Christophe Gourdel

Surface self-diffusion of silicon during high temperature annealing

The atomic-scale mechanisms driving thermally activated self-diffusion on silicon surfaces are investigated by atomic force microscopy. The evolution of surface topography is quantified over a large spatial bandwidth by means of the Power Spectral Density functions. We propose a parametric model, based on the Mullins-Herring (M-H) diffusion equation, to describe the evolution of the surface topography of silicon during thermal annealing. Usually, a stochastic term is introduced into the M-H model in order to describe intrinsic random fluctuations of the system. In this work, we add two stochastic terms describing the surface thermal fluctuations and the oxidation-evaporation phenomenon. Using this extended model, surface evolution during thermal annealing in reducing atmosphere can be predicted for temperatures above the roughening transition. A very good agreement between experimental and theoretical data describing roughness evolution and self-diffusion phenomenon is obtained. The physical origin and time-evolution of these stochastic terms are discussed. Finally, using this model, we explore the limitations of the smoothening of the silicon surfaces by rapid thermal annealing.

Smoothening by self-diffusion of silicon during annealing in a rapid processing chamber

Atomic-scale mechanisms of thermal activated self-diffusion on crystal surfaces are investigated through AFM images. Surface evolution is studied by means of the Power Spectral Density (PSD) function over a large spatial bandwidth. We propose a parametric model based on the Mullins-Herring (MH) diffusion equation by adding two stochastic terms. Then, surface evolution during high temperature annealing in reducing ambient can be predicted. Very good agreement between experimental and theoretical roughness and diffusion parameters was observed. Origin and evolution of the stochastic terms, describing conservative and non-conservative noises, are discussed.