| The main objective of the work described in this dissertation was to develop a framework for integrating detailed chemical kinetics, heat transfer and fluid-flow modeling of reactors with aerosol dynamics models that predict the evolution of particle size distributions and, ultimately, particle morphology.; A numerical model has been developed to predict gas-phase nucleation, growth, and coagulation of silicon nanoparticles formed during thermal decomposition of silane. Solution of the aerosol general dynamic equation was handled by three approaches: (1) the efficient and reasonably accurate method of moments (MOM); (2) the quadrature method of moments (QMOM), which requires no prior assumption for the shape of the particle size distribution; and (3) a computationally more expensive sectional method (SM). The sectional method developed was then extended to include surface area concentration within each volume bin as well as number concentration and to explicitly account for the finite rate of sintering between coagulating particles.; A Computational Fluid Dynamics (CFD) model has been developed using both a commercial package, FIDAP, and a code developed at Sandia National Laboratories, MPSalsa, to model the fluid flow and heat transfer in a laser-driven aerosol synthesis reactor used for preparing nanoparticles of silicon and other materials. From the detailed 3-D model of the reactor, temperature and velocity profiles along the axis of the reactor, in the zone where particle formation takes place, have been extracted and coupled with the one-dimensional aerosol dynamics model developed earlier.; A data inversion program was written to obtain particle size distributions from differential mobility analyzer (DMA) data. Multiply charged particles have the same electrical mobility as smaller singly charged particles, such that there is not a unique relationship between particle size and electrical mobility. This ill-posedness was managed using a regularization algorithm that forces the solution (the size distribution) to be as smooth as possible while maintaining fidelity to the mobility data. The inversion program was tested with both synthetic data and experimental data and worked well for both cases. |
