Multi-Scale Modeling and Simulation of Aggregation Processes in Crystalline Semiconductor Materials

May 04, 2009 - 16:00

May 04, 2009 - 17:00

Creativity Hall (Room 118) Chemical Engineering

Seminars

Manish Prasad Ph.D. University of Pennsylvania

The design and control of microstructural evolution is the key to the processing of materials ranging from semiconductors to metals to polymers. In the case of crystalline silicon and its alloys which are commonly used in the microelectronics industry nucleation and aggregation of point defects and impurities are responsible for the formation of a wide variety of nano- and microstructures. While these microstructures often are detrimental to electronic devices they can also be useful if their formation can be precisely controlled. The aim of my work was to develop a quantitative and mechanistic understanding of atomic scale aggregation processes in solids with emphasis on defect evolution in silicon. In the first part of my talk I present a multiscale “internally consistent modeling framework” designed to study aggregation in crystalline materials. The key component of this method is a parametrically consistent comparison between atomistic and continuum representation of the aggregation process in which all parameters needed for the continuum model are derived from the same interatomic potential used to generate the atomistic aggregation data.This consistency allows for the direct probing of the mechanistic accuracy of the continuum rate equations without any ambiguity in the input parameters. The results demonstrate that existing models of vacancy cluster aggregation exclude important dynamic and structural effects that enhance the aggregation rate by providing additional aggregation pathways. In the second part of my talk I present a new method for extending the scope of molecular dynamics simulations of clustering in crystals. The method is referred to as “Feature Activated Molecular Dynamics” or FAMD. FAMD exploits the local nature of lattice disturbances around certain types of defects and is shown to greatly reduce the computational burden associated with MD simulation of aggregation. The computational cost of this method is shown to scale linearly with the number of defect entities simulated and not with the overall size of the system. The method is shown to be useful in several different applications beyond homogeneous nucleation. The modeling and simulation schemes developed in my work exemplify the growing trend in area of materials research – consistent exchange of mechanistic and parametric information across multiple scales from nanoscale to microscale to macro (process) scale. This approach has led to more fundamental understanding of materials physics along with improved efficiencies in the existing technologies and development of new ones.