Martha Grover Gallivan
Assistant Professor
Contact Information
Building: Ford ES&T
Office:
1228
Phone: 404.894.2878
Fax: 404.894.2866
email
Mailing Address
Georgia Institute of Technology
School of Chemical &
Biomolecular Engineering
311 Ferst Drive, N.W.
Atlanta, GA 30332-0100
Links
Martha Grover Gallivan
Education
B.S. 1996, University of Illinois
M.S. 1997, California Institute of Technology
Ph.D. 2002, California Institute of Technology
Research Interests
Dr. Gallivan's research brings together the fields of process control and materials processing. Of particular interest are systems in which the dynamics require explicit consideration of discrete atomic interactions. The overall goal is to optimize material properties by computing nominal processing conditions and designing feedback control strategies using a mathematical representation of the relevant physics. Systematic approaches for control and optimization have become an integral component in macroscopic chemical processes, improving efficiency, quality, and safety. In contrast, a more empirical approach is often taken in applications like microelectronics processing, in which conditions are typically fixed in time. However, in some cases time-varying processing parameters can produce altered properties, especially when the process is governed by kinetics.
Building on continuing improvements in atomic-scale modeling and in computational capacity, a new systematic approach is becoming increasingly practical, using a predictive model to develop control strategies. New tools must now be developed for model reduction, system identification, and control of the particular classes of equations seen in atomic-scale and multiscale models. For example, the evolution of surface roughness during thin film deposition may be described by an atomistic kinetic Monte Carlo simulation. The underlying structure for these simulations is a high-dimensional probabilistic equation, which is linear in the state, but nonlinear in the input. We have developed more compact representations, based on the original probabilistic structure and generated with simulation data. The resulting model is then used to compute control strategies. For more information on this work and other projects, visit Dr. Gallivan's Research Web Page.