Modeling of Semiconductor Manufacturing using Goal-Oriented Error Estimation and Adaptive Multi-scale Modeling

Paul Bauman, University of Texas at Austin

Photo of Paul Bauman

As technological advances are made to decrease the scales of the features in semiconductor devices to sub-50 nm scales, the costs to generate practical manufacturing tools become considerable.

Step and Flash Imprint Lithography (SFIL), currently a topic of intensive research, shows promise as an economically viable alternative to deliver features of less than 50 nm in size. As the technology is in early stages of development, computational modeling can provide a vital tool for understanding the processes critical to a robust fabrication paradigm.

While the effects of these processes may span micron to millimeter length scales, the governing physical mechanisms take place at atomic and molecular scales making notions of multi-scale modeling a necessity to realize practical, predictive modeling tools.

Recently, notions of goal-oriented adaptive modeling have provided a rigorous and systematic computational framework for adapting models of different scales to control the error in quantities of interest using so-called Goals algorithms. Here, the steps taken to develop computational models of imprint lithography based upon Goals approaches are given. Specifically, the imprint lithography process and models of the critical polymerization and densification steps are discussed. Special attention is given to the development of consistent surrogate models, based upon underlying molecular models, and successfully interfacing the base and surrogate models. Finally, progress towards realizing a computational tool using adaptive modeling to control error in quantities of interest for imprint lithography is presented.

Abstract Author(s): Paul T. Bauman