University of Victoria (Canada)
Feasibility of modelling liquid phase diffusion growth of SixGe₁-x with smoothed particle hydrodynamics
Abstract
dc:description.abstractMathematical and numerical modelling of silicon germanium (SixGe1-x) crystal growth by liquid phase diffusion (LPD) was carried out in the Crystal Growth Laboratory at the University of Victoria. The crystal growth process was modelled using a relatively new. meshless. Lagrangian technique called smoothed particle hydrodynamics (SPH). The purpose of this modelling was to determine the suitability of the SPH method for simulating general crystal growth processes. Crystal growth simulations in the literature use a mesh-dependent Eulerian finite volume method, which requires complicated treatment of mesh adaptation at the crystal solidification surface. In contrast, the crystal solidification process can be modelled with relative ease using SPH. as this approach is automatically adaptable, and does not require a mesh. Several benchmark problems were also solved using SPH in order to investigate the problematic aspects of SPH simulation. It was found that. for simulations involving flows with low Reynolds numbers, SPH performed well even with relatively low particle number densities when compared with analytical and finite difference solutions. A simulation time study was also conducted for the two-dimensional driven cavity problem, in order to determine the simulation time as a function of particle resolution. This result showed that simulation times in the SPH method increase almost exponentially as the number of SPH particles is increased. The drastic increase in simulation times is due to the particle averaging procedure involving particle "neighbor. determination. in addition to the necessary reduction in time-step for a stable solution. The results of the LPD simulation using SPH were compared to the previously obtained solution found using the finite volume method. Although the thermal field computed by SPH agreed well with the finite volume results, the SPH velocity field broke down when domains having 738 and 2754 particles were used. The SPH concentration field was simulated correctly when the fluid flow was neglected. After an extensive, systematic review of the SPH coding. it was determined that the SPH method requires several thou-sands of particles to produce an accurate crystal growth simulation. where realistic Reynolds numbers are involved. Unfortunately. such a large number of particles requires unreasonable simulation times based on available computational power, and therefore SPH does not seem feasible for crystal growth simulations at this time. Future advances in computational power may render the SPH method comparable to the mesh-dependent approaches in terms of computational simulation times. At. that point in time. the SPIT approach would have significant advantages over mesh-dependent techniques when simulating crystal growth.
Degree
thesis:*- Level thesis:degree_level
- Master of Applied Science M.A.Sc.
- Department thesis:degree_department
- Department of Mechanical Engineering
- Year dc:date.issued
- 2010
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Rook, Rusty Allen
- Advisor dc:contributor.supervisor
-
- Dost, Sadik
Subjects
dc:subject × 3Rights
dc:rights- Statement dc:rights
-
- Available to the World Wide Web
- Language dc:language.iso
- en, English
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- http://hdl.handle.net/1828/2144