(image) (image) [ Home ]

Emulation and Simulation of
Microelectronic Fabrication Processes

4.6 Summary

Several independent libraries were implemented during the course of this work, culminating in a complete process simulation library, ViennaPS.

ViennaHRLE provides an implementation of a hierarchical run-length encoded data structure for the efficient storage of sparse data. It provides sequential data access with constant time complexity and random access with \(\mathcal {O}(\log {N})\). In addition to defined values, any number of undefined values can be stored, which occupy more than one grid point and can therefore be used conveniently to store background values describing large regions of the simulation domain. Several iterators were implemented, giving highly efficient access to neighbouring grid points, the number and arrangement of which depends on the specific iterator.

ViennaLS is a high-performance level set library based on the ViennaHRLE data structure. It provides all the necessary features for topography simulations, such as fast geometry creation, and conversions from and to several other material representations, such as triangular or tetrahedral meshes. An algorithm for the extraction of disc meshes for efficient transport modelling, as described in Section 3.2.3.3, has been developed and is provided with the library. Several algorithms for the analysis of the stored level set surface have been developed. These include the analysis of connected regions of the level set used for the detection of voids inside the surface or disconnected stray points outside of the surface. Feature detection directly on the level set was developed and is implemented using different underlying detection algorithms.

An efficient algorithm for iterative level set advection of multiple materials has been improved from the original ViennaTS framework and implemented in ViennaLS. The geometric advection algorithm developed during the course of this work is implemented using efficient algorithms for candidate and contribute point identification. This feature can be used highly efficiently for process emulations of multiple materials.

ViennaCS provides an efficient wrapper for storing volumetric data using ViennaHRLE. Monte Carlo ray tracing for top-down particle transport modelling is provided within the ViennaRay library, which provides an intuitive interface for the modelling approach presented in Section 3.2.4.

Finally, ViennaPS contains the implementation of a modelling framework for the emulation and simulation of semiconductor fabrication processes. This library employs all of the above software tools to provide the functionality required to combine emulation and simulation models within an intuitive interface.