Intellectual Merit: Particulate contamination is a leading cause of yield loss in semiconductor processing. As integrated circuits become smaller, and as improved cleanroom technology eliminates external sources of particles, homogeneous nucleation of particles within the processing environment is rapidly becoming the most important source of particulate contamination. In most cases, these particles are generated via the chemical nucleation processes that will be considered in this project, rather than by condensation of the supersaturated vapor of a pure material. Fundamental understanding of the chemical nucleation process is important if one is to control particle formation. This same understanding can help in design methods for aerosol synthesis of nanoparticles and nanostructured materials that are the building blocks of nanoscale science and engineering. The most detailed and informative approach to modeling chemical nucleation is found at the mechanistic level. In this approach, particle nucleation is described by a network of elementary chemical reactions whose rates can be related to properties of the participating species. This project continues the PIs' collaboration on this topic in which they have applied two complementary methodologies, automated reaction mechanism generation and kinetic Monte Carlo (KMC) simulation, to the development of mechanistic understanding of silicon nanoparticle nucleation. In recent work, they have (1) developed improved algorithms for determining species uniqueness and identifying rings in complex polycyclic clusters, (2) carried out extensive quantum chemical calculations on silicon-hydrogen clusters and generalized the results as a group additivity scheme, (3) developed improved methods for selective generation of reaction pathways, and applied these methods to identify the critical cluster size for particle nucleation and key reaction pathways for silicon nanoparticle nucleation, (4) applied kinetic Monte Carlo simulation to identify cluster growth probabilities and pathways, and (5) constructed a framework for linking detailed chemical reaction mechanisms to reacting flow and aerosol dynamics simulations that can predict particle concentrations and size distributions. From this recent work they have identified the most important areas for continued research on this problem as (1) improved descriptions of the chemistry of polycyclic silicon hydrogen molecules and silicon-hydrogen molecules with multiple functional groups, (2) improved incorporation of such molecules into both deterministic and KMC simulations, and (3) linking of these detailed models of nucleation to aerosol dynamics models that provide results for experimentally accessible quantities like particle concentration and size distribution.
Broader Impacts: Undergraduates, including members of traditionally underrepresented groups, will have opportunities to participate in this project and related work through an REU site on nanostructured seminconductors in Buffalo, for which Swihart is the PI, and through additional targeted programs such as the McNair Scholars program, the Louis Stokes Alliance for Minority Participation (LS-AMP) program, and the Collegiate Science and Technology Entry (C-STEP) program. Examples from this project will be used in Broadbelt's Applied Molecular Modeling course at Northwestern, for which a new course module on kinetic Monte Carlo simulations will be added, and in Swihart's Aerosol Science and Technology course at SUNY Buffalo, which is a new offering, taught as a special topics course in spring 2003, and being permanently added to the curriculum in spring 2005. Both of these courses attract both graduate and undergraduate students, broadening the impact of this project on education.
