John Lowengrub
(School of Mathematics, University of Minnesota)
Theory and Simulation of 3-D Crystal Growth
Abstract
In this talk, we consider the quasi-steady evolution of growing crystals
in 3-d. The case of quasi-steady crystal growth is a fundamental problem
both in phase transitions and in diffusion dominated growth. The growth
of a spherical germ from a supercooled melt or supersaturated solution
(with isotropic surface tension) was first analyzed by Mullins and Sekerka
in 1963. It was found that a growing sphere is linearly unstable to large
wavelength perturbations. Moreover, as the sphere becomes larger, the unstable
wavelengths become smaller and smaller. This provides a heuristic explanation
for the dendritic and highly complex shapes typically observed in freezing
processes in nature (e.g. snowflakes). A re-examination of this problem,
however, reveals that the Mullins-Sekerka instability may be suppressed
by appropriately varying the undercooling (far-field temperature) in time.
For example, in 3 dimensions, by imposing the far-field temperature flux
(rather than a temperature condition), a class of asymptotically self-similar,
non-spherical growing crystals can be found. Simulations show that this
class of solutions is robust with respect to perturbations and is well-predicted
by solutions of the linearized equations. To simulate the problem numerically,
we use a boundary element method with a fully adaptive surface triangulation.
This enables us to simulate three dimensional crystals stably and accurately
well into the nonlinear regime. Simulations of both stable and unstable
crystal growth will be presented. This work has important implications
for shape control in processing applications; in fact, experiments are
currently being designed (by Stefano Guido at the University of Naples)
to test this possibility.
This work is joint with Dr. Vittorio Cristini (School of Math, Dept
Chem. Eng. and Mat. Sci, U. Minn.) |
|