelligence structure," These smart components change in the working conditions of the time and have a goal of the initiative to optimize the mechanical properties of regulation. Both mainstream technologies have laid an important foundation for large-scale complex computing. This paper is using the two mainstream technologies to simulate the new nanostructure semiconductor devices and to solve the high degree of coupling complex nonlinear systems.At present, the main problems of semiconductor device simulation include: device model increasingly complex, so need to complement and develop the classic drift diffusion model; coupled partial differential equations system will inevitably lead to the difficulty of the discrete increase in the numerical solution of the stability and reliability; with the invention of the new structure and the new materials devices, device physics model and parameters are growing complexity, thus increasing the difficulty of analyze devices character; increasing in the number of equations and taking into account characteristics of multi-dimensional and transient, there are very large systems of nonlinear equations to be solved, thus must rely on advanced technology and parallel iterative method to be effective in solving the problems.This paper proposes a new adapive parallel computation algorithm used for new nanostructure semiconductor devices. With the method used for device simulation, it can automatically track and estimated physical and electrical characteristics of the device, refine automatically mesh, calculate and simulate automatically. Experimental results show that the method provide a high level of accuracy for analysis of the device character. The methods can hold parallel distributed computing on personal computer or workstation, and achieve high accuracy, high performance, fast computer-aided design simulation requirements.In this thesis, we present two parallel algorithms. The first one is a PDD approach and the second is a PIVPS. With the applied adaptive computing methodology, the novel parallel algorithms have been developed and successfully implemented on a Linux cluster with MPI library. Compared with the measured data, numerical results on various devices have been presented to show the accuracy and efficiency of the method. It gives the speedup, efficiency, and maximum difference to show the performance of the parallel algorithms. The parallel and adaptive computing methods proposed in this paper show that cluster parallel computing is a feasible alternative that can be used to provide fast characterization of semiconductor devices.Then we proposed the adaptive computing procedure that contains, such as Gummel decoupled algorithm, finite volume method, monotone iteration, error estimation and 1-irregular mesh refinement methodologies in device simulation. Then use the adaptive technology to simulate organic light emitting devices. First simulate a two dimension dual layer organic light emitting device. For the difference of material parameters, different materials thickness and length of the OLED device structure, the simulation has get the intrinsic and terminal voltage and current characteristics. Through comparison with the experimental results, it shows that this two-dimensional simulation technology for OLED device gets the accuracy and numerical robustness. Further simulate the optical properties and electrical characteristics of multilayer organic light emitting device, given Ag/LiF/Alq3/EML/TPD/Ag device structure, as the EML thickness changes, get the details of simulation results including the combination of carrier rates, the exciton luminescence rate and the current-voltage characteristics.At last this article proposed the matrix eigenvalue problem parallelization method for nanostructure semiconductor devices. As semiconductor device dimensions shrink to lead to quantum effect of devices, this paper introduced the Schrodinger -Poisson model which include all the quantum mechanisms. However, the solution of Schrodinger-Poisson model needs to spend a lot of time, the model can not be widely used in engineering applications. This new simulation technique provides a fast, efficient, and accurate approach to extract device parameters in which quantum confinement effects are significant. The results are also found to compare quite well with experimental data. The speedup and efficiency are also presented to show the computational performance.This dissertation has proposed integrity and uniform method for computer simulation semiconductor devices and nano-structure devices. The presented algorithms have been developed a computer simulation program that can be used for kinds of different sizes and different structure of nano-semiconductor devices analyzing character. From the theoretical and practical test results have demonstrated their good convergence, accuracy, efficiency and the effectiveness of high-speed computing characteristics.