Einladung zum Vortrag im Kolloquium Technische Kybernetik
Modeling and Control of Atomic-Scale Process Dynamics
Professor Martha Gallivan
Georgia Institute of Technology
Zeit: Dienstag · 24. 02. 2004 · 16:00 Uhr
Ort: Raum V 9.31 · Pfaffenwaldring 9 · Campus Stuttgart-Vaihingen
Molecular dynamics and Monte Carlo methods describe the discrete interactions among many atoms or molecules. These dynamic models are nonlinear, stochastic, and high-dimensional, and therefore are not compatible with most tools and algorithms for optimization and control. In most macroscopic systems, continuum models describe the dynamics. However, continuum assumptions are not always valid, such as in the fabrication of nanometer-scale features in integrated circuits, or in the bulk processing of polymers with nonlinear architecture. This seminar will focus on the processing of thin films by chemical vapor deposition, which we are investigating via experiment and simulation in our research group. Dynamic models are needed to interpret the limited sensor data obtained by optical techniques, and several approaches have been taken to develop compact process models that approximate the dynamics of the atomic-scale simulations. Two approaches that will be discussed in this seminar are nonlinear principal component analysis via neural nets, and aggregation of Markov chains. Both are data driven, and rely on the large simulations to identify the dominant dynamics in a more compact representation.
Martha Gallivan is an assistant professor in the School of Chemical and Biomolecular Engineering at the Georgia Institute of Technology. She obtained her bachelors degree in mechanical engineering from the University of Illinois, and her Ph.D. in mechanical engineering from the California Institute of Technology. Martha has been at Georgia Tech for one year, and is working on process modeling and control. Her research program combines theory, simulation, and experiment to address problems in microelectronics fabrication and materials development.