AVS 47th International Symposium
    Plasma Science and Technology Thursday Sessions
       Session PS1+TF+SE-ThM

Paper PS1+TF+SE-ThM3
Surface Transport Kinetics in Plasma Deposition of Hydrogenated Amorphous Silicon

Thursday, October 5, 2000, 9:00 am, Room 310

Session: Fundamentals of Plasma Enhanced Chemical Vapor Deposition
Presenter: K.R. Bray, North Carolina State University
Authors: K.R. Bray, North Carolina State University
A. Gupta, North Carolina State University
G.N. Parsons, North Carolina State University
Correspondent: Click to Email

The concept of dynamic scaling was developed to help understand the role of kinetic phenomena that occur on surfaces during non-equilibrium processes (such as film deposition). Plasma deposition of a-Si:H is particularly intriguing because it is well known that over a wide temperature range, kinetic growth process results in very smooth (non-random) surface texture indicating significant surface species transport, but the growth rate is not thermally activated. We have used rf plasma deposition to form a-Si:H films with both helium and argon diluted silane, and used dimensional and frequency analyses to analyze surface topography obtained from AFM images. Surface fractal scaling parameters, including static (a) and dynamic (b) scaling coefficients, Fourier index, saturation roughness, and lateral correlation length (Lc), were determined as a function of film thickness and temperature. After film coalescence (15-20 s) the scaling coefficients are consistent with the surface topology being described as a self-similar structure: a is constant with growth time and is ~1.0, b is ~4.0, and the saturation roughness value increases exponentially with time as tb/a. Based on Herring's models of surface transport, the scaling coefficient values are consistent with surface smoothening being driven by diffusion. In this picture, the lateral correlation length can be equated with the surface diffusion length. We find that Lc ranges from ~50 to 200nm, and is thermally activated, corresponding to a diffusion activation energy of ~0.2eV. This result has important implications for current growth models, where diffusion length is proposed to decrease with increasing temperature because of increasing density of diffusion-terminating dangling bond sites. Possible modifications to current models, consistent with our observed data, will be discussed and presented.