next up previous contents
Next: Bibliography Up: Peter Fleischmann's Dissertation Previous: 7.5 CMOS Inverter


8. Outlook

The present state of the meshing tool forms a foundation for the future to cope with coming issues and already important aspects of mesh generation for semiconductor process and device simulation. Further development is required from a scientific as well as software engineering point of view. The chosen approach -- the type of Delaunay kernel algorithm and the a priori boundary integration -- seem promising to meet the demands. The ultimate goal is a robust tetrahedralization engine which satisfies the special boundary triangle requirements of a Voronoi type box integration and which handles extremely thin oxide layers without compromising the mesh in the silicon bulk region. The engine must be versatile to allow a combination with special techniques for semiconductor device simulation, as for example elliptical grid generation techniques. Ideally, the engine is to some extent itself capable to produce boundary-fitted meshes at e.g. the channel of a semiconductor device. For none-Voronoi type box integration the strict boundary triangle requirements can be disabled to not become a limitation of the mesh generator and to utilize the efficiency of a Delaunay kernel for the generation of hybrid, prismatic meshes or generally meshes where not all elements possess the Delaunay property. This is especially important for the optimization of meshes with regard to slivers for finite element computations. In summary the following are important further steps:


next up previous contents
Next: Bibliography Up: Peter Fleischmann's Dissertation Previous: 7.5 CMOS Inverter
Peter Fleischmann
2000-01-20