Paper SS+AS-WeM12
Molecular Dynamics Simulation of Ge Deposition and Islanding on Amorphous Silica Substrates

Wednesday, November 12, 2014, 11:40 am, Room 312

Selective epitaxial growth (SEG) of Ge on Si substrates has proven to be a versatile pathway for producing Ge substrates to enable III-V device integration on Si. However, persistent problems remain, including dislocation formation and high stresses due to lattice parameter and thermal expansion coefficient mismatches between Si and Ge. Further optimization of the SEG process may be significantly assisted by atomistic simulation. Here, we present an atomistic analysis of Ge deposition on SiO₂. We begin by describing a validation process for a Tersoff-based model for the ternary Si-Ge-O system [1,2], in which we compare simulation predictions to detailed experimental data [3,4] for a variety of properties. Using this validated interatomic potential, Ge deposition and islanding on an amorphous SiO₂ surface is studied with direct molecular dynamics and the results are compared to experimental measurements [4] of island size distributions as a function of deposition rate and temperature. A key aspect of our modeling approach is a procedure to accelerate the simulations. While direct molecular dynamics simulations of Ge deposition on SiO₂ are able to capture Ge island nucleation, growth and coarsening, the very fast deposition rates necessary makes difficult direct comparison to experimental measurements of island density and size distributions. In particular, we show that direct molecular dynamics simulations are able to approach, but not quite reach, the deposition conditions in experiment. The accelerated simulations are based on “equation-free” coarse projective integration [5]. Here, measures of the island size distribution dynamics are obtained from short molecular dynamics simulations and then used to evolve numerically the size distribution over large time intervals. The new island size distribution is then used to reconstruct consistent atomic configurations that are subsequently evolved further with molecular dynamics and the process is repeated. Here, we show that the reconstruction of atomic configurations from size distribution moments represents the key challenge in deposition simulations and we propose approaches for achieving this in a computationally tractable manner.

[1] J. Tersoff, Phys. Rev. B39, 5566 (1989).

[2] S. Munetoh, T. Motooka, K. Moriguchi and A. Shintani, Comput. Mater. Sci39, 334 (2007).

[3] Q. Li, J. L. Krauss, S. Hersee, and S. M. Han, J. Phys. Chem. C 111, 779 (2007).

[4] D. Leonhardt and S. M. Han, Surf. Sci. 603, 2624 (2009).

[5] M.E. Kavousanakis, R. Erban, A.G. Boudouvis, C.W. Gear, I.G. Kevrekidis, . (2007) 382-407.